Gliding motility proteins GldJ and SprB contribute to Flavobacterium columnare virulence

ABSTRACT

Flavobacterium columnare causes columnaris disease in fish. Columnaris disease is incompletely understood, and adequate control measures are lacking. The type IX secretion system (T9SS) is required for F. columnare gliding motility and virulence. The T9SS and gliding motility machineries share some, but not all, components. GldN (required for gliding and for secretion) and PorV (involved in secretion but not required for gliding) are both needed for virulence, implicating T9SS-mediated secretion in virulence. The role of motility in virulence is uncertain. We constructed and analyzed sprB, sprF, and gldJ mutants that were defective for motility but that maintained T9SS function to understand the role of motility in virulence. Wild-type cells moved rapidly and formed spreading colonies. In contrast, sprB and sprF deletion mutants were partially defective in gliding and formed nonspreading colonies. Both mutants exhibited reduced virulence in rainbow trout fry. A gldJ deletion mutant was nonmotile, secretion deficient, and avirulent in rainbow trout fry. To separate the roles of GldJ in secretion and in motility, we generated gldJ truncation mutants that produce nearly full-length GldJ. Mutant gldJ₅₆₃, which produces GldJ truncated at amino acid 563, was defective for gliding but was competent for secretion as measured by extracellular proteolytic activity. This mutant displayed reduced virulence in rainbow trout fry, suggesting that motility contributes to virulence. Fish that survived exposure to the sprB deletion mutant or the gldJ₅₆₃ mutant exhibited partial resistance to later challenge with wild-type cells. The results aid our understanding of columnaris disease and may suggest control strategies.

IMPORTANCE

Flavobacterium columnare causes columnaris disease in many species of freshwater fish in the wild and in aquaculture systems. Fish mortalities resulting from columnaris disease are a major problem for aquaculture. F. columnare virulence is incompletely understood, and control measures are inadequate. Gliding motility and protein secretion have been suggested to contribute to columnaris disease, but evidence directly linking motility to disease was lacking. We isolated and analyzed mutants that were competent for secretion but defective for motility. Some of these mutants exhibited decreased virulence. Fish that had been exposed to these mutants were partially protected from later exposure to the wild type. The results contribute to our understanding of columnaris disease and may aid development of control strategies.

INTRODUCTION

Flavobacterium columnare causes columnaris disease in many species of wild and cultured freshwater fish and is a major problem for freshwater aquaculture worldwide (1–4). Infections and mortalities occur at many stages of development, from larval to adult (4–6). Infections typically occur on the skin, fins, and gills and result in substantial mortality. Lipopolysaccharides, proteins involved in iron acquisition, and secreted proteins such as proteases and adhesins are potential virulence factors (4, 7–13). The virulence mechanisms of F. columnare are incompletely understood, and control measures are inadequate.

F. columnare cells move across fish tissues by gliding motility, as observed over 100 years ago in the first publication on this bacterium (14). Gliding motility is widespread among members of the phylum Bacteroidota, to which F. columnare belongs (15, 16). F. columnare cells also glide on glass, agar, and other surfaces, and they form thin spreading colonies on agar (4, 7, 17, 18). Some mutants that form nonspreading colonies are avirulent, suggesting a link between motility and virulence (7, 17–20). However, the role of gliding in virulence is uncertain because motility mutants, such as gldN deletion mutants, are often also defective for protein secretion, which is known to impact virulence (7, 18).

Gliding has been well studied in the related bacterium, Flavobacterium johnsoniae (15). Gliding does not involve flagella or pili. Rather, the rapid movement of the major motility adhesin SprB along the outer membrane and the simultaneous attachment of SprB to the substratum move the cell across surfaces (21–23). Cells also have other motility adhesins that together with SprB are thought to facilitate gliding on many different surfaces (22). GldL and GldM are thought to constitute the proton gradient-driven rotary motor that powers movement of SprB and gliding motility (24–26). Additional Gld and Spr proteins are also involved in gliding (Fig. S1) (15). Among these, the periplasmic protein GldN and the outer membrane lipoprotein GldK are thought to form a complex with the cytoplasmic membrane-spanning GldLM motor, as the analogous proteins do in Porphyromonas gingivalis (27, 28). Another outer membrane lipoprotein, GldJ, may form the track along which SprB and the other motility adhesins are propelled (26, 29–31).

The gliding motility apparatus appears to be intertwined with the type IX protein secretion system (T9SS). Proteins such as GldK, GldL, GldM, and GldN, originally identified as required for gliding motility (32), were later determined to also be core components of the T9SS (33, 34). T9SSs are common in, but apparently confined to, the phylum Bacteroidota (16). T9SSs secrete proteases, adhesins, and many other proteins (15, 35, 36). Some of these are released in soluble form, whereas others, such as SprB, remain attached to the cell surface. Proteins secreted by T9SSs typically have conserved C- terminal domains (CTDs) of type A (TIGR04183) or type B (TIGR04131) (18, 37). Where it has been studied, cell surface localization of type A CTD proteins involves covalent attachment to lipopolysaccharide (38). Cell surface localization of SprB and other type B CTD proteins has not been as well studied, but it appears to involve interaction with specific outer membrane proteins (39). The GldLM motor is thought to function directly in both protein secretion and in cell movement (24–26), and loss of GldL or GldM or of some other Gld proteins results in failure to glide and defects in secretion (32, 33).

Although the Gld proteins mentioned above are required for both gliding and secretion, other proteins are specific to secretion or to gliding. PorV is involved in secretion of proteins that have type A CTDs, which are the most common proteins secreted by the F. johnsoniae and F. columnare T9SSs (7, 18, 40). Type A CTD proteins are diverse and include predicted enzymes and adhesins (18). PorV does not appear to be required for secretion of proteins that have type B CTDs, such as the motility adhesin SprB (40). porV mutant cells secrete SprB and propel it along the cell surface resulting in cell movement (40). Thus, deletion of porV eliminates secretion of most proteins by the T9SS but does not eliminate gliding. F. columnare porV deletion mutants are avirulent, suggesting that the T9SS is required for virulence (7, 18). It is not known which of the many proteins secreted by the T9SS are most critical for virulence. In contrast to PorV, SprB has a role in motility but not in secretion. F. johnsoniae sprB mutants are thus defective for motility but competent for secretion. Strains with mutations in other genes of the sprB operon such as sprC, sprD, and sprF (which are each thought to support SprB function) behave similar to sprB mutants in that they are defective for motility but competent for secretion (41). F. johnsoniae sprB mutants retain some residual motility because of the presence of other semiredundant motility adhesins, such as RemA (22).

Motility is a virulence factor of many pathogenic bacteria (42–44). Nonmotile gldN deletion mutants of F. columnare are avirulent (7, 18), but these mutants lack T9SS function in addition to lacking motility. Because of this, it is unknown if gliding motility contributes to F. columnare virulence. Here, we constructed and analyzed F. columnare sprB, sprF, and gldJ mutants that had motility defects, but that appeared to maintain T9SS function, to better understand the role of motility in virulence. Zebrafish (Danio rerio) and rainbow trout (Oncorhynchus mykiss) at multiple stages of development were challenged with wild-type and mutant F. columnare to determine the importance of motility in virulence for multiple fish species at different stages of development.

RESULTS

F. columnare cells lacking sprB or sprF are defective in gliding but competent for secretion

To investigate the role of motility in virulence, we constructed F. columnare deletion mutants ΔsprB and ΔsprF. In F. johnsoniae, SprB is secreted to the cell surface by the T9SS (33, 34, 45) and is the major motility adhesin that is propelled along the cell resulting in gliding motility (21, 23). SprF is thought to connect SprB to the rest of the motility machinery (39, 45). In F. johnsoniae, SprF is needed for secretion of SprB, but absence of SprF does not affect secretion of other proteins by the T9SS (41, 45).

The predicted F. columnare sprB operon (Fig. S2) contains six genes arranged in the order, remF, remG, sprC, sprD, sprB, sprF. This gene order is identical to that found in the characterized sprB operons of F. johnsoniae and Flavobacterium psychrophilum (41, 46, 47). F. johnsoniae, remF, remG, sprC, sprD, sprB, and sprF are each involved in gliding motility (23, 41, 47), but the corresponding F. columnare genes have not been studied.

The F. columnare ΔsprB and ΔsprF mutants were examined for gliding motility and for the ability to secrete peptidases. The previously described ΔporV mutant, which is deficient in T9SS-mediated secretion but is competent for motility, and ΔgldN mutant, which lacks both T9SS-mediated secretion and motility (18) were also examined. F. columnare ΔsprB and ΔsprF mutants were severely but incompletely compromised for motility. They formed nonspreading colonies similar to those of the completely nonmotile ΔgldN mutant (Fig. 1). Individual cells of the ΔsprB and ΔsprF mutants exhibited short movements (less than one cell length) on glass before reversing their direction and thus made little progress (Fig. 2; Movie S1). In contrast, wild-type cells and cells of the ΔporV mutant moved longer distances, and cells of the ΔgldN mutant failed to move. Complementation of the ΔsprF mutant by introducing a wild-type copy of sprF on a plasmid (ΔsprF_PC) or on the chromosome (ΔsprF_C) restored movement of cells on glass and spreading of colonies on agar (Fig. 1 and 2). Because of difficulties cloning sprB, which is very large and highly repetitive, an alternative approach was used to help determine if the observed phenotypes were due to the sprB mutation rather than to some unexpected spontaneous mutations elsewhere on the genome. For this purpose, the ΔsprB mutant was reconstructed starting from wild-type cells. The same defects in colony spreading and movement on glass were observed for reconstructed ΔsprB₂ mutant and for the original ΔsprB mutant (Fig. 1 and 2). The motility behaviors of the F. columnare mutants described above were similar to those of analogous F. johnsoniae mutants, except that F. johnsoniae porV deletion mutants move as well as the wild type (40), whereas the F. columnare porV mutants are partially defective in gliding (7, 18). This defect is illustrated by the decreased spreading of colonies of the F. columnare porV mutant compared to those of the wild type (Fig. 1) and by the observation that not all F. columnare porV mutant cells moved within 40 s (Fig. 2).

Fig 1.

Spreading of wild-type, mutant, and complemented F. columnare strains on agar. Cells were spotted on 1/4 tryptone yeast extract salts agar and incubated for 48 h at 30°C to form colonies. Strains examined were wild type (WT), ΔgldN, ΔporV, ΔsprB, ΔsprB₂ (reconstruction of ΔsprB), ΔsprF, ΔsprF complemented with wild-type sprF on plasmid pNT10 (∆sprF_PC), and ΔsprF complemented with wild-type sprF restored to the chromosome (∆sprF_C). Photomicrographs were taken with a Photometrics Cool-SNAP_cf² camera mounted on an Olympus SZ40 dissecting microscope. Bar indicating 1 mm applies to all images. Identical images for WT and ΔgldN appear in Fig. 4. WT and ΔgldN were previously characterized for spreading colonies (18), and they serve as controls that demonstrate the presence (WT) and absence (ΔgldN) of gliding motility.

Fig 2.

Gliding of wild-type and mutant F. columnare cells on glass coverslips. Cells were grown in tryptone yeast extract salts at 28°C to mid-exponential phase (OD₆₀₀ approximately 0.5). Ten microliters of cultures were introduced into a glass tunnel slide and observed for motility using an Olympus BH-2 phase-contrast microscope. Strains examined are indicated in the figure, with “ΔsprB₂” indicating a reconstruction of ΔsprB mutant and “∆sprF_PC” indicating ΔsprF complemented with wild-type sprF on plasmid pNT10. In each case, a series of images were taken using a Photometrics Cool-SNAP_cf² camera. Individual frames were colored from red (time 0) to yellow, green, cyan, and finally blue (40 s) and integrated into one image, resulting in “rainbow traces” of gliding cells. The top seven images (in grayscale) show the first frame for each strain. The bottom seven images show the corresponding 40-s rainbow traces. White cells correspond to cells that exhibited little if any net movement. Bars indicate 10 µm. Some of the rainbow traces correspond to the sequences in Movie S1 in the supplemental material. Identical images for WT appear in Fig. 5. WT and ΔgldN were previously characterized for gliding (18), and they serve as controls that demonstrate the presence (WT) and absence (ΔgldN) of gliding motility.

Many peptidases are secreted by the F. columnare T9SS (18), and thus, extracellular proteolytic activity is a convenient measure of T9SS function. Wild-type cells and cells of the ΔsprB and ΔsprF mutants produced similar levels of secreted proteolytic activity (Fig. 3A). In contrast, cells of ΔgldN and ΔporV mutants were deficient in secretion of peptidases (Fig. 3A), as has previously been reported (7, 18). Deletions of sprB and sprF thus affect motility but appear to have little impact on peptidase secretion. These results are similar to those obtained for F. johnsoniae sprB and sprF mutants, which were also partially defective for gliding but competent for secretion (41).

Fig 3.

Secreted proteolytic activity of wild-type and mutant strains. (A) Proteolytic activity in cell-free spent media of the strains indicated, with “ΔsprB₂” indicating a reconstruction of ΔsprB mutant. (B) Proteolytic activity in cell-free spent media of the strains indicated, with subscript “C” (∆gldJ_C and gldJ_558C) indicating mutants complemented with wild-type gldJ restored to the chromosome. Statistics correspond to one-way analysis of variance with Tukey post-test comparing all conditions to WT. ****P < 0.0001; ***P < 0.001; *P < 0.05; absence of star or “ns,” non-significant.

Involvement of F. columnare GldJ in gliding and secretion

T9SS-mediated secretion and gliding are intertwined, and both require core components of the T9SS (29). The results above suggest that F. columnare sprB and sprF mutants retain T9SS function but are partially defective in gliding motility. To further separate the roles of motility and secretion in F. columnare, we constructed mutants that were more severely defective for motility but that retained secretion. These were designed based on studies of F. johnsoniae that demonstrated that a ΔgldJ mutant lacked gliding and secretion, but that strains producing truncated but nearly full-length GldJ were defective in gliding and retained T9SS function (29). We constructed an F. columnare gldJ deletion mutant (ΔgldJ) and two truncation mutants (gldJ₅₅₈ and gldJ₅₆₃), which produce proteins containing the first 558 and 563 amino acids of GldJ, respectively (full-length F. columnare GldJ contains 573 amino acids). F. columnare gldJ₅₅₈ and gldJ₅₆₃ mutants correspond to F. johnsoniae mutants CJ2386 and CJ2443, respectively (29).

Each of the F. columnare gldJ mutants produced nonspreading colonies on agar (Fig. 4) indicating defects in gliding. The ΔgldJ and gldJ₅₅₈ mutants were completely nonmotile as assessed by lack of movement of cells on glass (Fig. 5; Movie S2). Cells of the gldJ₅₆₃ truncation mutant were also severely defective in gliding but exhibited slight movements. These phenotypes were also observed for the corresponding F. johnsoniae mutants (29). Complementation by introducing wild-type gldJ on a plasmid or by inserting gldJ into the chromosome restored gliding of cells and colony spreading to each mutant (Fig. 4 and 5; Movie S2).

Fig 4.

Spreading of wild-type, gldJ mutant, and complemented F. columnare strains on agar. Cells were spotted on 1/4 tryptone yeast extract salts agar and incubated for 48 h at 30°C to form colonies. Strains examined were WT, ΔgldN, ΔgldJ, truncation mutant gldJ₅₅₈, truncation mutant gldJ₅₆₃, gldJ mutants with wild-type gldJ restored to the chromosome (∆gldJ_C, gldJ_558C, and gldJ_563C), and gldJ mutants complemented with wild-type gldJ on plasmid pNT60 (∆gldJ_PC, gldJ_558PC, and gldJ_563PC). Photomicrographs were taken as in Fig. 1. Bar indicating 1 mm applies to all images. Identical images for WT and ΔgldN, which serve as motile (WT) and nonmotile (ΔgldN) controls, appear in Fig. 1.

Fig 5.

Gliding of wild-type and gldJ mutant cells on glass coverslips. Cells were cultured, introduced into glass tunnel slides, and observed for motility as in Fig. 2. Strains examined are indicated, with subscript “C” indicating mutants complemented with wild-type gldJ restored to the chromosome (∆gldJ_C, gldJ_558C, and gldJ_563C). The bottom seven images are 40-s rainbow traces of gliding cells, obtained as described in Fig. 2. White cells correspond to cells that exhibited little if any net movement. The top seven images (in grayscale) show the first frame for each strain. Bars indicate 10 µm. The rainbow traces correspond to the sequences in Movie S2 in the supplemental material. Identical images for WT, which serves as motile control, appear in Fig. 2.

The F. columnare gldJ mutants were examined for secretion defects (Fig. 3B). The ΔgldJ mutant was deficient for extracellular proteolytic activity but to a lesser extent than the ΔgldN mutant (Fig. 3B). In contrast, the gldJ truncation mutant gldJ₅₆₃ displayed extracellular proteolysis similar to wild type, and the gldJ₅₅₈ mutant exhibited levels intermediate between those exhibited by the wild type and the ΔgldJ mutant. Complementation of gldJ in both secretion-defective F. columnare mutants (ΔgldJ and gldJ₅₅₈) restored extracellular proteolytic activity to near wild-type levels (Fig. 3B).

Effect of mutations in sprB, sprF, and gldJ on adhesion

The T9SS secretes adhesins to the cell surface (7, 15, 18, 36). Cells of the T9SS-deficient mutants ∆gldN and ∆porV were defective in attachment to polystyrene (Fig. 6A), as previously reported (18). In contrast, the ∆sprB mutant cells attached to polystyrene similarly to the wild type (Fig. 6A). ∆sprF mutant cells showed a partial reduction in attachment to polystyrene, and complementation restored adhesion to wild-type levels (Fig. 6A). This suggests that SprF contributes directly or indirectly to adhesion. Cells of the ∆gldJ mutant were partially defective for secreted proteolytic activity (Fig. 3), but they attached to polystyrene similarly to wild-type cells (Fig. 6B). This contrasts with the severe attachment defect observed for the T9SS-defective ∆gldN and ∆porV mutants. The gldJ truncation mutants (gldJ₅₅₈ and gldJ₅₆₃) attached similarly to the wild type and the ∆gldJ mutant (Fig. 6). The T9SS is needed for efficient attachment to polystyrene, but we do not know what proteins or other molecules are responsible for this attachment.

Fig 6.

Adhesion of wild-type and mutant F. columnare strains. Adhesion to polystyrene after 3 h of incubation at 30°C without shaking as determined by staining with crystal violet and measuring absorbance at 595 nm. Adhesion shown in relation to WT, which was set as 100. Strains are indicated, with ΔsprB₂ indicating reconstruction of ΔsprB mutant, subscript “C” (∆sprF_C) indicating complementation with wild-type sprF restored to the chromosome, and the subscripts “558” and “563” indicating the length in amino acids of GldJ proteins produced by gldJ truncation mutants. Statistics correspond to one-way analysis of variance with Tukey post-test comparing all conditions to WT. ****P < 0.0001; absence of star: non-significant.

F. columnare cells also adhere to fish tissues (18, 48). Attachment of wild-type cells, T9SS-defective mutants (∆gldN and ∆porV), and gldJ mutants, each expressing GFP, to adult zebrafish pectoral fins was assessed by fluorescence microscopy (Fig. 7). As previously reported (18), many wild-type cells attached to fins, whereas ∆gldN mutant cells, which are completely deficient in T9SS function, and ∆gldJ mutant cells did not. Of six fins examined for each strain, hundreds or thousands of wild-type cells attached per fin, whereas no ∆gldN or ∆gldJ mutant cells were observed attached to fins. Previously, we showed that ∆porV mutant cells also failed to attach to fins (18). Here, using improved experimental methods (see Materials and Methods), we detected a few ∆porV mutant cells attached to fins (Fig. 7). porV mutants are partially defective in secretion, but they secrete type B CTD proteins, some of which may be adhesins (18, 40). Diverse adhesins secreted by the T9SS (including type A and type B CTD proteins) may contribute to attachment to fish fins. In contrast to the ∆gldJ mutant cells, cells of the truncation mutants, gldJ₅₅₈ and gldJ₅₆₃, readily attached to fins. The gldJ₅₆₃ mutant appeared to attach better than the gldJ₅₅₈ mutant. Complementation of ∆gldJ and gldJ₅₅₈ mutants resulted in increased attachment (Fig. 7). Cells of ∆sprB and ∆sprF mutants were also examined for attachment to fish fins. Fewer ∆sprB, ∆sprB₂, and ∆sprF mutant cells attached to fins in comparison to wild-type cells (Fig. S3) suggesting that SprB may play a role in attachment to fish fins. One or more of the many other proteins secreted by the T9SS (18) are also likely involved in attachment to fish fins because the T9SS-deficient ∆gldN mutant cells failed to attach.

Fig 7.

Adhesion to adult zebrafish fins of F. columnare wild-type, mutant, and complemented strains. Strains examined are indicated, with subscript “C” indicating mutants complemented with wild-type gldJ restored to the chromosome (∆gldJ_C and gldJ_558C), and subscripts “558” and “563” indicating the length in amino acids of GldJ proteins produced by gldJ truncation mutants. Cells of each strain carried pNT67, which expresses GFP. Live fish were exposed to the F. columnare strains for 60 min and rinsed twice in clean water, and pectoral fins were examined for attached cells by fluorescence microscopy. Fluorescence microscopy images and fluorescence overlayed on phase-contrast images are shown for each sample. Six fins were examined for each strain, and the experiment was repeated at least twice. Representative images are shown. Bar indicating 100 µm applies to all images.

SprB and GldJ are important for virulence in rainbow trout fry and zebrafish

Motility mutants were examined for virulence in juvenile rainbow trout (fry) (Fig. 8). As expected from previous studies (7, 18), rainbow trout fry were susceptible to challenge by wild-type F. columnare but were resistant to challenge with the ΔgldN mutant, which lacks T9SS function and gliding motility. Challenge with the ΔgldN mutant thus serves as a negative control, and lack of mortalities for fish challenged with the ΔgldN mutant provides confidence that mortalities observed for fish challenged with the other F. columnare strains occurred because of exposure to those strains. Furthermore, as indicated in the methods, 16% of fry mortalities were examined for F. columnare, and in each case, F. columnare was identified. The ΔsprB mutant, ΔsprB₂ mutant (reconstruction of ΔsprB), and ΔsprF mutant, which were each competent for secretion but partially defective for gliding, showed virulence defects. Complementation of the ΔsprF mutant restored virulence, with the caveat that the final challenge dose, as measured by colony-forming-units (CFU) per milliliter, was higher for the complemented strain (Fig. 8). ΔgldJ and gldJ₅₅₈ mutants exhibited strong virulence defects, killing almost no fish in challenges of rainbow trout fry. The gldJ₅₆₃ truncation mutant also caused few mortalities in fry. Complementation of both truncation mutants restored virulence to wild-type levels, with the caveat that the final challenge dose of the gldJ₅₆₃ truncation mutant was slightly lower than for the complemented strain (Fig. 8).

Fig 8.

SprB and GldJ are important for virulence in rainbow trout fry. Rainbow trout fry were exposed by immersion to F. columnare strains, and percent survival was monitored for 21 days. Strains examined and final challenge concentrations were: (A) wild type (WT; 1.4 × 10⁷ CFU/mL), ΔgldN mutant (1.4 × 10⁷ CFU/mL), ΔsprB (9.8 × 10⁶ CFU/mL), ΔsprB₂ (7.5 × 10⁶ CFU/mL), ΔsprF (5.5 × 10⁶ CFU/mL), ΔsprF complemented with wild-type sprF restored to the chromosome (∆sprF_C; 1.4 × 10⁷ CFU/mL); and (B) WT (1.4 × 10⁷ CFU/mL), ΔgldN (1.4 × 10⁷ CFU/mL), ΔgldJ (1.3 × 10⁷ CFU/mL), gldJ₅₅₈ truncation mutant (1.8 × 10⁷ CFU/mL), gldJ₅₅₈ with wild-type gldJ restored to the chromosome (gldJ_558C; 9.7 × 10⁶ CFU/mL), gldJ₅₆₃ truncation mutant (7.5 × 10⁶ CFU/mL), gldJ₅₆₃ with wild-type gldJ restored to the chromosome (gldJ_563C; 1.1 × 10⁷ CFU/mL). In both panels, fry were also mock challenged with growth medium lacking F. columnare (not shown), and these fish all survived. Statistical significance is indicated as P < 0.0001 (****) and P < 0.001 (***). “ns” indicates not significant.

Adult zebrafish were also challenged with motility mutants to test for virulence defects (Fig. 9). The ΔsprF mutant caused similar numbers of mortalities as wild type, whereas the ΔsprB mutant caused fewer deaths. Challenge with ΔgldJ mutant and with the truncation mutants gldJ₅₅₈ and gldJ₅₆₃ resulted in no fish deaths, similar to challenge with the ΔgldN mutant.

Fig 9.

SprB and GldJ are important for virulence in adult zebrafish. Zebrafish were exposed by immersion to F. columnare strains for 30 min at 26°C and transferred to fresh water, and percent survival was monitored for 10 days. Strains examined and final challenge concentrations were: (A) wild type (WT; 7.5 × 10⁵ CFU/mL), ΔgldN (5.2 × 10⁶ CFU/mL), ΔsprB (2.5 × 10⁶ CFU/mL); (B) WT (3.0 × 10⁶ CFU/mL), ΔgldN (6.0 × 10⁶ CFU/mL), ΔsprF (1.9 × 10⁶ CFU/mL); and (C) WT (2.5 × 10⁶ CFU/mL), ΔgldN (4.2 × 10⁶ CFU/mL), ΔgldJ (3.0 × 10⁶ CFU/mL), gldJ₅₅₈ truncation mutant (2.1 × 10⁶ CFU/mL), gldJ₅₆₃ truncation mutant (1.6 × 10⁶ CFU/mL). No mortalities were observed for any of the mutants examined in panel C. Fifteen fish were challenged with each strain as indicated in the Materials and Methods. Statistical significance is indicated as follows: P < 0.01 (**). “ns” indicates not significant.

F. columnare wild type and mutants were also examined for virulence in immature, developing fish, approximately 3 days post hatch. Fish in this age range are referred to as rainbow trout alevin (sac-fry) and zebrafish larvae, respectively. As previously reported (18), rainbow trout alevin were susceptible to challenge by the wild type, but not by the ΔgldN mutant (Fig. 10). The ΔsprB mutant showed virulence defects in rainbow trout alevin (Fig. 10), confirming that SprB is important for virulence. The ΔgldJ and gldJ₅₅₈ mutants were also deficient in virulence in alevin. In contrast, the ΔsprF mutant and the gldJ₅₆₃ mutant were as virulent as the wild type in rainbow trout alevin, whereas they exhibited virulence defects in fry (Fig. 8). Rainbow trout alevin are younger and less developed than fry, which may account for differences in the results of these challenge experiments.

Fig 10.

SprB and GldJ are important for virulence in rainbow trout alevin. Rainbow trout were exposed by immersion to F. columnare strains, and percent survival was monitored for 8 days. Strains examined and final challenge concentrations were: (A) wild type (WT; 6.9 × 10⁶ CFU/mL), ΔgldN (1.1 × 10⁷ CFU/mL), ΔsprB (6.1 × 10⁶ CFU/mL), ΔsprF (6.4 × 10⁶ CFU/mL); and (B) WT (6.9 × 10⁶ CFU/mL), ΔgldN mutant (1.1 × 10⁷ CFU/mL), ΔgldJ mutant (4.6 × 10⁶ CFU/mL), gldJ₅₅₈ truncation mutant (9.9 × 10⁶ CFU/mL), gldJ₅₆₃ truncation mutant (6.2 × 10⁶ CFU/mL). Statistical significance is indicated as P < 0.0001 (****) and P < 0.05 (*).

Germ-free zebrafish provided another window on F. columnare virulence. Germ-free zebrafish are highly sensitive to F. columnare infection, and they allow high-throughput analyses. They were previously used to determine the importance of microbiota in resistance to infection (5). Germ-free zebrafish larvae were challenged with wild-type and mutant F. columnare strains as previously described (18). Larvae exposed to ∆gldN and ∆gldJ mutants survived for the full 9 days post exposure, whereas larvae exposed to all other mutants or to the wild type died within 1 day after exposure (Fig. S4). ∆gldN and ∆gldJ mutants have major defects in T9SS secretion, which may explain their lack of virulence in zebrafish larvae. In contrast, the gldJ truncation mutants, ∆sprB mutant, and ∆sprF mutant, which are defective for motility but competent for secretion, retained virulence similar to the wild type when used to challenge otherwise germ-free zebrafish larvae.

Exposure to ΔsprB mutants or to truncation mutant gldJ₅₆₃ provided partial protection against later challenge with wild-type F. columnare

Survivors from the rainbow trout fry challenges shown in Fig. 8 were maintained for 31 days and examined for resistance to challenge with wild-type F. columnare. Survivors of exposure to ΔsprB and ΔsprB₂ mutants exhibited increased resistance to subsequent exposure to wild-type F. columnare (Fig. 11). Previous exposure to truncation mutant gldJ₅₆₃ also appeared to provide some protection against later challenge with wild-type cells. In contrast, fish that survived exposure to ΔsprF, ΔgldJ, or gldJ₅₅₈ were not resistant to later exposure to wild-type cells.

Fig 11.

Previous exposure of rainbow trout fry to ΔsprB and gldJ₅₆₃ mutants provided partial protection against later challenge with wild-type cells. Naïve fish and fish that survived exposure to mutant strains as shown in Fig. 8 were maintained for 31 days and then examined for resistance to challenge with wild-type F. columnare. (A) Naïve fish, ΔsprB survivors, ΔsprB₂ survivors, and ΔsprF survivors. (B) Naïve fish, ΔgldJ survivors, ΔgldJ₅₅₈ survivors, and ΔgldJ₅₆₃ survivors. In both panels, “Control” indicates exposure of naïve fish to sterile tryptone yeast extract salts (TYES) growth medium instead of to wild-type F. columnare in TYES growth medium. Wild-type F. columnare dose used was 3.8 × 10⁷ CFU/mL. Statistical significance is indicated as P < 0.0001 (****) and P < 0.05 (*). “ns” indicates not significant.

DISCUSSION

The Flavobacterium gliding motility apparatus and T9SS are intertwined, and many mutations that alter one process also affect the other. This complicated initial attempts to determine the roles of F. columnare secretion and motility in virulence. However, some mutations more specifically alter either secretion or motility. For example, mutations in porV disrupt secretion without having a major impact on motility. Analysis of F. columnare porV mutants suggested that the T9SS is required for virulence (7, 18). In contrast, the role of gliding in virulence remained uncertain. Here, we constructed mutations that compromised motility while apparently having little effect on secretion and examined the mutants for virulence defects. The phenotypes of the resulting mutants are summarized in Table S1.

SprB is the major Flavobacterium motility adhesin, and SprF is needed for SprB function (21, 23, 41). We deleted F. columnare sprB and sprF and demonstrated that each mutant was defective for motility but competent for secretion of peptidases. The ΔsprB and ΔsprF mutants exhibited reduced virulence in rainbow trout fry, and the ΔsprB mutant also exhibited reduced virulence in rainbow trout alevin. The gldJ₅₆₃ truncation mutant encodes nearly full-length GldJ, lacking just 10 amino acids. gldJ₅₆₃ was defective for motility but competent for secretion, as is the F. johnsoniae mutant (CJ2443) that has the equivalent gldJ truncation (29). Cells of the gldJ₅₆₃ mutant exhibited reduced virulence in rainbow trout fry, and complementation restored virulence. Other gldJ mutants (ΔgldJ and gldJ₅₅₈) were avirulent in rainbow trout fry, but these had defects in secretion and thus were less informative regarding the role of motility in virulence. Defects in secretion were expected for ΔgldJ and gldJ₅₅₈ because similar mutations in F. johnsoniae result in instability of the T9SS protein GldK, and thus loss of T9SS function (29). Cells of the gldJ₅₆₃ mutant attached readily to zebrafish fins, similarly to the wild type, whereas cells of the ΔgldJ mutant and of other T9SS-defective mutants (ΔgldN and ΔporV) were defective in attachment to zebrafish fins. Overall, these results indicate that the motility proteins SprB and GldJ are important for virulence. Moreover, since the sprB deletion and the gldJ₅₆₃ truncation did not appear to impact secretion and the gldJ₅₆₃ mutation did not disrupt attachment of cells to polystyrene or to fish fins, the decreased virulence in rainbow trout fry for these mutations may be a consequence of the lack of motility.

The sprB mutant and the gldJ mutants also exhibited decreased virulence in adult zebrafish, but the ΔsprF mutant was fully virulent. We do not know why the ΔsprF mutant was virulent in adult zebrafish. Whereas the ΔsprB mutant produces no SprB protein, the ΔsprF mutant is expected to accumulate SprB unnaturally in the periplasm, as do F. johnsoniae ΔsprF mutants (41). This could have unpredictable effects on the cell surface and on virulence of ΔsprF mutants. It might also explain other phenotypic differences between F. columnare ΔsprB and ΔsprF mutants, such as the differences in attachment to polystyrene.

Regarding the differences in susceptibility of adult zebrafish and rainbow trout fry to the ΔsprF mutant, there are many possible explanations. Most obviously, they are different species, and the many genetic differences between them could affect susceptibility. Furthermore, rainbow trout are coldwater fish, and they were challenged at 16°C, whereas zebrafish are tropical, and they were necessarily challenged in warm water (26°C). In addition, rainbow trout fry and adult zebrafish may differ in susceptibility to F. columnare because of the differences in age and maturity. Nevertheless, the results for rainbow trout fry and adult zebrafish indicate that the ΔsprB and gldJ₅₆₃ mutants are compromised for motility and for virulence, suggesting that motility may have a role in virulence.

Larval zebrafish were sensitive to all F. columnare mutants tested except for those that were severely deficient in T9SS function (ΔgldN and ΔgldJ). This was similar to the results observed for rainbow trout alevin, except that the ΔsprB mutant exhibited reduced virulence in alevin but not in zebrafish larvae. Both rainbow trout alevin and larval zebrafish have incompletely developed immune systems, and they are thus expected to be more sensitive to pathogens. A major difference between the rainbow trout alevin and zebrafish larvae used in this study is that the zebrafish larvae were germ free, whereas the rainbow trout alevin (and all other fish used here) had their normal microbiota. Germ-free zebrafish larvae are more susceptible to F. columnare presumably because they lack the normal microbiota that compete with pathogens and induce the developing immune system (5). It was recently reported that germ-free yolk sac fry of rainbow trout and Atlantic salmon (Salmo salar) are also more susceptible to F. columnare (49, 50). In addition to the difference that only the zebrafish larvae used in our studies were germ free, the challenge conditions were also different for zebrafish larvae and rainbow trout alevin, which could have affected susceptibility to F. columnare strains. Despite the susceptibility of zebrafish larvae to most mutants, they were completely resistant to the ΔgldN and ΔgldJ mutants, which are each compromised for T9SS activity. In contrast to adult zebrafish, zebrafish larvae were susceptible to the ΔsprB mutants and to the gldJ truncation mutants, each of which was compromised for motility. This suggests that motility was not needed to kill these immature fish but that a functional T9SS was required.

Rainbow trout fry that survived exposure to either the ΔsprB mutant or the gldJ₅₆₃ truncation mutant, which are both competent for secretion, were partially protected against later challenge with wild-type F. columnare. In contrast, fish exposed to the avirulent nonmotile mutants ΔgldJ and gldJ₅₅₈, which each had reduced levels of secretion, were not protected against later challenge with the wild type. This could be because these mutants did not secrete immunogenic proteins to the bacterial cell surface. Both of the mutants (ΔsprB and gldJ₅₆₃) that provided some protection against later exposure to the wild type were at least partially competent for attachment to fish fins. Attachment to fish tissues may be important to generate an immune response from the fish. However, other mutants that attached to fish fins (ΔsprF and gldJ₅₅₈) did not provide protection against later exposure to wild type, so attachment is not the only factor needed to provide protection. It may be possible to build on the vaccine potential of F. columnare mutants such as ΔsprB and gldJ₅₆₃ by reducing the virulence of these strains further while maintaining or increasing antigenicity.

Our studies implicating motility in virulence may be relevant to bacteriophage therapy, which is a promising strategy to control diseases caused by Flavobacterium fish pathogens, including F. columnare and F. psychrophilum (51, 52). Mutations in Flavobacterium gliding motility and T9SS genes often result in resistance to phages (33, 53, 54). Furthermore, many of the F. columnare and F. psychrophilum phage-resistant mutants that arise following exposure to phage are deficient for motility and/or T9SS function (55–58). Fortunately, these mutants also exhibit reduced virulence. The T9SS is required for virulence of both bacteria (7, 18, 59), explaining the reduced virulence of phage-resistant T9SS mutants. We provide evidence here that F. columnare motility also appears to contribute to virulence and suggest that defects in motility may explain the reduced virulence of some phage-resistant mutants that retain T9SS function. Phage-resistant T9SS or motility mutants that exhibit decreased virulence are unlikely to decrease the efficacy of phage therapy because they effectively steer the surviving F. columnare cells to a less virulent state.

Phage may also be used to select for spontaneous motility mutants to aid development of vaccines. The genetically engineered sprB and gldJ₅₆₃ motility mutants constructed here retained some virulence and provided partial protection of rainbow trout fry against later exposure to F. columnare. Phage selection followed by screening for defects in gliding, and maintenance of T9SS function, could result in a larger and more diverse collection of spontaneous motility mutants, some of which may have improved characteristics as potential vaccines.

The role of the Bacteroidota T9SS in virulence is well documented for diverse members of the phylum (7, 18, 34, 59–61). It has also been known for many years that some forms of bacterial motility have roles in virulence (42–44). However, it was not previously known if the gliding motility displayed by F. columnare and many other members of the phylum Bacteroidota is involved in virulence. The results presented here demonstrate that several F. columnare mutants that are compromised for motility but that appear to have functional T9SSs are partially deficient in virulence, suggesting that motility plays a role in virulence. Since gliding motility is widespread in the phylum Bacteroidota, including among pathogens, these results may have implications beyond F. columnare.

MATERIALS AND METHODS

Bacterial strains and growth conditions

The wild-type strain F. columnare MS-FC-4 was used in this study, and all mutants were derived from this strain. Strain MS-FC-4 was previously isolated from an outbreak in rainbow trout and belongs to F. columnare genetic group 1 (62–64). F. columnare strains were grown at 28°C for liquid cultures and at 30°C for agar cultures in tryptone yeast extract salts (TYES) medium as previously described (18, 65). F. columnare cultures used for rainbow trout challenges were grown in TYES-2xMg (9), which is identical to TYES except that it contains twice as much MgSO₄. E. coli strains were grown at 37°C in lysogeny broth (66). Antibiotics were used in the following concentrations when needed, unless otherwise specified: ampicillin, 100 µg/mL; tetracycline, 5 µg/mL for initial selection of strains and 1 µg/mL for routine culture of strains exhibiting tetracycline resistance; and tobramycin, 1 µg/mL. The strains and plasmids used in this study are listed in Table 1, and primers are listed in Table 2.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Descriptiona	Source or reference
E. coli strains
DH5αMCR	Strain used for general cloning	Life Technologies (Grand Island, NY)
S17-1 λ pir	Strain used for conjugation	(67)
F. columnare strains (all derived from the wild-type strain MS-FC-4)
MS-FC-4	Wild type	(62, 68)
FCB8	ΔporV; deletion of T9SS gene C6N29_02900	(18)
FCB14	ΔgldN; deletion of T9SS and motility gene C6N29_09600	(18)
FCB80	ΔsprF; deletion of C6N29_06255	This study
FCB82	ΔsprB; deletion of motility adhesin C6N29_06250	This study
FCB121	gldJ558; C6N29_02910 truncated at 558AA with engineered stop codon	This study
FCB129	ΔgldJ; deletion of T9SS and motility gene C6N29_02910	This study
FCB213	gldJ563; C6N29_02910 truncated at 563AA with engineered stop codon	This study
FCB214	ΔsprB2; reconstruction of ΔsprB (FCB82)	This study
FCB233	ΔgldJC; gldJ restored to the native site of ΔgldJ by chromosomal insertion with pNT74	This study
FCB234	gldJ558C; gldJ restored to the native site of gldJ558 by chromosomal insertion with pNT74	This study
FCB235	gldJ563C; gldJ restored to the native site of gldJ563 by chromosomal insertion with pNT74	This study
FCB236	ΔsprFC; sprF restored to the native site of ΔsprF by chromosomal insertion with pNT75	This study
Plasmids
pCP23	E. coli-F. columnare shuttle plasmid; Apr (Tcr)	(69)
pMS75	Suicide vector carrying sacB used to construct gene deletion mutants; Apr (Tcr)	(70)
pNT1	2.2-kbp region upstream of sprF amplified with primers 2138 and 2139 and inserted into pMS75; Apr (Tcr)	This study
pNT2	1.1-kbp region upstream of sprB amplified with primers 2134 and 2135 and inserted into pMS75; Apr (Tcr)	This study
pNT3	2.1-kbp region downstream of sprF amplified with primers 2140 and 2141 and inserted into pNT1; Apr (Tcr)	This study
pNT4	2.1-kbp region downstream of sprB amplified with primers 2136 and 2137 and inserted into pNT2; Apr (Tcr)	This study
pNT10	Plasmid for complementation of ∆sprF; 1.0-kbp region containing sprF amplified with primers 2187 and 2185 and inserted into KpnI and SphI sites of pCP23; Apr (Tcr)	This study
pNT48	2.0-kbp region upstream of gldJ for truncation at 558AA amplified with primers 2431 and 2432 and inserted into KpnI and BamHI sites of pMS75; Apr (Tcr)	This study
pNT49	2.0-kbp region upstream of gldJ for deletion amplified with primers 2435 and 2436 and inserted into KpnI and BamHI sites of pMS75; Apr (Tcr)	This study
pNT55	2.5-kbp region downstream of gldJ for truncation at 558AA amplified with primers 2433 and 2434 and inserted into BamHI and PstI sites of pNT48; Apr (Tcr)	This study
pNT56	2.6-kbp region downstream of gldJ for deletion amplified with primers 2437 and 2434 and inserted into BamHI and PstI sites of pNT49; Apr (Tcr)	This study
pNT60	Plasmid for complementation of ∆gldJ; 1.8 kbp region containing gldJ amplified with primers 2490 and 2491 and inserted into KpnI and PstI sites of pCP23; Apr (Tcr)	This study
pNT66	2.1-kbp region upstream of gldJ for truncation at 563AA amplified with primers 2431 and 2519 and inserted into KpnI and BamHI sites of pMS75; Apr (Tcr)	This study
pNT67	Expresses GFP in F. columnare. pOmpA and GFPMut1 from pAS43 amplified with primers 2578 and 2579A and inserted into BamHI and PstI sites of pCP23; Apr (Tcr)	(18)
pNT68	2.5-kbp region downstream of gldJ for truncation at 563AA amplified with primers 2433 and 2434 and inserted into BamHI and PstI sites of pNT66; Apr (Tcr)	This study
pNT74	Plasmid for chromosomal complementation of gldJ mutants; 6.2-kbp fragment spanning gldJ and 2-kbp regions upstream and downstream amplified with primers 2435 and 2434 and inserted into KpnI and PstI sites of pMS75; Apr (Tcr)	This study
pNT75	Plasmid for chromosomal complementation of ΔsprF; 5.0-kbp fragment spanning sprF and 2-kbp regions upstream and downstream amplified with primers 2138 and 2141 and inserted into XmaI and SphI sites of pMS75; Apr (Tcr)	This study

^a Antibiotic resistance phenotypes: ampicillin, Ap^r; tetracycline, (Tc^r). Unless indicated otherwise, the antibiotic resistance phenotypes are those expressed in E. coli. The antibiotic resistance phenotypes given in parentheses are those expressed in F. columnare but not in E. coli.

TABLE 2.

Primers

Primers used to construct plasmids
Primer	Sequence (5′ to 3′)a	Plasmids constructed
2134	GCTAGCCCGGGGAAATAGACAATTTGTCAAACGTT	pNT2
2135	GCTAGGGATCCATCTGCATTAAGACCATTTGTCTG	pNT2
2136	GCTAGGGATCCGGTAAAGAATTACCTTCTGGTGAT	pNT4
2137	GCTAGGCATGCACATCTATTAACTCAGTATCTGCTG	pNT4
2138	GCTAGCCCGGGAACAGGTGGTTATAATGTTCC	pNT1
2139	GCTAGGGATCCCAGGTGTAACTCCTGAGCAAT	pNT1
2140	GCTAGGGATCCGGTTACGCTTATCAAGTTACAACC	pNT3
2141	GCTAGGCATGCTACGGCACTAGATCGTAATGC	pNT3
2185	GCTAGGCATGCCTAATCTTCAGATAAAGAAT	pNT10
2187	GCTAGGGTACCTTTAACTTTTGAGTTATGACAG	pNT10
2431	GCTAGGGTACCTTCTGGCTGAAAATTAGGAAT	pNT48, PNT66
2432	GCTAGGGATCCCTA GCGAGACATAGCACATCTAAA	pNT48
2433	GCTAGGGATCCCATACTTTCGTTTAAAATAAAATCCCG	pNT55, pNT68
2434	GCTAGCTGCAGTTTGAATATACAGCACCTGCGTAA	pNT55, pNT56, pNT68, pNT74
2435	GCTAGGGTACCATAGCATCTAAGGTCACAACCT	pNT49, pNT74
2436	GCTAGGGATCCAGCAGATAAAAGCGCTCTTACT	pNT49
2437	GCTAGGGATCCTCTCGCGTAGGTCCTAAGTCAG	pNT56
2490	GCTAGGGTACCTTGCATCACTTTAAATAATTACCTACTG	pNT60
2491	GCTAGCTGCAGAGTATGTTATTTTCTATTTCTTGGTGT	pNT60
2519	GCTAGGGATCCCTATGACTTAGGACCTACGCGAGA	pNT66

^a Underlined sequences indicate introduced restriction enzyme sites.

Construction of deletion and truncation mutants

In-frame gene deletion mutants were constructed as previously described (7, 18, 70). These deletions leave upstream and downstream regions unaltered to limit the possibility of polar effects on downstream genes. To delete sprF, a 2.2-kbp region upstream of sprF was amplified by PCR using Phusion DNA polymerase (New England Biolabs, Ipswich, MA) and primers 2138 (adding an XmaI site) and 2139 (adding a BamHI site). The product was digested with XmaI and BamHI and ligated into pMS75 that had been digested with the same enzymes to produce pNT1. A 2.1-kbp region downstream of sprF was amplified using primers 2140 (adding a BamHI site) and 2141 (adding an SphI site). The product was digested with BamHI and SphI and ligated into pNT1 that had been digested with the same enzymes to generate pNT3. Plasmid pNT3 was transferred into F. columnare MS-FC-4 by conjugation from E. coli S17-1 λ pir as previously described (18), and colonies with the plasmid recombined into the chromosome were obtained by selecting for tetracycline resistance. Colonies were streaked for isolation on TYES containing tetracycline, and isolated colonies were grown in liquid without tetracycline to allow for plasmid loss. The cells were plated on TYES media containing 5% sucrose, and the mutant was obtained by selecting for sucrose resistance. PCR was performed to confirm the deletion. The other mutants were constructed in the same way, using the primers listed in Table 2 to obtain the appropriate upstream and downstream regions to construct the plasmids listed in Table 1.

Plasmid complementation of mutants

A 1.0-kbp fragment spanning sprF was amplified used primers 2187 (adding a KpnI site) and 2185 (adding a SphI site). The product was digested with KpnI and SphI and ligated into the shuttle vector pCP23 that had been digested with the same enzymes to produce pNT10. pNT10 was transferred into F. columnare by conjugation from E. coli as previously described (18). Plasmid complementation of other mutants was conducted in the same way, using the plasmids listed in Table 1 and the primers listed in Table 2.

Complementation of mutants by chromosomal insertion

A 5.0-kbp fragment spanning sprF and 2-kbp regions upstream and downstream was amplified using Q5 DNA Polymerase (New England Biolabs, Ipswich, MA) and primers 2138 (adding an XmaI site) and 2141 (adding an SphI site). The product was digested with XmaI and SphI and ligated into the suicide vector pMS75 that had been digested with the same enzymes to produce pNT75. pNT75 was transferred into the F. columnare sprF deletion mutant by conjugation. Colonies containing plasmid integrated into the chromosome were selected using tetracycline resistance (5 µg/mL tetracycline) and then grown without antibiotic to allow for plasmid loss. Sucrose-resistant colonies were screened using PCR to confirm the insertion of sprF into the native site. Chromosomal complementation of other mutants was conducted in the same way, using the plasmids listed in Table 1 and the primers listed in Table 2.

Analysis of colony spreading and cell motility

F. columnare cultures in mid-exponential phase of growth (OD₆₀₀ approximately 0.5) were spotted on 1/4 TYES agar (TYES medium diluted fourfold and solidified with 15 g/L agar) at a volume of 5 µL. Colonies of wild-type, mutant, and complemented strains were grown for 48 h at 30°C and then examined using a Photometrics Cool-SNAP_cf² camera mounted on an Olympus SZ40 dissecting microscope. Gliding of individual cells was also examined. Cells were grown with shaking at 28°C in TYES to mid-exponential phase. Tunnel slides were constructed using double-sided tape, glass microscope slides, and glass coverslips, as previously described (71). Ten microliters of cultures were introduced into the tunnel slides, incubated for 5 min, and cells attached to the cover slips were observed for motility using an Olympus BH2 phase-contrast microscope at 25°C. Images were recorded with a Photometrics CoolSNAP_cf² camera and analyzed using MetaMorph software (Molecular Devices, Downingtown, PA). Rainbow traces of cell movements were made using ImageJ version 1.45s (http://rsb.info.nih.gov/ij/) and macro Color FootPrint (21).

Adhesion assays

Adhesion of F. columnare cells to polystyrene was determined using crystal violet to stain attached cells, as previously described (18). In brief, cells were grown in half-strength TYES to an OD₆₀₀ of 0.6. For each sample, 1 mL of culture was centrifuged, the supernatant was removed, and the cells were suspended in 1-mL sterile Milwaukee municipal tap water. A 100-µL volume of suspended cells was added to a 96-well flat-bottom polystyrene plate (product no. 9017; Corning, Inc., Kennebunk, ME). The plate was incubated at 30°C for 3 h without shaking. Wells were washed twice with sterile distilled water, and adherent cells were stained with 100 µL of crystal violet (10 g/L) for 30 min at room temperature. The wells were washed four times with sterile distilled water. The remaining crystal violet was dissolved in 100 µL of absolute ethanol and detected by measuring OD₅₉₅ using a BioTek Synergy HTX microplate reader (BioTek Instruments, Inc., Winooski, VT). The level of adhesion observed for each strain was compared with the adhesion of the wild-type strain, which was set as 100. All assays were performed in quadruplicate, and the entire experiment was repeated three times with similar results. The absorbance of an uninoculated negative control was subtracted from the absorbance of each strain.

Attachment of F. columnare cells to zebrafish fins was observed using strains carrying pNT67, which expresses GFP. These experiments were performed as previously described (18) with the slight modifications described below. Strains were plated on TYES agar from −80°C stocks and incubated at 28°C for 48 h. Cells were grown overnight in 50-mL TYES with 1 µg/mL tetracycline at 28°C with shaking at 200 rpm. Five milliliters of overnight culture were diluted into 25 mL of fresh TYES and grown until mid-exponential phase (OD₆₀₀ = 0.5). Cells were examined by phase-contrast and fluorescence microscopy before use, and greater than 99% of cells were fluorescent. Adult Ekkwill zebrafish were immersed in 99-mL water to which 1-mL F. columnare cells (OD₆₀₀ = 0.5) were added. After 60 min at 26°C, the fish were transferred to 100 mL of fresh water, incubated for 10 min, transferred again to fresh water, and incubated for another 10 min. Fish were euthanized with buffered tricaine (MS-222), and pectoral fins were removed and examined by fluorescence microscopy as previously described (18). Six fins were examined for each strain, and the experiment was repeated at least twice.

Secreted proteolytic activity

Secreted proteolytic activity was measured using azocasein as a substrate as described (9). In brief, triplicate cultures were grown in TYES broth at 28°C, cells were removed by centrifugation, and the supernatant was filtered to remove the remaining cells. Protein concentration was determined for bacterial cell pellets, and proteolytic activity was determined on the cell-free spent culture fluid.

Rainbow trout challenges

Challenges of alevin and juvenile rainbow trout were performed as described in reference (18). Rainbow trout (Oncorhynchus mykiss) were reared from certified disease-free eggs from Troutlodge, Inc., Sumner, WA. Trout were maintained at the USDA-ARS National Center for Cool and Cold Water Aquaculture research facility in Kearneysville, WV, in flow through water at a rate of 1 L/min, at 12.5°C, until the challenge weight of ~1.3 g was met. The fish in this facility are checked yearly for multiple diseases including columnaris disease, and except for fish in the challenge room, they are certified disease free. No signs of disease were observed prior to challenge, and no indications of F. columnare or columnaris disease were observed in the uninfected control tanks or in the maintenance tanks. Fish were moved to challenge aquaria 1 week prior to immersion challenge to acclimate to the elevated water temperature of 16°C. F. columnare strains for challenges were cultured as previously described (63, 72). Briefly, cultures were incubated at 30°C with shaking at 150 rpm until OD₅₄₀ of 0.5–0.6 was reached, at which point the cells were used for the challenge. The wild-type and mutant F. columnare strains used in this study all grew at similar rates, and the inoculation doses were standardized and cultures monitored so that they reached the target OD₅₄₀ near the same time.

Challenges of fry were performed using triplicate 3-L tanks with restricted water flows (~200 mL/min) at 16°C. Each tank contained 40 fish of approximately 1.35 g each. Water flows were stopped for the immersion challenge, and tanks were inoculated with bacterial cultures and incubated for 0.5 h after which water flows were resumed. Control tanks were inoculated with sterile TYES-2xMg broth. Dilutions of water samples from each tank after inoculation were plated on TYES-2xMg agar to determine CFU per milliliter. Mortalities were removed and counted daily. The data for triplicate tanks of each strain were pooled, and survivor fractions for each strain were calculated. Challenges continued for 21 days or until 3 days without recorded mortalities post-exposure. In some cases, fry that survived an initial challenge with F. columnare mutants were maintained for 496 degree-days (31 calendar days) and were then challenged with wild-type F. columnare to determine if they had developed protective immunity. These challenges were performed in 3-L tanks essentially as described above, with approximately 30 fish (1–2 g each) per tank.

Challenges of alevin were performed as previously described (6). In brief, 100 eyed eggs were seeded for each 3-L tank, and alevin were challenged 3 days post hatch. For each strain, virulence was tested in triplicate tanks giving a starting n = 300. Hatch rates are >95% (according to producers’ standards and our logging of unhatched eggs) giving a final n ≥ 285, split between three tanks. Total CFUs are given in the figure legends. Mortalities were removed daily.

For alevin and fry challenges, approximately 16% of mortalities were randomly tested by homogenizing whole alevin or homogenizing fry gill tissue and streaking on TYES-2xMg agar to determine if F. columnare was present. Confirmation of F. columnare was determined by morphological observation of yellow, rhizoid, adherent colonies and by amplifying 16S rRNA genes and confirming the genomovar by enzymatic digestion (HaeIII) and gel electrophoresis as previously described (73–75). F. columnare was detected in all mortalities tested, except that F. columnare was not detected in the few mortalities that occurred in alevin that were not exposed to F. columnare or in alevin that were exposed to the F. columnare ΔgldN mutant, which has been repeatedly demonstrated to be completely avirulent (7, 18). A small amount of death unrelated to F. columnare infection is expected for alevin and explains these mortalities. The F. columnare detected from F. columnare-associated mortalities of fry and alevin were all genomovar I and genetic group 1, as expected for strain MS-FC-4 (62).

Challenges of adult zebrafish

Challenges of adult zebrafish were performed as described in reference (9). In brief, F. columnare strains were grown in TYES medium at 28°C until OD₆₀₀ reached 0.5. Cultures were diluted and plated on TYES agar to determine the number of live cells per milliliter. Naïve adult Ekkwill zebrafish were immersed in a solution of 0.5-mL F. columnare cells and 99.5-mL dechlorinated tap water for 30 min at 26°C. Control fish were exposed to 0.5-mL growth medium without F. columnare in 99.5 mL of water. After exposure, fish were moved to tanks with fresh water at 26°C and observed for up to 10 days for signs of infection. Each treatment was performed in triplicate tanks, with each tank containing five zebrafish in 2-L water. Mortalities were recorded daily. A minimum of 20% of the fish that died were examined for the presence of F. columnare by swabbing gills, fins, and skin; streaking on TYES agar containing tobramycin; and incubating for 2 days at 30°C. F. columnare colonies are rhizoid, yellow, and tobramycin resistant. F. columnare was detected in all mortalities tested. No signs of disease were observed prior to challenge, and no indications of F. columnare or columnaris disease were observed in the uninfected control tanks.

Challenge of germ-free zebrafish larvae

Germ-free zebrafish larvae were challenged with F. columnare strains at 28°C as previously described (5, 18). Briefly, 10–12 germ-free larvae (6 days post fertilization) were exposed to 10⁴ CFU/mL of washed F. columnare cells for 3 h in 25-cm³ culture flasks with vented caps containing 20 mL of sterile mineral water. The larvae were then transferred individually to 24-well plates containing 2-mL sterile water per well. Larvae were fed every 48 h with 50 µL of germ-free Tetrahymena thermophila per well. Mortalities were counted daily and measured in days post infection (dpi) with 0 dpi corresponding to the infection day. Zebrafish larval experiments were stopped at 9 dpi, and zebrafish were euthanized with tricaine (MS-222) (Sigma-Aldrich; catalog no. E10521). Each challenge included multiple replicates (n = 10–12 zebrafish for each condition) and was repeated twice.

Statistical analyses

A value of P < 0.05 was considered significant. For characterization assays, a one-way analysis of variance with Tukey’s post-test was used to analyze differences between treatment groups, unless otherwise noted. Error bars represent standard error of the mean. Kaplan-Meier survival analyses (76) were performed on fish challenge data. GraphPad Prism version 9.4.1 (GraphPad Software, LLC) was used to compute statistical tests. The results of statistical analyses are shown in the figures and figure legends where appropriate.

ETHICS APPROVAL

Experiments with adult zebrafish followed protocols approved by the University of Wisconsin-Milwaukee Institutional Animal Care and Use Committee. Larval zebrafish experiments were conducted according to European Union guidelines for handling of laboratory animals (http://ec.europa.eu/environment/chemicals/lab_animals/home_en.htm) and were approved by the Institut Pasteur Animal Health and Care Committee under permit #dap200043. Rainbow trout challenges were performed as described in Protocol #176, which was approved by the U.S. Department of Agriculture NCCCWA Institutional Animal Care and Use Committee.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/jb.00068-24.

Supplemental material. jb.00068-24-s0001.pdf.

Supplemental figures, table, and movie legends.

Movie S1. jb.00068-24-s0002.mov.

Movement of cells of wild-type F. columnare strain MS-FC-4, gldN mutant, porV mutant, sprB mutant, sprF mutant, and sprF mutant complemented with wild-type sprF on pNT10 on glass cover slips.

Movie S2. jb.00068-24-s0003.mov.

Movement of cells of wild-type F. columnare strain MS-FC-4, gldJ deletion mutant, truncation mutant gldJ558, truncation mutant gldJ563, and gldJ mutants complemented with wild-type gldJ restored to the chromosome were examined on glass cover slips.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.