Use of OmpU porins for attachment and invasion of Crassostrea gigas immune cells by the oyster pathogen Vibrio splendidus

V. splendidus LGP32 Invades and Survives in Oyster Hemocytes, Impairing Immune Defense Functions.

A GFP-expressing LGP32 was constructed that carries the gfp3 gene under the P_TRC constitutive promoter in a genomic locus. After a 2-h in vitro contact between the GFP-expressing LGP32 and oyster hemolymph, 31.7 ± 5.6% hemocytes were GFP⁺, as determined by flow cytometry (Fig. 1A). A massive engulfment of LGP32 was observed on confocal microscopy optical sections (Fig. 1 B and C). Although less intense, the engulfment of GFP-expressing LGP32 by hemocytes was also observed in vivo 1 h after injection to oysters (Fig. 1D). Interestingly, counts of intracellular Vibrios monitored by plating hemocyte lysates from in vitro contacts were stable at 10⁶ cfu/mL over 16 h (Fig. 2A). This finding showed that LGP32 remains viable and culturable within oyster hemocytes. In contrast, the avirulent control LMG20012^T (6, 28), which was also avidly engulfed, did not survive within hemocytes, with cfu counts dropping by 4 logs over time (Fig. 2A). The bacterial load related to the oyster-associated environmental microflora remained below 10 cfu/mL over the time course.

Hemocyte responses were measured 1 h after engulfment of LGP32 or LMG20012^T, when similar intracellular bacterial loads (10⁵ to 10⁶ cfu/mL) are present in both samples (Fig. 2A). Unchallenged hemocytes were used as a control. Interestingly, the lysotracker fluorescence indicative of acidic vacuole formation was similar in control and LGP32-containing hemocytes but significantly higher in LMG20012^T-containing hemocytes (P < 0.05) (Fig. 2B). Similarly, the 2',7'-dichlorfluorescein-diacetate (DCFH-DA) fluorescence indicative of ROS production was significantly higher in LMG20012^T- than in LGP32-containing hemocytes (P < 0.05), although both were higher than in control hemocytes (Fig. 2B). Thus, defense reactions involved in the efficient control of phagocytosed bacteria, such as LMG20012^T, are significantly attenuated in hemocytes invaded by LPG32.

LGP32 Uses β-Integrin Recognition and Subverts the Host-Cell Actin Cytoskeleton for Hemocyte Invasion.

The role of β-integrin in recognition and binding of LGP32 was studied by incubating oyster hemolymph with anti–β-integrin or an irrelevant antibody before contact with GFP-expressing LGP32. Strikingly, flow cytometry counts of GFP⁺ hemocytes dropped by 72% upon anti–β-integrin treatment (P < 0.001), whereas they remained stable with the irrelevant antibody (Fig. 3A). Similarly, incubation of hemolymph with an Arg-Gly-Asp (RGD) tripeptide inhibited invasion by 72% (P < 0.01), but an irrelevant Glu-Gly-Phe (EGF) tripeptide had no significant effect (Fig. 3B). This result showed that the RGD-recognition domain of β-integrin is essential for hemocyte invasion.

Contacts between oyster hemolymph and GFP-expressing LGP32 were then performed in the presence of hypertonic concentrations of sucrose, which inhibit receptor-mediated phagocytosis by inducing abnormal clathrin polymerization (29). Sucrose inhibited hemocyte invasion in a dose-dependent manner, from 24% (P < 0.01) at 0.1 M up to 68% (P < 0.001) at 0.5 M (Fig. 3C). Similarly, cytochalasin D used as an inhibitor of actin polymerization inhibited hemocyte invasion from 13% at 1 μg/mL (P < 0.05) to 71% at 10 μg/mL (P < 0.001) (Fig. 3D). This finding showed that hemocyte invasion by LGP32 uses host cell clathrin and actin polymerization.

Hemocyte Invasion Requires the OmpU Porin and a Plasma Opsonin That Specifically Binds OmpU.

The ability of an avirulent ΔompU mutant (6) to invade hemocytes was studied by constructing a GFP-expressing derivative. Hemocyte–Vibrio interactions were monitored by flow cytometry. Strikingly, the number of GFP⁺ hemocytes shifted from 31.7 ± 5.6% when exposed to LGP32 and down to 2.5 ± 1% when exposed to the ΔompU mutant (P < 0.01) (Fig. 4A). In agreement, confocal microscopy showed little to no intracellular GFP-expressing bacteria (Fig. 4B), indicating that OmpU is required for hemocyte invasion. Furthermore, invasion of hemocytes by wild-type LGP32 dropped dramatically from 31.7 ± 5.6% to 7.5 ± 2.3% (P < 0.05) upon plasma withdrawal, and returned to 32.0 ± 6.0% when plasma was added back (Fig. 4A and Fig. S1). This finding strongly suggests that a plasma component operates as an opsonin. To determine how OmpU and plasma are involved in the host–pathogen interaction, we tested the ability of LGP32 and ΔompU to bind to immobilized plasma proteins. Results showed that OmpU is required for the dose-dependent binding of LGP32 to plasma (Fig. S2). Taken together, these data strongly suggest that a plasma opsonin mediates recognition and promotes phagocytosis of LGP32 by binding to OmpU.

Oyster Cg-EcSOD Binds to OmpU in an RGD-Dependent Manner and Promotes Hemocyte Invasion by LGP32.

To identify OmpU-binding plasma proteins, plasma was fractionated by HPLC and fractions were tested for binding to GFP-expressing LGP32 and ΔompU. Only three HPLC fractions, referred to as f6, f7, and f8, showed differential binding (P < 0.05) (Fig. 5 A and C). On SDS/PAGE, they displayed a 30 kDa band predominant in f6 and f7, which was absent from the fractions displaying no differential binding (Fig. 5B). An LC-MS/MS analysis of f6 and f7 unambiguously identified the proteins as variants of Cg-EcSOD (GenBank AAY60161.1) (Fig. S3). The aberrant migration of Cg-EcSOD variants at ∼30 kDa instead of 20 kDa is in agreement with previous studies (19, 30). Fractions f6 and f7, which could contain different variants or foldamers of Cg-EcSOD, as presumed from their different chromatographic behavior (31), displayed similar binding to LGP32 (Fig. 5C). The binding of LGP32 to immobilized Cg-EcSOD was OmpU-dependent, dose-dependent, and linear in the range of concentrations tested (Fig. 6). Preincubation of LGP32 with RGD tripeptide resulted in a significant loss of binding by more than twofold (P < 0.01), but an EGF tripeptide (control) had no significant effect (Fig. 6). Taken together, these data show that the RGD sequence of Cg-EcSOD is essential for the OmpU-mediated binding of Cg-EcSOD to LGP32.

This finding prompted us to investigate the role of Cg-EcSOD in the plasma opsonisation previously observed. GFP-expressing wild-type or ΔompU LGP32 were treated with purified Cg-EcSOD before being exposed to washed hemocytes, unable to be invaded by LGP32. This process fully restored the ability of LGP32 to invade oyster hemocytes, as evidenced by both flow cytometry (Fig. 4A) and confocal microscopy (Fig. S1). Taken together, these results identify Cg-EcSOD as an oyster opsonin recognizing OmpU of LGP32.

Expression of Cell Remodeling Genes Is OmpU-Dependent in LGP32 Experimentally Infected Oysters.

To determine the role of OmpU on hemocyte responses in vivo, oysters were infected with 10⁸ cfu of wild-type or ΔompU LGP32, or sterile sea water (SSW) (control). Expression of oyster hemocyte genes involved in actin dynamics and cell motility was monitored 24 h postinjection. Among them, β-actin is involved in cell motility, structure, and integrity through microfilament building. Intersectin-1 is a cytoplasmic membrane-associated protein that coordinates endocytic membrane traffic with the actin assembly machinery, and regulates the formation of clathrin-coated vesicles (32). Fascin is an actin cross-linking protein involved in the assembly of actin filaments into bundles (33). Finally, calreticulin is a multifunctional protein that serves as a cell-surface receptor for the C1q component of complement and mediates phagocytosis upon binding to major plasma proteins (34, 35). Although the number of actin transcripts was similar in Vibrio- and SSW-injected oysters, all other three genes had a significantly higher number of transcripts in LGP32-injected oysters (P < 0.1) (Fig. 7). Interestingly, ΔompU-injected oysters did not differ significantly from the SSW-injected controls (Fig. 7). That this lack of response is caused by OmpU-expression is indicated by the statistical difference observed for intersectin-1 and fascin between LGP32- and ΔompU-injected oysters (P < 0.1). Although calreticulin responded similarly, the response to an injection by the wild-type or the ΔompU mutant was not statistically different (P = 0.16) (Fig. 7). Taken together, these data show that OmpU activates the expression of genes involved in cell remodeling. Conversely, the expression of Cg-EcSOD and β-integrin, which mediate recognition, was not statistically different between LGP32- and ΔompU-injected oysters (Fig S4).

Bacterial Strains.

Vibrio strains (Table 1) were grown at 20 °C in Zobell medium. When needed, 12.5 μg/mL chloramphenicol (Cm) was added. GFP-expressing derivatives of LGP32 and its ΔompU isogenic mutant were obtained by allelic exchange as described previously (6). The gfp gene under the PTRC promoter was integrated in a non essential transposase gene of LGP32 using the pSW3654T suicide vector (Table 1). Primers for pSW3654T construction were Tr32s-Pst1 and Tr32as-BamH1 (Table 1).

Animals and Hemolymph Collection.

Adult diploid C. gigas were purchased from local oyster farms in Mèze and Palavas-les-Flots (Gulf of Lion, France). Hemolymph was collected from the posterior adductor muscle sinus using a 2-mL syringe equipped with a 23-G needle. Cell-free hemolymph (plasma) was obtained by centrifugation (3,000 × g, 15 min, 4 °C) and filtration through a 0.22-μm pore-size filter.

Oyster Bacterial Challenge.

Experimental infections were performed at 20 °C, as previously described (6). Oysters were injected with 1 × 10⁸ cfu per animal of wild-type or ΔompU LGP32. Control animals were injected with SSW. For every condition, oysters were placed for 24 h in three separate in 50-L tanks of sea water. For gene-expression analysis, hemolymph was collected from 30 oysters per experimental conditions (10 oysters per tank). Hemocytes were collected from pools of 10 hemolymphs (one pool per tank) by centrifugation (15 min, 1,000 × g, 4 °C) and directly processed for RNA extraction. For confocal microscopy, hemolymph was collected 1 h after injection of GFP-expressing LGP32 (1 × 10⁸ cfu).

Hemocyte Invasion Assays.

GFP-labeled LGP32 in stationary phase of growth and oyster hemolymph or hemocytes were incubated in a 50:1 ratio. When needed, a 1-h preincubation was performed with either plasma, purified Cg-EcSOD, or various inhibitors, including sucrose (Fluka), cytochalasin D (Sigma), 0.4 mg/mL of RGD or EGF tripeptides (Sigma), or 8 μg/mL of anti–β-integrin antibody (generous gift from Marie-Christine Lebart, University of Montpellier 2, Montpellier, France). After fixation with 3.7% formaldehyde, hemocytes were washed twice with SSW and counter stained with Evan's blue (Sigma), Alternatively, hemocytes were stained with phalloidin-tetramethylrhodamine B isothiocyanate and DAPI (Sigma). Hemocyte invasion was monitored after a 30-min contact by confocal microscopy, and after 2 h by flow cytometry.

Confocal Microscopy.

Images were acquired on a Leica SPE confocal laser scanning system connected to a Leica DM 2500 upright microscope. Lasers were used at λex 488 nm for GFP (λem 505–530 nm), λex 532 nm for rhodamine-phalloidin (λem 560–630 nm), λex 405 nm for DAPI (λem 420–480 nm), and λex 405 nm for Evan's blue (λem 670–750 nm). A 40× ACS APO 1.15 oil Leica objective was used. Images were collected sequentially to avoid cross-contamination between fluorochromes and scanned at a 1,024 × 1,024 pixel resolution. Series of optical sections were collected.

Flow Cytometry.

The green fluorescence of LGP32 in hemocytes was monitored in a FACscan apparatus using a 488-nm argon-ion laser. Hemocyte invasion was monitored using at least a total of 30,000 events. Data were analyzed with the Cell Quest software.

Viability of Intracellular Bacteria.

Monolayers of 2.5 × 10⁵ hemocytes were exposed to LGP32 or LMG20012^T in a 50:1 ratio. After 1, 4, and 16 h, extracellular bacteria were eliminated with trypsin-EDTA 0.02% and hemocytes were lysed in 0.05% Triton X-100 before plating on Zobell medium for cfu counting.

Quantification of Hemocyte Responses.

Hemocyte suspensions exposed for 1h to LGP32 or LMG20012^T in a 50:1 ratio were treated with 0.1 mM green-lysotracker (Invitrogen) or 1 mM DCFH-DA (Sigma). Fluorescence indicative of acidic vacuole formation and ROS production was monitored on a Tecan microplate reader at λex 504/λem 511 and λex 492/λem 530, respectively. Unchallenged hemocytes were used as controls.

Plasma Fractionation.

Pooled plasma from 10 oysters was acidified to pH 3.8 with 2 M HCl. Proteins (2 mg) were separated on a UP5NEC-25QS HPLC column (Interchim) equilibrated in 0.05% trifluoroacetic acid (TFA). Fractionation was performed with a linear gradient of 0 to 70% acetonitrile (ACN) in TFA 0.05% over 35 min at a flow rate of 0.7 mL/min. Absorbance was monitored at 225 nm. Fractions were lyophilized and reconstituted in PBS before SDS/PAGE analysis or binding assays.

LC-MS/MS Analysis.

Protein bands were excised from SDS/PAGE, dehydrated, and incubated with 0.15 μg trypsin (Promega) in 25 mM NH₄HCO₃ for 16 h at 37 °C. Peptides were extracted from gel pieces on a robot (EVO150; Tecan) with three sequential 30 μL-extractions in 50% ACN, 5% formic acid, and 100% ACN. Peptide extracts were pooled and dried under vacuum before LC-MS/MS analysis on an Ultimate 3000 HPLC (Dionex) coupled with a LTQ-ORBITRAP discovery (Thermo Fisher Scientific) mass spectrometer. The MS-MS sequenced peptides were analyzed by the Mascot search engine for protein identification, against the 29,745 unique ESTs available for C. gigas (http://public-contigbrowser.sigenae.org:9090/Crassostrea_gigas/index.html) and the NCBI database (http://www.ncbi.nlm.nih.gov).

Plasma- and Cg-EcSOD-Binding Assay.

Plasma, HPLC fractions, or purified Cg-EcSOD were coated on a microplate for 16 h at 4 °C. Bound proteins were then overlaid for 1 h at room temperature with GFP-expressing bacteria in stationary phase of growth (10⁸ per well). In specific assays, bacteria were preincubated for 30 min with 400 μg/mL RGD or EGF tripeptides (Sigma) before overlay. Unbound bacteria were washed off with SSW before GFP reading (λex 395 nm/λem 509 nm) on a Tecan Infinite200 microplate reader.

RNA Isolation and qPCR Analysis of Oyster Gene Expression.

RNAs were extracted in TRIzol (Invitrogen) from oyster hemocytes (three pools of 10 oysters per condition). Complementary DNA was synthesized with M-MLV reverse transcriptase (Invitrogen) and diluted for qPCR reactions with LightCycler 480 master mix (Roche). Primers used for intersectin-1 (GenBank CU995003), fascin (GenBank FP007130), calreticulin (GenBank CU993006), and β-actin (GenBank EW779066) are listed in Table 1. Quantitative PCR were carried out in triplicates on the LightCycler 480 System (Roche). Relative expression was calculated using the 2^-ΔΔCq method (47), with normalization to the C. gigas RPL40 (GenBank FP004478). Statistics used the nonparametric ANOVA-Kruskal-Wallis test (STATISTICA software).