The Pseudomonas aeruginosa Lectin LecB Causes Integrin Internalization and Inhibits Epithelial Wound Healing.

Pseudomonas aeruginosa is a ubiquitous environmental bacterium that is one of the leading causes of nosocomial infections. P. aeruginosa is able to switch between planktonic, intracellular, and biofilm-based lifestyles, which allows it to evade the immune system as well as antibiotic treatment. Hence, alternatives to antibiotic treatment are urgently required to combat P. aeruginosa infections. Lectins, like the fucose-specific LecB, are promising targets, because removal of LecB resulted in decreased virulence in mouse models. Currently, several research groups are developing LecB inhibitors. However, the role of LecB in host-pathogen interactions is not well understood. The significance of our research is in identifying cellular mechanisms of how LecB facilitates P. aeruginosa infection. We introduce LecB as a new member of the list of bacterial molecules that bind integrins and show that P. aeruginosa can move forward underneath attached epithelial cells by loosening cell-basement membrane attachment in a LecB-dependent manner.

to crawl underneath epithelial cells. Importantly, knocking out LecB diminished the number of P. aeruginosa bacteria found underneath epithelial cells, thus implicating LecB as a virulence factor enabling bacteria to colonize host tissue along the interface between cells and the basement membrane.

RESULTS
Differential effects of LecB at the apical and basolateral side of polarized epithelial cells. When P. aeruginosa infects a human body, it typically encounters first the apical pole of epithelial cells. Through induced or preexisting damages, the bacterium can access the basolateral cell pole of epithelial cells. Since the apical and basolateral plasma membranes of individual epithelial cells harbor distinct sets of membrane proteins and lipids, we investigated if LecB causes different effects when applied to the apical or basolateral side. We chose Madin-Darby canine kidney (MDCK) cells as a model system because they reliably form polarized monolayers in vitro (32,33) and have been already successfully used in P. aeruginosa infection studies (14,34). Purified LecB was able to bind apical and basolateral plasma membranes of MDCK cells (see Fig. S1A in the supplemental material). Interestingly, apical application of LecB resulted in completely different responses of the host cells than did basolateral application (Fig. 1A). After 6 h and 12 h of apical treatment with 50 g/ml (4.3 M) LecB, the overall morphology of the cells was intact as evidenced by staining of ␤-catenin (red) that remained basolateral and green fluorescent protein with a glycophosphatidylinositol anchor (GPI-GFP) (green) that remained apical. In addition, tight junction integrity was not disturbed as demonstrated by the unchanged staining of ZO-1 (white in Fig. 1A) and the preserved transepithelial electrical resistance (TEER) (Fig. 1B). In contrast, 6 h and 12 h of basolateral treatment with 50 g/ml LecB resulted in rounded cell morphologies and severely disturbed epithelial polarity. GPI-GFP became localized all around the cells, and tight junctions almost disappeared (Fig. 1A), which was corroborated by a drastic reduction of the TEER (Fig. 1B). Importantly, these effects cannot be explained by potential LecB-mediated apoptosis or necrosis ( Fig. S1B and C). Yet, the observed consequences seem to be specific for LecB, because another fucosebinding lectin, Ulex europaeus agglutinin I (UEA-I) (35), which also bound to apical and basolateral plasma membranes of MDCK cells (Fig. S1D), did not cause apparent changes in cell morphology (Fig. S1E), nor did it influence the TEER (Fig. S1F).
Taken together, basolateral application of LecB dissolves epithelial polarity in MDCK cells, whereas another fucose-binding lectin, UEA-I, does not cause such effects.
Basolaterally applied LecB binds ␤1-integrin and causes its internalization. To uncover the mechanisms of LecB-induced loss of epithelial polarity, we monitored the localization of cell adhesion receptors upon basolateral LecB stimulation. This revealed a rapid and efficient internalization of ␤1-integrins (Fig. 1C). Interestingly, this effect was reversible after washout of LecB after 6 h (Fig. 1D), and the timing of ␤1-integrin internalization and return to the cell surface correlated well with decrease and increase of the TEER, respectively (Fig. 1E).
To elucidate LecB-triggered ␤1-integrin internalization, we first investigated if LecB binds to ␤1-integrin. To this end, we used LecB-biotin to precipitate LecB-receptor complexes with streptavidin beads. Western blot analysis of the precipitates showed that LecB-biotin is able to bind to ␤1-integrins only when applied to the basolateral side ( Fig. 2A). The binding of LecB-biotin to ␤1-integrin appeared to be strong, since LecB-biotin was able to extract approximately 75% of total ␤1-integrin when applied to the basolateral side, as quantified from the band intensities of the Western blot. In addition, fluorescently labeled LecB colocalized with internalized ␤1-integrins (Fig. 2B) and was able to bind immunoprecipitated ␤1-integrin in a far-Western assay depending on the ␤1-integrin glycosylation status (Fig. 2C), which provides complementary evidence for the capacity of LecB to bind to glycosylated ␤1-integrin.
In summary, LecB recognizes ␤1-integrin at the basolateral cell surface and causes its rapid internalization.
␣3-Integrin and laminin are also internalized and degraded upon basolateral LecB application. From the far-Western assay in Fig. 2C, it can be seen that LecB recognized not only glycosylated ␤1-integrin but also other receptors that were presumably coprecipitated during ␤1-integrin immunoprecipitation (blue arrows). Hence, in a next step, we identified basolateral interaction partners of LecB by LecBbiotin coprecipitation followed by mass spectrometry (MS) analysis and found 65 profoundly enriched proteins (Table S1). This analysis revealed that LecB is able to pull down virtually all integrins expressed by MDCK cells (36), and also, many proteins known to interact with integrins, such as tetraspanins, basigin, and epidermal growth factor receptor (EGFR), were detected. From this, it appears that integrins are major cellular receptors of LecB. We focused our further analysis on ␣3␤1-integrin and were able to demonstrate that ␣3-integrins are cointernalized with ␤1-integrins upon baso- filter-grown MDCK cells stably expressing the apical marker GPI-GFP (green) were left untreated (ctrl) or treated apically (AP) or basolaterally (BL) with 50 g/ml LecB for the indicated time periods, fixed, and stained with antibodies recognizing the basolateral marker ␤-catenin (red) and the tight junction marker ZO-1 (white); nuclei were stained with DAPI (blue). Representative sections along the apicobasal axis (x-z sections) extracted from confocal image stacks are shown. (B) Time course of the transepithelial electrical resistance (TEER) of MDCK monolayers treated AP or BL with LecB. The mean values from n ϭ 3 experiments are displayed. For evaluating statistical significance, a paired two-sided t test using GraphPad Prism 5 was applied; **, P Ͻ 0.01. (C) LecB was applied BL to MDCK cells stably expressing PH-Akt-GFP (green) for the indicated time periods. Cells were fixed and stained for ␤1-integrin (red); nuclei were stained with DAPI (blue). Representative x-z sections extracted from confocal image stacks are depicted. (D) MDCK cells were treated with LecB as indicated, fixed, and stained for ␤1-integrin (red) and ␤-catenin (green). Maximum-intensity projections of confocal image stacks covering total cell heights are shown. (E) The time course of the TEER of MDCK cells treated BL with LecB as indicated and after washout was measured.
Thuenauer et al. lateral LecB application (Fig. 3A). Surprisingly, also the major ligands of ␣3␤1-integrin expressed by MDCK cells, laminin-332 and/or -511 (37), were cointernalized (Fig. 3B). This suggests that LecB is able to cause endocytosis of intact ␣3␤1-integrin-laminin complexes. To measure the dynamics of ␣3␤1-integrin internalization, we carried out surface biotinylation experiments (38). These experiments confirmed the rapid LecBtriggered internalization of ␣3and ␤1-integrin subunits ( Fig. 3C to E). In addition, the surface biotinylation experiments revealed that also the intracellular amount of ␣3and ␤1-integrin subunits decreased upon LecB stimulation, suggesting a degradation of internalized integrins. Consistently, LecB-mediated reduction of integrins was also detected when whole-cell lysates were subjected to Western blot analysis (Fig. S2A). Loss of integrins by degradation is further supported by our finding that internalized integrins after basolateral LecB treatment showed a time-dependent increase in colocalization with the late endosome marker Rab9 ( Fig. S2B and C) and the lysosome marker Lamp1 (Fig. S2D and E).
In the surface biotinylation experiments, we were not able to distinguish between active and inactive ␤1-integrins. Thus, we devised an alternative strategy in which we applied activation-specific ␤1-integrin antibodies to the basolateral surface of live cells. This approach revealed that LecB internalizes active and inactive ␤1-integrins at similar kinetics ( Fig. 3F to H and Fig. S3F), which indicates that the activation status of ␤1-integrins does not play an important role in LecB-mediated integrin internalization.
Taken together, LecB binds integrins, including ␣3␤1-integrin, and causes their internalization and degradation regardless of their activation status and bound basement membrane ligands.
Membrane invagination by LecB and LecB-mediated cross-linking of fucosylated lipids with ␤1-integrin can explain LecB-triggered integrin internalization. Endogenous lectins, like galectin-3, were previously shown to be able to mediate integrin internalization (39,40). The proposed mechanism for galectin-3-mediated integrin internalization is that galectin-3 is able to cause plasma membrane invagina- to polarized filter-grown MDCK cells, or cells were mock treated AP or BL. Cells were lysed, and LecB-biotin-receptor complexes were precipitated with streptavidin beads. Afterward, the presence of ␤1-integrin was probed by Western blotting in the precipitate and the remaining supernatant of the precipitation. (B) LecB-Cy3 (red) was applied basolaterally to MDCK cells for 6 h. Cells were fixed and stained for ␤1-integrin (green). A confocal section (x-y section) crossing the cells in the subapical region is displayed, since most internalized vesicles were concentrated in this region. (C) MDCK cells were lysed, and ␤1-integrins were immunoprecipitated and treated or left untreated with peptide-N-glycosidase F (PNGase F) to remove N-linked glycans. Western blot analysis of the immunoprecipitated ␤1-integrins was performed, and ␤1-integrin presence was proven by staining with anti-␤1-integrin antibodies (white arrows). Also, bands from the antibody used for ␤1-integrin precipitation (white arrowheads) and proteins that putatively coprecipitated with ␤1-integrin (blue arrows) are visible. To probe the binding of LecB to ␤1-integrin, LecB-Cy5 was incubated with membranes (far-Western assay).
tions by binding to glycolipids and also drags integrins into the invaginated membrane regions by functioning as a cross-linker between glycolipids and integrins.
Since LecB is a tetramer with four opposing fucose-binding sites (41), which represents an ideal geometry for a potential cross-linker, we investigated if a galectin-like mechanism could explain LecB-mediated ␤1-integrin internalization.
In a first step, we examined if binding of LecB to fucosylated glycosphingolipids is sufficient to induce membrane invaginations in giant unilamellar vesicles (GUVs) (42). Indeed, GUVs containing glycosphingolipids that bear the fucosylated Lewis a antigen ( Fig. 4A) or other fucosylated glycosphingolipids (Fig. S3) showed invaginations immediately after LecB application, whereas control GUVs with the nonfucosylated glyco- LecB-Cy3 (red) as indicated, fixed, and stained for panlaminin (green) and ␤1-integrin (ITG-b1; magenta); nuclei were stained with DAPI (blue); x-y confocal sections through the middle of the cells are depicted. (C to E) MDCK cells were treated with LecB as indicated, and surface biotinylation from the basolateral side was performed. After precipitation of biotinylated proteins, the precipitates representing the surface fraction (S) and the supernatant representing the intracellular fraction (I) were subjected to Western blot analysis, and ␣3-integrins and ␤1-integrins were probed, as well as actin to control for purity of the surface fractions. Quantification for ␣3 (D) and ␤1 (E) subunit-composed integrins from n ϭ 3 independent experiments. (F to H) LecB was applied basolaterally to MDCK cells for the indicated time periods followed by basolateral application of activation-specific anti-␤1-integrin antibodies to live cells. After fixation, the signal from bound anti-␤1-integrin antibodies in randomly chosen regions of interest was measured and normalized to the cell number in the regions (n ϭ 5 for one experiment). The graphs show the mean value from n ϭ 3 experiments with the activating anti-␤1-integrin antibody 9EG7 (F) and the inhibitory anti-␤1-integrin antibodies monoclonal antibody (MAb) 13 (G) and AIIB2 (H).

Thuenauer et al.
® sphingolipid lactotetraosylceramide (Lc4cer) did not (Fig. 4A). To investigate the relevance of this effect, we carried out experiments with energy-depleted cells, because under these conditions cellular machineries cannot pinch off vesicles, which previously led to easily visible membrane invaginations when other lipid-binding lectins like Shiga toxin were applied (42,43). Indeed, LecB was able to induce plasma membrane invaginations in energy-depleted MDCK cells (Fig. 4B). Importantly, ␤1-integrin colocalized with fluorescently labeled LecB at invaginations (Fig. 4B, magnification), thus implying that LecB can recruit integrins to invaginations. Furthermore, we observed that basolateral LecB application led to marked clustering of endogenous galectin-3 ( Fig. S4), which could suggest that LecB outcompetes galectin-3-integrin interaction.
In summary, LecB is able to cause membrane invaginations upon binding to fucose-bearing glycosphingolipids. Since LecB is also able to bind integrins and thus to cross-link lipids in membrane invaginations with integrins, this provides a mechanistic explanation for LecB-mediated integrin internalization.
LecB inhibits cell migration and epithelial wound healing. As an opportunistic pathogen, P. aeruginosa mainly relies on, and exploits, extrinsic circumstances-like a wound-to gain access to the basolateral side of epithelia. In addition, integrin blocking, e.g., through antibodies, has been previously shown to inhibit cell migration in wound healing assays (44). These considerations motivated us to investigate the effect of LecB on epithelial wound healing. Indeed, the presence of LecB strongly inhibited collective cell migration and wound healing in MDCK monolayers (Fig. 5A). Importantly, this effect was blockable with L-fucose, demonstrating that LecB needs to bind to host cells to cause migration defects. Moreover, we established that the blockage of wound healing by LecB occurred in a dose-dependent manner, with concentrations larger than 50 g/ml completely blocking cell migration ( Fig. 5B and C),  whereas another fucose-binding lectin, UEA-I (50 g/ml), did not induce suppression of wound healing (Fig. S5A). The inhibitory effect of LecB on wound healing was-like other LecB-mediated effects before-reversible by washing out LecB (Fig. S5B).
To explain the abrogation of cell migration by LecB, we carried out live-cell imaging experiments with MDCK cells stably expressing the plasma membrane marker ML-GFP (GFP tagged with a Lyn-derived myristoylation motif) (45). ML-GFP allowed visualization of lamellipodia formed by migrating MDCK cells (Fig. 5D, ctrl, arrows). Interestingly, when cells were treated with LecB right after wounding, no lamellipodia formed (Fig. 5D, LecB), whereas washout of LecB was sufficient to reinstate lamellipodium formation and cell migration (Fig. S5C). When migrating cells were treated with LecB, lamellipodia "froze" and LecB strongly bound to lamellipodia (Fig. S6). It is also interesting that cells deeper within the monolayer, which expose only their apical membranes to LecB in this assay, did still move (Fig. S6). In subsequent experiments, we stained for ␤1-integrins in wound healing assays. This revealed that cells at the wound edge take up large amounts of LecB, and in the same cells, pronounced ␤1-integrin internalization was evident (Fig. 5E, arrows), which can explain why these cells are not able to migrate any more.
Taken together, LecB inhibits epithelial wound healing in a reversible manner, which is presumably caused by the fact that integrins in wound edge cells are accessible by LecB and are internalized.
Depending on LecB expression, P. aeruginosa is able to crawl underneath cells. The additional and probably dominant cytotoxic effects caused by the numerous toxins produced by P. aeruginosa prevented us from directly quantifying an effect of LecB knockdown in in vitro wound healing assays with live P. aeruginosa. However, we observed that P. aeruginosa (PAO1-wt) is able to crawl underneath exposed cells (Fig. 6A). We postulated that this requires at least local loosening of potentially integrin-mediated cell-substrate adhesion. We tested this hypothesis by investigating the influence of LecB on "P. aeruginosa crawling." To this end, we used a LecB-deficient PAO1 strain (PAO1-dLecB), which exhibited the same growth kinetics as PAO1-wt (Fig. S7A). After overnight culture, the PAO1-wt strain showed clear expression of LecB, whereas the LecB-deficient P. aeruginosa strain (PAO1-dLecB) did not ( Fig. S7B and C). In accordance with our hypothesis, PAO1-dLecB was visibly found in smaller numbers underneath cells (Fig. 6A). To substantiate the experimental procedure, we established that increasing the multiplicity of infection (MOI) (Fig. 6B and C) and increasing the duration of incubation ( Fig. 6B and D) also increased the number of bacteria crawling underneath cells. Importantly, for all investigated conditions, the number of bacteria crawling underneath per cell was lower for the PAO1-dLecB strain than the PAO1-wt strain.
In summary, our study reveals a novel mechanism in which LecB, through impairing cell-to-basement membrane attachment, allows for the creation of microniches enabling subepithelial colonization by P. aeruginosa.

DISCUSSION
Integrins are major receptors of LecB. Lectins bind to carbohydrates and can therefore have multiple receptors that express the appropriate glycosylation. Hence, it is interesting that our experiments suggest a favored binding of LecB to integrins. LecB binding to ␤1-integrin appeared to be strong, because LecB was able to recover 75% Integrin Internalization by the Lectin LecB ® of total ␤1-integrin from cells when it was basolaterally applied and had access to ␤1-integrin only at the cell surface. As expected, LecB binding to ␤1-integrin solely occurred through carbohydrates, since removal of ␤1-integrin glycosylation abolished binding. Furthermore, mass spectrometry analysis of all basolateral LecB receptors revealed that virtually all integrins expressed by MDCK cells were among the top hits, including many integrin-associated proteins that were presumably recovered through coprecipitation with integrins.
This introduces LecB as a new member of the list of bacterial molecules that bind integrins. Another fucose-binding lectin, UEA-I, was not able to cause any of the cellular effects that LecB caused. This shows that binding to fucose is not sufficient by itself and suggests that LecB has additional features that bring about its specific capabilities. First,  LecB and UEA-I prefer different fucose linkages in oligosaccharides (UEA-I binds ␣ 1,2linked fucose but only weakly binds ␣ 1,3 -and ␣ 1,6 -linked fucose [46,47]; LecB shows a slight preference for ␣ 1,4 -linked fucose but also binds to ␣ 1,2 -and ␣ 1,3 -linked fucose [48][49][50]). In addition, LecB is a tetramer with four fucose-binding sites, but UEA-1 is a dimer offering only two binding sites (35).
It is also worth noting that the sequence of LecB varies slightly between different strains of P. aeruginosa (51,52). Since ligand binding among LecB variants is conserved, we expect that these LecB variants are utilized in similar ways as we reported here for LecB in the P. aeruginosa strain PAO1.
Mechanism of LecB-mediated integrin internalization. Since LecB recognizes only the carbohydrate moieties of integrins, it is able to manipulate integrins via unique mechanisms. The structure of LecB with four opposing binding sites is ideal to cross-link receptors. This has been demonstrated before in vitro by showing that LecB is able to cross-link GUVs that contain LecB receptors (53).
We hypothesize that the cross-linking capacity of LecB is key for the observed rapid internalization of integrins. In addition to integrins, LecB was also able to recognize fucose-bearing glycosphingolipids and to cause membrane invaginations without the need of additional energy input, which replicates effects caused by other glycosphingolipid-binding lectins like Shiga toxin subunit B (StxB) (42), cholera toxin subunit B 9 (CtxB) (54), or Ralstonia solanacearum lectin (RSL) (55). In addition, ␤1integrin was also found on LecB-generated membrane invaginations on energydepleted cells. This suggests that LecB on the one hand causes membrane invaginations and on the other hand is able to recruit integrins to these invaginations, thus constituting a potent endocytic mechanism. A similar mechanism was suggested for the host cell-endogenous protein galectin-3 (40). This means that bacteria have evolved molecules like LecB that can hijack this endogenous uptake route. Importantly, LecBmediated lipid-integrin cross-linking for internalization represents a mechanism that can explain the observed integrin internalization that occurred regardless of integrin activation status and carried also basement membrane ligands like laminins with it.
LecB-triggered inhibition of wound healing. LecB binding to basolateral cell surfaces caused cellular effects that could be causally linked to integrin internalization. In fully polarized epithelial cells, binding of LecB to basolateral cell surfaces but not to apical cell surfaces, which contain only minute amounts of integrins, led to loss of apicobasal polarity. This indicates the need of integrin internalization for dissolution of polarity. In addition, loss of polarity was reversible after washout of LecB and coincided well with the return of ␤1-integrin to the basolateral plasma membrane. Our data also suggest that LecB-mediated integrin internalization is responsible for the observed block in cell migration in epithelial wound healing assays. First, integrins were preferentially internalized in wound edge cells, and LecB prominently bound to lamellipodia in edge cells. This makes sense, since edge cells offer more cellular surface area for LecB binding and expose their integrins, in contrast to cells deeper within the intact monolayer, which display only their apical surfaces to LecB. Second, edge cells rapidly and efficiently stopped moving upon LecB treatment, whereas other cells deeper within the monolayer preserved their capacity to move within the monolayer.
The effects that LecB exerts through integrin internalization require basolateral access of LecB. In this context, it is interesting that P. aeruginosa possesses strategies to convert apical into basolateral plasma membranes (34). However, the direct way for P. aeruginosa and LecB to access integrins is through wounds in an epithelial cell layer.
P. aeruginosa preferentially colonizes wounds, and diminished healing of P. aeruginosa-infected wounds was reported previously (10). Importantly, P. aeruginosa possesses multiple mechanisms to manipulate and to intoxicate host cells. We therefore anticipate that, in order to inhibit wound healing, P. aeruginosa will apply different combinations of its arsenal, including LecB-mediated epithelial depolarization and inhibition of cell migration, depending on the host tissue (25).

LecB has a role in enabling bacteria to crawl underneath host cells.
Our experiments revealed a novel feature of P. aeruginosa behavior. We observed that bacteria frequently crawled underneath host cells. For this, at least local loosening of cell-substrate adhesion is required, which could be achieved by LecB-mediated integrin internalization. We therefore investigated the contribution of LecB to crawling of the P. aeruginosa strain PAO1 underneath cells. Indeed, knocking out LecB significantly decreased crawling events. Based on our results, we suggest that P. aeruginosa uses LecB to manipulate integrin-basement membrane interaction to proceed along the interface between epithelial cells and the basement membrane.
In summary, our work brings integrins into focus as targets of P. aeruginosa and provides additional rationales for the ongoing efforts to develop LecB inhibitors as a treatment strategy in addition to antibiotics (26)(27)(28)(29)(30)(31).

MATERIALS AND METHODS
Antibodies, plasmids, and reagents. Used antibodies are listed in Table S2 in the supplemental material. The plasmid pPH-Akt-GFP encoding PH-Akt-GFP was a gift from Tamas Balla (Addgene plasmid no. 51465). The plasmid encoding GFP tagged with a glycosylphosphatidylinositol (GPI) anchor (pGPI-GFP) was described before (56). The plasmid encoding GFP tagged with a Lyn-derived myristoylation motif (pML-GFP) was a gift from Christian Wunder (Curie Institute, Paris, France).
Recombinant LecB was produced in Escherichia coli BL21(DE3) cells and purified with affinity columns as previously described (41). LecB and fluorophore-conjugated LecB were used at a concentration of 50 g/ml (4.3 M) unless stated otherwise. UEA-I and UEA-I-fluorescein isothiocyanate (FITC) were from Vector Labs. Cycloheximide, and L-fucose (6-deoxy-L-galactose) were from Sigma-Aldrich.
Mammalian cell culture and creation of stable cell lines. MDCK strain II cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal calf serum (FCS) at 37°C and 5% CO 2 . Unless stated otherwise, 3 ϫ 10 5 MDCK cells were seeded per transwell filter (0.4-m pore size, polycarbonate membrane, catalog no. 3401 from Corning) and cultured for 4 days before experiments. For the creation of stable MDCK cell lines, plasmids encoding the proteins of interest and G418 resistance (pPH-Akt-GFP, pML-GFP, and pGPI-GFP) were transfected into cells with Lipofectamine 2000 (Thermo Fisher). After the cells were allowed to express the proteins overnight, they were trypsinized and plated sparsely in medium containing 1 mg/ml G418. After single colonies had formed, GFP-positive colonies were extracted with cloning rings. At least 6 colonies were extracted for each cell line, grown on transwell filters for 4 days, fixed, and stained against the basolateral marker protein ␤-catenin and the tight junction marker protein ZO-1 to assay their polarized morphology. Based on these results, we chose one colony for each cell line for further experiments. TEER measurements were carried out using an EVOM2 equipped with chopstick electrodes (World Precision Instruments).
Immunofluorescence. Cells were washed two times with phosphate-buffered saline without Ca 2ϩ and Mg 2ϩ (PBS) and then fixed with 4% (wt/vol) formaldehyde (FA) for 15 min at room temperature. Samples were treated with 50 mM ammonium chloride for 5 min to quench FA and then permeabilized with a SAPO medium (PBS supplemented with 0.2% [wt/vol] bovine serum albumin [BSA] and 0.02% [wt/vol] saponin) for 30 min. Primary antibodies were diluted in SAPO medium and applied on the samples for 60 min at room temperature. After three washes with PBS, secondary dye-labeled antibodies and, if required, 4=,6-diamidino-2-phenylindole (DAPI) and dye-labeled phalloidin were diluted in SAPO medium and applied to the cells for 30 min at room temperature (details for the antibodies used are listed in Table S2). After 5 washes with PBS, cells were mounted for microscopy. Since ␣3-integrin antibodies did not work in FA-fixed cells, methanol fixation was applied in this case. Briefly, cells were incubated with precooled methanol for 15 min at Ϫ20°C. After washing with PBS, cells were permeabilized with 0.05% (vol/vol) Triton X-100 for 10 min at room temperature and blocked with 3% (wt/vol) bovine serum albumin (BSA) for 1 h at room temperature. Staining with primary and secondary antibodies was then carried out as described before but with a 3% (wt/vol) BSA solution.

Microscopic imaging of fixed cells and live-cell experiments.
For microscopic imaging, an A1R confocal microscope (Nikon) equipped with a 60ϫ oil immersion objective (numerical aperture [NA] ϭ 1.49) and laser lines at 405 nm, 488 nm, 561 nm, and 641 nm was utilized. Image acquisition and analysis were performed with NIS-Elements 4.10.04 (Nikon).
Live-cell experiments were carried out at 37°C, and MDCK cells grown as polarized monolayers for 3 days on Lab-Tek II chambered cover glasses (8 well, 1.5 borosilicate glass) were used. The medium was changed to recording medium (Hanks' balanced salt solution [HBSS] supplemented with 1% FCS, 4.5 g/liter glucose, and 20 mM HEPES).
Wound healing assays. MDCK cells were seeded on 12-well plates or, for live-cell microscopy of cell migration, on 8-well Lab-Tek II chambered cover glasses and allowed to form confluent monolayers for 3 days. Then, cells were scratched with a 200-l pipette tip to inflict a wound. On 12-well plates, marker lines were drawn on the bottom to ensure that always the same position of the wound was imaged.
Western blot analysis. Cells were washed twice with PBS and lysed in RIPA buffer (20 mM 1 mM sodium orthovanadate). Protein concentrations were analyzed using a bicinchoninic acid (BCA) assay kit (Pierce). Equal amounts of protein per lysate were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was blocked with Tris-buffered saline (TBS) supplemented with 0.1% (vol/vol) Tween 20 and 3% (wt/vol) BSA for 1 h and incubated with primary and horseradish peroxidase (HRP)-linked secondary antibodies diluted in the blocking solution. In some cases, TBS supplemented with 0.1% (vol/vol) Tween 20 and 5% (wt/vol) milk powder was used (details for the antibodies and conditions used are listed in Table S2). Detection was performed by a chemiluminescence reaction using the Fusion-FX7 Advance imaging system (Peqlab Biotechnologie GmbH).
Energy depletion. Energy depletion was carried out as described before (42). Briefly, MDCK cells were washed 2 times with PBS supplemented with 100 mg/liter CaCl 2 and 100 mg/liter MgCl 2 ·6H 2 O (PBS ϩϩ ) and then treated with PBS ϩϩ supplemented with 10 mM 2-deoxy-D-glucose and 10 mM NaN 3 for 30 min at 37°C.
Bacterial culture and crawling experiments. GFP-tagged P. aeruginosa PAO1 wild-type (PAO1-wt) and LecB-deficient (PAO1-dLecB) strains were provided by S. de Bentzmann (CNRS, Aix Marseille University, France). The generation of LecB-deficient PAO1 is described in reference 23, and GFP tagging was carried out according to the procedure described in reference 57. For experiments, bacteria were cultured overnight (approximately 16 h) in LB-Miller medium containing 60 g/ml gentamicin in a shaker (Thriller; Peqlab) at 37°C and 650 rpm. The bacteria reached an optical density (OD) measured at 600 nm of approximately 5. Using these growth conditions, PAO1-wt and PAO1-dLecB strains showed comparable growth kinetics (Fig. S7A) and harvested PAO1-wt efficiently expressed LecB, whereas PAO1-dLecB did not ( Fig. S7B and C).
For crawling experiments, MDCK cells were sparsely seeded on 8-well Lab-Tek II chambered cover glasses and cultured for 1 day, so that clusters of maximally 10 cells in diameter formed to ensure that all cells were exposed to P. aeruginosa similarly as at a wound edge. Then, PAO1-wt or PAO-dLecB was applied for the indicated MOI and duration. Bacteria crawling under cells were counted manually per cell from confocal image stacks of whole cells to ensure that only bacteria located directly underneath cells at the level of the glass coverslip were counted.
Cell surface biotinylation and immunoprecipitation. For cell surface biotinylation, all following steps were carried out in a cold room (4°C). Sulfo-NHS-biotin (Thermo Fisher) was freshly diluted in PBS ϩϩ (concentration, 0.3 mg/ml) and applied to the apical or basolateral plasma membrane of transwell filter-grown MDCK cells for 20 min. Afterward, the reaction was quenched for 20 min with PBS ϩϩ supplemented with 50 mM ammonium chloride. Cells were lysed with RIPA buffer, and biotinylated proteins were precipitated with streptavidin-agarose beads (Thermo Fisher). Elution was carried out with Laemmli buffer (2% [wt/vol] SDS, 10% [vol/vol] glycerol, 60 mM Tris-Cl [pH 6.8] in water) and boiling at 98°C for 5 min.
Mass spectrometry-based identification of LecB interaction partners. MDCK cells were cultured in medium for stable-isotope labeling by amino acids in cell culture (SILAC medium) for 9 passages and then seeded on transwell filters and allowed to polarize for 4 days. For the first sample, biotinylated LecB was applied to the apical side of light-SILAC-labeled cells and on the basolateral side of medium-SILAClabeled cells, whereas heavy-SILAC-labeled cells received no stimulation and served as control. For the second sample, the treatment conditions were permuted. After lysis with IP lysis buffer, the different SILAC lysates were combined and LecB-biotin-receptor complexes were precipitated using streptavidinagarose beads as described before. Eluted LecB-biotin-receptor complexes were then prepared for mass spectrometry (MS) analysis using SDS-PAGE gel electrophoresis. Gels were cut into pieces, proteins therein were digested with trypsin, and resulting peptides were purified by stop-and-go-extraction (STAGE) tips.
For mass spectrometry analysis, samples were fractionated by nanoscale high-pressure liquid chromatography (HPLC) on a 1200 HPLC (Agilent Technologies, Waldbronn, Germany) connected online to an LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). Fused silica HPLC-column tips with a 75-m inner diameter were self-packed with ReproSil-Pur 120 ODS-3 (Dr. Maisch, Ammerbuch, Germany) to a length of 20 cm. Samples were directly injected into the mass spectrometer (for details see reference 58). The raw data files were uploaded into the MaxQuant software. Database searches were performed against a full-length dog database containing common contaminants such as keratins and enzymes used for in-gel digestion. Carbamidomethylcysteine was set as fixed modification, and oxidation of methionine and protein amino-terminal acetylation were set as variable modifications. Triple SILAC was used as quantitation mode. The enzyme specificity was trypsin/ PϩDP with three allowed miscleavages. The MS/MS tolerance was set to 0.5 Da, and the mass precision of identified peptides after recalibration was in general less than 1 ppm. For identification and quantitation, the following settings were used: peptide and protein false-discovery rates (FDR) were set to 0.01; maximum peptide posterior error probability (PEP) was set to 0.1; minimum peptide length was set to 7; minimum number peptides for identification and quantitation of proteins was set to two, of which one must be unique; minimum ratio count was set to two; and only unmodified peptides and the variable modification were used for protein quantification. The "match between run" option was used with a time window of 2 min.
From the generated list of MS-identified proteins, we defined those proteins as LecB interaction partners that showed more than 2-fold enrichment on a log 2 scale over controls in both SILAC samples (see Table S1 in the supplemental material).
Statistics. If not stated otherwise, data obtained from n ϭ 3 independent experiments were used to calculate arithmetic means. Error bars represent standard errors of the means (SEM). Statistical significance analysis was carried out using GraphPad Prism 5.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.