Identification of the Flavobacterium johnsoniae cysteate-fatty acyl transferase required for capnine synthesis and for efficient gliding motility

Summary

Sulfonolipids (SLs) are bacterial lipids that are structurally related to sphingolipids. Synthesis of this group of lipids seems to be mainly restricted to Flavobacterium, Cytophaga and other members of the phylum Bacteroidetes. These lipids have a wide range of biological activities: they can induce multi-cellularity in choanoflagellates, act as von Willebrand factor receptor antagonists, inhibit DNA polymerase, or function as tumour suppressing agents. In Flavobacterium johnsoniae, their presence seems to be required for efficient gliding motility. Until now, no genes/enzymes involved in SL synthesis have been identified, which has been limiting for the study of some of the biological effects these lipids have. Here, we describe the identification of the cysteate-fatty acyl transferase Fjoh_2419 required for synthesis of the SL precursor capnine in F. johnsoniae. This enzyme belongs to the α-oxoamine synthase family similar to serine palmitoyl transferases, 2-amino-3-oxobutyrate coenzyme A ligase and 8-amino-7-oxononanoate synthases. Expression of the gene fjoh_2419 in Escherichia coli caused the formation of a capnine-derived molecule. Flavobacterium johnsoniae mutants deficient in fjoh_2419 lacked SLs and were more sensitive to many antibiotics. Mutant growth was not affected in liquid medium but the cells exhibited defects in gliding motility.

Introduction

Flavobacterium johnsoniae (formerly Cytophaga johnsonae) is a member of the large and diverse phylum of Gram-negative bacteria known as Bacteroidetes (McBride et al., 2009). Flavobacterium johnsoniae is a model organism to study gliding motility and cells of F. johnsoniae glide at rates of 2–10 μm s⁻¹ (Pate, 1988). These bacteria lack flagella or pili, and their movement relies on the presence of motility adhesins (Nelson et al., 2008; Shrivastava et al., 2012; Nakane et al., 2013). Several genetic screens have been performed to understand this type of motility. Members of this group of organisms frequently form membranes whose composition is unusual compared with the model bacterium Escherichia coli. Escherichia coli membranes are mainly composed of the phospholipids phosphatidylethanolamine (PE), phosphatidylglycerol (PG) and cardiolipin (CL) and this membrane composition is relatively stable even under different growth conditions (Raetz and Dowhan, 1990; Sohlenkamp and Geiger, 2016). Flavobacterium johnsoniae membranes when grown in common laboratory media are composed of sulfonolipids (SL), ornithine lipids (OLs), glycine lipids, serineglycine lipids (flavolipin) and PE (Pitta et al., 1989; Kawazoe et al., 1991; Sohlenkamp and Geiger, 2016).

SLs can be described as an unusual class of sphingolipids with a sulfonic acid group in the sphingoid base and are structurally related to ceramides. Based on this structural similarity, it was suggested earlier that the biosynthesis might work in an analogous manner (Geiger et al., 2010; Walker et al., 2017). The first step in sphingolipid synthesis is catalysed by serine palmitoyl transferase (Spt) using the substrates serine and acyl coenzyme A leading to the formation of 3-oxosphinganine (Hanada, 2003). Spt belongs to a sub-family of pyridoxal 5′-phosphate-dependent enzymes that includes 5-aminolevulinic acid synthase (HemA), 2-amino-3-oxobutyrate ligase (Kbl) and 8-amino-7-oxononanoate synthase (BioF) (Geiger et al., 2010). The 3-oxosphinganine is then reduced in an NADPH-dependent reaction to sphinganine and an N-acyltransferase catalyses the formation of N-acylsphinganine (dihydroceramide). Capnine is probably formed by the condensation of cysteate with isofatty acyl-CoA with the release of CO₂, in a reaction similar to the first step in sphingolipid synthesis (White, 1984; Geiger et al., 2010). Analogously, it has been proposed that SLs are produced by N-acylation of capnine with different isofatty acids (Geiger et al., 2010). It was observed that SLs are predominantly located in the outer membrane (Pitta et al., 1989). Two different SLs, sulfobacin A and sulfobacin B, which differ by the presence of a hydroxyl group, have been described in Flavobacterium sp. (Fig. 1) (Takikawa et al., 1999; Walker et al., 2017). SLs were reported in members of the genera Capnocytophaga, Cytophaga, Flexibacter, Flavobacterium and Sporocytophaga, but SLs were also reported in diatoms (Anderson et al., 1979; Godchaux 3rd and Leadbetter, 1983). A structurally related O-acetylated SL subclass has been described in the halophilic bacterium Salinibacter ruber (Corcelli et al., 2004; Baronio et al., 2010).

Fig 1.

Proposed biosynthesis of sulfonolipids in F. johnsoniae. CoA: coenzyme A.

Several biological activities have been ascribed in recent years to SLs. Sulfobacin B was described to inhibit DNA polymerase and to be an antagonist of inflammation. Sulfobacin A was shown to have potential as an anticancer agent and both seem to be von Willebrand factor receptor antagonists (Kamiyama et al., 1995a; Kamiyama et al., 1995b; Chaudhari et al., 2009; Maeda et al., 2010; Alegado et al., 2012; Woznica et al., 2016). A bacterial SL was shown to trigger multicellular development in choanoflagellates, the closest living relatives of animals (Alegado et al., 2012). Probably because of these diverse biological activities and the lack of knowledge of the genes/proteins involved in SL synthesis protocols for chemical synthesis of SLs were developed (Beemelmanns et al., 2014).

As a result of a screen of an MNNG-mutagenized population, a mutant deficient in SL synthesis was described that exhibited defects in gliding motility (Abbanat et al., 1986). By adding cysteate to the culture medium, SL formation and gliding motility were restored in this mutant. The genes involved in SL formation have not been described, but it is clear that the earlier-described mutants were affected in the synthesis of the SL precursor cysteate (White, 1984). In the absence of SLs, the amounts of OLs were drastically increased (Pitta et al., 1989). Furthermore, this mutant lacked an outer membrane polysaccharide (Godchaux 3rd et al., 1990), but polysaccharide accumulation in the outer membrane was also restored by adding cysteate to the growth medium (Godchaux 3rd et al., 1990).

Here, we present the identification of a cysteate-fatty acyl transferase Fjoh_2419 required for capnine synthesis. Capnine is an intermediate in the formation of the SLs sulfobacin A and B that are formed by F. johnsoniae and related bacteria. This is the first identification of a gene/protein involved in SL synthesis. Upon expression of this gene in E. coli small amounts of a capnine-like molecule were formed. Mutants lacking fjoh_2419 are deficient for gliding motility on agar and thus form non-spreading colonies, but they grow as wildtype in liquid media. Our identification of a gene involved in SL synthesis should make studies of the biological activity of this group of lipids easier.

Results

Membrane lipid analysis of F. johnsoniae

As a starting point for this project, we confirmed the membrane lipid composition of F. johnsoniae DSM2064 (also called UW101 or ATCC17061) that had been published (Holt et al., 1979). Cells were grown in liquid medium and [¹⁴C] acetate or [³⁵S] sulfate were added for labelling. Lipids were extracted according to Bligh and Dyer (1959) and separated by two-dimensional TLC as described in Experimental procedures. Acetate-labelling revealed six major lipids (Fig. 2A and C), two of which were labelled specifically with [³⁵S] (Fig. 2B and D). Total lipids were also analysed by LC–MS/MS confirming the presence of two different SLs, which had been named earlier sulfobacin A and sulfobacin B (data not shown). Based on earlier studies and on the relative mobilities the other lipid spots observed in the TLC were assigned as PE, OLs, glycine lipids and flavolipin (serineglycine lipid). To confirm that the [³⁵S]-labelled lipids were likely to be SLs, further control experiments were performed. Upon treatment of total lipids with phospholipase A, PE was hydrolyzed, whereas the putative SLs and aminolipids were not affected (Fig. 2C and D) and treatment with 0.5 N KOH also did not affect the putative SL (data not shown).

Fig 2.

Separation of [¹⁴C]acetate (A, C)- or [³⁵S]sulfate-labelled (B, D) lipids from F. johnsoniae wildtype DSM2064 grown in NM medium at 30°C by two-dimensional thin-layer chromatography (TLC). Samples (C) and (D) were treated with phospholipase A before TLC analysis. The lipids phosphatidylethanolamine (PE), sulfonolipids (SF, SF-a, SF-b), glycine lipids (GL), ornithine lipids (OL) and serineglycine lipids (SGL) are indicated.

Identification of candidate genes for the first step of F. johnsoniae sulfonolipid synthesis

SLs are structurally closely related to sphingolipids. The first step in sphingolipid synthesis leading to the formation of 3-oxosphinganine is catalysed by serine palmitoyltransferase (Spt) which belongs to the α-oxoamine synthase family of proteins (Geiger et al., 2010). Considering the structural similarities between capnine and sphinganine it is very likely that both molecules are synthesized by analogous reactions. Spts have been identified in the α-proteobacteria Sphingomonas wittichii RW1 (Raman et al., 2010) and we considered that a capnine synthase should belong to the α-oxoamine synthase family of proteins. The amino acid sequence of S. wittichii Spt was used to search the F. johnsoniae genome for homologous genes encoding putative α-oxoamine synthases and three candidate genes fjoh_0698, fjoh_0814 and fjoh_2419 were identified. They are predicted to code for proteins of 397, 378 and 419 amino acids respectively. Fjoh_0698 is annotated as 2-amino-3-oxobutyrate coenzyme A ligase (Kbl) and the other two candidate genes are annotated as 8-amino-7-oxononanoate synthases (BioF). We expected that one of three genes encodes the cysteate-fatty acyl transferase we are looking for and that the other two genes encode BioF, which catalyses the formation of 8-amino-7-oxononanoate from 6-carboxyhexanoyl-CoA and L-alanine during biotin synthesis and Kbl, which cleaves 2-amino-3-oxobutyrate into acetyl-CoA glycine during threonine degradation (Geiger et al., 2010). A phylogenetic tree of selected α-oxoamine synthase sequences was constructed to see if one of the three candidate genes from F. johnsoniae would group with known Spt sequences (Fig. 3). Indeed, Fjoh_2419 (ABQ05446) clustered with Spt sequences that have been shown to catalyse the first step in sphingolipid synthesis, such as the Spt from Sphingomonas wittichii.

Fig 3.

Unrooted phylogenetic tree of selected bacterial serine palmitoyltransferases, other α-oxoamine synthases and possible candidate genes encoding capnine synthase from F. johnsoniae. Distances between sequences are expressed as 0.05 changes per amino acid residue. Accession numbers are as follows: Bacteroides thetaiotaomicron VPI-5482 (NP_809783, NP_810284, NP_810356); Bacteriovorax stolpii (BAF73753); Caulobacter vibrioides CB15 (NP_419978, NP_420168, NP_420387); Escherichia coli B7A (EDV60350, WP_001297522, WP_001213834); Gluconobacter oxydans 621H (AAW61792, WP_011253160, WP_011252295); Granulibacter bethesdensis (WP_011630947, ABI61206, ABI61396); Nitrosomonas eutropha (WP_011633580); Porphyromonas gingivalis (WP_012458484, WP_012458311, WP_012457897); Sphingobacterium multivorum (BAF73751); Sphingomonas paucimobilis (BAB56013); Sphingomonas wittichii (WP_012050084, ABQ70245); Sphingomonas sp. (WP_012050007); Zymomonas mobilis (AAV89894, WP_011241021, WP_011241642); F. johnsoniae UW101 (Fjoh_0698: ABQ03733, Fjoh_0814: ABQ03848, Fjoh_2419: ABQ05446). Annotations are 8-amino-7-oxononoate synthase (BioF), 2-amino-3-oxobutyrate coenzyme A ligase (Kbl), 5-aminolevulinate synthase (HemA), and serine palmitoyltransferase (Spt).

Expression of gene fjoh_2419 in E. coli leads to capnine accumulation

The coding sequences of the three candidate genes were amplified by PCR, cloned in an expression plasmid and expressed in E. coli BL21(DE3).pLysS. Lipids were extracted and separated by 2D-TLC. In the E. coli wildtype strain only the phospholipids PE, PG and CL were present. The E. coli strains expressing Fjoh_0698 or Fjoh_0814 showed a wildtype-like lipid profile when analysed by TLC, whereas in the E. coli strain expressing Fjoh_2419 apparently an additional lipid was present in very minor amounts (data not shown). Using LC–MS/MS the presence of a capnine-like molecule was detected in the lipid sample from the E. coli strain expressing Fjoh_2419, which is probably derived from capnine by the loss of water (Fig. 4).

Fig 4.

Collision-induced dissociation mass spectra of sulfonolipids detected in lipid extracts of E. coli BL21(DE3).pLysS harbouring an empty vector control (A) or expressing Fjoh_2419 (B). The structure of a major fragment ion derived from sulfonolipids is indicated. Complete structures of the SLs are shown in Fig. 1.

Flavobacterium johnsoniae mutant deficient in fjoh_2419 lacks two different SLs, sulfobacin A and sulfobacin B

We constructed an F. johnsoniae mutant deficient in fjoh_2419. To confirm that Fjoh_2419 is involved in SL synthesis, we compared the lipid composition of the mutant and the wildtype grown in MM medium and in NM medium. Six major lipids could be detected by TLC in the wildtype, whereas the lipids earlier identified as SLs were missing in the mutant (Fig. 5). We also did a relative quantification of the membrane lipids based on [¹⁴C] labelling experiments (Table 1). In the wildtype grown in MM medium, OLs were the major lipids (35.3%), followed by SLs (23.2%), flavolipin (18.8%) and the unknown lipid Y (12.0%). Minor amounts were detected of PE (3.6%), glycine lipid (5.5%) and free fatty acids. In the wildtype grown in NM medium OLs were the major lipid (35.5%), but PE was now the second most abundant lipid (24.7%), followed by SLs (19.0%) and glycine lipids (10.5%). Minor amounts were detected of flavolipin (4.7%), free fatty acids and unknown lipid Y. The absence of SLs in the mutant was apparently compensated by a drastic increase of OLs under both growth conditions. The other remarkable change was a decrease in flavolipin in the mutant grown in MM medium. Total lipids were also analysed by LC–MS/MS confirming the presence of two SLs, which were named earlier sulfobacin A and sulfobacin B, in lipids extracted from the wildtype (Fig. 5E) (Takikawa et al., 1999). Sulfobacin A has an additional hydroxyl group in the C3 position of the amidified fatty acid (Fig. 1). No SLs were detected in the F. johnsoniae mutant deficient in fjoh_2419 (Fig. 5F), but introducing a plasmid-borne copy of fjoh_2419 in trans into the mutant strain led again to SL formation (Fig. 5G).

Fig 5.

Analysis of the membrane lipid composition of F. johnsoniae wildtype and mutant ΔFjoh_2419 and complementation of mutant ΔFjoh_2419 by a fjoh_2419-haboring plasmid. Separation of [¹⁴C]acetate-labelled lipids from F. johnsoniae wildtype DSM2064 (A, C) or F. johnsoniae mutant ΔFjoh_2419 (B, D) grown in NM medium (A, B) or in MM medium (C, D) at 30°C by two-dimensional thin-layer chromatography (TLC). A quantification of the lipid is presented in Table 1. (E–G) Total lipids of F. johnsoniae wildtype (E), F. johnsoniae mutant ΔFjoh_2419 (F) or F. johnsoniae mutant ΔFjoh_2419 and complemented by a copy of fjoh_2419 in trans (G) were separated by LC. Summed-up full scan mass spectra obtained in negative ionization mode from the elution interval from 2 to 6 min are shown. This corresponds to the m/z area where sulfonolipids are detected if present. The lipids phosphatidylethanolamine (PE), sulfonolipids (SF), glycine lipids (GL), ornithine lipids (OL) and serineglycine lipids (SGL) are indicated. Y: unknown lipid.

Table 1.

Membrane lipid composition of F. johnsoniae DSM2064 wildtype and the mutant ΔFjoh_2419 deficient in SL formation after growth in either NM medium or in MM medium at 30°C.

Lipids	F. johnsoniae NM	F. johnsoniae Δfjoh_2419 NM	F. johnsoniae MM	F. johnsoniae Δ fjoh_2419 MM
SGL	4.7 ± 0.1	6.7 ± 0.3	18.8 ± 0.1	7.8 ± 0.1
OL	35.5 ± 1.3	52.2 ± 0.1	35.3 ± 0.3	63.7 ± 0.2
PE	24.7 ± 1.5	19.9 ± 0.4	3.6 ± 1	4.7 ± 0.4
SL	19.0 ± 0.8	ND	23.2 ± 0.5	ND
GL	10.5 ± 0.5	11.9 ± 0.8	5.5 ± 0.5	7.5 ± 0.1
FA	3.0 ± 0.1	4.7 ± 0.4	1.6 ± 0.3	3.5 ± 0.5
Y	2.6 ± 0.1	4.6 ± 0.2	12.0 ± 1	12.8 ± 0.9

The values shown are mean values ± standard deviation derived from at least three independent experiments. SGL, serineglycine lipid (flavolipin); OL, ornithine lipid, PE, phosphatidylethanolamine; SL, sulfonolipid; GL, glycine lipid; FA, fatty acids; Y, unknown lipid; ND, not detected. ND in the context of this experiment means that no radioactivity above background could be detected.

Mutants deficient in fjoh_2419 are not affected during growth in liquid medium but lose gliding motility

Flavobacterium johnsoniae wildtype and the mutant deficient in fjoh_2419 were grown in two different liquid media to test if the lack of SL and the derived modifications of the membrane composition would affect growth. In both media tested, the mutant deficient in fjoh_2419 grew similar as the wildtype (Fig. 6A). Modification in the membrane lipid composition frequently lead to changes in resistance towards antibiotics and we, therefore, compared the sensitivity of mutant and wildtype to a range of antibiotics. The mutant was clearly more sensitive to many of the antibiotics tested: it was sensitive to vancomycin, neomycin, kanamycin and gentamicin (Fig. 6B), all of which did not affect the growth of the wildtype under the concentrations used. Also, the mutant showed an increased inhibition of growth compared with the wildtype in the presence of chloramphenicol, erythromycin, streptomycin and nalidixic acid. Finally, we were interested to see if the mutant deficient in the capnine synthase Fjoh_2419 would be affected in gliding motility. Dilutions of cell suspensions were plated onto MM plates and the plates were incubated for 5 days. The wildtype grew to form large spreading colonies, whereas the mutant only grew to very small non-spreading colonies (Fig. 6C and D). The lack of spreading of the mutant probably limited its access to nutrients contributing to the observed growth defect on the low nutrient agar medium.

Fig 6.

Characterization of mutant ΔFjoh_2419 deficient in SL formation.</p>A. Growth of F. johnsoniae DSM2064 wildtype and F. johnsoniae mutant deficient in the cysteate synthase ΔFjoh_2419 in liquid culture in nutrient medium (NM) or mobility medium (MM). Wildtype DSM2064 in NM medium-horizontal bars ( ), wildtype DSM2064 in MM medium-squares (■), mutant ΔFjoh_2419 in NM medium-diamonds (♦), mutant ΔFjoh_2419 in MM medium-triangles (▲). The optical density was measured at 620 nm. The experiment was repeated three times independently.</p>B. The mutant ΔFjoh_2419 is more sensitive to several antibiotics than the F. johnsoniae wildtype DSM2064. Inhibition halos were measured after 3 days. The data shown are the average of three independent experiments. Error bars indicate standard deviations.</p>W5: trimethoprim 5; C30: chloramphenicol 30; VA30: vancomycin 30; N30: neomycin 30; K30: kanamycin 30; E15: erythromycin 15; DA2: clindamycin 2; S10: streptomycin 10; F300: nitrofurantoin 300; TOB10: tobramycin 10; NA30: nalidixic acid 30; CN10: gentamicin 10; TE30: tetracyclin 30; B10: bacitracin 10; AMP10: ampicillin 10. The numbers indicate the amount per antibiotic per disc in micrograms. Black columns: F. johnsoniae DSM2064 wildtype, grey columns: ΔFjoh_2419 mutant. Error bars indicate standard deviation.</p>C, D. Gliding phenotype on plate. Diluted cells suspensions of F. johnsoniae wildtype DSM2064 (C) and F. johnsoniae mutant ΔFjoh_2419 (D) were plated on MM plates and incubated for 5 days at 30°C.

We then looked at the differences in the gliding phenotype between the mutant and wildtype colonies in more detail. Both strains were spotted onto a MM agar plate and photos were taken every 10 min and compiled into a movie to observe the behaviour of both colonies over time (Supplementary movie 1). Photos from 12, 24, 36, 48 and 72 h are shown in Fig. 7. The mutant shows the growth of the colony and possibly a very minor irregular shape, whereas the wildtype shows a flame- or whip-like movement and irregular shape. When the mutant was complemented with the plasmid pCP.fjoh_2419, the colony spreading phenotype of the wildtype was recovered (Fig. 8A), confirming the importance of SLs for gliding motility in F. johnsoniae.

Fig 7.

Montage obtained from a time-lapse movie (Supplementary Movie 1) comparing growth and motility patterns of F. johnsoniae wildtype (WT) and the F. johnsoniae mutant ΔFjoh_2419 (Δ). Each frame corresponds to a different time-point and illustrates the distribution of both strains in time and space. Optical density values were obtained by subtracting the intensity of each pixel to the corresponding value of the first image (i.e. black corresponds to a background pixel and white to the largest optical density value observed in the experiment).

Fig 8.

Gliding motility of the mutant strain complemented with a plasmid-borne copy of fjoh_2419.</p>A. Image acquired after 72 h of growth in a semi-solid MM media petri dish.</p>B. Microscopy image showing complemented mutant cells mounted directly into agar pads from the edge of the growing colony to be observed using differential interface contrast.</p>C. Trajectories of individual cells tracked in a 90 s time-lapse movie show gliding motility is characterized by back-and-forth movement with occasional pivoting that changes the direction of movement (Supplementary movie 2). Each cell is represented with a different colour.

We also examined the motility of individual cells under the microscope. We followed the ‘tunnel slide’ protocol and observed gliding motility (McBride, 2014), both in the wildtype cell and the complemented mutant, but the mutant strain apparently almost had completely lost its motility with only a very small number of cells being motile. We also mounted cells from the edge of the spreading colony formed by the complemented mutant directly from solid media into agar pads, following a previously published protocol (Heering et al., 2017) and observed motile cells under the microscope. In both experimental protocols, cells attached to the glass exhibited gliding motility characterized by back-and-forth movements with occasional pivoting that suddenly changes the direction of movement, as shown in Fig. 8C and the Supplementary Movie 2.

Discussion

Flavobacterium johnsoniae (formerly Cytophaga johnsonae) is a member of the large and diverse phylum of Gram-negative bacteria known as Bacteroidetes. In addition to being a model organism to study gliding motility, F. johnsoniae presents a very unusual membrane lipid composition with only one phospholipid among the major membrane-forming lipids. Its membranes are composed of SLs, OLs, glycine lipids, serineglycine lipids (flavolipin) and PE. Earlier, chemical mutants deficient in SL synthesis had been described and characterized (Abbanat et al., 1986). Based on the phenotypes, two of these mutants were affected in cysteate formation and one mutant was affected in the N-acylation step leading to SL formation. These mutants had not only lost the ability to form SL, but they were also deficient in gliding motility and apparently formation, transport or maturation of an outer membrane polysaccharide was affected. For the mutant deficient in cysteate formation, all these phenotypes could be restored simply by adding cysteate to the growth medium. The site of the mutations and the genes affected were not determined.

Walker et al. (2017) had identified SLs in their metabolomics data describing the gut metabolome of mice. Within their metagenomics data, two bacterial gene homologues including 2-amino-3-oxobutyrate coenzyme A ligase (K00639) and 8-amino-7-oxononanoate synthase (K00652), but no genes encoding Spt were identified. Following the idea that the capnine synthase should belong to oxo-amine synthase family, we identified candidate genes in F. johnsoniae. The expression of fjoh_2419 in E. coli led to the accumulation of a new lipid that upon mass spectrometry analysis appeared to be a capnine having lost a molecule of water. There might be several reasons why only a low production of capnine (or of a capnine-derived molecule) was observed in E. coli: (i) Low availability of cysteate inside the E. coli cells, (ii) Absence of the possibly preferred branched-chain fatty acids in E. coli and (iii) Requirement of a specific donor of activated fatty acids not present in E. coli. According to the literature, cysteate is not taken up into E. coli BL21(DE3), and also, we assume that E. coli has few branched-chain fatty acids. The mutant deficient in Fjoh_2419 showed similar phenotypes as the previously described chemically induced mutant deficient in the unknown gene encoding cysteate synthase. It lacks SLs, it grows as the wildtype in liquid media and it is deficient in gliding motility (Abbanat et al., 1986). The deficiency in gliding motility was observed as a lack of colony spreading at the macroscopic level and also as reduced motility at the microscopic level. Earlier, it had been observed that mutants lacking SLs did not accumulate a polysaccharide in the outer membrane. We did not look for differences in the outer membrane between fjoh_2419 and wildtype, but we noticed that the mutant was very sensitive to a wide range of antibiotics which is consistent with the lack of an outer membrane polysaccharide leading to increased permeability of the membrane. It had been concluded that the absence of SL somehow prevents the synthesis, transport or assembly of this polysaccharide or its retention in the membrane. This polysaccharide might be indirectly involved in gliding motility. For example, its presence might be needed for proper functioning of outer membrane components of the motility machinery, or of the type IX secretion system, which is required for motility. Our observations are consistent with this hypothesis.

Diverse biological roles have been assigned to SLs such as an antagonistic action on von Willebrand factor receptors, cytotoxicity in cancer cells, inhibition of DNA polymerase and anti-inflammatory effects. Our identification of the first gene involved in SL synthesis is an important step forward: it should allow the construction of mutants for example in the bacteria that causes the multi-cellularity of choanoflagellates and the selective synthesis of pathway intermediates.

Experimental procedures

Strains and growth media used

The bacterial strains and plasmids used, and their relevant characteristics, are shown in Supplementary Table S1. Flavobacterium johnsoniae DSM2064/UW101/ATCC17061 was obtained as a freeze-dried stock from DSM in Braunschweig, Germany. Flavobacterium johnsoniae was grown in nutrient medium (NM) or motility medium (MM) media at 30°C. NM (pH 7) contains 10 g peptone, 10 g meat extract and 15 g agar per litre (Stanier, 1947), MM contains 1.7 g yeast extract, 3.3 g casitone, 3.3 mM Tris–HCl, pH 7.5 and 15 g agar per litre (Liu et al., 2007). Escherichia coli strains were grown in Luria–Bertani (LB) medium at 37°C (Green et al., 2012). When needed, antibiotics were added at the following final concentrations (μg ml⁻¹): kanamycin (Km) 50; carbenicillin (Cb) 100; tetracycline (Tc) 10; chloramphenicol (Cm) 60; erythromycin (Em) 60 and streptomycin (Sm) 50.

DNA manipulations

Recombinant DNA techniques were performed according to standard protocols (Green et al., 2012). Oligonucleotide primer sequences are listed in Supplementary Table S2.

In vivo labelling of F. johnsoniae with [¹⁴C]acetate or [³⁵S]sulfate and quantitative analysis of lipid extracts

The lipid compositions of bacterial strains were determined following labelling with [1-¹⁴C]acetate or [³⁵S]sulfate (Amersham Biosciences). Cultures (1 ml) of wildtype and mutant strains were inoculated from pre-cultures grown in the same medium. After addition of 0.5 μCi of [¹⁴C]acetate (60 mCi mmol⁻¹) or [³⁵S]sulfate to each culture, the cultures were incubated for 4 h. The cells were harvested by centrifugation, washed with 500 μl of water once, resuspended in 100 μl of water and then lipids were extracted according to Bligh and Dyer (1959). Aliquots of the lipid extracts were spotted on high-performance TLC silica gel 60 plates (Merck, Poole, UK) and were separated in two dimensions using chloroform/methanol/ammonium hydroxide (140:60:10, vol./vol./vol.) as a mobile phase for the first dimension and chloroform/methanol/glacial acetic acid/acetone/water (130:10:10:20:3, vol./vol./vol./vol./vol.) for the second dimension. To visualize the membrane lipids, developed two-dimensional TLC plates were exposed to autoradiography film (Kodak) or to a PhosphorImager screen (Amersham Biosciences). The individual lipids were quantified using ImageQuant software (Amersham Biosciences) (Vences-Guzman et al., 2011).

Liquid chromatography/tandem mass spectrometry analysis of lipid samples

A 1 L culture of E. coli BL21(DE3).pLysS expressing Fjoh_2419 was grown to an OD of 1.2 at 620 nm in LB medium. Cells were harvested by centrifugation and lipids extracted according to Bligh and Dyer (1959). When F. johnsoniae lipids were studied, 1 L cultures F. johnsoniae wildtype, F. johnsoniae Δfjoh_2419 and F. johnsoniae Δfjoh_2419 complemented with a copy of fjoh_2419 in trans were grown to an optical density of 1.2 at 620 nm in LB medium, and again lipids were extracted according to Bligh and Dyer (1959).

Normal phase LC-ESI MS of the lipid extracts was performed using an Agilent 1200 Quaternary LC system coupled to a high-resolution TripleTOF5600 mass spectrometer (Sciex, Framingham, MA). Chromatographic separation was performed on an Ascentis Silica HPLC column, 5 μm, 25 cm × 2.1 mm (Sigma-Aldrich, St Louis, MO). Elution was achieved with mobile phase A, consisting of chloroform/methanol/aqueous ammonium hydroxide (800:195:5, vol./vol./vol.), mobile phase B, consisting of chloroform/methanol/water/aqueous ammonium hydroxide (600:340:50:5, vol./vol./vol./vol.) and mobile phase C, consisting of chloroform/methanol/water/aqueous ammonium hydroxide (450:450:95:5, vol./vol./vol./vol.), over a 40 min-long run, performed as follows: 100% mobile phase A was held isocratically for 2 min and then linearly increased to 100% mobile phase B over 14 min and held at 100% B for 11 min. The mobile phase composition was then changed to 100% mobile phase C over 3 min and held at 100% C for 3 min, and finally returned to 100% A over 0.5 min and held at 100% A for 5 min. The LC eluent (with a total flow rate of 300 μl min⁻¹) was introduced into the ESI source of the high-resolution TF5600 mass spectrometer. MS and MS/MS were performed in negative ion mode, with the full-scan spectra being collected in the m/z 200–2000 range. The MS settings are as follows: Ion spray voltage (IS) = −4500 V (negative ion mode), Curtain gas (CUR) = 20 psi, Ion source gas 1 (GS1) = 20 psi, De-clustering potential (DP) = −55 V and Focusing Potential (FP) = −150 V. Nitrogen was used as the collision gas for tandem mass spectrometry (MS/MS) experiments. Data analysis was performed using Analyst TF1.5 software (Sciex).

Cloning and expression of F. johnsoniae candidate genes in E. coli BL21(DE3).pLysS strain

The candidate ORFs fjoh_0698, fjoh_0814, and fjoh_2419 were amplified using genomic DNA from F. johnsoniae wildtype DSM2064 as a template. Specific oligonucleotide primers Fjoh_0698F, Fjoh_0698R, Fjoh_0814F, Fjoh_0814R, Fjoh_2419F and Fjoh_2419R incorporating NdeI and BamH1 sites into the PCR product were used. Oligonucleotide used in this study is listed in Supplementary Table S2. After digestion with the respective enzymes, the PCR products were cloned as NdeI/BamH1 fragment into pET9a yielding (pET9a. fjoh_0698, pET9a.fjoh_0814 and pET9a.fjoh_2419).

Construction of F. johnsoniae mutants ΔFjoh_2419 and complementation of the mutant

Oligonucleotide primers Fjoh241901 and Fjoh241902 were used in a PCR to amplify about 1.0 kb of genomic DNA upstream of gene fjoh_2419 encoding the cysteate synthase from F. johnsoniae, introducing EcoRI, Not1 and BamHI sites into the first PCR product. Similarly, primers Fjoh241903 and Fjoh241904 were used to amplify about 1.0 kb of genomic DNA downstream of fjoh_2419 gene, introducing BamHI and XbaI sites into the second PCR product. Oligonucleotide used in this study is listed in Supplementary Table S2. After digestion with the respective enzymes, PCR products were cloned as EcoRI/BamHI or BamHI/XbaI fragments into pUC19 to yield the plasmids pUC01-SL and pUC02-SL respectively. Then, the BamHI/XbaI fragment from pUC02 was subcloned into pUC01 to yield pUC03-SL. Plasmid pUC03-SL was digested with NotI and XbaI to subclone the regions usually flanking the fjoh_2419 gene into the suicide vector pYT354 (Zhu et al., 2017) to yield pYT01. Via diparental mating using E. coli S17–1 as a mobilizing strain, pYT01 was introduced into the wildtype strain F. johnsoniae. Transconjugants were selected on NM medium containing erythromycin to select for single recombinants in a first step. The plasmid pYT354 contains the sacB gene (Zhu et al., 2017), which confers sucrose sensitivity to many bacteria. Growth of the single recombinants on high sucrose will therefore select for double recombinants and the loss of the vector backbone of pYT354 from the bacterial genome. Single recombinants were grown under non-selective conditions in complex medium for 1 day before being plated on NM medium containing 12.5% (wt./vol.) sucrose. Several large and small colonies grew after 4 days, and the membrane lipids of eight candidates were analysed by in vivo labelling during growth on complex medium with [¹⁴C]acetate and subsequent TLC (data not shown). PCR analysis confirmed that the gene fjoh_2419 was deleted in the mutant (data not shown).

The gene fjoh_2419 was amplified by PCR using the oligonucleotides Fjoh_2419Fc and Fjoh_2419R, which introduced KpnI and BamHI restriction sites. The PCR fragment was digested with the respective enzymes and cloned into the expression vector pCP23 (Agarwal et al., 1997) yielding plasmid pCP.fjoh_2419. This plasmid was mobilized via diparental mating into the mutant ΔFjoh_2419.

Construction of a phylogenetic tree by the Neighbour-Joining method

The evolutionary history was inferred by using the Maximum Likelihood method and JTT matrix-based model (Jones et al., 1992). The tree with the highest log likelihood (−9271.58) is shown. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbour-Join and BioNJ algorithms to a matrix of pairwise distances estimated using a JTT model, and then selecting the topology with superior log likelihood value. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. This analysis involved 31 amino acid sequences. There were a total of 218 positions in the final dataset. Evolutionary analyses were conducted in MEGA X (Kumar et al., 2018).

Antibiogram

Flavobacterium wildtype and mutant strains were grown with agitation at 30°C in 5 ml cultures in liquid NM medium to an optical density of 1.1 at 620 nm. From these cultures, 100 μl were transferred to tubes with 5 ml semisolid NM agar (8 g of agar per litre), the temperature of which had been adjusted to 42°C in a water bath. The tubes were briefly vortexed and the medium was poured onto pre-warmed standard NM plates. After the semi-solid medium had been solidified, four antibiotics containing paper discs (Oxoid) were placed onto each plate. The plates were incubated at 30°C and the diameter of the halos surrounding each paper disc was measured after 3 days. W5: trimethoprim 5; C30: chloramphenicol 30; VA30: vancomycin 30; N30: neomycin 30; K30: kanamycin 30; E15: erythromycin 15; DA2: clindamycin 2; S10: streptomycin 10; F300: nitrofurantoin 300; TOB10: tobramycin 10; NA30: nalidixic acid 30; CN10: gentamicin 10; TE30: tetracyclin 30; B10: bacitracin 10; AMP10: ampicillin 10. The numbers indicate the amount per antibiotic per paper disc in micrograms. The data shown in Fig. 6B are the average of three independent experiments.

Macroscopic imaging of gliding motility

Time-lapse movies were obtained using a standard DSLR camera (Canon EOS Rebel T6i with a 100 mm macro lens) controlled by an open-source, low-cost, opto-electronic device that maintains temperature constant (30°C) and produces uniform illumination conditions throughout a long-term experiment (using an addressable strip of RGB LEDs). Design files and instructions to build the macroscope are available online (https://github.com/ccg-esb-lab/BAFFLE). The experiment was performed separately four times, with images acquired at a high-resolution image every 10 min for 72 h. In order to make a comparison of the spatio-temporal distribution of the mutant, complemented and wildtype strains, we inoculated all strains into a semi-solid MM media Petri dish using sterile laser-cutted acrylic cylinders (with a radius of 2 mm and a 4 mm depth) and transferred directly from clonal colonies growing in agar plates. Image analysis was performed using bespoke ImageJ macros (https://imagej.nih.gov/ij/) and consisted of (i) quantifying the optical density by subtracting the intensity values of each pixel to the corresponding value of the first image and, (ii) normalizing the obtained values with respect to the maximum difference observed in the experiment (black corresponds to a background pixel and white to the largest optical density value). All ImageJ scripts are available upon request.

Microscopy and image analysis

All strains were cultured overnight in MM liquid media and collected by centrifugation at 6000 rpm for 2 min at room temperature. After washing and resuspending in PBS, an aliquot was inserted into a tunnel chamber with 2-mm width, 40-mm length and <1 mm thickness fabricated as described in McBride (2014). The tunnel chamber was constructed with a coverslip and a glass slide, assembled with double-sided tape and filled with MM liquid media. We also mounted cells directly from MM agar plates into agar pads (2 g of agarose to 100 ml of a 20% vol./vol. PBS and 10% vol./vol. MM broth solution), following a previously published protocol (Heering et al., 2017). In both experimental setups, motile cells were observed using a Nikon Eclipse Ti-E inverted microscope equipped with Perfect Focus System and differential interface contrast. The microscope was controlled with Nikon NIS-Elements AR 4.20. Time-lapse movies were acquired at 6 fps using a 9v DIA-lamp intensity with exposure of 100–200 ms. Trajectories were visualized in Matlab from spatial coordinates obtained by tracking individual cells using bespoke scripts implemented in ImageJ (http://rsb.info.nih.gov/ij/).

Supplementary Material

Table

Table S1. Strains and plasmids used in this study; Abbreviations: Ap^r-ampicillin resistant; Em^r –erythromycin resistant; Km^r-kanamycin resistant; Tc^r-tetracyclin resistant.

Table S2. Oligonucleotides used in this study. Introduced restriction sites are underlined.

Video-1

Video S1. Supporting information

Video-2

Video S2. Supporting information