Bacterial strains and plasmids used in this study are listed in Table 1. RA-YM was isolated and deposited at the Department of Veterinary Microbiology and Immunology of Huazhong Agricultural University. RA strains were grown on TSA (Difco, Detroit, MI, United States) or TSB (Difco) at 37°C in 5% CO₂. Escherichia coli was grown at 37°C on a Luria-Bertani plate or in Luria-Bertani broth. Antibiotic concentrations were as follows: ampicillin, 100 μg/mL; chloramphenicol, 50 μg/mL; spectinomycin (Spc), 100 μg/mL; and diaminopimelic acid, 50 μg/mL.

The mutant strain ΔsprT was constructed by allelic exchange using the recombinant suicide plasmid pRE112. Briefly, a 585-bp left flanking region of the sprT gene (Leftarm) was amplified from RA-YM genomic DNA using primers SprT-F1 (introducing a KpnI site) and SprT-R1. A 1086-bp Spc^R cassette was amplified from plasmid pIC333 using primers Spc-F and Spc-R. A 867-bp right flanking region of sprT gene (Rightarm) was amplified from RA-YM genomic DNA using primers SprT-F2 and SprT-R2 (introducing a SacI site). The left flanking region, Spc^R cassette, and right flanking region were fused by overlap extension PCR using SprT-F1 and SprT-R2. The fused DNA fragment was inserted into pMD18-T to generate pMD18T-Leftarm-Spc-Rightarm, which was digested along with the pRE112 plasmid with KpnI and SacI to obtain the recombinant suicide plasmid pRE112-Leftarm-Spc-Rightarm (pRE112-LSR). This was transformed into E. coli X 7213. The transformants were used as the donor strain in conjugal transfer. The mutant strain was selected on TSA containing 100 μg/mL Spc. The sprT gene deletion mutant strain (ΔsprT) was confirmed by PCR.

The shuttle plasmid pRES-JX-bla was constructed as previously described (Guo et al., 2017) by adding the putative replication region of the RA plasmid RA-JX to the suicide vector. The promoter and coding sequences of sprT were amplified using primers Promoter-sprT1 (introducing a KpnI site) and Promoter-sprT2, SprT-CF, and SprT-CR (introducing a SphI site). The promoter and coding fragments were joined by overlap extension PCR. The product was digested with KpnI and SphI and inserted into pRES-JX-bla to generate plasmid pRES-JX-bla-sprT, which was transferred by conjugation into the mutant strain RA-YMΔsprT. Transconjugants were screened on TSA supplemented with ampicillin and verified by PCR.

Growth curves were generated for mutant strain ΔsprT, complemented strain CΔsprT, and wild-type strain RA-YM to determine whether sprT deletion influenced bacterial growth. The three strains were grown for about 48 h at 37°C in 5% CO₂ on TSA plates and were inoculated into TSB. At mid-exponential phase, the culture was diluted with fresh medium to an optical density at 600 nm (OD₆₀₀) of 1.0, and equal amounts of each bacterial culture were transferred to fresh TSB medium at a ratio of 1:1000 (v/v) and cultured at 37°C with shaking at 200 rpm. OD₆₀₀ was measured at 1 h intervals for 20 h using a spectrophotometer (Bio-Rad, Hercules, CA, United States).

Biochemical testing of the three strains was carried out using bacterial biochemical tubes (Hope Bio, Qingdao, China) with glucose, arabinose, sucrose, citrate, hydrogen sulfide, nitrate, carbamide, and gelatin. To visually characterize the presence of liquefied gelatin, RA-YM, ΔsprT, and CΔsprT cells were inoculated into TSB overnight. Cultures (5 μL) were spotted on nutrient gelatin plates and cultivated at 37°C for 24 h. Then, 2–3 mL acid mercuric chloride was added to the plates and color development was assessed after 5 min. Cells capable of liquefying gelatin, were identified by a transparent halo around the bacterial lawn (Frazier, 1926; Mcdade and Weaver, 1959; Thomas et al., 2014).

Blood was drawn from 12 healthy 30-day-old Cherry Valley ducks without anti-RA antibody and immediately placed on ice. After 2 h, the clotted blood was centrifuged at 2000 × g and 4°C for 10 min. The serum was collected as normal duck serum and stored at -80°C until use.

For the serum survival assay, mutant strain ΔsprT, complemented strain CΔsprT, and wild-type strain RA-YM were cultured in TSB to an OD₆₀₀ of 0.8. The cells were washed and resuspended to 10⁶ CFU/mL in HBSS (Gibco, Grand Island, NY, United States) with 0.15 mM calcium and 1 mM magnesium (HBSS⁺⁺). After incubation at 37°C for 30 min with or without duck serum, 10-fold serial dilutions of the mixture were spread onto TSA plates. Colonies were counted after incubation for 48 h. The serum survival assay was performed with bacteria resuspended in HBSS⁺⁺ to prevent replication so that the viability of all strains could be evaluated (Rosadini et al., 2013). The reaction was also performed in the presence or absence of 10 mM Mg²⁺ EGTA to block the classical complement and lectin pathways and selectively activate the alternative complement pathway. Heat-inactivated serum used in this assay was generated by incubating normal duck serum at 56°C for 30 min. The survival rate was calculated as follows: (number of cells that survived serum treatment/number of cells that survived control treatment) × 100%. Differences among groups were evaluated by analysis of variance using SPSS v.18.0 software (SPSS Inc., Chicago, IL, United States). Data are presented as mean ± standard error of the mean, and statistical significance was determined with the Student’s t-test. A P-value of <0.05 was defined as statistically significant, and P-value of <0.01 was defined as extremely significant.

Gelatin zymography was used to evaluate the difference in gelatinase secretion between mutant strain ΔsprT, CΔsprT and wild-type strain RA-YM as previously described (Pickett et al., 1991; Toth and Fridman, 2001). Briefly, cells were grown to mid-log phase in TSB at 37°C with shaking, then centrifuged at 10,000 × g for 15 min. The culture medium was passed through a 0.22 μm polyvinylidene difluoride filter to remove residual cells, and the supernatant was precipitated with 70% (v/v) saturated ammonium sulfate solution. The precipitate was dissolved with phosphate buffer and samples were resolved by SDS-PAGE on a 10% polyacrylamide gel copolymerized with gelatin (1%) as the substrate at 4°C. The gel was washed for 2 min, proteins were renatured by incubation for 1 h at room temperature in 2.5% Triton X-100 solution, and then incubated at 37°C for 18–20 h in 50 mmol/L Tris-HCl buffer (pH 7.4) containing 5 mmol/L CaCl₂. The gel was stained with 0.05% Coomassie Brilliant Blue R-250 and then destained with 30% methanol and 10% acetic acid. Gelatinolytic activity was detected as unstained bands against the background of Coomassie-stained gelatin.

To detect proteins involved in gelatin hydrolysis, bands that differed between the RA-YM and ΔsprT gels were analyzed by LC-MS/MS. The bands were excised and digested overnight in 12.5 ng/μL trypsin in 25 mM NH₄HCO₃. The peptides were extracted three times with 60% acetonitrile/0.1% trifluoroacetic acid, and the extracts were pooled and dried completely in a vacuum centrifuge.

Experiments were performed on a Q Exactive mass spectrometer coupled to Easy nLC (Thermo Fisher Scientific, Waltham, MA, United States). A 6 μL volume of each fraction was injected for nanoLC-MS/MS analysis. The peptide mixture (5 μg) was loaded onto a C18 reversed-phase Easy Column (length, 10 cm; inner diameter, 75 μm; and 3 μm resin) (Thermo Fisher Scientific) in buffer A (0.1% formic acid) and separated with a linear gradient of buffer B (80% acetonitrile and 0.1% formic acid) at a flow rate of 250 nL/min controlled by IntelliFlow technology over 140 min. MS data were acquired using a data-dependent top 10 method by dynamically selecting the most abundant precursor ions from the survey scan (300–1800 m/z) for higher-energy collisional dissociation fragmentation. Determination of the target value was based on predictive Automatic Gain Control. The duration of dynamic exclusion was 60 s. Survey scans were acquired at a resolution of 70,000 at m/z 200, and the resolution for HCD spectra was set to 17,500 at m/z 200. Normalized collision energy was 30 eV and the underfill ratio, which specifies the minimum percentage of the target value likely to be reached at maximum fill time, was defined as 0.1%. The instrument was run with peptide recognition mode enabled.

MS/MS spectra were searched using MASCOT engine v.2.2 (Matrix Science, London, United Kingdom) against the non-redundant International Protein Index Arabidopsis sequence database v.3.85 (released September 2011; 39,679 sequences) from the European Bioinformatics Institute¹. The following options were used for protein identification: peptide mass tolerance = 20 ppm, MS/MS tolerance = 0.1 Da, enzyme = trypsin, missed cleavage = 2, fixed modification: carbamidomethyl (C); and variable modification: oxidation (M).

The coding sequence of the RAYM_01812 gene (without the signal peptide) was amplified using primers P01812-1 (introducing a BamHI site) and P01812-2 (introducing an XhoI site). The amplified fragment and vector pGEX-6P-1 were digested with BamHI and XhoI and ligated to generate the expression vector pGEX-6P-1_01812, which was transformed into E. coli BL21 (DE3). Recombinant RAYM_01812 protein was purified as previously described (Ksiazek et al., 2015a). Transformed E. coli was grown in Luria-Bertani medium at 37°C to an OD₆₀₀ of 0.8, and isopropyl β-D-1-thiogalactopyranoside was added to a final concentration of 0.2 mM. After cultivation at 25°C for 6 h, cells were harvested by centrifugation and resuspended in sodium phosphate buffer (50 mM, pH7.5). Cells were disrupted twice in a French press at 100 MPa. After centrifugation at 6000 × g and 4°C for 15 min, the supernatant was separated from the cell pellet and filtered through a 0.45 μm pore size polyvinylidene difluoride membrane to remove residual cell debris. The protein fraction in the supernatant was purified using an ÄKTA purifier instrument and GST Trap HP crude affinity chromatography column (GST Trap FF, GE, China). The protein was eluted using Tris-HCl (50 mM, pH 8.0) containing 10 mM reduced glutathione. The purified RAYM_01812 protein was incubated at 37°C for 0.5, 1, 4, 8, 24, and 48 h. Protein purity was evaluated by 12% SDS-PAGE and Coomassie staining.

Gelatin zymography of the purified RAYM_01812 protein was performed as previously described (Ksiazek et al., 2015a). The protein was incubated at 37°C for 0.5, 1, 4, 8, 24, and 48 h, then mixed with non-reducing SDS sample buffer (1:1) and incubated at 20°C for 15 min. The mixture was resolved by 10% SDS-PAGE on gels containing gelatin at a final concentration of 1 mg/mL at 4°C. The following operations were analogous to those performed in the gelatin enzyme spectrum experiment. Gelatinolytic activity was visible unstained zones against a blue background.

Cherry Valley ducklings (10 days old) were obtained from Chunjiang Duck Farm (Wuhan, China) and allowed to adapt to the environment under controlled temperature (30°C) for 2 days. The ducklings had free access to food and water. The experiments were approved by the Institutional Animal Experiment Committee of the Veterinary Faculty of Huazhong Agriculture University.

Ducklings (12 days old) were randomly divided into 18 groups (n = 10 per group) and challenged by intramuscular injection of bacteria at 1.0 × 10⁴, 1.0 × 10⁵, 1.0 × 10⁶, 1.0 × 10⁷, 1.0 × 10⁸, or 1.0 × 10⁹ CFU of mutant strain ΔsprT, complemented strain CΔsprT, or wild-type strain RA-YM. The mortality of the ducklings was monitored for 1 week post infection. The LD₅₀ was calculated using an improved Karber’s method (Irwin and Cheeseman, 1939).

To evaluate bacterial invasion into organs, we compared bacterial load in the blood and target organs (spleen, liver, heart, and brain) and examined pathological lesions. Cherry Valley ducklings (12 days old) were randomly divided into two groups (n = 15 per group) and injected intramuscularly with 1 × 10⁷ CFU bacteria (ΔsprT or RA-YM). The organs and blood were collected 24 or 48 h post infection and diluted appropriately. Colonies were counted with the plate pouring method. Data are presented as mean ± standard error of the mean, and statistical significance was determined with the Student’s t-test. A P-value of <0.05 was defined as statistically significant, and P-value of <0.01 was defined as extremely significant. Spleen, liver, heart, and brain tissue sections were immersed in 10% formalin solution for histopathological analyses.

Cherry Valley ducklings (7 days old) were randomly divided into five groups (n = 20 per group). Groups 1–3 received a single injection of the mutant strain ΔsprT at doses of 4 × 10⁵, 2 × 10⁶, or 1 × 10⁷ CFU. Group 4 was injected with phosphate-buffered saline. Group 5 was injected with trivalent RA oil-inactivated vaccine at the recommended dose (1 × 10¹⁰ CFU/mL, 0.5 mL) (Tianbang, Chengdu, China). Groups 4 and 5 were served as negative and positive controls, respectively. On day 14 post-immunization, the ducks in each group were injected with 2 LD₅₀ or 10 LD₅₀ of RA-YM to evaluate the level of protection conferred by ΔsprT. The protection rate was calculated using the following formula: [1 - (dead ducklings per group/total ducklings per group)] × 100.

The deletion of the sprT gene after allelic exchange was confirmed by PCR amplification of the smaller fragment (1752-bp) in RA-YM as compared to the larger Leftarm-Spc-Rightarm fragment (2538-bp) in RA-YMΔsprT (Figure 1A). The deleted part of the sprT gene (300-bp) was not detected (Figure 1B). The identity of the complemented strain was also confirmed by PCR.

There were no differences in biological characteristics between wild-type RA-YM, mutant ΔsprT, and complemented CΔsprT strains except in the early logarithmic growth phase, with ΔsprT exhibiting more rapid growth as compared to RA-YM, which had a similar growth rate as CΔsprT (Figure 1C). In biochemical tests, there was no difference among ΔsprT, RA-YM, and CΔsprT except in terms of gelatin liquefaction. All three strains showed negative test results for glucose, arabinose, sucrose, citrate, hydrogen sulfide, nitrate, and carbamide. RA-YM and CΔsprT were able to liquefy gelatin whereas ΔsprT was not. The capacity of different strains to liquefy gelatin on nutrient gelatin plates was in accordance with the results of biochemical tests. As shown in Figure 1D, there were transparent zones around RA-YM and CΔsprT, but not around ΔsprT, demonstrating that RA-YM and CΔsprT were able to liquefy gelatin whereas ΔsprT was not.

The serum survival assay was carried out to ascertain the ability of the mutant strain ΔsprT to resist complement-mediated killing as compared to the wild-type strain RA-YM. To exclude potential effects of variable growth rates among strains, the assay was performed with the cells resuspended in HBSS⁺⁺, which prevents replication without affecting viability (data not shown). A range of serum concentrations was evaluated for wild-type strain RA-YM; the survival rate of RA-YM was 93.8% in 6.25% normal duck serum. In contrast, the survival rate of ΔsprT was 1.3% in 6.25% normal duck serum, whereas that of complemented strain CΔsprT was 95% in 6.25% normal duck serum, which was comparable to the wild-type value (Figure 2A). Heat inactivation abrogated the bactericidal effect of serum in ΔsprT, consistent with complement-mediated killing. Together, these results indicate that the T9SS is required for complement resistance in wild-type RA cells.

The ΔsprT mutants were examined for their sensitivity to alternative complement pathway activation. Strains were treated with normal duck serum in buffer in the presence or absence of 10 mM Mg²⁺ EGTA, which inhibits classical/mannose-binding lectin pathway activation, and then examined for survival. Incubation with Mg²⁺ EGTA-containing buffer alone did not affect the viability of any of the strains (data not shown). When cells were incubated in 12.5% normal duck serum in the presence of Mg²⁺ EGTA, the survival rates of RA-YM ΔsprT, and CΔsprT were 61.7, 24.3, and 57.4%, respectively (Figure 2B). Taken together, these data indicate that the proteins secreted by T9SS suppress the alternative complement pathway in RA, possibly by binding factor H.

A gelatin zymography assay was performed to compare gelatinase secretion between mutant strain ΔsprT and wild-type strain RA-YM (Figure 3A). Distinct bands in the RA-YM culture supernatant were excised and analyzed by MS. The raw data were converted to XML files for an MS/MS ion search using Mascot and were assigned to the protein sequences of RA (strain ATCC 11845/DSM 15868/JCM 9532/NCTC 11014), as annotated by the Uniprot database. Positive hits were accepted with a Mascot score of at least 20. A total of 52 proteins were identified (Supplementary Table S1). T9SS proteins were predicted by running a Function Search analysis of the conserved CTDs of the TIGRfam families TIGR04131 and TIGR04183 (Sato et al., 2013; Kharade and McBride, 2014).

Among the 52 proteins, seven had CTDs belonging to TIGR04183 or TIGR04131, and may therefore be secreted by the T9SS (Table 2). RAYM_09380 and RAYM_02622 were hypothetical proteins and RAYM_05530 had a periplasmic ligand-binding sensor domain related to signal transduction. The four other proteins all had a Por_Secre_tail, but their functions differed according to their specific structural domains. RAYM_08375, containing a polycystic kidney disease (PKD) and two Sialidase_non-viral domains, was predicted to encode glycosyl hydrolase. RAYM_04099, containing a Peptidases_S8_Kp43_protease domain, encodes peptidase S8/S53 subtilisin kexin sedolisin. RAYM_03382, containing a metallophosphatase superfamily and calcineurin-like phosphoesterase domains, was predicted to encode phosphohydrolases. RAYM_01812, also containing Peptidases_S8_S53 superfamily and peptidase_S8 domains, encodes a subtilisin-like serine protease.

RAYM_01812 was expressed as a fusion protein with a molecular mass of 102 kDa and the capacity for autoproteolytic cleavage (Figure 3B). The full-length recombinant RAYM_01812 protein (without GST label) had a mass of 76 kDa. Following removal of the NTP, preCTD and CTD regions, the active protease comprising only CD regions was left with a molecular mass of 38 kDa. Gelatin zymography analysis of purified RAYM_01812 revealed that three forms of the protein were capable of degrading gelatin (Figure 3C).

The LD₅₀ for the mutant strain ΔsprT was 1.61 × 10⁹ CFU, which was 4.21 × 10⁴-fold lower than that of the parent strain RA-YM (3.82 × 10⁴ CFU) and of CΔsprT (4.37 × 10⁴ CFU). This indicated that the virulence of ΔsprT was potently suppressed, an effect that was rescued by complementation. Since the difference between RA-YM and CΔsprT was negligible, in subsequent experiments we compared only RA-YM and ΔsprT.

To examine the invasiveness of the bacteria in greater detail, bacterial load in the blood, spleen, liver, heart, and brain was quantified. The bacterial burden of ΔsprT in the blood, spleen, and liver was much lower than that of RA-YM at 24 and 48 h post challenge (Figure 4). We also compared tissue damage caused by RA-YM and ΔsprT. In the RA-YM group, there was obvious congestion in the hepatic sinusoid and central vein, fatty degeneration of hepatocytes, a large quantity epicardial fibrin exudate, and inflammatory cells; notably, the latter two were present also in the subarachnoid space. Additionally, lymphocyte proliferation of splenic white pulp was also observed. This phenotype was greatly attenuated in the ΔsprT and negative control groups at 24 h (Figure 5A) and 48 h (Figure 5B) post challenge.

Finally, we evaluated whether the mutant strain ΔsprT could be used as a live attenuated vaccine. Ducks were administered 2 LD₅₀ and 10 LD₅₀ of virulent RA-YM by intramuscular injection 14 days after vaccination. At 24 h post-injection, clinical symptoms and death were observed in Group 4, whereas ducks in Groups 1–3 that were vaccinated with ΔsprT showed only mild symptoms. When ΔsprT was injected at a dose of 1 × 10⁷ CFU per duck (Group 3), immune protective ratios were 100 and 88.9% as compared to 80 and 70%, respectively, in the vaccine group (Table 3).