Edwardsiella piscicida employs C3 derivative to mediate internalization and infection via the teleost-specific receptor

ABSTRACT

Edwardsiella piscicida is a prominent fish pathogen, and spreads the infection to humans and mammals. It can resist serum killing and invade the host phagocytic cells. However, the precise mechanisms underlying E. piscicida infection remain unclear. In this study, we found that E. piscicida could bind to complement factor H and recruit complement factor I from Paralichthys olivaceus (Po) serum, leading to the degradation of complement recognition molecule PoC3b to PoC3dg on the bacterial surface. E. piscicida also bound to peripheral blood leukocytes, depending on the teleost-specific interaction between PoC3dg and the integrin beta VWA (INB) domain of PoCD18, whereby facilitating bacterial internalization and infection. When PoCD18 was knocked down or blocked, bacterial infection and tissue lesions were significantly decreased. Mechanistically, the intracellular domain of PoCD18 subsequently interacted with PoCytohesin and triggered the phosphorylation of the phosphatidylinositol 3-kinase (PI3K)/serine-threonine protein kinase (Akt) signalling pathway. Furthermore, inhibition of the PI3K/Akt pathway impaired the internalization and pathogenicity of E. piscicida. Overall, we demonstrated for the first time that E. piscicida utilized the C3dg-CD18 axis to facilitate internalization, thereby targeting immune evasion and promoting systemic infections. These results provide new insights into the pathogen–host interaction mechanism and a potential target to control edwardsiellosis.

Introduction

Edwardsiella piscicida, formerly known as Edwardsiella tarda, is a gram-negative pathogen belonging to the Enterobacteriaceae family [1,2]. Its infection can lead to the development of edwardsiellosis in a diverse range of hosts, including fish, amphibians, mammals, and humans [2]. Edwardsiella can be transmitted to humans and other animals through food and waterborne infections, resulting in gastroenteritis in patients with other underlying diseases [3,4]. Among the Edwardsiella species, E. piscicida can infect various fish, such as flounder, turbot, tongue sole, and tilapia [4,5]. E. piscicida infection leads to significant economic losses in the global aquaculture industry and raises potential risks to human health [4,5].

The virulence of E. piscicida is characterized by its ability to invade and replicate within leukocytes, epithelial cells, or macrophages, as well as evade host serum killing [4,5]. E. piscicida attaches and replicates in leukocytes and macrophages depending on diverse virulence factors, such as Type III secretion system (T3SS), type VI secretion system (T6SS), adhesins, haemolysin EthA, thioredoxin, iron uptake regulators, translocation and assembly module (TAM), flagellin, and outer membrane proteins [4,6–10]. Additionally, virulence factors of E. piscicida, such as TraT, Eta1, Inv1, Sip1, Sip2, and MliC, play important roles on resistance serum killing [4,11–13]. E. piscicida can invade the mouse macrophage cell line RAW264.7 via phagocytosis of host cells [14]. However, the pivotal mechanism underlying E. piscicida entry into phagocytes during systemic infection remains unclear.

The complement system plays a crucial role in host defence against pathogens and damaged cells, and different recognition molecules initiate three distinct pathways: the classical, lectin, and alternative pathways [15]. The classical and lectin pathways are generally activated via antibody or lectin [16]. When pathogens invade the host for the first time, C3b recognizes the pathogen and initiates the alternative pathway to generate C3bBb (C3 convertase) for cleavage of C3 [15–17]. The combination of C3b and C3 convertases forms C5 convertase, leading to the cleavage of C5 into C5a and C5b [15,18]. Ultimately, this cascade results in the formation of membrane attack complex (MAC) that exhibits lytic activity against pathogens [18,19]. To avoid complement activation to damage host cells, complement system exists several inhibitors to regulate complement excessive activation [20]. CFH, a soluble inhibitor, is a cofactor that regulates CFI to cleave C3b, which generates C3c and C3dg, thus preventing C3 convertase formation and complement activation [21]. In teleosts, most complement components structurally resemble those in mammals, and the complement system comprises all three activation pathways [22]. In Cyprinus carp, C3d fragments are identified and a possible CR2-like C3d receptor is detected in carp lymphocytes [23]. However, the specific functions of C3dg in teleosts remain unclear.

In mammals, cleaved C3 fragments undergo opsonization of the pathogen and can be recognized by complement receptor (CR) [24]. CRs are divided into three categories according to different structural characteristics: CR1 and CR2 (containing the CCP domain), CR3 and CR4 (integrin family), and CRIg (immunoglobulin superfamily) [19,25]. CR1 (CD35) is expressed in T and B cells and serves as a receptor for C3b and C4b, which regulate complement activation [25]. CR2 (CD21) binds to the terminal degradation fragments C3dg and C3d of C3b, thereby providing a key interface between innate and adaptive immunity [26]. The integrin families CR3 (CD11b/CD18) and CR4 (CD11c/CD18) are heterodimeric transmembrane complexes that are expressed on the surface of phagocytes to induce an immune response [27]. Both the receptors CR3 and CR4 bind to iC3b, recognizing different sites of iC3b [27]. The immunoglobulin superfamily receptor CRIg is expressed on macrophages and Kupffer cells to facilitate the binding to C3b and iC3b [27,28].

Previous studies have reported that E. piscicida evades serum killing by preventing the complement of alternative pathway activation, and can attach and invade host cells [11,13,29]. However, the mechanism of E. piscicida in systematic dissemination remains to be investigated. In this study, by using a Paralichthys olivaceus (Po) infection model, we found that serum treatment significantly improved the pathogenicity of E. piscicida. Based on these observations, we investigated the effects of serum on E. piscicida infection. We have elucidated the mechanism by which E. piscicida evades serum killing and invades the host cells. We have discovered that PoC3dg, a PoC3 degradation fragment, plays an important role in the progression of E. piscicida infection, and E. piscicida attaches to cells via a unique interaction between PoC3dg and PoCD18 in teleosts. Our results indicate that PoC3dg initiates the phosphatidylinositol 3-kinase (PI3K)/serine-threonine protein kinase (Akt) pathway in phagocytes and enhances bacterial internalization. These findings provide insights into the immune evasion mechanisms of E. piscicida.

Materials and methods

Ethics statement

In this study, the Ethics Committee of Tianjin Normal University approved every animal experiment, which was conducted in accordance with specific regulations and guidelines.

Animals

Clinically healthy P. olivaceus was purchased from a local fish farm. Fish were fed in the laboratory as reported previously [9]. Before the experiment, the presence of bacteria in tissues (spleen, kidney, liver, blood, and intestine) of P. olivaceus were examined randomly to ensure the health and requirement of experiment fish as reported previously [9]. Fish were euthanized for tissue collection experiments using an overdose of tricaine methane sulphonate (Sigma, St. Louis, MO, USA).

Cell culture and transfection

HEK293T cells (American Type Culture Collection) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (Corning, NY, USA) supplemented with 10% (v/v) foetal bovine serum (Gibco, Renfrewshire, UK), penicillin (100 U/mL), and streptomycin (0.1 mg/mL) at 37°C with 5% CO₂. E. piscicida was cultured in Luria – Bertani broth (LB) medium at 28°C [29]. The Escherichia coli strains (TransGen Biotech., Beijing, China) were cultured in LB medium at 37°C. Lipofectamine 3000 Transfection Reagent (Invitrogen, Carlsbad, CA, USA) was used to transiently transfect the plasmids into HEK293T cells. For gene overexpression, pmCherry (Clontech, Mountain View, CA, USA), pEGFP (Clontech, Mountain View, CA, USA), pCAGGS-flag-N (MiaoLing Plasmid Platform, Wuhan, China), pCAGG-HA (Solarbio, Beijing, China) expressing PoCD18 variants, and PoCytohesin were transfected into HEK293T cells as above. The primers are listed in Supplementary Table S1.

Sequence and evolution analysis

Sequence alignment was performed with DNAMAN and BLAST program at NCBI. Domain prediction was performed with the SMART and conserved domain search program at NCBI. The phylogenetic tree was constructed with the MEGA 6.0. The crystal structural and interaction simulation was generated by the ZDOCK server, Alphafold2, or SWISS-MODEL. Subcellular localization of protein was predicted by CELLO. The sequences are listed in Supplementary Table S2 and Supplementary Table S3.

Purification of recombinant proteins and preparation of antibodies

The coding sequences of PoC3dg (residues 948–1292 from PoC3), PoC3dgM (lacking residues 1201–1261 based on PoC3dg), PoCFB (residues 484–782), PoCFH (residues 267–480), and the integrin beta VWA (INB) domain of PoCD18 (residues 32–454) were amplified by PCR from the cDNA of P. olivaceus. The sequence was then cloned into pET259 [30]. E. coli BL21 (DE3) was transformed with the corresponding expression plasmid or pET32a (expressing the Trx tag). Transformants were cultured in LB medium at 37°C to mid-log phase. The expression of recombinant proteins was induced by adding IPTG (Solarbio, Beijing, China) at 16°C for 12 h. After induction, recombinant proteins were prepared and purified as reported previously [30]. Protein concentrations were measured by NanoPhotometer (Implen GmbH, Munich, Germany). Mouse antibodies against PoC3dg, PoCFB, PoCFH, or Trx were prepared and purified using rProtein G Beads (Solarbio, Beijing, China) as reported previously [13].

Immunoblotting

E. piscicida and E. coli were cultured to OD₆₀₀ of 0.8 and resuspended to 2 × 10⁹ CFU/mL in PBS. Cells were treated with or without P. olivaceus serum at 22°C for 1 h. Cells were then washed with PBS, and whole-cell proteins were extracted and subjected to immunoblotting as reported previously [13] with antibodies against PoC3 [31], PoC5 [32], PoCFB, PoCFH, and PoCFI [31]. RNA polymerase β (RNAP) was used as a loading control [10] with anti-RNAP beta antibody (Abcam, Cambridge, MA, USA).

Depleted serum preparation

P. olivaceus serum was incubated with anti-PoC3 antibody, anti-PoCFH antibody, or anti-PoCFI antibody at 4°C for 4 h. Serum was transferred to the spin column bound with protein-A/G and incubated at 4°C for 4 h. After incubation, the elution was collected by centrifugation as PoC3-, PoCFH-, or PoCFI-depleted serum.

Phagocytosis assay

The caudal veins of P. olivaceus (~1000 g) were used for blood collection. Peripheral blood leukocytes (PBL) were extracted using the Percoll gradient method, as described previously [33]. PBL were cultured in L–15 medium (Solarbio, Beijing, China) at 22°C. E. piscicida was cultured as above and resuspended in 100 mg/mL FITC (Solarbio, Beijing, China) dissolved in dimethyl sulphoxide. After incubation, bacteria were washed with PBS and resuspended to 10⁸ CFU/mL. FITC-labelled E. piscicida (10⁷) was treated with or without serum, 56°C heat-inactivated (HI) serum, PoC3-, PoCFH-, or PoCFI-depleted serum at 22°C for 1 h, and added to 1 mL PBL (~10⁶ cells) and incubated in the dark at 22°C for 1 h. After incubation, cells were washed and treated with trypan blue to quench extracellular fluorescence as described previously [34]. Flow cytometry analysis was performed by Flow Cytometer (Millipore, Massachusetts, USA). Data were analysed with IDEAS Application software (Millipore, Massachusetts, USA).

Microscopy

For TEM, E. piscicida was cultured and resuspended to 10⁸ CFU/mL as above. E. piscicida was incubated with or without P. olivaceus serum at 22°C for 1 h. After incubation, bacteria were observed using a TEM (HT7700, Hitachi, Japan). To determine the interaction between rPoC3dg and PBL, cells were prepared as above and resuspended to 10⁸ cells/mL. The cell suspension was transferred onto a glass slide (Citotest, Jiangsu, China) to achieve cellular adherence. The slide was added with or without rPoC3dg (40 µg/mL) or rTrx (40 µg/mL) containing 10 µg/mL CaCl₂ at 22°C for 1 h. After incubation, 4% paraformaldehyde was added to fix cells at 22°C for 0.5 h. Then, the slide was blocked with 6% skim milk powder at 22°C for 1 h. Cells were incubated with anti-His tag mouse monoclonal antibody (Abcam, Cambridge, UK) and FITC-labelled goat anti-mouse IgG (Abcam, Cambridge, UK) in the dark at 37°C for 1 h. Cells were stained with DAPI and observed under a confocal microscope (Nikon, Tokyo, Japan). To determine the interaction between rPoC3dg and PoCD18 or its mutants, the cells were seeded into 35 mm glass-bottom culture dishes (Nest Scientific, Rahway, NJ, USA) (1 × 10⁶ cells) and transfected as above to express PoCD18 or PoCD18 mutants. The cells were incubated with E. piscicida (10⁷ CFU) containing 40 µg/mL rPoC3dg and 10 µg/mL CaCl₂ at 22°C for 1 h. Cells were immobilized and blocked as above. Antibody against His-tag targeting rPoC3dg and FITC-labelled second antibody were added to cells as above. The rPoC3dg treated-bacteria bound cell was observed as above. Furthermore, the fluorescence intensity was measured with Multi-Mode Microplate Reader (Tecan, Mannedorf, Switzerland).

Cell death and cytotoxicity assay

To investigate cell death, E. piscicida (10⁷ CFU) containing 40 µg/mL rPoC3dg, 40 µg/mL rPoC3dgM, 40 µg/mL rTrx, or PBS (control) was added to PBL (~10⁶ cells) prepared as above at 22°C for 3 h. Subsequently, PBL were stained with propidium iodide (PI) (Invitrogen, Carlsbad, CA, USA) for 10 min and observed with confocal microscopy as above. Lactate dehydrogenase (LDH) release in cell culture supernatant was detected using a CytoTox96 LDH release kit (Promega, Leiden, Netherlands) according to the protocol as described previously [35].

In vitro infection

To determine the effect of rPoC3dg on bacterial infection, PBL (10⁵ CFU/well) were cultured in 96-well plates at 22°C as described previously [34]. E. piscicida (10⁶ CFU) was mixed with or without rPoC3dg, rPoC3dgM, or rTrx (40 µg/mL) containing 10 µg/mL CaCl₂ at 22°C for 1 h. Subsequently, the mixture was added to each well of PBL plates and incubated at 28°C for 1, 2, or 4 h. After incubation, the cells were washed and lysed with 100 µL PBS containing 1% Triton X-100. The cell lysate was diluted and plated on LB agar plates as described previously [34]. For intracellular bacteria, fresh L-15 medium containing 100 µg/mL gentamicin (Solarbio, Beijing, China) was added to the cells and incubated at 28°C for 1 h to kill the extracellular bacteria. After incubation, bacterial recovery was detected as above.

In vivo infection

E. piscicida was cultured and incubated with rPoC3dg (40 µg/ml), rTrx (40 µg/ml), or PBS as above. After incubation, P. olivaceus were randomly divided into three groups and intramuscularly injected with E. piscicida (2 × 10⁶ CFU/fish). At 12, 24, and 48 h post infection (hpi), the kidney, spleen, and blood were collected from the fish (five fish/time points). Bacterial recovery was determined as above report [33]. To determine the mortality of fish, three groups (20 fish/group) of P. olivaceus were infected as above, and the mortality of fish was recorded daily for 10 days. For histopathological analysis, tissues were collected from fish at 24 hpi and fixed with 4% paraformaldehyde. Tissue sectioning and HE staining were performed by Zuocheng Bio. Co., Ltd. (Shanghai, China). Tissue sections were treated and observed as described previously [13].

Immunoprecipitation

To investigate the interaction between rPoC3dg and PoCD18, HEK293T cells were cultured in 24-well plates (2 × 10⁵ cells/well) and transfected as above to express Flag-tagged PoCD18. After 24 hours post-transfection, cells were lysed with RIPA lysis buffer (Solarbio, Beijing, China). rPoC3dg (20, 40, or 80 µg/mL), rTrx (80 µg/mL), or PBS were incubated with cell lysates containing 10 µg/mL CaCl₂ at 22°C for 1 h and then treated with anti-His antibody at 4°C for overnight. Immunoprecipitation was performed using the immunoprecipitation kit (Sangon Biotech, Shanghai, China) according to the protocol. The immunoprecipitated complex was subjected to immunoblotting as above with anti-His antibody or anti-Flag antibody.

qRT-PCR

For qRT-PCR analysis of gene expression under normal physiological conditions, the liver, intestine, kidney, spleen, blood, brain, heart, gill, and muscle were obtained aseptically from five P. olivaceus (13.6 ± 1.2 g), and total RNA was extracted with total RNA extraction reagent (Vazyme, Nanjing, China). The cDNA was prepared with the cDNA synthesis kit (Toyobo, Osaka, Japan). qRT-PCR was performed with ChamQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China) by 7500 real-time PCR System (Applied Biosystems, California, USA). Gene expression was detected with primers, and β-actin was used as internal reference gene (Supplementary Table S1). The data was analysed as the previous report [32].

For qRT-PCR analysis of gene expression during bacterial infection, E. piscicida was cultured at 28 °C to an OD₆₀₀ of 0.8. P. olivaceus (13.6 ± 1.2 g) were randomly divided into three groups (n = 20) and intramuscularly injected with or without (control) 100 µL E. piscicida. Blood, spleen, and kidney were collected from five fish at 6, 12, 24, and 48 hpi, respectively. The gene expression was evaluated by qRT-PCR as above.

Gene knockdown

Gene knockdown was achieved through DNA vector-based small interfering RNA (siRNA) technology as reported previously [36]. Briefly, to select gene-specific siRNA, three different siRNA targeting gene were inserted into the siRNA expression vector pRNAT-CMV3.1 (GenScript, Piscataway, USA) at the BamHI/AlfII site, which was contained a T7 promoter and used as templates to express siRNA. Additionally, the control plasmid, which expresses a scramble siRNA, was similarly constructed. To determine the efficiency of the siRNA plasmids, five groups (N = 5) of P. olivaceus as above were injected intramuscularly with each plasmid (10 µg/fish) or PBS. At 7 days post-plasmid administration, the expression of the gene in the spleen, kidney, and blood was evaluated by qRT-PCR as described above. The plasmid with the strongest inhibitory effect on gene expression was used. The primers are listed in the Supplementary Table S1.

Statistics and reproducibility

The experiments were conducted in triplicate. Statistical analysis was performed using the SPSS software (version 17.0; SPSS Inc., Chicago, IL, USA). Data were subjected to analysis of variance (ANOVA), and the statistical significance was defined as p < 0.05.

Results

C3 derivative in serum improves the ability of E. piscicida into host cells

When treated with P. olivaceus serum, E. piscicida showed a high survival rate (87.6%) [29]. However, the control non-pathogenic species, E. coli, showed no survival in serum [29]. Transmission microscopy revealed that E. piscicida maintained its intact cellular structure after serum treatment; however, serum treatment severely damaged E. coli (Figure 1a). Following incubation with serum, PoC3, PoC3b, and PoC3dg were observed on the surface of E. piscicida and E. coli (Figure 1b). PoC5 and PoBb (the cleaved PoCFB fragment of ~ 80 kDa) were detected on E. coli, but not on E. piscicida (Figure 1c,d). PoCFH and PoCFI heavy chain were detected abundantly on E. piscicida, but scantily on E. coli (Figure 1e,f,g). Consistently, the amount of C5a in E. piscicida-treated serum was significantly lower than that in E. coli-treated serum (Figure 1h). However, following incubation with human serum, C3 fragments and CFI heavy chain were not detected on the surface of E. piscicida, indicating that E. piscicida could resist human serum killing via a different mechanism (Supplementary Figure S1). Phagocytosis analysis indicated that PBL displayed stronger phagocytosis in serum treated-E. piscicida compared with E. piscicida alone (45.2% vs 17.8%) (Figure 1i). Similarly, PBL significantly increased the phagocytic rate for E. coli in the presence of serum (Suplementary Figure S2a, b). Compared to normal serum treatment, PoC3-, PoCFH-, or PoCFI-depleted serum treatment was significantly reduced phagocytosis of PBL for E. piscicida (Figure 1i). In vitro infection indicated that, compared to HI serum treatment, normal serum treatment was significantly increased the attachment and invasion of E. piscicida to PBL (Figure 1j,k). Furthermore, treatment with PoC3-depleted serum significantly impaired the effect of normal serum (Figure 1j,k). However, E. coli from serum treatment did not attach to cells (Supplementary Figure S2c), due to the bactericidal effect of serum.

Figure 1.

The effect of serum treatment on Edwardsiella piscicida. (a) Microscopic examination the effect of serum. E. piscicida and Escherichia coli were treated with or without (control) Paralichthys olivaceus serum for 1 h, and then subjected to transmission microscopy. Bar, 1 μm. (b–f) Interactions of E. piscicida with serum complement components. Bacteria was treated with serum as above, and western blot was performed to detect bacteria-bound PoC3 (b), PoC5 (c), PoCFB (d), PoCFH (e), and PoCFI (f) in serum. RNAP was used as a loading control. (g) The serum-treated bacteria were incubated with the antibody against PoC3, PoC5, PoCFB, PoCFH, or PoCFI, and then treated with FITC-labelled secondary antibody. The amount of the bacteria-bound antibody was determined by measuring fluorescence intensity. (h) E. piscicida and E. coli were incubated with serum, and C5a concentration in the serum was determined. (i) PBL from P. olivaceus were incubated with FITC-labelled E. piscicida in the presence or absence (control) of serum, heat-inactivated (HI) serum, PoC3-depleted serum, PoCFH-depleted serum, or PoCFI-depleted serum. Phagocytosis of cells was analysed by flow cytometry. FITC fluorescence intensity was performed by using IDEAS software. (j, k) E. piscicida was treated with or without (control) serum, PoC3-depleted serum, or HI serum for 1 h. PBL were infected with E. piscicida from different treatment at different time points, and the bacterial recovered from attachment to cells (j) and invasion into cells (k) were detected. Data are the means of three experiments and shown as means ± SEM. In panel g, **P < 0.01. For panels h, i, j, and k, different letters indicate statistical significance among the samples, P < 0.05.

PoC3dg exhibits positive effects on the infectivity of E. piscicida

The deduced PoC3dg protein is composed of 345 amino acid residues, with a predicted molecular mass of 38 kDa and a predicted isoelectric point (pI) of 7.1. It possessed a receptor-binding region (Figure 2a). In humans, C3dg is produced via the cleavage the Arg₉₃₂-Glu₉₃₃ and Arg₁₂₈₁-Ser₁₂₈₂ sites of C3b by CFI, which subsequently binds to the receptor-binding region of CR2 on the cellular surface [26,37]. The predicted CFI and cofactor cleavage sites of PoC3 were identified at Gly₉₄₇-Glu₉₄₈ and Arg₁₂₉₂-Ser₁₂₉, resulting in the production of PoC3dg (Figure 2a). To investigate the effect of PoC3dg on E. piscicida, recombinant PoC3dg (rPoC3dg) or rPoC3dgM (a mutant lacking the receptor-binding region) was purified from E. coli using a His-tag (Supplementary Figure S3a). Recombinant thioredoxin (rTrx) was used as a control protein and purified under the same conditions (Supplementary Figure S3a). ELISA and western blot assay revealed that rPoC3dg and rPoC3dgM exhibited the distinct concentration-dependent binding to E. piscicida (Figure 2b,c). Following the incubation of proteins with E. piscicida or PBL, immunofluorescence microscopy revealed that rPoC3dg or rPoC3dgM was found on the surface of E. piscicida (Figure 2d). However, the PBL surface only showed the presence of rPoC3dg (Figure 2e), suggesting that rPoC3dg bound to PBL, but not rPoC3dgM. We further determined whether rPoC3dg could affect the pathogenicity of E. piscicida. The results revealed that the presence of rPoC3dg induced significant cell death and increased the production of LDH, whereas the presence of rPoC3dgM or rTrx had no apparent effect (Figure 2f,g,h). When PBL were infected with E. coli, no cell death was observed, and low LDH levels were found in the PBL (Supplementary Figure S4a, b). While the presence of rPoC3dg significantly enhanced the attachment of E. coli to PBL (Supplementary Figure S4c). Thus, when bacteria were incubated with PBL, the survival rate of E. piscicida was no significant change, whereas the survival rate of E. coli was significantly decreased (Supplementary Figure S5).

Figure 2.

The binding of PoC3dg to Edwardsiella piscicida and peripheral blood leukocytes (PBL). (a) Diagram of predicted CFI and cofactor cleavage sites for PoC3. Binding region of PoC3dg to receptor is marked in red box. (b, c) Binding of rPoC3dg to E. piscicida. E. piscicida was incubated with or without (control) different concentrations of rPoC3dg or rPoC3dgM, and the binding index of E. piscicida and protein was detected by ELISA or western blot with HRP-labelled anti-His antibody. RNAP was used as a loading control. (d) E. piscicida was incubated with rPoC3dg, rPoC3dgM, rTrx, or PBS (control), and treated with FITC-labelled antibody against His-tag and stained with DAPI. Bacteria were observed by confocal microscopy. Bar, 10 μm. (e) PBL from Paralichthys olivaceus were incubated with protein as above, and cells were treated and observed. Bar, 10 μm. (f-h) The effect of rPoC3dg on E. piscicida inducing cell death. E. piscicida was treated with rPoC3dg, rPoC3dgM, rTrx, or PBS (control), and PBL were infected with E. piscicida from different treatment for 1 h. The cells were stained with PI and subjected to microscopy (f), counted the PI-positive cells (g), or determined the LDH release (h). For panel f, Bar, 20 μm. For panels b, g, and h, data were obtained from three independent experimental and presented as means ± SEM. *P < 0.05, **P < 0.01.

In vitro analysis revealed that the rPoC3dg was significantly enhanced the attachment and invasion of E. piscicida to PBL (Figure 3a,b). However, E. piscicida treated with rPoC3dgM or rTrx showed comparable levels to those of control bacteria (Figure 3a,b). Flow cytometry analysis revealed that rPoC3dg significantly increased the phagocytic activity of PBL towards E. piscicida (Figure 3c). In addition, the presence of rPoC3dg significantly enhanced the capacity of E. piscicida to invade cells (Figure 3d), suggesting PoC3dg promoted E. piscicida internalization by binding to the phagocyte surface. To determine the effect of PoC3dg on the pathways of E. piscicida entry into PBL, rPoC3dg-treated E. piscicida infection was performed in the presence of inhibitors against clathrin (chlorpromazine), caveolin (nystatin), and macropinocytosis (rottlerin). Subsequent analysis revealed that, in the presence of chlorpromazine, the internalization ratio of rPoC3dg-treated E. piscicida was significantly decreased, whereas the presence of nystatin and rottlerin had no significant effect (Figure 3e). Consistent with the above results, in vitro analysis revealed that the presence of chlorpromazine severely decreased bacterial internalization into PBL (Figure 3f). In the presence of chlorpromazine, rPoC3dg-treated E. coli significantly decreased the phagocytic rate of PBL (Supplementary Figure S6), suggesting PoC3dg mediated the same endocytic pathway for E. piscicida and E. coli. Examination of immune gene expression showed that, compared to E. piscicida, rPoC3dg-treated E. piscicida significantly lowered the interleukin (IL)-1β, IL2, IL-6, IL-8, IL-12, IL-17, IL34, and tumour necrosis factor α (TNFα), and significantly higher induction of IL-10 in PBL (Figure 3g). When inoculated into P. olivaceus, rPoC3dg significantly decreased the survival rate of fish (Figure 3h) and increased tissue dissemination in the blood, spleen, and kidney of P. olivaceus following E. piscicida infection (Figure 3i).

Figure 3.

Effect of PoC3dg on Edwardsiella piscicida infection. (a, b) Effect of rPoC3dg on attachment and invasion. Peripheral blood leukocytes (PBL) were infected with E. piscicida in the presence or absence (control) of rPoC3dg, rPoC3dgM, or rTrx, and the bacterial recovered from attachment to cells (a) and invasion into cells (b) were detected. (c, d) PBL were incubated with FITC labelled E. piscicida for 1 h as above. Cells were detected the phagocytosis (c). The image of single cell was shown (d). Bar, 20 μm. (e, f) PBL were treated with chlorpromazine, nystatin, or rottlerin for 1 h, and cells were infected with FITC labelled E. piscicida in the presence of rPoC3dg. Control was infected cells without treatment. Phagocytosis of cells was analysed by flow cytometry (e), and the bacterial recovered from invasion to cells were detected (f). (g) PBL were infected with E. piscicida in the presence or absence (control) of rPoC3dg, rPoC3dgM, or rTrx for 1 h. Control was uninfected cells. The immune gene expression was detected by qRT-PCR. (h, i) In vivo effect of rPoC3dg on E. piscicida infection. Paralichthys olivaceus were infected with E. piscicida in the presence or absence (control) of rPoC3dg or rTrx. The survival rate of fish was recorded for 10 days (h), and bacterial numbers in the kidney, spleen, and blood of fish were determined (i). Data were obtained from three independent experimental and presented as means ± SEM. *P < 0.05, **P < 0.01.

PoC3dg binds to the INB domain of PoCD18

In humans, ligands bind to the VWA domain of CR3 (CD11b/CD18) α subunit [19]. To examine the mechanism underlying the binding of PoC3dg to the cellular receptor, PoCD18 and PoCD11b were overexpressed in HEK293T cells, and PoC3dg was immunoprecipitated with cell lysis. We found a direct interaction between rPoC3dg and PoCD18, whereas rPoC3dg exhibited no interaction with PoCD11b from P. olivaceus (Figure 4a). In addition, rPoC3dg interacted with PoCD18 in a concentration-dependent manner (Figure 4b). Further examine the location of PoCD18, we found PoCD18 expressed on the cellular surface (Supplementary Figure S7). The presence of Ca²⁺ increased the binding ability of rPoC3dg-treated E. piscicida to HEK293T cells expressing PoCD18 (Supplementary Figure S8). Sequence analysis revealed that PoCD18 possessed INB, integrin beta tail (tail), and integrin beta cytoplasmic (Cyt) domains (Figure 4c). Confocal microscopy observation showed that PoCD18 was located on the cellular surface of PBL (Supplementary Figure S9). To examine the functional role of domains, three mutants of PoCD18, i.e. PoCD18ΔINB, PoCD18Δtail, and PoCD18ΔCyt, were constructed, which bear a deletion of the INB (residues 32–454), tail (residues 455–621), and Cyt (residues 705–774) domains, respectively (Figure 4c). These mutants were expressed as Flag-tagged protein in HEK293T cells. Subsequent analysis showed that rPoC3dg exhibited no immunoprecipitation with PoCD18ΔINB (Figure 4d), suggesting that rPoC3dg interacted with PoCD18 depending on the INB domain. Surface plasmon resonance (SPR) analysis showed the specific binding of rPoC3dg to the INB domain of PoCD18, with a KD value of 0.486 μM (Supplementary Figure S10). Consistent with the above results, microscopic observations revealed that PoCD18ΔINB-expressing cells significantly affected the attachment of rPoC3dg-treated E. piscicida (Figure 4e,f). Consistent with the results of immunoprecipitation and microscopic observations, in vitro infection analysis revealed that rPoC3dg significantly decreased the attachment ability of E. piscicida to PoCD18ΔINB-expressing cells (Figure 4g). A 3D structural model was constructed to understand the molecular basis underlying the interaction between PoC3dg and PoCD18, and confidence scores derived from the AlphaFold2 structures of PoC3dg and PoCD18 were 87 and 91, respectively (Figure 4h). Combining the results shown in Figure 2, the interaction model predicted that PoC3dg could bind to the INB domain of PoCD18, depending on the Lys₂₅₄-Ala₃₁₄ domain in the extracellular region (Figure 4h).

Figure 4.

Interaction between PoC3dg and PoCD18. (a) The lysate of HEK293T cells expressing Flag-tagged PoCD18 and HA-tagged PoCD11b were incubated with or without His-tagged rPoC3dg, and then subjected to IP with anti-His antibody. The immunoprecipitation was analysed by western blot with the indicated antibody. Actin was used as a loading control. (b) The lysate of HEK293T cells expressing Flag-tagged PoCD18 was incubated with different concentrations of His-tagged rPoC3dg (20 to 80 µg/ml), rTrx (80 µg/ml), or PBS (control), and then subjected to IP as above. (c) Schematic presentation of the PoCD18 variants. The domain structures of PoCD18 and its mutants are schematically drawn. (d) The lysate of HEK293T cells expressing Flag-tagged PoCD18 or its mutants was incubated with His-tagged rPoC3dg, and then subjected to IP with anti-Flag antibody. (e, f) E. piscicida was pre-treated with rPoC3dg, and the bacteria were then used to infect HEK293T cells expressing mCherry-tagged PoCD18 or its mutants. After infection, the cells were treated with FITC-labelled antibody targeting rPoC3dg. The cells were then stained with DAPI and subjected to microscopy (e) or measurement of green fluorescence intensity (f). Bar, 20 μm. (g) The effect of PoCD18 overexpression on the cellular infection of E. piscicida. HEK293T cells expressing Flag tag alone, Flag-tagged PoCD18, or its mutants were infected with E. piscicida in the presence or absence (control) of rPoC3dg or rTrx, and the bacterial numbers from attachment to cells were determined. (h) Interaction simulation between PoC3dg and PoCD18. The 3D structures of PoC3dg and PoCD18 were predicted by Alpha fold 2. Interaction simulation was moulded by ZDOCK. The receptor-binding region of PoC3dg was marked with red. The sequences are listed in Supplementary Table S2 and Supplementary Table S3. For panels e and f, data were obtained from three independent experimental and presented as means ± SEM. **P < 0.01.

Sequence and evolution analysis of C3dg and CD18

To investigate the characteristics of C3dg and CD18 in the model species, sequence alignment and phylogenetic trees derived from the amino acid sequences were constructed to analyse evolutionary relationships (Figure 5a, Supplementary Figure S11). Sequence alignment showed that PoC3dg and PoCD18 shared 23.7–50.3% and 35.3–54.4% of the overall sequence identities, respectively, with the indicated species (Supplementary Figure S11). Phylogenetic analysis showed that PoC3dg and PoCD18 were clustered with homologs from teleosts, suggesting a close evolutionary distance among teleosts (Figure 5a). Additionally, PoC3dg or PoCD18 shared similar tertiary structures of C3dg or CD18 from the indicated species, respectively (Figure 5b), and possessed a conserved functional domain (Figure 5c). To determine the interactions between C3dg and CD18 among the model species, C3dg proteins were expressed and purified (Supplementary Figure S3b), and CD18 overexpressing vectors were constructed (Figure 5e). Subsequent binding analysis revealed that only C3dg from P. olivaceus (PoC3dg) and C3dg from D. rerio exhibited significantly binding ability to corresponding CD18 (Figure 5d,e).

Figure 5.

Evolutionary conservation of the interaction between C3dg and CD18. (a) Phylogenetic analysis of C3dg and CD18. The trees were constructed using the neighbour-joining (NJ) algorithm with the MEGA 6.0. Ticks at the base of the graph represent the values of the average branch length of each phylogenetic tree used. (b, c) The structures prediction (b) and domain analysis (c) of C3dg or CD18 from indicated species were performed by SWISS-MODEL and SMART. The accession numbers of selected sequences are listed in Supplementary Table S2 and Supplementary Table S3. For C3dg, thiol-ester was marked in green box. For CD18, INB domain was marked in light box, EGF-like domain was marked in dark green box, integrin tail was marked in grey box, transmembrane region was marked in blue box, and Cyt domain was marked in purple box. (d, e) HEK293T cells expressing Flag-tagged CD18 from indicated species were incubated with C3dg of corresponding species for 1 h. The cells were treated with anti-His antibody targeting C3dg, and then measured for fluorescence intensity. The normal HEK293T cells was used as control (d). Immunoprecipitation was analysed by western blot with indicated antibodies (e). For panel d, data was obtained from three independent experimental and presented as means ± SEM, different letters indicate statistical significance among the samples, P < 0.05.

PoCD18 as a host cell receptor is required for E. piscicida infection

qRT-PCR analysis revealed that constitutive PoCD18 expression occurred, in increasing order, in the muscle, gill, heart, brain, blood, spleen, kidney, intestine, and liver of P. olivaceus (Figure 6a). When the fish were infected with E. piscicida, significant inductions of PoCD18 expression were detected in the blood, spleen, and kidney (Figure 6b). To explore the role of PoCD18 associated with bacterial infection, PoCD18 expression in blood, spleen, and kidney of P. olivaceus was knocked down via RNA interference (RNAi) (Supplementary Figure S12a). When P. olivaceus was administered with pPoCD18si, rPoC3dg was significantly decreased the binding capability to PBL (Figure 6c,d). When PBL were infected with rPoC3dg-treated E. piscicida, the presence of anti-PoCD18 antibody was significantly decreased the bacterial attachment and invasion to cells (Figure 6e,f). However, the presence of a preimmune antibody or anti-rTrx antibody was unaffected the attachment and invasion of E. piscicida (Figure 6e,f). To determine whether PoCD18 knockdown affected bacterial infection, P. olivaceus exposed to pPoCD18si was infected with E. piscicida. The results showed that, compared with the control or pPoCD18siC-administered fish, the survival rates of pPoCD18si-administered fish were significantly higher (45%) than those of control fish (Figure 6g). Histopathological analysis showed that P. olivaceus was induced marked the reduction in spleen lymphocytes and vacuolation of intestinal epithelial cells and kidney connective tissue following E. piscicida infection, whereas pPoCD18si-administered fish was induced no apparent histological abnormality in various tissues (Figure 6h). In the line with above results, the bacterial replication in fish administered with pPoCD18si was significantly reduced compared to that in the control fish (Figure 6i).

Figure 6.

PoC3dg regulation Edwardsiella piscicida infection through PoCD18. (a) PoCD18 expression in the indicated tissues of Paralichthys olivaceus was determined by qRT-PCR. (b) PoCD18 expression in blood, spleen, and kidney was determined by qRT-PCR after E. piscicida infection. (c, d) Peripheral blood leukocytes (PBL) were collected from P. olivaceus with or without pPoCD18si treatment, and, respectively, incubated with rPoC3dg. The cells were treated with FITC-labelled antibody targeting rPoC3dg and stained with DAPI, and subjected to microscopy (c) or measurement of green fluorescence intensity (d). Bar, 20 μm. (e, f) PBL were pre-treated with or without (control) rPoCD18 antibody or preimmune antibody. Cells were infected with E. piscicida in the presence or absence of rPoC3dg or rTrx, and the bacterial recovered from attachment to cells (e) and invasion into cells (f) were detected. (g-i) P. olivaceus was administrated with or without (control) pPoCD18si or pPoCD18siC and infected with E. piscicida. The survival rate of fish was monitored daily for 10 days (g). Histopathological changes in the spleen, kidney, and intestine of the fish were observed at 24 h post infection (h). Bar, 100 μm. Bacterial numbers in the blood, spleen, and kidney were determined (i). Data were obtained from three independent experimental and presented as means ± SEM. *P < 0.05, **P < 0.01.

Interaction between PoC3dg and PoCD18 initiates the PI3K/Akt pathway by PoCytohesin

Since PoCD18 can activate downstream pathways via the adaptor protein cytohesin-1 in mammals [38], we have investigated the downstream signalling molecule of PoCD18 in P. olivaceus. PoCytohesin possesses Sec7 and pleckstrin homology (PH)-like domain (Supplementary Figure S13). When HEK293T cells were co-transfected with PoCD18 mutants and PoCytohesin-expressing vectors in the presence of rPoC3dg, PoCytohesin exhibited the immunoprecipitation with WT, ΔINB, and Δtail of PoCD18, whereas no apparent the interaction with PoCD18ΔCyt (Figure 7a). Subsequent study found that rPoC3dg was significantly increased the phosphorylation of PI3K/Akt following E. piscicida infection at 1 or 2 h (Figure 7b). When PoCytohesin was knocked down in P. olivaceus (Supplementary Figure S12b), the phosphorylation of PI3K/Akt was impaired during E. piscicida infection (Figure 7c). When PoCytohesin was knocked down, bacterial replication in the tissue of P. olivaceus was significantly decreased, compared to that from the control fish (Figure 7d). To elucidate the mechanism underlying the rPoC3dg-triggered pathway, we examined the phosphorylation levels of PI3K/Akt in the process of E. piscicida infection by using Pictilisib (GDC-094), a PI3K/Akt-specific inhibitor. Inhibitor significantly impaired rPoC3dg-triggered PI3K/Akt phosphorylation and bacterial internalization in PBL following E. piscicida infection (Figure 7e,f,g). Furthermore, PI3K/Akt inhibitor significantly decreased E. piscicida-induced tissue dissemination and fish death in P. olivaceus (Figure 7h,i).

Figure 7.

PI3K/Akt pathway involvement in PoC3dg mediation Edwardsiella piscicida infection. (a) PoCD18 interaction with PoCytohesin. HEK293T cells were co-transfected with the indicated Flag-tagged PoCD18 variants and HA-tagged PoCytohesin in the presence of rPoC3dg for 24 h, and IP was performed with anti-Flag antibody. The immunoprecipitation was analysed by western blot with the indicated antibody. Actin was used as a loading control. (b) Peripheral blood leukocytes (PBL) were infected with E. piscicida in the presence or absence of rPoC3dg or rTrx. The phosphorylation levels of PI3K and Akt were determined by western blot at 1 or 2 h post infection. (c) The phosphorylation levels of PI3K and Akt in the PBL from pPoCytohesinsi or pPoCytohesinsiC treated fish were determined at 2 h after E. piscicida infection. (d) Fish administrated with or without (control) pPoCytohesinsi or pPoCytohesinsiC was infected with E. piscicida, and bacterial numbers in the blood, spleen, and kidney were determined. (e-g) The phosphorylation levels of PI3K/Akt (e), bacterial numbers of invasion to cells (f), or phagocytic rate (g) in inhibitor-treated PBL were determined at 2 h after E. piscicida infection. (h, i) Fish were infected with E. piscicida in the presence or absence (control) of PI3K/Akt inhibitor. Bacterial numbers in the indicated tissues were determined (h), and the survival rate of fish was recorded for 10 days (i). Data were obtained from three independent experimental and presented as means ± SEM. **P < 0.01.

Discussion

In previous studies, E. piscicida exhibited a high survival rate in various serum samples, including P. olivaceus, Cynoglossus semilaevis, Scophthalmus maximus, and Mus musculus serum [11–13,29]. Microscopic studies revealed that the wild-type strain of E. piscicida maintained an intact bacterial structure following tongue sole or mouse serum treatment [11,13]. Similarly, transmission electron microscopy (TEM) observations in this study showed that E. piscicida maintained also kept structural integrity after P. olivaceus serum treatment. However, cellular lysis was observed in the serum-sensitized strain E. coli. In the alternative complement activation pathway, C3 and C3b can bind to bacteria through interaction with bacterial surface molecules [15,39]. Hence, the activation of CFB by C3b leads to the formation of C3 and C5 convertases, resulting in the generation of lytic MAC [16,40]. CFH regulates alternative pathway activation by competing with CFB for binding to C3b [41]. It recruits CFI to cleave C3b to iC3b, and further cleaves iC3b to C3c and C3dg [41]. In mammals, Yersinia pestis, Borrelia burgdorferi, Bacillus anthracis, and Streptococcus pneumonia can recruit complement negative regulators, such as CFH, CFH-related protein 1, or C4b-binding protein, to survive and disseminate in the host [42–44]. Previous study reports that E. piscicida-incubated serum still possessed strong complement activity, which suggests the possibility that the E. piscicida prevents the complement activation [29]. In this study, we found that E. piscicida exhibited apparent binding to PoC3, PoC3dg, PoCFH, and PoCFI, but not binding to PoCFB and PoC5, implying that C3 convertase (C3bBb) and C5 convertase (C3bBbC3b) no formation on the bacterial surface. These findings suggest that E. piscicida recruits PoCFH from P. olivaceus serum to bacterial surface, which leads to the subsequent binding of PoCFI to degrade PoC3b to PoC3dg. Hence, PoCFH and PoCFI are probably the key targets for E. piscicida to prevent complement activation, which is consistent with the reduced the internalization of E. piscicida in the PoCFH- or PoCFI-depleted serum. Consistently, the amount of C5a was no significantly increased in E. piscicida-treated serum. These results indicate that E. piscicida can resist serum attack by preventing complement activation. In contrast, E. coli activates the complement system and leads to serum killing, resulting in E. coli exhibiting no serum survival. Therefore, we speculated that E. piscicida prevents the formation of C3 convertase, C5 convertase, and MAC on the bacterial surface. This prevention mechanism contributes to the high survival rate of E. piscicida in P. olivaceus serum. In human infection model, spores produced on medium supplemented with serum increased the virulence of Mucor circinelloides [45]. A mutant strain of Streptococcus mutans, CBP⁺/PA⁻, utilizes serum to improve the pathogenicity in infective endocarditis [46]. In our study, serum treatment was significantly increased the phagocytosis of PBL for E. piscicida, and improved the bacterial ability of attachment and invasion. However, the abilities of E. piscicida were compromised by treatment with PoC3-depleted serum, implying that PoC3 might play an important role on the regulation function of serum. These results indicate that E. piscicida exhibits resistance serum killing and improves infectivity through interaction with PoC3 fragments in P. olivaceus serum.

In humans, C3b is downregulated by CFI and one of several cofactor molecules (MCP, CR1, or CFH) [16]. The α chain of C3b in humans is cleaved by CFI and a cofactor at Arg₉₃₂-Glu₉₃₃ and Arg₁₂₈₁-Ser₁₂₈₂ sites, resulting in the production of C3dg [16]. Additionally, the binding site of CR2 to its ligand C3dg is localized within C3 at residues 1201–1214 [16]. Our analysis shows that, like human C3, PoC3dg possesses a comparable CR binding domain and cleavage sites of CFI. The activated complement produces multiple opsonins (C3b, iC3b, and C3dg), which attaches to target surfaces through covalent bonding to facilitate transportation and promote the elimination of target cells or immune complexes [16]. The interaction between C3dg fragment and CR2 facilitates lymphocyte activation, thereby augmenting the cellular immune response [47]. In this study, we found that rPoC3dg and rPoC3dgM exhibited equivalent-binding abilities in E. piscicida. However, only rPoC3dg could bind to PBL, suggesting that PoC3dg may recognize the receptor on the surface of PBL depending on the specific-binding sites. In addition, we found that, in the presence of rPoC3dg, E. piscicida induced rapid cell death, and the infected cells released LDH. Previous studies have reported that pathogenic microorganisms, such as Salmonella, can utilize phagocytosis of host to infect target cells [48]. The entry of Edwardsiella tarda into mouse macrophages is a passive process of phagocytosis that no depend on virulence factors [14]. FACS analysis showed that rPoC3dg stimulated PBL to display remarkably higher phagocytic activity for E. piscicida, and induced a mass of bacterial internalization. Consistently, we found that, when PBL was infected with E. piscicida, rPoC3dg enhanced bacterial attachment and invasion. The mechanisms by which macrophages engulf microorganisms and foreign particles include micropinocytosis, clathrin- and caveolin-mediated endocytosis, as well as clathrin- and caveolin-independent endocytosis [49]. In this study, the results indicated that, among the three inhibitors examined, the inhibitor targeting clathrin significantly inhibited the ability of rPoC3dg-treated E. piscicida to internalize and invade cells. This suggests that clathrin-mediated phagocytosis is crucial for PoC3dg-involved bacterial internalization. Furthermore, we found, when P. olivaceus were infected with E. piscicida, rPoC3dg exhibited significantly increased bacterial tissue dissemination and mortality rates, along with decreased expression of immune genes. These results suggested that PoC3dg promotes bacterial replication and plays a positive role in E. piscicida infection. Fish death was mostly attributed to PoC3dg-induced cell death and tissue dissemination caused by E. piscicida, resulting in a weak immune response that was fatal to fish. These results indicate the possibility that E. piscicida promotes internalization and infection through PoC3dg regulation host phagocytosis.

In mammals, complement activation on bacterial surfaces causes extensive deposition of C3b, iC3b, and C3dg, which are primary complement opsonins [50]. Previous study has shown that CR2 functions as a receptor for C3dg, while CR3 acts as a receptor for iC3b [51]. In erythrocytes, C3dg can serve as a ligand for the binding domain of CR3 and facilitate erythrophagocytosis [52]. CR3, also known as CD11b/CD18, belongs to the β2-integrin receptor family [53]. It contains a common β chain (CD18) and a unique α chain (CD11b) [53]. The binding sites of iC3b and C3d are located on the VWA-A domain of CR3 α chain [54]. However, in this study, we found that rPoC3dg exhibited no immunoprecipitation with cell lysis from overexpressing PoCD11b, suggesting that rPoC3dg might bind to other receptor in the host cell. Since PoCD18 own similar domain with PoCD11b, subsequent study found that rPoC3dg could directly bind to the INB domain of PoCD18. Microscopic observation and in vitro analysis consistently showed that the attachment of E. piscicida to the host cells was mediated via the interaction between PoC3dg and PoCD18. This process significantly depends on the functional domain of PoCD18. Furthermore, alpha-fold analysis revealed a specialized region within the INB domain of PoCD18 that specifically interacted with certain residues in PoC3dg. These findings indicate that the evasion of E. piscicida from serum killing into cells is regulated via the interaction between PoC3dg and PoCD18, with the structural integrity of the INB domain playing a crucial role.

In the progress of complement regulation, iC3b is cleaved by factor I to form surface-anchored C3dg and free C3c, and C3dg contains a thioester-containing domain [55]. In humans, CR2 is a surface-associated glycoprotein that binds to C3dg [56]. CR3 is a heterodimeric type I membrane protein that consists of a CD18 β chain, and binds to a variety of endogenous ligands, such as C3 fragments iC3b and C3dg [50]. CD18 contains an INB domain, which serves as the primary-binding site for the ligand and contributes to the interaction with iC3b [56]. At present, it is strange that CR2 has not been characterized in teleosts. Our study showed that PoC3dg and PoCD18 shared moderate sequence identities with C3dg and CD18 from other species, respectively, and displayed a similar primary domain to that of other species, suggesting that PoC3dg and PoCD18 possess conserved structures during evolution. Interestingly, we found that the interaction between C3dg and CD18 only occurred in P. olivaceus and D. rerio, and was not found in mammals or other species. This indicated that C3dg-mediated phagocytosis via specific binding to CD18 may only exist in teleosts. These results speculate that E. piscicida, as a fish pathogen, can utilize the unique complement-binding mechanism of C3dg-CD18 in teleosts to regulate internalization and infection.

CD18 integrin, the β chain of CR3, is a conserved phagocytic receptor that mediates adhesion, phagocytosis, and chemotaxis in various immunocytes [56]. Entry into the host phagocytes by silent phagocytosis is a crucial virulence characteristic for intracellular pathogens [57]. Previous study has demonstrated that E. piscicida can invade into various types of host cells, such as fish and mouse macrophages [14]. The conidia of the medically important pathogenic fungus Aspergillus fumigatus can internalize into alveolar epithelial cells and disseminate within host [58]. Mouse neutrophils lacking CR3 or treatment with anti-CR3 antibody impaired the phagocytosis against Neisseria gonorrhoeae [59]. Paracoccidioides brasiliensis was observed to occur a slower internalization rate in macrophages from mice with low CD18 expression [60]. Previous study identified that CD18 is a potential cellular surface receptor to uptake of live Pseudomonas aeruginosa in the phagocytes from a patient with leukocyte adhesion deficiency type 1 [61]. In this study, we found that constitutive PoCD18 expression was ubiquitous in all nine examined tissues, and the expression in immune related tissues was significantly induced by E. piscicida challenge. When PoCD18 was knocked down in P. olivaceus, microscopy and fluorescence analysis revealed that rPoC3dg was decreased the binding capability to PBL, suggesting that PoC3dg binds and regulates host cells via interaction with PoCD18. These results indicate that the utilization PoC3dg-PoCD18 into host cells is a unique immune evasion mechanism for E. piscicida. Furthermore, knockdown or antibody blocking of PoCD18 significantly impaired the ability of E. piscicida to induce cellular infection, tissue damage, and fish mortality. These results indicate that PoCD18 is essential for E. piscicida to effectively infect the host cells by utilization complement opsonization of PoC3dg, and the targeting of CD18 could be a potential therapeutic approach to control edwardsiellosis in teleosts.

The effective internalization of opsonin-coated particles by CR3 requires an inside-out activation signal [62]. CR3-dependent phagocytosis is regulated by intracellular signalling cascades initiation from multiple receptors, and the production of membrane ruffles that capture opsonized target particles [63,64]. In leukocytes, CR3 activation triggers the phosphorylation of several intracellular signalling molecules, such as ERK1/2, p38 MAPK, NF-κB, and JNK [58,59,65]. Previous study revealed that cytohesin-1, functioning as an adaptor protein, interacts with the cytoplasmic tail of CD18, thus inducing changes in the functional characteristics of β2 integrins [66]. Knockdown of cytohesin-1 significantly impairs the macrophage to uptake of mycobacteria [38]. This study found that E. piscicida bound to PoC3dg and triggered PoCytohesin activation via the intracellular domain of PoCD18, thereby subsequently initiating the host PI3K/Akt signalling pathway and facilitating bacterial infection. Cytohesin-1 possesses a pleckstrin homology domain that binds to the PI3K metabolite, leading to protein properties changes [67–70]. In this study, we found that the infection and internalization of E. piscicida mediated by PoC3dg were specifically inhibited by PI3K/Akt inhibitor, implying that PoC3dg mediated internalization by activation PI3K/Akt pathway. These results demonstrate that E. piscicida binds to PoC3dg and initiates PoCytohesin involving PI3K/Akt signalling pathway, ultimately leading to enhanced bacterial internalization dependent on PoCD18. These observations suggest that hijacking the complement components and mediating cellular phagocytosis are unique immune evasion mechanisms that may have evolved for intracellular pathogens.

In conclusion, E. piscicida exhibited binding affinity towards PoCFH and PoCFI in P. olivaceus serum, leading to the degradation of PoC3b to PoC3dg on the bacterial surface. Subsequently, E. piscicida bound to the PBL via the specific interaction between PoC3dg and the INB domain of PoCD18 in teleosts (Figure 8). PoC3dg-PoCD18 mediates the activation of PI3K/Akt signalling pathway by PoCytohesin, thereby enhancing clathrin-dependent phagocytosis and facilitating the internalization of E. piscicida (Figure 8). These findings provide the insight into E. piscicida infection mechanism via complement-mediated internalization.

Figure 8.

Schematic representation of C3dg mediation the internalization and infection of Edwardsiella piscicida by binding CD18 and phosphorylating PI3K/Akt. In the infection progress, E. piscicida bound CFH in serum, whereby inactivating C3b to C3dg via recruitment CFI. PoC3dg on the bacterial surface interacted with the INB domain of CD18 to attach to the cell. The interaction between C3dg and CD18 triggered the intracellular domain of CD18 binding to Cytohesin, and inducing the phosphorylation of PI3K/Akt. Phosphorylation of PI3K/Akt internalized E. piscicida into cell by clathrin-mediated phagocytosis, resulting in intracellular bacterial infection.

Funding Statement

This study was financially supported by grants from the National Natural Science Foundation of China [42476094], the Natural Science Foundation of Tianjin [23JCYBJC00240], the Social Development and Agricultural Science and Technology Project of Tianjin [23YFZCSN00140], and the Young Scholar Program of Tianjin City Distinguished Professor.

Author contributions

MFL and JSS conceived and designed the experiments. MFL, XYJ, HL, ZF, MW, HQZ, YTD, YYS and XPL performed the experiments and analysed the data. MFL wrote the manuscript. JSS edited the manuscript. All the authors have read and approved the final manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data and supplementary materials that support the findings of this study are openly available in zenodo at http://doi.org/10.5281/zenodo.14954856.

ARRIVE statement

In this study, all experiments involving live animals were conducted by ARRIVE guidelines. And the ARRIVE checklist was uploaded as a supplementary file.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21505594.2025.2526783