Swarming motility and biofilm formation of Paenibacillus larvae, the etiological agent of American Foulbrood of honey bees (Apis mellifera)

American Foulbrood is a worldwide distributed, fatal disease of the brood of the Western honey bee (Apis mellifera). The causative agent of this fatal brood disease is the Gram-positive, spore-forming bacterium Paenibacillus larvae, which can be classified into four different genotypes (ERIC I-IV), with ERIC I and II being the ones isolated from contemporary AFB outbreaks. P. larvae is a peritrichously flagellated bacterium and, hence, we hypothesized that P. larvae is capable of coordinated and cooperative multicellular behaviors like swarming motility and biofilm formation. In order to analyze these behaviors of P. larvae, we firstly established appropriate functional assays. Using these assays we demonstrated that P. larvae ERIC II, but not P. larvae ERIC I, was capable of swarming. Swarming motility was hampered in a P. larvae ERIC II-mutant lacking production of paenilarvin, an iturin-like lipopeptide exclusively expressed by this genotype. Both genotypes were able to form free floating biofilm aggregates loosely attached to the walls of the culture wells. Visualizing the biofilms by Congo red and thioflavin S staining suggested structural differences between the biofilms formed. Biofilm formation was shown to be independent from paenilarvin production because the paenilarvin deficient mutant was comparably able to form a biofilm.


Results
Swarming motility of P. larvae. To test the hypothesis that P. larvae is able to perform a multicellular flagellum-driven surface movement known as swarming motility, we first established an appropriate experimental procedure based on soft agar plates (0.5%) made from brain heart infusion broth (BHI) and on inoculating 10 µl of a bacterial suspension with an OD 600 of 0.1 centrally onto the agar plate. While P. larvae ERIC I did not show swarming motility (Fig. 1A), P. larvae ERIC II reproducibly exhibited swarming motility under these conditions (Fig. 1A). Time course experiments performed by measuring the swarm radii at different time points (Fig. 1B) revealed that P. larvae ERIC II was able to cover the entire surface of the plate within three days (swarm radius 47 mm). In contrast, the macrocolonies formed by P. larvae ERIC I had reached a size of only 12.91 mm ± 0.29 mm (mean ± SEM) even after seven days. The radii of the area covered by P. larvae ERIC II were significantly different from those covered P. larvae ERIC I for all time points (p < 0.0001, unpaired Student's t-test). Next, we randomly selected another five field strains of each P. larvae genotype and tested their swarming behaviour. All P. larvae ERIC I strains formed macrocolonies but were not able to swarm while all tested P. larvae ERIC II strains nearly covered the entire surfaces of the plates within two days (Fig. 2). These results suggested that swarming behaviour is another feature distinguishing these two genotypes.
Biofilm formation of P. larvae. To test the hypothesis that P. larvae is capable of forming multicellular aggregates known as biofilms we established culture conditions that resulted in the formation of free floating bacterial aggregates that were visible to the naked eye and resembled biofilms (Fig. 3A,B). These aggregates reproducibly formed after five to six days when bacterial suspensions were incubated without agitation (static culture) at 37 °C whereas no such aggregates were visible in the negative controls (Fig. 3C). While the aggregates formed by P. larvae ERIC I appeared "hairy" with protrusions attaching all around to the wall of the well (Fig. 3A), the P. larvae ERIC II aggregates had a rather laminar appearance (Fig. 3B).
A characteristic hallmark of biofilms is that the bacteria are surrounded by an extracellular matrix which contains i.a. exopolysaccharides and amyloid fibers 36,37 . Hence, dyes specifically staining these extracellular matrix Figure 1. Swarming motility of P. larvae ERIC I (ATCC9545) and P. larvae ERIC II (DSM25430). Bacterial suspensions (10 µl) with an OD 600 of 0.1 were inoculated centrally onto brain heart infusion (BHI; 0.5%) agar plates and incubated for two days under aerobic conditions at 37 °C. For both genotypes, three biological replicates with three technical replicates each were performed. (A) A representative plate is shown for each strain. Uncolonized agar is black and bacterial biomass is milky white to white. (B) Swarming motility was monitored over time and swarm radii were measured for each strain at one, two, three, five, and seven days post inoculation. Three biological replicates with three technical replicates each were performed. Values are given as mean ± SEM. Statistical analysis of the difference in swarming motility between ATCC9545 (light gray columns) and DSM25430 (dark gray columns) was performed with an unpaired Student's t-test for each time point (not significantly different: n.s., p ≥ 0.05; significantly different: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001).
components can be used to specifically visualize biofilms [47][48][49] . To determine whether the observed P. larvae aggregates (Fig. 3A,B) indeed represented biofilms, that is that the bacteria were surrounded by an extracellular matrix, we used the amyloidophilic dyes Congo red and thioflavin S. Both dyes stain polysaccharides and amyloid fibers 50-53 present in the extracellular matrix produced by bacteria during biofilm formation and for stabilizing the biofilm 48,54-58 and, therefore, are routinely used to test for bacterial biofilm formation. Swarming motility of field isolates of P. larvae ERIC I and P. larvae ERIC II. Swarming assays were performed with five strains each of P. larvae ERIC I and ERIC II which had been isolated from contemporary AFB outbreaks (field isolates). For all strains, three biological replicates with three technical replicates each were performed and a representative plate is shown for each strain. Uncolonized agar is black and bacterial biomass is milky white to white. Swarming was only evident for P. larvae ERIC II strains. Upon Congo red staining, the multicellular aggregates (Fig. 3A,B) stained red while the surrounding medium remained unstained (Fig. 3D,E) like the negative control which represented CR-stained medium (Fig. 3F). These results indicated that both P. larvae ERIC I and ERIC II aggregates were embedded in extracellular matrix and, hence, that the bacteria formed biofilms under the chosen conditions. The biofilms though floating again appeared to have contact to the walls of the wells. Consequently, bacterial cells adherent to the well walls could be visualized by Crystal violet staining (Fig. 3G,H). No staining was evident in the negative controls (Fig. 3I).
In order to further substantiate that CR specifically stained the extracellular matrix surrounding P. larvae cells organized in a biofilm, we next tried to stain planktonic P. larvae cells, which were obtained by culturing P. larvae under constant agitation thereby preventing biofilm formation. Incubating such a non-static culture of P. larvae, hence, of planktonic P. larvae cells, with CR ( Fig. 4A,B) did not reveal any stained aggregates (for comparison see Fig. 3D,E) but instead the wells looked like the negative controls (Figs 3F, 4C,D). These results suggested that the aggregates which formed when P. larvae was cultivated in static liquid and which were visualized by using the biofilm specific dye CR (Fig. 3A-F) were indeed P. larvae biofilms.
In order to provide further evidence for P. larvae biofilm formation, we next examined the three-dimensional structures of the bacterial aggregates by thioflavin S staining and semi-confocal fluorescence microscopy allowing to generate Z-stacks by incrementally stepping through the samples. Green fluorescence indicative for the presence of amyloid fibers was observed in a defined structured zone of the well as floating mirror-image ellipsoidal clouds each having a more densely fluorescing "axis of reflection" (Fig. 5A,B) which seemed to be more Influence of paenilarvin on swarming motility and biofilm formation. Paenilarvins are iturin-like secondary metabolite specifically produced by P. larvae ERIC II 28 . Because iturin-like secondary metabolites are known to influence the motility of their bacterial producers 59,60 , we tested whether paenilarvin has an influence  on swarming motility or biofilm formation of P. larvae ERIC II. We used the newly established assays for swarming motility and biofilm formation and compared the behaviour of wild-type P. larvae DSM25430 and of an inactivation mutant for the paenilarvin gene cluster, P. larvae DSM25430 Δitu, which is deficient in paenilarvin production 29 .
We first analyzed the involvement of paenilarvin in swarming motility of P. larvae ERIC II. One day post inoculation, neither the wild-type nor the mutant bacteria had yet started to swarm (Fig. 6). However, after two days, the wild-type bacteria showed the expected swarming behaviour which resulted already after three days in a rather featureless mat covering the entire plate and with the original macrocolony still visible. At day six post inoculation, the inoculation site in the centre of the plate was no longer visible as area of higher cell density and instead a pattern resembling zones of consolidation or terraces (Bull's eye 30 ;) had developed (Fig. 6A, upper row). In contrast, P. larvae DSM25430 Δitu showed delayed swarming motility: Only after three days, some swarming started at the edge of the central macrocolony and the entire plate was covered not until after six days; the central macrocolony remained clearly visible even after seven days (Fig. 6A, lower row). Measuring the swarming radii at different time points allowed a quantitative analysis of the data and confirmed statistically significant differences (day 2: p value = 0.002; day 3: p < 0.0001) in swarming motility between the wild-type and mutant bacteria (Fig. 6B). These results indicated that in the absence of the iturin-like paenilarvin swarming was significantly delayed suggesting a role of this secondary metabolite in swarming motility of P. larvae ERIC II. Further experiments are necessary to examine the role of paenilarvins in swarming motility in more detail.
When we tested the involvement of paenilarvin in biofilm formation of P. larvae ERIC II, we did not observe any difference between wild-type P. larvae DSM25430 and the paenilarvin gene cluster inactivation mutant P. larvae DSM25430 Δitu (Fig. 7A-D). The biofilms formed by DSM25430 and DSM25430 Δitu contained 8.21 ± 2.15 µg bound CR and 8.22 ± 0.16 µg bound CR, respectively (Fig. 7E). The difference was not significant (p = 0.9889, Student's t-test). Therefore, biofilm formation was unaffected by the absence of paenilarvin production indicating that the iturin-like lipopeptide has no role in this process.

Discussion
The honey bee pathogen P. larvae is the etiological agent of AFB, the most destructive bacterial disease of honey bees 4 . P. larvae is considered an obligate killer because death of the infected larva and subsequent degradation of the larval cadaver to a slimy mass that dries out to the so-called foulbrood scale consisting of P. larvae spores only are both necessary prerequisites for efficient spore production and transmission (for recent reviews see 61,62 ). Hence, the pathogenic phase (until the death of the larva) and the saprophytic phase (degradation of the larval cadaver) are equally important phases in the P. larvae lifecycle wherein the death of the infected larva does not represent the end of the vegetative growth of P. larvae but rather marks the transition from the pathogenic to the saprophytic phase. In the recent past, the availability of the genome sequences of the P. larvae genotypes ERIC I and ERIC II 13 and of protocols for the genetic manipulation of this pathogen 19,63 has led to considerable progress in elucidating P. larvae virulence factors. These results improved our understanding of the bacterial strategies employed to kill the larvae, hence, strategies mainly for the pathogenic phase of the lifecycle of P. larvae 62 . In contrast, the saprophytic phase still remained elusive due to the lack of appropriate functional assays. We are convinced that for degrading the larval cadaver, P. larvae has to be capable of some kind of coordinated or cooperative multicellular behaviour like swarming motility and/or biofilm formation. Hence, the here presented swarming and biofilm assays for P. larvae will allow identifying factors relevant for the success of this bee pathogen after the infected larva has been killed. This will further our understanding of the entire lifecycle of P. larvae, which in turn is a prerequisite for the future development of sustainable treatment strategies.
Swarming motility is a coordinated bacterial activity and widespread amongst flagellated bacteria 64 . Several species of the genus Paenibacillus have been shown to be capable of swarming [32][33][34] . P. larvae harbors peritrichously arranged flagella, needs to rapidly colonize the larval cadaver which can be considered a nutrient-rich environment, and needs to convert larval to bacterial biomass quite fast. Hence, capability of swarming suggested itself for P. larvae. However, so far we could not demonstrate a general ability to swarm for the species P. larvae. Instead, only the P. larvae genotype ERIC II exhibited swarming motility in our assays. One possible reason for this difference might be the differential expression of paenilarvin, an iturin-like lipopeptide, most probably acting as biosurfactant. It has been described that in addition to flagellae and an extracellular matrix the production of biosurfactants is often required for swarming motility to occur 35 . Biosurfactants are surface active agents (hence the name) produced by microorganisms. They are amphiphilic and thus lower the surface tension or interfacial tension between either two liquids or a liquid and a solid. On agar plates, they might increase the wettability thereby facilitating spreading behaviour and swarming motility. The role of lipopeptide biosurfactants for the motility of their bacterial producers on agar plates is well characterized and it has been shown for many swarming bacteria that surface motility is lost or reduced in mutants deficient in producing lipopeptides 59,60 . In accordance with these results, the P. larvae ERIC II Δitu mutant lacking the production of paenilarvins 29 exhibited a significantly delayed swarming behaviour. Because this mutant does not differ in growth rate from its corresponding wild type strain 29 , growth defects cannot explain the delayed swarming phenotype. Our results rather indicate that the paenilarvins are involved in facilitating swarming motility of P. larvae ERIC II. Hence, we propose that the biological roles of paenilarvins comprise not only their anti-fungal activity 28 but also their activity as biosurfactant.
It has already been demonstrated that P. larvae ERIC II -but not ERIC I -produces the lipopeptide paenilarvin 13,23,28 . We here showed that P. larvae ERIC II swarming was hampered in the absence of paenilarvin production. Hence, the inability of P. larvae ERIC I to swarm might be related to the lack of paenilarvin production in this genotype. However, since surface motility was only delayed but not totally lost in the paenilarvin deficient P. larvae ERIC II Δitu mutant, paenilarvin obviously alleviated swarming but was not essential for swarming. Therefore, the lack of paenilarvin production cannot be the only explanation for the lack of swarming of P. larvae ERIC I. Further research is needed to elucidate the regulatory networks that control swarming in P. larvae and to understand the role of swarming during AFB pathogenesis and degradation of the larval cadaver.
Biofilm formation is another coordinated activity of a cooperatively acting group of bacteria. Normally, biofilms are considered sessile communities composed of one or several bacterial species embedded within an extracellular matrix and attached to a solid surface or formed as pellicles at liquid-air interfaces 35 . Recently, the additional existence of free floating biofilm aggregates has been demonstrated for Pseudomonas aeruginosa 40 . The ability to form biofilms had not been proposed for P. larvae so far, although P. larvae-specific FISH analysis of diseased larvae had already revealed the formation of bacterial clusters or aggregates akin to floating biofilms in the larval midgut lumen during early stages of infection 16 . Bacterially produced extracellular matrix is a hallmark and prerequisite of biofilms and, indeed, genes encoding exopolysaccharide biosynthesis proteins have been annotated in the genome of P. larvae (ERIC I: GenBank ETK30098.1; ERIC II: GenBank AHD06625.1) 13 . Furthermore, the slimy consistence of the bacterial mass which remains after total degradation of larval cadaver points to biofilm formation by P. larvae at least at the end of its lifecycle. Using biofilm specific staining methods we here demonstrated that both genotypes of P. larvae were able to form biofilms when incubated in broth culture without agitation, hence, in static liquid. The biofilms did neither adhere to the bottom of the wells nor did they form pellicles at the air liquid interface but instead they appeared as free floating biofilms as already described for Pseudomonas aeruginosa 40 . These free floating biofilms seemed to be loosely attached to the walls of the wells. For P. larvae degrading the larval cadaver, this cadaver does neither provide a proper "surface" to adhere to nor a liquid-air interface for the formation of a pellicle. Therefore, a free floating biofilm at the beginning of the saprophytic phase is a likely cooperative activity to ensure optimal colonization of the cadaver and access to all nutrients.
Although lipopeptides have been shown to play an important role in surface attachment and biofilm formation in Pseudomonas and Bacillus 59 , we could not demonstrate a role for paenilarvin in biofilm formation of P. larvae. Future research is needed to analyze P. larvae biofilm formation in detail. We need to unravel the steps and the regulatory mechanisms in biofilm formation, to identify the components of the extracellular matrix, and to understand the role of biofilm formation for the success of P. larvae as bee pathogen.

Experimental Procedures
Bacterial strains and culture conditions. P. larvae wildtype strains ATCC9545 and DSM25430 representing genotypes ERIC I and ERIC II were used in this study (Genersch et al. 4 ). Both strains have been extensively characterized before; both are virulent and harbor all tested phenotypic features characteristic for the respective genotype 4,[9][10][11]14,15 . In addition, the recently published mutant strain DSM25430 Δitu was used which carries a paenilarvin gene cluster inactivation 29 generated via the TargeTron Gene Knockout System (Sigma-Aldrich,Germany) following a protocol originally established for P. alvei 65 and modified for use in P. larvae 19 . DSM25430 wt and DSM25430 Δitu do neither differ in their growth characteristics nor in virulence for infected larvae 29 . Furthermore, five field strains of P. larvae ERIC I (00-087, 01-440, 02-130, 03-159, 03-189) and five of P. larvae ERIC II (00-897, 00-1163, 03-194, 03-522, 03-525), all of them isolated from contemporary AFB outbreaks, were employed in this study. These strains have already been extensively characterized and used in other studies 4,9,10,66 . Wildtype strains as well as mutant strains were cultivated on Columbia sheep blood agar plates (CSA; Oxoid, Hampshire, UK) at 37 °C for 3-5 days or according to the conditions given for each assay (see below). Swarming assay. Swarming assays were performed with P. larvae ERIC I strains (ATCC9545, 00-087, 01-440, 02-130, 03-159, 03-189) and ERIC II strains (DSM25430, DSM25430 Δitu, 00-897, 00-1163, 03-194, 03-522, 03-525). To obtain pre-cultures, 2.0 ml of BHI (brain heart infusion; Merck KGaA, Darmstadt, Germany) medium were inoculated with one single colony of each P. larvae strain and incubated overnight at 35 °C with agitation. The next day, BHI broth (Merck, Darmstadt, Germany) agar plates (0.5%) were poured from relatively cool (~50 °C 30 ;) agar. The plates were dried with the lid open at room temperature for 15 min. The pre-cultures were diluted to an OD 600 of 0.1, and 10 µl of diluted culture were carefully pipetted into the center of each plate. The plates were then dried for another 15 min and subsequently incubated at 37 °C for several days. Pictures of the plates were taken after one, two, three, six, and seven days. The pictures were analyzed via ImageJ 1.44p to measure the radii of the areas covered by swarming P. larvae. Three biological replicates with three technical replicates each were performed. Data represent mean values ± SEM and statistical analysis was performed by the unpaired Student's t-test with GraphPad Prism 6.
Biofilm assay. Biofilm assays were performed with the P. larvae ERIC I strain ATCC9545 and the wildtype ERIC II strain DSM25430 and the corresponding mutant DSM25430 Δitu. Pre-cultures of P. larvae were prepared as described above and incubated at 35 °C with agitation until the stationary phase was reached (after about 42 h). The pre-cultures in the stationary phase were diluted to an OD 600 of 0.1 and 6.0 ml of diluted culture were transferred to the wells of a Cellstar 6-well tissue culture plate (Greiner Bio-One GmbH, Frickenhausen, Germany). Wells for negative controls contained medium only. The plates were incubated at 37 °C for five days without agitation (static culture). After five days, pictures of the unstained wells containing either cultivated (static culture) P. larvae or medium only (negative controls) were taken. By then, the cultures of ATCC9545 and DSM25430 had reached an OD 600 of about 0.4 and 0.5, respectively. For better visualizing the biofilm, Congo red (CR; Merck KGaA, Darmstadt, Germany), a direct dye for staining amyloid fibres and extrapolysaccharides in biofilms [47][48][49][50][51] was used. CR was added to all wells to reach a final concentration of 30 µg/ml and the plates were incubated for another 24 h at 37 °C before pictures of all wells (static culture of P. larvae or medium only) were taken.
To compare biofilm formation of P. larvae DSM25430 wildtype and the mutant strain DSM25430 Δitu, binding of CR and, hence, the production of extracellular matrix was quantified. To this end, CR bound by the P. larvae biofilm was removed by centrifugation (5 min, 13,000 g) and unbound CR was determined by measuring the absorbance of the supernatant at 490 nm 67,68 . BHI medium which was incubated with CR in the same way was used as reference value for calculating the amount of bound CR 67,68 . The amounts of bound CR can be compared directly since the growth of the mutant does not differ from the wild-type 29 . Data represent mean values ± SD of three biological replicates and were analyzed by the unpaired Student's t-test (GraphPad Prism 6).
To further substantiate that CR specifically stains the extracellular matrix and does not stain planktonic bacterial cells, the dilute pre-cultures of P. larvae (OD 600 = 0.1) were incubated in the presence of CR (30 µg/ml) for 24 hours at 37 °C with constant agitation to prevent biofilm formation. Under these non-static conditions, the cultures of ATCC9545 and DSM25430 reached an OD 600 of about 1.0 and 1.2, respectively. Subsequently, a volume of 6 ml of planktonic cells was transferred to the wells of a 6-well culture plate and pictures were taken.
To demonstrate P. larvae cells adherent to the walls of the culture wells, Crystal violet staining was performed by slightly modifying established protocols 69 . P. larvae cultures were grown overnight in BHI broth and diluted to an OD 600 of 0.1. A volume of each 150 µl of the diluted bacterial culture was pipetted into the wells of a 96-well cell culture plate (Greiner Bio-One GmbH, Frickenhausen, Germany). Negative controls were left uninoculated. Cultures were incubated at 37 °C for six days. The plates were washed three times with MilliQ-water, stained with 0.1% Crystal violet (Merck KGaA, Darmstadt, Germany) and again washed three times before pictures were taken.
For visualizing the three-dimensional structure of the biofilms via fluorescence microscopy, thioflavin S (Sigma-Aldrich Chemie GmbH, Munich, Germany), a fluorescent dye for staining amyloid fibres and extrapolysaccharides in biofilms, was used. Pre-cultures were obtained by inoculating Sf-900 II SFM medium (Thermo Fisher Scientific, Darmstadt, Germany) with one single colony of each P. larvae strain followed by incubation at 37 °C with agitation overnight. The pre-cultures were diluted with SF-900 II SFM medium to an OD 600 of 0.01 and transferred to a 96-well tissue culture plate (Greiner Bio-One GmbH, Frickenhausen, Germany). After addition of 30 µg/ml thioflavin S (Sigma-Aldrich Chemie GmbH, Munich, Germany), the plates were sealed with parafilm and incubated at 37 °C for 6 days. Thioflavin S fluorescence organized in a clearly defined 3D-structure was visualized using fluorescence inverted microscopy (Eclipse Ti, Nikon GmbH, Düsseldorf, Germany). Z-stack processing was performed to obtain the three-dimensional images. Biofilms grown in the absence of thioflavin S were included in the experiments as controls in order to determine the background fluorescence.
Data availability. All data generated or analyzed during this study are included in this published article.