Bacterial matrix metalloproteases and serine proteases contribute to the extra-host inactivation of enteroviruses in lake water

Virus propagation and enumeration

Echovirus-11 (E11, Gregory strain, ATCC VR737) and Coxsackievirus-A9 (CVA9, environmental strain from sewage, kindly provided by the Finnish National Institute for Health and Welfare) stocks were produced by infecting sub-confluent monolayers of BGMK cells as described previously [7]. Viruses were released from infected cells by freezing and thawing the culture flasks three times. To eliminate cell debris, the suspensions were centrifuged at 3000 × g for 5 min. Each stock solution was stored at −20 °C until use. Infectious virus concentrations were enumerated by a most probable number (MPN) infectivity assay as described in the Supplementary Information. The assay limit of detection (LoD), defined as the concentration corresponding to one positive cytopathic effect in the lowest dilution of the MPN assay under the experimental conditions used, corresponding to 2 MPN/mL.

Inactivation of enteroviruses by bacterial consortia from lake water

To study the inactivation of CVA9 and E11 by a bacterial consortium from lake water, four surface water samples were collected from Lake Geneva (Ecublens, Switzerland) during the summer 2021. Each sampling event was conducted on warm and sunny days, to minimize biological variation. Immediately after sampling, large particles of the sample were removed by filtering 500 mL of water on a 8 μm nitrocellulose filter membrane (Merck Millipore, Cork, Ireland). The sample was then filtered through a 0.8 μm nitrocellulose filter membrane (Merck Millipore) to remove large microorganisms such as protists. The resulting water sample corresponds to the bacterial fraction used to study virus inactivation.

For inactivation experiments, each virus was spiked into individual 1 mL aliquots of fractionated lake water to a final concentration of 10⁶ MPN/mL, and samples were incubated for 48 h at 30 °C without shaking. Duplicate experiments were conducted for each virus and each lake water sample. Experiments to control for thermal inactivation were conducted using the same procedure but by replacing the fractionated lake water with sterile milliQ water. Viral infectivity at times 0 h and 48 h was determined by MPN as described above. Virus decay was calculated as log₁₀ (C/C₀), where C is the residual titer after 48 h of incubation, and C₀ is the initial titer. The experimental LoD was approximately 5-log₁₀.

These same experiments were conducted for three new water samples in the presence of four protease inhibitors with the following final concentrations: E64—10 μM (E3132, Sigma–Aldrich, Saint-Louis, MO, USA), GM6001—4 μM (CC1010, Sigma–Aldrich), Chymostatin—100 μM (C7268, Sigma–Aldrich), and PMSF—100 μM (P7626, Sigma–Aldrich). Each inhibitor was added to 1 mL of fractionated lake water, vortexed for 30 seconds, and incubated at room temperature for 15 min, before adding the two viral strains under the same conditions as described above.

Bacterial isolation, cultivation, and storage

Bacteria were isolated from two water samples from Lake Geneva’s Ecublens beach, taken in November 2019 (Fall, 89 isolates) and May 2020 (Spring, 47 isolates). Bacteria recovery was performed on R2A agar plate (BD Difco, Franklin Lakes, NJ, USA) as described previously [15]. Briefly, successive dilutions from 10⁻¹ to 10⁻⁵ were carried out in sterile water for each sample. For each dilution, a volume of 1 mL was deposited on three separate R2A plates, before being incubated at 22, 30, and 37 °C. After 5 days of incubation, each colony was picked and enriched on a new R2A plate. To ensure purity, each isolate was successively plated five times on R2A plate and incubated at the same temperature as the initial isolation. Each purified isolate was cryopreserved in R2A / 20% glycerol at −80 °C. The isolates were named based on the water body (Lake (L)), isolation temperature, and the isolation order (L-T°C-number).

Bacterial identification

The identification of each isolate was performed by 16 S rRNA gene sequencing using the pair of primers 27 F (5’- AGA GTT TGA TCM TGG CTC AG- 3’, Microsynth AG, Balgach, Switzerland) / 786 R (5’- CTA CCA GGG TAT CTA ATC – 3’, Microsynth AG), following a methodology previously described [15]. The thermocycling conditions and the purification of PCR products are described in the Supplementary Information. The complete list of isolated bacteria and associated accession numbers is given in Supplementary Table 1.

Phylogenetic inference and metadata visualization

The consensus from 16 S rRNA gene sequences of the 136 isolates was aligned using the MUSCLE algorithm [16]. The phylogenetic analysis of 566 bp aligned sequences from the V2-V4 16 S rRNA gene regions (Positions: 152–717) was performed using Molecular Evolutionary Genetics Analysis X software [17]. Phylogeny was inferred by maximum likelihood, with 1000 bootstrap iterations to test the robustness of the nodes. The resulting tree was uploaded and formatted using iTOL [18].

Virus incubation with bacterial isolates

For the preparation of the bacteria before co-incubation, each one was first cultured on R2A agar for 48 h at their initial isolation temperature. Overnight suspensions of each bacterial isolate were grown in R2A broth at room temperature under constant agitation (180 rpm). For co-incubation experiments, 200 μL of each bacterial suspension were mixed with 100 μL of a 10⁵ MPN/mL stock of E11 or CVA9. Then, each condition was supplemented with 600 μL of R2A broth. Incubation was carried out for 96 h at room temperature, without shaking. At the end of the co-incubation, each tube was centrifuged for 15 min at 9000 × g (4 °C) to eliminate bacteria, and the residual infectious viral titer was enumerated by MPN assay as described above [7]. Each co-incubation experiment was carried out in triplicate. Control experiments were performed under the same conditions but using sterile R2A. Virus decay was quantified as log₁₀ (C_exp/C_ctrl), where C_exp is the residual titer after a co-incubation for 96 h, and C_ctrl is the titer after incubation of the virus in sterile R2A for 96 h. The experimental LoD was 3-log_10.

Protease activity measurement using casein and gelatin agar plates

Casein agar was prepared as follows: 20 g of skim milk (BD Difco), supplemented with 1 g glucose were reconstituted with 200 mL of distilled water. Likewise, a 10% bacteriological agar solution was prepared in a final volume of 200 mL. Finally, a solution consisting of 0.8% NaCl, 0.02% KCl, 0.144% Na₂HPO₄, and 0.024% KH₂PO₄ was reconstituted in 600 mL of water. All solutions were autoclaved for 15 min at 110 °C. The solutions were mixed, and 25 mL were poured into each Petri dish. Gelatin agar was composed of 0.4% peptone, 0.1% yeast extract, 1.5% gelatin and 1.5% bacteriological agar. The mixture was autoclaved 15 min at 120 °C, and 25 mL of medium was poured into each Petri dish.

For each isolate, an overnight suspension was performed in R2A broth at room temperature, before spotting 15 μL of each suspension at the center of both gelatin and casein agar plates. Each plate was incubated at 22, 30, or 37 °C for 72 h, depending on the initial isolation temperature of the bacteria. Casein-degrading activity (cas), which is exerted by many different protease classes, and gelatin-degrading activity (gel), which is mostly caused by MMPs, were revealed by a hydrolysis halo around the producing bacteria. Hydrolysis diameters were measured in millimeters (mm) to report the extent of the proteolytic effect of each strain on both substrates.

Protease activity quantification in cell-free supernatant

Using the same bacterial suspensions as for bacterial/virus co-incubation, 200 μL of each suspension was inoculated into 600 μL of R2A broth and incubated without shaking for 96 h at room temperature. Each culture was centrifuged for 15 min at 9000 × g at 4 °C. The resulting cell-free supernatants (CFS) were stored at −20 °C until use. For each CFS, protease activity was measured using the Protease Activity Assay Kit (ab112152, Abcam, Cambridge, UK), which measures general protease activity (pgen) except MMPs, and the MMP Activity Assay Kit (ab112146, Abcam), which selectively measures MMP activity (mmp). Briefly, for the Protease Activity Assay kit, 50 μL of the substrate was added into each well of a dark-bottom plate containing 50 μL of each CFS. Standard trypsin provided by the kit was used as a positive control. For the MMP Activity Assay kit, 50 μL of each CFS was incubated with 50 μL of 2 mM APMA for 3 h at 37 °C, prior to the activity test. Collagenase I (C0130, Sigma–Aldrich) was used as a positive control. R2A broth was used as a negative control for each assay. Protease activity was measured at time 0 and after 60 min, using a Synergy MX fluorescence reader (BioTek). The excitation and emission wavelengths were set to 485 and 530 nm, respectively. The emitted fluorescence, generated by proteolytic cleavage of the substrate of each kit, was calculated as follows: ∆_RFU = RFU (60 min) − RFU (0 min). Proteolytic activity was calculated in mmol/min/μL based on the emitted fluorescence measured for trypsin and collagenase I at known proteolytic activities.

Data analysis

Statistical analyses to compare inactivation data were performed by one-way t-test or one-way ANOVA with Dunnett’s post-hoc test in GraphPad Prism v.9. An alpha value of 0.05 was used as a threshold for statistical significance. For each dataset we confirmed that data were normally distributed.

To analyze a potential correlation between protease activity and viral decay, the decay values for each virus strain was related to the four protease activity tests of this study using a scatterplot combined with a Kernel density estimation. The analyses were performed with R v.3.6.1 using the SmoothScatter function of the R Base package.

A Left-Censored Tobit model (CTM) with mixed effects was chosen to investigate interactions between protease activity and the decay measured for each virus strain. Briefly, the CTM with mixed effect was chosen for three reasons: (1) The protocol used to measure viral decay had a limit of quantification of −3-log₁₀, and 152 measurement points reached the detection limit, requiring the use of this value as the left-censored value of the model; (2) The two virus strains used in the study showed distinct responses after exposure to environmental bacteria, preventing the use of a multiple linear regression model; (3) Among biological replicates of co-incubation experiments, inactivation variability was observed, suggesting the concomitant action of random biological effects (e.g., production of other compounds than proteases by bacteria, or differences in protease production rate between replicates for each bacterial isolate). The resulting statistical model was then formulated as follows:

for which β₀ defines the model intercept, (beta _1{rm I}_{{{{{{{{mathrm{virus}}}}}}}}_i = 2}) corresponds to the main effect of the virus factor on the viral decay, (beta _2,;beta _3,;beta _4,;{{{{{{{mathrm{and}}}}}}}};beta _5) corresponds to the main effects of the different protease activity measurements on viral decay, (beta _6I_{{{{{{{{mathrm{virus}}}}}}}}_i = 2},;beta _7I_{{{{{{{{mathrm{virus}}}}}}}}_i = 2},;beta _8I_{{{{{{{{mathrm{virus}}}}}}}}_i = 2},{{{{{{{mathrm{and}}}}}}}};beta _9I_{{{{{{{{mathrm{virus}}}}}}}}_i = 2}) corresponds to the interaction effects between each of these variables and the viral decay, (alpha _{{{{{{{{mathrm{id}}}}}}}}_i}) corresponds to the mixed effect of the model and (varepsilon _i) corresponds to the error term of the model. The selection of the model is further detailed in the Supplementary Information (Supplementary Material and Figs. S1 and S2).

The full dataset included in the correlation analysis and the CTM is provided in Supplementary Table 2. A description of the variables used is given in the Supplementary Information. The dataset was analyzed using the censReg package in R [19]. The R code is given in the Supplementary Information.

Source: Ecology - nature.com