Collective behavior and virulence arsenal of the fish pathogen Piscirickettsia salmonis in the biofilm realm

Abstract

Piscirickettsiosis is a fish disease caused by the Gram-negative bacterium Piscirickettsia salmonis. This disease has a high socio-economic impact on the Chilean salmonid aquaculture industry. The bacterium has a cryptic character in the environment and their main reservoirs are yet unknown. Bacterial biofilms represent a ubiquitous mechanism of cell persistence in diverse natural environments and a risk factor for the pathogenesis of several infectious diseases, but their microbiological significance for waterborne veterinary diseases, including piscirickettsiosis, have seldom been evaluated. This study analyzed the in vitro biofilm behavior of P. salmonis LF-89^T (genogroup LF-89) and CA5 (genogroup EM-90) using a multi-method approach and elucidated the potential arsenal of virulence of the P. salmonis LF-89^T type strain in its biofilm state. P. salmonis exhibited a quick kinetics of biofilm formation that followed a multi-step and highly strain-dependent process. There were no major differences in enzymatic profiles or significant differences in cytotoxicity (as tested on the Chinook salmon embryo cell line) between biofilm-derived bacteria and planktonic equivalents. The potential arsenal of virulence of P. salmonis LF-89^T in biofilms, as determined by whole-transcriptome sequencing and differential gene expression analysis, consisted of genes involved in cell adhesion, polysaccharide biosynthesis, transcriptional regulation, and gene mobility, among others. Importantly, the global gene expression profiles of P. salmonis LF-89^T were not enriched with virulence-related genes upregulated in biofilm development stages at 24 and 48 h. An enrichment in virulence-related genes exclusively expressed in biofilms was also undetected. These results indicate that early and mature biofilm development stages of P. salmonis LF-89^T were transcriptionally no more virulent than their planktonic counterparts, which was supported by cytotoxic trials, which, in turn, revealed that both modes of growth induced important and very similar levels of cytotoxicity on the salmon cell line. Our results suggest that the aforementioned biofilm development stages do not represent hot spots of virulence compared with planktonic counterparts. This study provides the first transcriptomic catalogue to select specific genes that could be useful to prevent or control the (in vitro and/or in vivo) adherence and/or biofilm formation by P. salmonis and gain further insights into piscirickettsiosis pathogenesis.

Keywords: RNA sequencing; aquaculture; biofilm cell viability; cytotoxicity; exopolymeric matrix; piscirickettsiosis; virulence.

Figure 1

Crystal violet-stained biofilms and specific biofilm formation (SBF) index. The bulk fraction of the biofilm structure formed by P. salmonis (A) LF-89^T and (B) CA5 was visualized under 1,000X magnification. (C) Box plots show the medians, upper/lower quartiles (boxes), value ranges (vertical lines), and extreme values (circles) of SBF index data sets (×). Statistically significant differences between SBF index boxes (Tukey HSD test; P < 0.05) are indicated with different letters. Data were collected from three independent experiments.

Figure 2

Automated imaging of biofilm development in 96‐well microplates and SCLM micrographs of 48-h-old biofilms. (A) Biofilms of P. salmonis LF-89^T and CA5 on the bottom of wells (including negative-control wells containing sterile AUSTRAL-SRS broth) were stained with the LIVE/DEAD bacterial viability kit at different times. For SCLM, 48-h-old biofilms of P. salmonis (B, C) LF-89^T and (D, E) CA5 were stained with the (B, D) LIVE/DEAD reagent or (C, E) fluorophore-conjugated WGA and DAPI stain. Bacteria stained with DAPI are depicted in orange only for visualization purposes. Automatically captured and SCLM images include a white scale bar in the lower‐right (100 µm) and lower‐left (15-20 µm) corners, respectively. All images are representative of three independent experiments.

Figure 3

Fluorescence signal of live sessile bacteria and biofilm growth prediction. Fluorescence signal was measured in relative fluorescence units (RFU) in biofilms formed by P. salmonis (A) LF-89^T and (B) CA5 previously stained with the LIVE/DEAD reagent. Each box plot shows the median of fluorescence data sets (×), upper/lower quartiles (boxes), value ranges (vertical lines), extreme values (circles), and general patterns (dotted lines). Statistically significant differences between boxes (Tukey HSD test; P < 0.05) are denoted with different letters. (C) The biofilm mode of growth was predicted by fitting a four-parameter lognormal distribution to fluorescence data. The proportion of variance explained by this model accounted for ~88% on average. Data are representative of three independent experiments.

Figure 4

Temporal variations in fluorescence ratio of live-to-dead bacteria in biofilms formed by P. salmonis LF-89^T and CA5. Box plots show the medians of data sets (×), upper/lower quartiles (boxes), value ranges (vertical lines), and extreme values (circles). Statistically significant differences between boxes (Tukey HSD test; P < 0.05) are depicted with different letters. Data were obtained from three independent experiments.

Figure 5

Cytotoxic effect on embryo cells from Chinook salmon (CHSE-214) induced by biofilm-derived and planktonic bacteria of P. salmonis. (A) Micrographs of CHSE-214 cells infected with 48-h-old biofilm-derived bacteria or planktonic counterparts. The cytotoxic effect was measured as the lactate dehydrogenase (LDH) release from CHSE-214 cells infected with P. salmonis (B) LF-89^T and (C) CA5. Box plots show the median of data sets (×), upper/lower quartiles (boxes), and range of percentages (vertical lines). Statistically significant differences between boxes (Tukey HSD test; P < 0.05) are depicted with different letters. Data were taken from three independent experiments.

Figure 6

Comparative gene ontology (GO) functional classification of differentially expressed genes (DEGs) in P. salmonis LF-89^T obtained from four comparisons between different conditions. (A) The gene transcript dataset was analyzed with the WEGO tool set to second-level GO terms for the three major categories (Biological Process, Cellular Component, and Molecular Function). (B) The heatmap depicts significant DEGs between different conditions (i.e., modes of growth and/or incubation times) grouped at the ‘Cellular process’ GO term (see arrow).