Journal list menu

Volume 286, Issue 4 p. 710-725
Original Article
Free Access

The structure of Legionella effector protein LpnE provides insights into its interaction with Oculocerebrorenal syndrome of Lowe (OCRL) protein

Kevin A. Voth

Kevin A. Voth

Department of Biochemistry, University of Saskatchewan, Saskatoon, Canada

Search for more papers by this author
Ivy Yeuk Wah Chung

Ivy Yeuk Wah Chung

Department of Biochemistry, University of Saskatchewan, Saskatoon, Canada

Search for more papers by this author
Karin van Straaten

Karin van Straaten

Department of Biochemistry, University of Saskatchewan, Saskatoon, Canada

Search for more papers by this author
Lei Li

Lei Li

Department of Biochemistry, University of Saskatchewan, Saskatoon, Canada

Search for more papers by this author
Michal T. Boniecki

Michal T. Boniecki

Department of Biochemistry, University of Saskatchewan, Saskatoon, Canada

Search for more papers by this author
Miroslaw Cygler

Corresponding Author

Miroslaw Cygler

Department of Biochemistry, University of Saskatchewan, Saskatoon, Canada


M. Cygler, Department of Biochemistry, University of Saskatchewan, 107 Wiggins Road, Saskatoon, Saskatchewan S7N 5E5, Canada

Tel: +1 306 966 4361

E-mail: [email protected]

Search for more papers by this author
First published: 26 November 2018
Citations: 7


Legionella pneumophila is a freshwater bacterium that replicates in predatory amoeba and alveolar macrophage. The ability of L. pneumophila to thrive in eukaryotic host cells is conferred by the Legionella containing vacuole (LCV). Formation and intracellular trafficking of the LCV are governed by an arsenal of effector proteins, many of which are secreted by the Icm/Dot Type 4 Secretion System. One such effector, known as LpnE (L. pneumophila Entry), has been implicated in facilitating bacterial entry into host cells, LCV trafficking, and substrate translocation. LpnE belongs to a subfamily of tetratricopeptide repeat proteins known as Sel1-like repeats (SLRs). All eight of the predicted SLRs in LpnE are required to promote host cell invasion. Herein, we report that LpnE(1-375) localizes to cis-Golgi in HEK293 cells via its signal peptide (aa 1–22). We further verify the interaction of LpnE(73-375) and LpnE(22-375) with Oculocerebrorenal syndrome of Lowe protein (OCRL) residues 10–208, restricting the known interacting residues for both proteins. To further characterize the SLR region of LpnE, we solved the crystal structure of LpnE(73-375) to 1.75Å resolution. This construct comprises all SLRs, which are arranged in a superhelical fold. The α-helices forming the inner concave surface of the LpnE superhelix suggest a potential protein–protein interaction interface.


Coordinates and structure factors were deposited in the Protein Data Bank with the accession number 6DEH.


  • LCV
  • Legionella containing vacuole
  • OCRL
  • Oculocerebrorenal syndrome of Lowe
  • T4SS
  • type 4 secretion system
  • Introduction

    Legionella pneumophila is a Gram-negative intracellular pathogen that infects free-living amoeba in aqueous environments [1]. Once transmitted to the human lung via contaminated aerosols, L. pneumophila colonizes alveolar macrophages and gives rise to a severe pneumonia known as Legionnaires’ disease or a milder illness called Pontiac fever [2]. Regardless of the eukaryotic host, the mechanisms facilitating intracellular replication of L. pneumophila are similar; the overall strategy is to evade phagosome–lysosome fusion by generating a replicative niche called the Legionella containing vacuole (LCV) [3-5]. Unlike autophagosomes, the LCV does not acidify [6]. Indeed, L. pneumophila subverts host autophagy by recruiting endoplasmic reticulum (ER)-derived secretory vesicles, mitochondria, and rough ER membranes to the LCV [3, 5]. Ultimately, the early secretory pathway is hijacked by the LCV to obtain safe passage to a functional ER [3]. In the final stages of infection, an ER-derived integral membrane protein, calnexin, is observed on LCV membranes. This recruitment of calnexin is correlated with a transition from ‘tight’ to ‘spacious’ LCVs [4, 7] and occurs only in mature vacuoles, which exist as stable replicative organelles. To accomplish these tasks, Legionella secretes many effector proteins predominantly through the Dot/Icm (defective in organelle trafficking/intracellular multiplication) type IV secretion system (T4SS) [8], which is evolutionarily related to the bacterial conjugation apparatus [9]. Effectors are highly variable in their structure and function, and interfere with various eukaryotic processes, including cytoskeleton rearrangement, signaling, cellular adherence, transcription, vesicular trafficking, and membrane biogenesis [8, 10].

    LpnE is an effector protein identified through genomic subtractive hybridization of L. pneumophila and Legionella micdadei genomes, with an unknown mechanism of export into the host that does not depend on either the Dot/Icm T4SS or the Lsp T2SS [11, 12]. The Legionella genome encodes several effectors with Sel1-like repeats (SLRs), among them LpnE, EnhC, LidL, Lpg1356, and Lpg1062. LpnE has been named for its role in facilitating entry (L. pneumophila entry) into macrophage and type II alveolar cells and is implicated in several roles related to L. pneumophila pathogenicity: (a) LpnE interacts with the N-terminal region of a eukaryotic polyphosphate-5 phosphatase known as OCRL (oculocerebrorenal syndrome of Lowe). At the onset of infection, OCRL is recruited to LpnE which, in turn, is associated with the LCV by an interaction with phosphatidylinositol-3-phosphate. Catalytically active OCRL inhibits L. pneumophila replication in host cells [13]. (b) LpnE binds Obscurin-like protein 1 via the SLR region. This was taken to be an indication that LpnE binds eukaryotic proteins with immunoglobulin type folds [11]. (c) Defective LpnE mutants affect vacuolar trafficking, as evidenced by increased lysosome-associated membrane protein-1 (LAMP-1) levels on the LCV [11]. (d) An additional SLR-family protein called EnhC complements LpnE in its ability to restore virulence to L. pneumophila [14]. This study showed that virulence defects displayed by Dot/Icm mutants can be reversed when stationary phase bacterial cultures are incubated in water prior to infection. In the case of lpnE deletion, this reversal occurs by induction of EnhC. Despite these advances, a detailed understanding of how LpnE promotes L. pneumophila entry and virulence still eludes us.

    Analysis of the amino acid sequence by SignalP ( [15] suggests that LpnE contains a signal peptide (aa 1–21) followed by the N-terminal domain (aa 22–51) of unknown function and eight predicted SLR regions (aa 52–338, InterPro entry These repeats were shown to play a critical role in Legionella host cell invasion [11]. Ectopic expression of LpnE lacking only two predicted C-terminal SLRs [LpnE(1-266)] was unable to complement the invasion defect seen in the cells lacking endogenous LpnE. Only a construct encoding all eight SLRs [LpnE(52-375)] partially restored the defect. This may suggest that a full complement of SLRs is required for LpnE to promote invasion [11].

    Sel1-like repeats are a subclass of tetratricopeptide (TRP) repeats, distinguished from the latter by their longer length (36–44 residues in SLRs; 34 residues in TRPs). SLRs fold into a pair of antiparallel α-helices, called helix-turn-helix motifs. These motifs are stacked together to produce a right-handed superhelix. Stabilization of the superhelical fold is achieved through conserved hydrophobic minicores within and between each motif, in some cases reinforced by disulfide bonds [16, 17]. The extended superhelix typically exhibits a continuous hydrophobic core [18], which may mediate protein–protein interactions.

    Herein, we present the crystal structure of LpnE(73-375). This construct comprises the SLR-containing domain. Two LpnE molecules were found in the asymmetric unit, with a structure exhibiting the anticipated superhelical fold. This architecture is stabilized by a hydrophobic core within each SLR centered around a conserved leucine in αA, as seen in the protein EsiB [17] from extra-intestinal pathogenic Escherichia coli (ExPEC). Interestingly, the hexa-histidine tag plays an essential role in crystallization by making extensive contacts with the tag of a neighboring molecule in the asymmetric unit, forming a ‘handshake’ interaction mediated by Ni2+ ions. We show further that while the LpnE(73-375) expressed in mammalian cells is localized predominantly to the cytosol, the full-length protein LpnE(1-375) localizes to the Golgi apparatus. This Golgi localization is dependent solely on the presence of the putative signal peptide LpnE(1-21).


    Analysis of the LpnE sequence indicates the presence of a signal peptide encompassing residues 1–21. We have cloned and expressed in E. coli the expected mature form of LpnE containing residues 22–375. Although we were able to crystallize LpnE(22-375), these crystals diffracted poorly. We applied limited proteolysis to identify a better candidate for crystallization. A limited digest of LpnE(22-375) showed the presence of a stable 36 kDa fragment. This fragment was identified by mass spectrometry to contain residues 73–375. Subsequently, this fragment containing most of the predicted SLR repeats was cloned with an N-terminal His6-tag, overexpressed in E. coli BL21(DE3)pLysS cells, and used for structural and binding studies.

    LpnE(1-375) localizes to Cis-Golgi in transfected mammalian cells

    First, we investigated the subcellular localization of various LpnE constructs in human cells. To this end, the C termini of these proteins were tagged with GFP and overexpressed in HEK293 cells. Cells transfected with a vector encoding only GFP served as a control. The subcellular localizations of GFP and LpnE(1-375)-GFP in HEK293 cells were examined by fluorescence microscopy. The GFP control was distributed throughout the cells (Fig. 1A, upper panels), whereas LpnE(1-375)-GFP predominantly showed a punctate pattern (Fig. 1A, lower panels). The subcellular localization of ectopically expressed LpnE(1-375) suggested that this protein may possess a hydrophobic membrane localization domain and bind-specific host organelles. To identify the nature of the punctate species, we analyzed the possible colocalization of ectopically expressed LpnE(1-375)-GFP with ER and Golgi markers in HEK293 cells. These experiments showed that LpnE(1-375)-GFP colocalizes with the cis-Golgi marker GM130 (Fig. 1A, middle), but not the ER marker calnexin (Fig. 1A, right) or 4′,6-diamidino-2-phenylindole (DAPI) identifying the nucleus (Fig. 1A, left). To quantify the degree of colocalization of LpnE and intracellular marker, we calculated the Pearson correlation coefficient for each marker. A Pearson coefficient of 0.72 and 0.18 was obtained for GM130 and Calnexin, respectively, supporting the colocalization of LpnE with Golgi. Conversely, overlap coefficients of 0.29 and 0.24 were obtained for GFP with GM130 and Calnexin (Fig. 1B).

    Details are in the caption following the image
    Localization of various LpnE fragments in HEK293 cells. (A) The plasmid containing either GFP or LpnE(1-375)-GFP was transfected to HEK293 cells. Left grid—nucleus visualized with DAPI. Middle grid—the cis-Golgi visualized with anti-GM130 antibody, right grid—the ER visualized with anti-calnexin antibody. Top panels—GFP alone, lower panels—LpnE(1-375)-GFP. The scale bar indicates the length of 5 μm; (B) quantitation of colocalization by Pearson coefficient. The Pearson correlation coefficient was determined with coloc 2 plugin of the imagej software [35] and was calculated with at least 30 cells (mean and standard deviation, SD, of the mean from each experiment, ***< 0.001); (C) LpnE(73-375)-GFP localization in HEK293 cells compared with DAPI staining of the nucleus (top panel) and the cis-Golgi staining with anti-GM130 antibody (lower panel); (D) LpnE(22-375)-GFP localization in HEK293 cells compared with GM130 marker; (E) fusion with LpnE signal peptide led Legionella pneumophila effector LegA15 and Lpir1 to Golgi apparatus. (F) quantitation of colocalization by Pearson coefficient of Fig. 1D with at least 30 cells was calculated (mean and standard deviation, SD, of the mean from each experiment, ***< 0.001).

    The putative signal peptide of LpnE is essential for its Golgi localization in mammalian cells

    To identify the segment of LpnE that encodes for Golgi localization, we first investigated localization of the LpnE(73-375) fragment, which behaved well and was successfully crystallized. In contrast to the punctate pattern observed for LpnE(1-375), LpnE(73-375) was distributed throughout the cell (Fig. 1C), indicating that the first 72 residues are guiding Golgi localization. Next, we investigated the localization of a predicted mature protein: LpnE(22-375). Surprisingly, this construct was also distributed throughout the cytosol of the HEK293 cells (Fig. 1D). Thus, neither of the ectopically expressed LpnE(73-375) or LpnE(22-375) colocalized with the cis-Golgi marker. In order to test whether the first 22 residues of LpnE serve as a localization peptide, we fused this LpnE peptide to the N terminus of two other effector proteins LegA15 and Lpir1, which we previously showed to have cytosolic localization. We expressed these proteins in HEK293 cells and observed their localization. Wild-type LegA15 and Lpir1 were distributed throughout the cytoplasm. When these proteins were fused to the LpnE peptide, however, they displayed a punctate pattern that colocalized with the Golgi marker (Fig. 1E). The Pearson correlation coefficients for wide-type LegA15 and Lpir1 with cis-Golgi marker GM130 were 0.34 and 0.29, respectively, while for the tagged proteins, it was 0.71 and 0.53, respectively (Fig. 1F). These results suggest that the putative signal peptide of LpnE is responsible for its cis-Golgi localization.

    LpnE–OCRL interactions

    LpnE has been shown by immunoprecipitation to interact with the N-terminal region of OCRL comprising residues 1–236 [13]. This region encompasses a pleckstrin homology (PH) domain (aa 1–119) [19] and a long flexible linker connecting the PH domain to the inositol polyphosphate 5-phophatase catalytic module [20]. Although PH domains have been implicated in protein–protein interactions, we suspected that the unstructured linker of OCRL may harbor binding elements essential for its interaction with LpnE.

    His-LpnE(73-375) eluted from the SEC column in two peaks corresponding to monomeric and dimeric species. The monomer was well behaved and we first explored if this fragment is sufficient to bind OCRL(1-236). We have shown using size-exclusion chromatography (SEC) that LpnE(73-375) does indeed interact with GST-OCRL(1-236), as evidenced by a distinct peak shift for the His-LpnE(73-375)–GST-OCRL(1-236) complex relative to His-LpnE(73-375) and GST-OCRL(1-236) alone (Fig. 2A). Secondary structure predictions (PsiPred, [21]) of OCRL show that the region 135–200 neighboring the PH domain is devoid of secondary structure and is followed by a short helix (aa 201-206). A polyproline motif (PPPPP) is present within this unstructured region (aa 177-181) and may also promote the binding of OCRL to LpnE [22]. Moreover, the N terminus of OCRL contains a string of nine hydrophobic residues. To define which part of OCRL(1-236) is crucial for interacting with LpnE, we designed the following constructs of OCRL based on the above information: OCRL(1-140) and OCRL(10-140) that correspond to the PH domain, OCRL(10-176), OCRL(10-183), OCRL(10-208), and OCRL(10-236). These constructs were cloned with a cleavable His6-tag. All these recombinant proteins were purified by affinity chromatography using cobalt Talon resin and eluted with a buffer containing 100 mm imidazole. The His6-tag was cleaved with TEV protease overnight at room temperature. All OCRL constructs, except the shortest OCRL(10-140), were found to have a degradation product corresponding to ~15 kDa after TEV cleavage. We found the longest construct, OCRL(10-236), to be poorly expressed and vulnerable to proteolytic degradation, making data interpretation difficult for this protein. We therefore concentrated our efforts on the other constructs. One of these, OCRL(1-140), includes the first nine residues to observe their role in binding LpnE. Our C-terminal truncations of OCRL systemically probe into the region without predicted secondary structure and should provide insight into the residues required to promote an interaction with LpnE.

    Details are in the caption following the image
    Interaction of LpnE with OCRL fragments. (A) SEC profile of His-LpnE(73-375)–GST-OCRL(1-236) complex and His-LpnE(73-375) alone from BioRad SEC650 column with the SDS/PAGE gel across the protein fraction of the complex. His-LpnE(73-375) elutes in two peaks, monomer (mon) and dimer (dim); (B) Combined SEC profiles and SDS/PAGE corresponding to each profile for LpnE(73-375) in blue, OCRL(10-208) in green and their mixture in red. The proteins were injected on Biorad SEC70 column. The same column was used in all the runs presented this figure and the proteins are marked using the same color scheme; (C) the ITC profile for binding of OCRL(10-208) to LpnE(73-375). Binding constant of 16.2 μm was calculated from the measured data; (D) combined SEC profiles and SDS/PAGE corresponding to each profile for LpnE(73-375), OCRL(10-183), and their mixture; (E) combined SEC profiles and SDS/PAGE corresponding to each profile for LpnE(73-375), OCRL(10-176) and their mixture; (F) combined SEC profiles and SDS/PAGE corresponding to each profile for LpnE(73-375), OCRL(10-140), and their mixture; (G) combined SEC profiles and SDS/PAGE corresponding to each profile for LpnE(158-375), OCRL(10-208), and their mixture; (H) combined SEC profiles and SDS/PAGE corresponding to each profile for LpnE(183-375), OCRL(10-208), and their mixture.

    The interaction of untagged LpnE and OCRL constructs was investigated by size-exclusion chromatography using BioRad SEC70 column. The proteins were mixed in approximate 1 : 1 ratio and incubated for 30 minutes on ice. When LpnE(73-375) was combined with OCRL(10-208), a clear shift in the elution profile was observed, with the main peak shifting to shorter elution time and containing both proteins, indicating that the two proteins form a complex (Fig. 2B). To quantify their binding, we have measured the binding constant by isothermal titration calorimetry and obtained a value of 16 μm (Fig. 2C) confirming a modest binding strength. Next, we tested if the shorter OCRL(10-183) interacts with LpnE(73-375). A shift in the profile is also visible for this complex, with fractions containing both proteins appearing at lower elution volumes (Fig. 2D). Additional deletion of the polyproline segment from OCRL (OCRL(10-176)) leads to a substantial weakening of the interaction with LpnE. This is evidenced by a small shift to lower volumes and limited overlap of fractions containing the proteins (Fig. 2E). We interpret the small shift of the peaks as indicating a dynamic formation and dissociation of the complex, which increases somewhat the apparent molecular weight of each protein. Finally, the shortest OCRL constructs, OCRL(1/10-140), do not interact with LpnE(73-375); they show no shift in the elution profile and no overlap of the fractions containing each protein is observed (Fig. 2F). We conclude that the polyproline region in OCRL significantly contributes to the binding with LpnE.

    We next investigated how far LpnE can be truncated on the N-terminal side and still retain OCRL binding. To this end, we have expressed His6-tagged LpnE with additional five helices [LpnE(158-375)] or six helices [LpnE(176-375)] removed from the LpnE(73-375) constructs. These proteins were purified and the tags were cleaved with TEV protease. We investigated if these constructs can bind OCRL(10-208), which showed good binding to LpnE(73-375). The SEC profile of LpnE(158-375) and OCRL(10-208) shows a shift of the main peak to lower volumes and the presence of both proteins in the peak fractions, indicating retention of binding (Fig. 2G). This LpnE(158-375) fragment still interacted with OCRL(10-183) (Fig. 3A) but not with OCRL(10-176) (Fig. 3B) or OCRL(10-140) (Fig. 3C). The shorter LpnE construct, LpnE(176-375), mixed with OCRL(10-208) showed almost no shift in peak position (Fig. 2H), suggesting a much weaker interaction between these two protein fragments. We conclude that the LpnE segment aa 158-177 containing an α-helix plays a key role in OCRL binding.

    Details are in the caption following the image
    The SEC elution profiles of individual constructs and their 1 : 1 mixture and corresponding SDS/PAGE gels of peak fractions. LpnE is in blue, OCRL in green, and the mixture in red. The proteins were injected on the Biorad SEC70 column. (A) LpnE(158-375), OCRL(10-183), and their mixture; (B) LpnE(158-375), OCRL(10-176), and their mixture; (C) LpnE(158-375), OCRL(10-140).

    Further analysis of the putative LpnE-OCRL complexes was carried out on untagged proteins using size-exclusion chromatography combined with multi-angle light scattering (SEC-MALS). LpnE(22-375) eluted in a peak corresponding to a molecular weight of 41.8 kDa, which indicates that this protein is monomeric (39.4 kDa theoretical) (Fig. 4A). Similar to its behavior on the SEC650 column, His-LpnE(73-375) eluted in two distinct peaks representing monomeric (35.7 kDa) and dimeric (71.5 kDa) states of the protein. Cleavage of the hexa-histidine tag abolished LpnE(73-375) dimerization, suggesting a role for the tag in promoting dimer formation (Fig. 4B). Indeed, untagged LpnE(73-375) gave a molecular weight estimate of 34.5 kDa on SEC-MALS (33.7 kDa theoretical monomer). OCRL(10-208) also proved to be monomeric and eluted in a 23.3 kDa peak on SEC-MALS (22.9 kDa theoretical monomer) (Fig. 4C). In agreement with the SEC data, the SEC-MALS interaction of LpnE(73-375) with OCRL(10-208) produced a peak with an accompanying molecular weight estimate of 69.6 kDa, which corresponds loosely to the expected molecular weight for a 1 : 1 complex of these proteins (56.6 kDa theoretical) (Fig. 4D). A ~10 kDa deviation from the expected value may suggest that the LpnE(73-375)-OCRL(10-208) complex is not spherical in solution, as MALS molar mass estimates are predicated on this assumption. The LpnE(22-375)-OCRL(10-208) mixture eluted together in a peak with a corresponding molecular weight of 147.7 kDa (Fig. 4E). This could represent an LpnE(22-375)2-OCRL(10-208)2 heterotetramer according to theoretical estimates (130.2 kDa). As for the LpnE(73-375)-OCRL(10-208) complex, the deviation of LpnE(22-375)2-OCRL(10-208)2 from its expected molecular weight may imply the complex is nonglobular.

    Details are in the caption following the image
    Interaction of LpnE(73-375) or LpnE(22-375) with OCRL(10-208) monitored by SEC-MALS and SDS/PAGE. (A) LpnE(22-375) elutes as a single peak with MW of 41.8 kDa. SDS/PAGE shows the fractions from the column; (B) LpnE(73-375) elutes as a single peak with Mw of 34.5 kDa. The middle part of the SDS/PAGE gel (shown in panel A) has been spliced out to move molecular weight markers close to the relevant lanes of the gel; (C) OCRL(10-208) elutes as a single peak with MW of 23.3 kDa. Degradation products of OCRL elute in the second peak; (D) LpnE(22-375) with OCRL(10-208). The complex elutes within fractions 40–43; (E) LpnE(22-375) with OCRL(10-208). The complex elutes as a broad peak within fractions 36–42. Excess of OCRL(10-208) elutes in the second peak and degradation products as the third peak.

    Overall structure

    The crystal structure of LpnE(73-375) was solved by molecular replacement and refined to a resolution of 1.75 Å. There are two molecules in the asymmetric unit. Each molecule consists of 17 antiparallel α-helices (Fig. 4A). The helices are twisted into a right-handed superhelix. While short segments of the two molecules superimpose with root-mean-square deviation (rmsd) of 0.2–0.3 Å, the superposition of molecule A and B gives rmsd of 1.2 Å and indicates a small difference in superhelical twist of the helices (Fig. 5). The loops connecting neighboring helices are short on one edge of the superhelix and long on the other edge (Fig. 6A). The helices are organized into helix-loop-helix pairs that are classified as Sel1-like repeats (SLRs). Two conventions have been used to describe multiple Sel1-like repeats. In a description by Lüthy et al. [16], the two helices of the Sel1-like repeat are joined by a short loop while the long loop connects two repeats. In the other description introduced by Mittl and Schneider-Brachert [18], the two helices of the Sel1-like repeat are connected by a long loop. Here, we follow the first definition as it better fits the LpnE structure. The 17 α-helices of LpnE(73-375) form eight SLRs (α1–α16), with the C-terminal helix α17 capping the repeats (Fig. 6A). The first helix of each SLR is designated as helix αA and the second as helix αB and are marked as n-αA, n-αB, where n = 1–8 and numbers the SLR in LpnE. The structural superposition of the eight repeats is shown in Fig. 6B together with their aligned sequences (Fig. 6C). The intra-SLR loops in LpnE are three residues long while the inter-SLR loops are longer, ~7 residues. SLRs in LpnE are 36 amino acids in length except SLR1 which has 33 amino acids (Fig. 6C).

    Details are in the caption following the image
    Superposition of the two independent molecules based only on SLR-1 and SLR-2. The difference in twist along the superhelix axis becomes clearly visible.
    Details are in the caption following the image
    Crystal structure of His-LpnE(73-375) comprising eight SLRs and a C-terminal capping helix. (A) Overall structure of His-LpnE(73-375). The SLR repeats are individually colored, with the N-terminal SLR1 blue and C-terminal SLR8 magenta. The capping helix is red. The N-terminal tag residues are colored gray; (B) structural superposition of SLR1 to SLR8 repeats of LpnE(73-375); (C) sequence alignment-based structural superposition of SLRs and the consensus sequence. The sequences of the predicted four helices in the 22–72 region are added to the alignment. The SLR0 sequence fits the consensus sequence quite well, but the sequence of putative SLR-1 diverges from the consensus; (D) the position of the capping helix differs from the expected position of the αA helix of the next SLR; (E) the packing of the two neighboring molecules involving the largest contact area. They interact through their N- (blue) and C-terminal (red) helices. The residues belonging to the affinity tag were omitted from this view.

    The first helices (αA) of Sel1-like repeats (SLR) are located on the convex, exterior side of the superhelix, while the second helices (αB) are on the concave, interior side. While both surfaces of the superhelix have hydrophilic character, they differ in that the convex side contains many charged side chains while the concave side is lined with uncharged polar side chains. Helix 1-αA of the N-terminal repeat deviates the most from the SLR consensus sequence (Fig. 6C). Hydrophilic residues Glu73, Lys74, Thr75, Glu76, Gln77, Ser82, and Asn84 are evenly dispersed throughout this helix. This arrangement comes in contrast with the amphipathic αA helices found in the rest of the molecule, which display distinct hydrophobic and hydrophilic sides. The final C-terminal capping helix α17 does not follow the Sel1-like arrangement pattern. While the loop between 7-αB and 8-αA is seven residues long, the loop connecting 8-αB to α17 is only three residues long. Thus, α17, instead of adjoining helix 8-αB on the opposite side to 8-αA as occurs between neighboring SLRs, folds toward helix 8-αA and provides the side chain of Tyr367 for stacking with Trp329 that would otherwise by exposed to the solvent and caps the SLR repeats (Fig. 6D).

    Crystal contacts

    The packing of molecules in the crystal involves multiple contacts. The largest intermolecular contacts are between two independent molecules in the asymmetric unit. They are related by a noncrystallographic twofold symmetry and pack in a head-to-head/tail-to-tail fashion through contacts between their SLR1s at the N termini and α17 helices at the C termini to form a loose, twisted doughnut-shaped dimer with large solvent-filled interior (Fig. 6E). The solvent accessible surface of His-LpnE(73-375) is ~16 500 Å2 and the interface area between two independent monomers is ~2500 Å2. However, much of the interface comes from the residues of the tag which extend along the opposite molecule and each provides ~915 Å2 contact area. Excluding the tag, the interface between the two monomers is only ~655 Å2. Of this, 364 Å2 comes from contacts between N termini and 279 Å2 from the C-terminal contact. The contacts with symmetry-related molecules are even smaller, indicating that the tagless protein would be monomeric in solution. The His-tag may promote LpnE(73-375) dimerization in solution as well, by adding to the contact area between monomers as it does in the crystal. Indeed, this would explain the presence of dimeric His-LpnE(73-375) in the size-exclusion chromatogram in addition to the major peak corresponding to a monomer.

    These dimers stack along the a-axis with a solvent channel in the center (Fig. 7A) and the resulting cylindrical stacks pack tightly in a pseudohexagonal arrangement (Fig. 7B). The packing of the dimer stacks is cemented by a ‘handshake’ between the His-tags from two neighboring dimers that involve Ni2+ ion coordination (Fig. 8A,B). All six histidine residues as well as the remainder of the tag preceding the first residue of LpnE(73-375) (Glu73) are well defined in the electron density. One Ni2+ ion rigidifies the His-tag through the coordination of the ND1His52 and backbone amides of Met49, His50, and His51 in a square planar coordination with an average Ni-N distance of 1.95 Å (Fig. 8C). The second Ni2+ ion bridges His-tags from two neighboring molecules by coordinating NE2His52 and NE2His53 from both His-tags, and with two additional water molecules establishing an octahedral coordination (Fig. 8D). The average Ni-N/O distance of 2.13 Å is ~0.2 Å longer than for the square planar coordination. These handshake interactions by the His-tags provide rationale for their role in crystal formation.

    Details are in the caption following the image
    Packing of LpnE(73-375) in the crystal. (A) The loose dimers form cylindrical stacks extending along the crystallographic a-axis with water channels in the middle; (B) the stacks are arranged in a hexagonal lattice in the b,c-plane viewed down the a-axis. This view is rotated 90o along the horizontal axis relative to the view in A.
    Details are in the caption following the image
    The ‘handshake’ formed by the His-tags from two molecules. (A) The His-tags join two dimers shown in Fig. 6B. The molecules contributing to the handshake are shown in wheat and cyan; their dimeric partners are semitransparent; (B) The close-up of the His-tags joined through tetrahedral coordination of a common Ni2+ ion. A second Ni2+ in each His-tag rigidifies the peptide through a square planar coordination with the His side chain and backbone atoms; (C) square planar coordination of Ni2+ ion within one His-tag; (D) tetrahedral coordination of the Ni2+ ion bridging two His-tags.


    LpnE is one of several Legionella effectors containing predicted SLR-like repeats and shares ~30% sequence identity with EnhC, LidL, Lpg1062, and Lpg1356. The effector plays a role in bacterial entry into the host cell, in phagosome acidification and helps the LCV evade fusion with the lysosome. Two host proteins, OCRL and OBSL1, were confirmed to be LpnE-interacting partners and the LpnE-binding regions on these two proteins have been roughly identified [11, 13]. We determined which segment of LpnE is directing its cellular localization, determined the 3-D structure of the SLR repeat containing domain, and refined the OCRL segment required for interaction with LpnE.

    LpnE localization

    Our localization studies showed that full-length LpnE is localized to the cis-Golgi in HEK293 cells. We then investigated if the SLR-containing fragment, LpnE(73-375), retains this localization pattern. Intriguingly, this construct was found predominantly in the cytosol. Since the first 22 residues were predicted as a signal peptide, we rationalized that the localization is associated with the segment 22–72. This is not the case, however, as LpnE(22-375) was also found in the cytosol. We, therefore, concluded that the first 21 amino acids direct LpnE to the cis-Golgi. LpnE is not delivered to the host cell through the Dot/Icm secretion system and the mechanism of its translocation is presently unknown [11]. The N terminus does indeed contain a signal peptide, but it is not known if this peptide is recognized inside Legionella or in the host cell and if or when it is cleaved off. While the mechanism of LpnE retention at the cis-Golgi remains unknown, it is possible that specific amino acid residues act as a sorting signal. Specific recruitment of LpnE to the cis-Golgi may depend on a preference for characteristic glycerophospholipid/sphingolipid ratios, as these lipids distinguish cis-Golgi from trans-Golgi [23].

    The biological significance of LpnE(1-375) retention at the cis-Golgi remains unclear. It is important to note that a punctate localization pattern was observed only in ~20% of HEK293 cells, while a cytosolic localization was observed in the remaining cells. This suggests cleavage of the signal peptide in some instances. LpnE is known to bind phosphatidylinositol-3-phosphate (PI3P) and localize to the Legionella-containing vacuole (LCV) [13]. The retention of LpnE at the LCV may be mediated by the signal peptide and/or the aa 23–72 segment. Cleavage of the signal peptide would liberate LpnE from the LCV, allowing it to interact with various eukaryotic proteins via the concave surface of its superhelix. Recruitment of OCRL to the LCV and its association with LpnE may render the latter inaccessible to host signal peptidases and thereby impede the infective role of LpnE. This notion is in keeping with previous studies showing that the first 51 residues (including the signal peptide) are not necessary for LpnE export and invasion [11]. In short, LpnE may interfere with host cell processes outside of its known role in uptake, with the signal peptide allowing its brief retention on the LCV.

    LpnE–OCRL interaction studies

    Our SEC-MALS data suggest that both LpnE(73-375) and LpnE(22-375) interact with OCRL(10-208), albeit with different stoichiometry. The 69.6 kDa peak containing both LpnE(73-375) and OCRL(10-208) agrees reasonably well with the theoretical molecular weight estimate of 56.6 kDa for a 1 : 1 complex of these proteins. Meanwhile, the 147.7 kDa peak observed for LpnE(22-375)-OCRL(10-208) cannot be rationalized in the same way, as a 1 : 1 stoichiometry would correspond to a theoretical molecular weight of 62.3 kDa. Thus, the observed peak at 147.7 kDa may represent an LpnE(22-375)2-OCRL(10-208)2 heterotetramer.

    It is tempting to speculate from these data that the N-terminal 22–72 residues of LpnE promote its own dimerization and thereby render a dimer-of-dimers in the presence of OCRL(10-208). This hypothesis, however, would also suggest that LpnE(22-375) readily dimerizes in the absence of OCRL(10-208). Our studies of LpnE(22-375) have shown this not to be the case, as LpnE(22-375) appears to be monomeric by SEC-MALS (39.4 kDa calculated vs. 41.8 kDa measured). An alternative explanation is that the binding of OCRL(10-208) to LpnE(22-375) induces a conformational change in either LpnE or OCRL, exposing a protein–protein interaction interface giving rise to dimerization. The fact that no dimer-of-dimers is observed for LpnE(73-375)-OCRL(10-208) may point to a special role of the N-terminal 22–72 residues of LpnE in promoting this process.

    To investigate which parts of LpnE and OCRL are essential for their interactions, we have expressed a series of LpnE constructs with N-terminal deletions and OCRL constructs with C-terminal deletions. We showed that while OCRL(10-208) and OCRL(10-183) bind LpnE, the slightly shorter construct lacking the polyproline sequence, OCRL(10-176) binds LpnE very weakly or not at all. This suggests that the polyproline segment of OCRL is critical for LpnE binding. The polyproline motif within OCRL likely adopts a left-handed PPII helix, as this is energetically favorable to PPI, and these helices are implicated in protein–protein interactions [24]. This extended structure could fit alongside the Sel1 helical motifs forming a concave surface on LpnE. To determine the region of LpnE critical for binding OCRL, we investigated binding of OCRL(10-208) to the shortened LpnE constructs. LpnE(158-375) is missing five helices from the crystallized LpnE construct and still binds OCRL. Moreover, similar to LpnE(73-375), this construct also binds OCRL(10-183) but not the shorter OCRL constructs. However, deleting one more helix from LpnE [LpnE(176-375)] abrogates binding, indicating that the 159–175 helix (3-αB) is critical for the interaction. This helix is at the center of the curved surface in the structure of LpnE. Taken together, our data suggest that the interaction is between the polyproline segment of OCRL and the 3-αB helix of LpnE. Our studies make clear that the N-terminal PH domain of OCRL is unlikely to be involved in binding LpnE and the primary interactions are predominantly with the unstructured linker between the PH and 5-phosphatase domains of OCRL.

    Structural comparison with other Sel1-like repeat proteins

    SLR proteins are prevalent in nature and many structures of proteins containing these domains are known. Comparison of LpnE(73-375) with other such proteins using the DALI server [25] identified EsiB (PDB code 4BWR [17]) from extra-intestinal pathogenic E. coli (ExPEC), HcpB (PDB code 1KLX [26]), HcpC (PDB code 1OUV [16]) of the Helicobacter pylori cysteine-rich (Hcp) family and mouse SEL1L (PDB code 5B26 [27]) of the ER-associated protein degradation (ERAD) machinery as the closest structural homologs. EsiB, HcpB, and HcpC are all SLR proteins from bacterial pathogens and their roles in virulence remain uncertain. Although LpnE has greater sequence identity with EsiB and SEL1L (36%) than with HcpB (29%) or HcpC (31%), structural superposition shows that repeats align best with HcpC (RMSD 2.3 Å). This difference is attributable to the long interrepeat loops, which lean in the direction of the convex outer face of the superhelix. These longer loops play an important role in defining the superhelical geometry of the structure and are noticeably better aligned between LpnE and HcpC. It is likely that this structural similarity made HcpC a good candidate model to obtain initial phases for LpnE by molecular replacement.

    Like EsiB, LpnE lacks the intrarepeat disulfide bonds seen in HcpC or its shorter homolog, HcpB to help stabilizing the repeats. Instead, the superhelical packing of LpnE is stabilized predominantly by its continuous hydrophobic core. This stabilization strategy is also observed in EsiB, which exhibits a strikingly similar pattern of SLR residues to that seen in LpnE. Despite sharing the same sequence identity with LpnE as EsiB, SEL1L forms a superhelix with a tighter twist and associates into dimers through C-terminal helical extension following the last SLR repeat.

    The secondary structure predictions for LpnE indicate the presence of four helices preceding the domain for which we have determined the structure (aa 1-72). Of these, helices α3 and α4 conform well to the consensus sequence of the SLR repeats in LpnE (Fig. 6D) and we suggest that they indeed form an additional SLR repeat (predicted SLR0). Trypsin cleavage of LpnE occurs within a nine residues long linker connecting SLR0 to SLR1 (the N-terminal repeat observed in our structure). The sequences of the predicted helices α1 and α2 do not correspond well to the consensus, and moreover, helix 1 falls within the predicted signal peptide. We, therefore, conclude that only helices 3 and 4 form a SLR repeat and that there are consequently nine SLRs in LpnE.

    Ni2+ coordination by LpnE hexa-histidine tag

    The presence of the His-tag promoted crystallization of LpnE. The six histidines of the tag acquired 1.5 Ni2+ ions per tag. One Ni2+ ion rigidifies the conformation of the tag while the other is shared between the tags from two different molecules and helped in crystal packing. These two Ni2+ adopt two different coordination environments, the intratag square planar coordination with Ni-N distances of 1.95 Å and the intertag octahedral coordination with Ni-N/O distances of 2.13 Å. The observation of two different coordination types in one protein crystal is rather unusual, but it appears to be serendipitous and unrelated to LpnE itself.

    Experimental methods

    Cloning of recombinant LpnE

    The entire lpnE (lpg2222) gene was amplified from Legionella pneumophila (Philadelphia) genomic DNA by PCR. Since residues 1–22 of LpnE were predicted to encode a signal peptide (SignalP,, the DNA sequence encoding residues 22–375 was also amplified. These constructs were designed using the ligation-independent cloning (LIC) method. Ultimately, the DNA sequences were cloned into the pMCSG7 expression vector [28] that incorporated a TEV-cleavable, N-terminal His6-tag. Additionally, a stable fragment, LpnE(73-375), was cloned into pMCSG7 as described above.

    Cloning of recombinant OCRL

    A plasmid containing the human OCRL gene was purchased from Addgene (pcDNA3-HA-human OCRL, plasmid #22207). The following constructs of OCRL were designed: OCRL(1-236), OCRL(10-236), OCRL(1-140), OCRL(10-140), OCRL(10-176), OCRL(10-208). Amplicons were placed in pMCSG7 and a pGEX derivative with a TEV cleavage site (pRL652) via ligation-independent cloning (LIC), as described above for LpnE.

    Protein expression and purification

    Initially, LpnE(1-375) and LpnE(22-375) were expressed in BL21(DE3) and purified as described below. LpnE(1-375) did not express from pMCSG7 in our hands and exhibited poor solubility as a GST fusion protein. His-LpnE(22-375) expressed well and was eluted from Ni-NTA in 100 mm imidazole and then loaded onto size-exclusion column (SEC650, Biorad/S200, GE Life Sciences, Mississauga, Canada) for separation of oligomeric states. Analysis of the eluted fractions by dynamic light scattering (DLS) showed LpnE(22-375) to have a high degree of polydispersity.

    Since LpnE(22-375) formed various aggregates/oligomers in solution, we investigated if the presence of OCRL(10-208) would improve the behavior of this protein. Cell pellets containing His-LpnE(22-375) and His-OCRL(10-208) were mixed, co-lysed, and purified on Ni-NTA followed by size-exclusion chromatography. This protocol resulted in a single narrow peak containing both proteins. An additional strategy was used to identify a fragment of LpnE(22-375) with good solubility was limited proteolysis. Digestion with trypsin produced a stable and soluble, which was identified by mass spectrometry to consist of residues 73–375. This trypsin digestion product gave a symmetrical peak on gel filtration and led to the initial crystal hits. Subsequently, LpnE(73-375) was cloned to replicate the trypsin digestion product and used in all further structural experiments.

    LpnE(73-375), LpnE(22-375), OCRL(10-140), OCRL(10-176), OCRL(10-208), and OCRL(10-236) plasmids were transformed into chemically competent BL21(DE3) cells and plated on LB agar containing ampicillin (100 μg·mL−1). A single transformant was inoculated into 20 mL of LB supplemented with ampicillin (100 μg·mL−1) and glucose (0.4%), and grown overnight at 37 °C. This overnight culture was subcultured into 1 L of terrific broth (TB) supplemented with ampicillin (100 μg·mL−1) and grown at 37 °C. Once the cell culture reached an optical density (A600) of ~1.0, the temperature was reduced to 18 °C, 1 mm of Isopropyl β-d thiogalactopyranoside (IPTG) was added to the culture to induce protein expression, and the cells were incubated for approximately 16 more hours. Cells were pelleted at 6900 g for 15 min in a Beckman JLA 8.1000 rotor and stored at −80 °C until further processed. Approximately 10 g of pellet was obtained from 1 L of culture. Cells were resuspended in 30 mL of a lysis buffer (50 mm Tris, pH 8.0, 10% (v/v) glycerol, 0.1% (v/v) Triton X-100) and lysed two times at 35 kPsi in a cell disruptor (Constant Cell Disruption Systems, Kennesaw, Georgia). The lysate was spun at 21 000 g for 30 min in a Beckman JA25.50 rotor. Supernatant was added to 5 mL of Qiagen Ni-NTA beads preequilibrated with three column volumes of a buffer containing 20 mm Tris, pH 8.0, and 50 mm NaCl, and the beads were washed with 50 mL of this buffer. Protein was eluted with the same buffer supplemented with 100 mm imidazole. Purified protein was concentrated to 16 mg·mL−1 in a 10 kDa molecular weight cutoff Millipore centrifugal filter span at 4000 g. Finally, the protein was loaded onto a Biorad SEC650 or S200 Increase (GE Healthcare, Mississauga, Canada) size-exclusion column for separation of monomeric from dimeric species.

    Crystallization of LpnE(73-375)

    Monomeric and dimeric fractions of LpnE(73-375) were screened for crystallization by the sitting drop method in 96-well plates using in-house and commercial screens. Only monomeric fractions of LpnE(73-375) showed low polydispersity by dynamic light scattering (DynaPro Plate Reader II, Wyatt Technology, Santa Barbara, CA, USA). The proteins were concentrated to ~60 mg·mL−1 in buffer containing 15 mm Tris-HCl, pH 8.0, and 50 mm NaCl, and preparations with and without the His-tag were screened. Untagged LpnE(73-375) gave extremely fine-needle clusters under a few conditions and optimization failed to improve the crystal size. Conversely, His-tagged LpnE produced single crystals in many different conditions. After optimization by the hanging drop vapor diffusion method, the best crystals were obtained at 20 °C in drops containing 1 μL protein mixed with 1 μL of reservoir solution [20% PEG 8000, 0.2 m ammonium sulfate, 0.1 m 2-(N-morpholino)ethanesulfonic acid (MES), pH 6.5] and suspended over 500 μL reservoir solution.

    The LpnE(22-375)-OCRL(10-208) complex displayed low polydispersity and was screened with the His-tag and with the tag removed by TEV protease. Only His-tagged preparations gave crystals after 5–7 days of growth at 20 °C under various conditions within the ProComplex Suite. Examination of the crystal content by SDS/PAGE revealed that His-LpnE(22-375) was the sole constituent. These crystals diffracted poorly and were recalcitrant to optimization.

    Data collection and structure determination

    The protein crystals were cryoprotected by transferring to 1 μL mother liquor containing 20% (v/v) glycerol, 0.2 m ammonium sulfate, 20% PEG8000, 0.1 m 2-(N-morpholino)ethanesulfonic acid (MES) pH 6.5. Diffraction data were collected to 1.75 Å resolution at the Canadian Macromolecular Crystallography Facility (CMCF) 08ID beamline, Canadian Light Source, using a Rayonix MX300_CCD Detector [29]. Integration and scaling were carried out using the xds software package [30] (Table 1).

    Table 1. Data collection and refinement statistics
    Data collection
    Space group P212121
    Unit cell (Å) 67.7, 84.1, 128.3
    Resolution (Å) 48.77–1.75 (1.81–1.75)
    Total reflections 1 058 671 (135 933)
    Unique reflections 74 365 (11 788)
    Completeness (%) 99.83 (98.68)
    Redundancy 14.2 (11.5)
    Mean I/σ(I) 21.12 (1.85)
    Rmerge (%) 8.3 (148.9)
    CC1/2 100 (73.4)
    Resolution (Å) 48.77–1.8 (1.86–1.80)
    No. of reflections 68 575 (6782)
    Rwork/Rfree (%) 18.7/21.4
    No. of atoms 5361
    No. of solvents 307
    RMSD from ideal values
    Bond lengths (Å) 0.013
    Bond angles (°) 1.39
    Ramachandran plot (%)
    Favored 97.37
    Allowed 1.70
    Outliera 0.93
    • Values in parentheses correspond to the highest resolution shell.
    • a The outliers are part of the linker between the His6 and Glu73, the first residue of LpnE, where the electron density is very weak.

    The structure was solved by molecular replacement using the molrep program [31]. Residues 28–285 of H. pylori cysteine-rich protein C (HcpC, PDB code 1OUV, [16]) were used as a search model, and two molecules were expected in the asymmetric unit based on Matthew's coefficient. This molecular replacement solution showed good fit to the electron density only for the N-terminal half of the protein, indicating a possibility for a somewhat different relative orientation of the N- and C-terminal segments between LpnE and the model structure. Therefore, the model protein was divided into two parts and used independently for molecular replacement with Phaser [32]. The final solution showed a small reorientation of the two fragments. This model was rebuilt with phenix.autobuild script and refined using phenix.refine [33] with intermittent manual rebuilding with Coot [34]. The coordinates and structure factors have been deposited with the Protein Data Bank with ID code 6DEH.

    Interaction studies using size-exclusion chromatography

    The interaction of LpnE(73-375), LpnE(158-375), and LpnE(176-375) with OCRL(10-140), OCRL(10-176), OCRL(10-183), and OCRL(10-208) was evaluated by size-exclusion chromatography. Recombinant proteins were purified as described previously and their hexa-histidine tags were cleaved using TEV protease. Complete digestion of the tag was monitored by SDS/PAGE. Purified and cleaved proteins were either injected onto a Biorad SEC70 column for analysis or mixed with a putative-binding partner at an approximate 1 : 1 molar ratio. Putative complexes were allowed a minimum of 30 minutes at 4 °C to form before being injected onto the SEC70 column.

    Size-exclusion chromatography with multi-angle light scattering

    The binding of LpnE(73-375) and LpnE(22-375) to OCRL(10-208) and OCRL(10-236) was evaluated using size-exclusion chromatography (SEC) with multi-angle light scattering (SEC-MALS) at room temperature. A 500 μL sample of size-exclusion purified (Biorad SEC650) LpnE and OCRL was injected at 2 mg·mL−1 on an analytical size-exclusion column, S200 Increase (Wyatt Technologies, Santa Barbara, CA), using AKTA Explorer FPLC system (GE Healthcare). Elution from the column was passed on the MALS system comprising MiniDawn TREOS (WTREOS-11, Serial # 869-TS). The column was equilibrated at room temperature with buffer comprised of 15 mm TrisHCl, pH 8.0, and 50 mm NaCl. Detector normalization was achieved using 2 mg·mL−1 BSA (Pierce, Burlington, Canada).

    Cloning LpnE for localization in human cells

    To obtain the GFP fusion expression vector, full-length LpnE and LpnE(73-375) were PCR-amplified from the genomic DNA of L. pneumophila strain Philadelphia. Fragments were digested with Xho I and BamH I and ligated to the same restriction sites of pEGFP-N1 (Clontech Laboratories, Mountain View, CA, USA) expression vector. Restriction enzymes and T4 DNA ligase were purchased from New England Biolabs (NEB, Ipswich, MA, USA). To obtain LpnE(22-375)-GFP, the signal peptide was looped out of pEGFP-N1 harboring full-length LpnE. All plasmids were confirmed by DNA sequencing (Table 2).

    Table 2. Comparison of theoretical and experimental molecular weight estimates for LpnE and OCRL obtained by size-exclusion chromatography coupled to multi-angle light scattering
    Proteins Theoretical molecular weight Calculated molecular weight
    OCRL(10-208) 22.9 kDa 23.3 kDa
    LpnE(73-375) 33.7 kDa 34.5 kDa
    LpnE(22-375) 39.4 kDa 41.8 kDa
    LpnE(73-375)-OCRL(10-208) 57.8 kDa (1 : 1) 69.6 kDa
    LpnE(22-375)-OCRL(10-208) 130.2 kDa (2 : 2) 147.7 kDa

    Transient transfection of HEK293 cells

    Human embryonic kidney cell line 293 (HEK293) was cultured in Dulbecco's Modified Eagle Medium (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% Fetal Bovine Serum (FBS) (Sigma-Aldrich) at 37 °C with 5% CO2. DNA constructs were transfected into HEK293 cells using the X-treme GENE™ HP DNA Transfection Reagent (Roche, Cat. 06366236001, Mississauga, Canada) according to the manufacturer's instructions.


    GFP-tagged LpnE(1-375), LpnE(22-375), or LpnE(73-375) expressing cells were grown on 12-mm diameter glass coverslips. Cells were fixed with 4% paraformaldehyde solution made up in PBS for 30 min at 20 °C, washed in PBS, permeabilized with 0.5% Triton X-100 in PBST, and blocked in 5% normal horse serum for 20 min. Cells were then incubated with primary anti-GM130 antibody (BD Transduction Laboratories, Mississauga, Canada) for 60 min at RT at a dilution of 1 : 200 in the blocking solution. Cells were washed with PBST, and a secondary Alexa Fluor 546 goat anti-mouse IgG (Invitrogen, Carlsbad, CA, USA) was overlaid on coverslips for 20 min at 20 °C at a dilution of 1 : 2000 in blocking solution. Slides were mounted and visualized on a Laser Scanner Confocal Microscope (Zeiss LSM700, Thornwood, NY, USA).

    Isothermal titration calorimetry

    Titrations were carried out using the Nano ITC instrument (TA Instruments, New Castle, DE). 200 μm OCRL(10-208) in a buffer containing 15 mm Tris-HCl, pH 8.0, and 50 mm NaCl was titrated into the calorimeter cell containing 50 μm LpnE(73-375) in the same buffer. Experiments were performed at 20 °C on untagged proteins. Data analysis was done with nanoanalyze software (TA Instruments, New Castle, DE, USA) using an independent binding model.


    Research described in this paper was supported by CIHR grant MOP-48370 to MC. Data collection was performed at the beamline 08ID-1, Canadian Light Source, which is supported by the Canada Foundation for Innovation, Natural Sciences and Engineering Research Council of Canada, the University of Saskatchewan, the Government of Saskatchewan, Western Economic Diversification Canada, the National Research Council Canada, and the Canadian Institutes of Health Research. We also acknowledge the Protein Characterization and Crystallization Facility, College of Medicine, University of Saskatchewan, for access to the crystallization robot and multi-angle light scattering instrumentation and support. We also thank Dr. G. Katselis for performing mass spectrometry analysis and determining molecular weight of the proteolytic fragment of LpnE.

      Authors contributions

      KV and MC designed the research. KV, KvS, LL, and MTB performed research and analyzed data. KV and MC wrote the paper with comments from KvS and LL.