Biofilm and capsule formation of the diatom Achnanthidium minutissimum are affected by a bacterium {#BF-form}
\chaptermark{\emph{A.~minutissimum} biofilm formation}
\noindent Miriam Windler\corrMW\eq\BioKN, Katrin Leinweber\eq\BioKN, Carolina Rio Bartulos\BioKN, Bodo\ Philipp\MMMu\ & Peter G. Kroth\BioKN
Photoautotrophic biofilms play an important role in various aquatic habitats and are composed of prokaryotic and/or eukaryotic organisms embedded in extracellular polymeric substances (EPS). We have isolated diatoms as well as bacteria from freshwater biofilms to study organismal interactions between representative isolates. We found that bacteria have a strong impact on the biofilm formation of the pennate diatom A.\ minutissimum. This alga produces extracellular capsules of insoluble EPS, mostly carbohydrates (CHO), only in the presence of bacteria (xenic culture). The EPS themselves also have a strong impact on the aggregation and attachment of the algae. In the absence of bacteria (axenic culture), A.\ minutissimum did not form capsules and the cells grew completely suspended. Fractionation and quantification of CHO revealed that the diatom in axenic culture produces large amounts of soluble CHO, whereas in the xenic culture mainly insoluble CHO were detected. For investigation of biofilm formation by A.\ minutissimum, a bioassay was established using a diatom satellite Bacteroidetes bacterium that had been shown to induce capsule formation of A.\ minutissimum. Interestingly, capsule and biofilm induction can be achieved by addition of bacterial spent medium, indicating that soluble hydrophobic molecules produced by the bacterium may mediate the diatom/bacteria interaction. With the designed bioassay, a reliable tool is now available to study the chemical interactions between diatoms and bacteria with consequences for biofilm formation.
\noindent \emph{Achnanthidium}, biofilm, capsule, diatom, EPS
\clearpage
Photoautotrophic biofilms are a typical feature in the littoral zones of lakes, streams and oceans. Stones or any other substrata can be covered by a brownish or greenish mucous layer, whenever sufficient light and water is available. These biofilms are a habitat of high primary production [@wetzel_comparative_1964] and may be responsible for sediment stabilization [@cooksey_adhesion_1995]. However, biofilms also have negative effects, as for instance biofouling on human made surfaces like ship hulls or pipes causes high costs in shipping and water management [@gaylarde_deteriogenic_1999; @schultz_economic_2010; @wingender_biofilms_2011]. In addition to the influence of exogenous factors like light, wave disturbance, temperature, water level fluctuations as well as grazing pressure [@hoagland_effects_1990; @schmieder_effects_2004; @rao_comparative_2010], the formation of photoautotrophic biofilms may be strongly influenced by the physiology of the inhabitants of the biofilms and their interactions [@bruckner_bacteria_2008]. Diatoms are common members and early colonisers of photoautotrophic biofilms [@cooksey_adhesion_1995; @wetherbee_first_1998] and their productivity may have a strong influence on the whole biofilm. They can produce copious amounts of extracellular polymeric substances (EPS; @myklestad_rate_1989) which are classified as cell bound EPS like stalks, tubes and capsules [@hoagland_diatom_1993] or soluble EPS (SE). Diatoms are generally associated with bacteria belonging mostly to Alpha-, Beta- and Gamma-Proteobacteria, to the Cytophaga-Flavobacteria-Bacteroides (CFB) group and to Actinobacteria [@knoll_bacterial_2001; @sapp_bacterial_2007; @stanish_bacteria_2013]. Interactions between diatoms and bacteria may occur on different levels and may span from synergy via competition to parasitism or defence reactions [@amin_interactions_2012]. Algae as primary producers provide organic substrates which serve as energy and carbon source for heterotrophic bacteria [@cole_interactions_1982]. @bell_chemotactic_1972 introduced the "phycosphere" concept, describing the zone around the algal cell "in which bacterial growth is stimulated by extracellular products of the alga." The diatoms may in turn require essential compounds from the bacteria (e.g. vitamins; @croft_algae_2005). Little is known about the molecular processes underlying diatom/bacteria interactions within these biofilm communities. Biofilm inhabitants may affect the physiology of other organisms by soluble molecules, indicating that many of these inter-species interactions are based on chemical signals released by diatoms and by bacteria. @thomas_role_1987 observed that the exudates of the xenic diatom Amphora\ coffeaeformis lead to enhanced tolerance of the diatom against copper and tributyltin fluoride. This suggests that either bacterial substances themselves or algal exudates induced by bacteria may trigger the stress response of A.\ coffeaeformis. Such unknown substances may act as signals that mediate recognition and communication between the interaction partners or directly cause a specific effect as toxic compounds. @amin_interactions_2012 suggested that substances which are used for intra-species communication, like autoinducers in bacterial quorum sensing or pheromones in the case of diatoms, might also be involved in interactions between diatoms and bacteria. Such interkingdom signalling was already described to play a role for the seaweed Ulva, where zoospores are attracted by bacterial biofilms via released N-acyl-L-homoserine lactones (AHLs; @joint_cross-kingdom_2007). These AHLs are common autoinducers of Gram-negative bacteria [@chhabra_extracellular_2005]. However, as @amin_interactions_2012 stated, a reliable bioassay comprising a diatom-bacterium pair with a stable interdependency is needed to elucidate the molecular and chemical basis of these interactions.
The goal of this study was to establish a model system for studying the interaction of benthic diatoms and bacteria during biofilm formation. A basic requirement for the investigation of a diatom/bacterium pair is the demonstration of a strong phenotype change in the diatom when cultivated in presence or absence of the bacterium. The model organisms utilised here, A.\ minutissimum (Kützing) Czarnecki and Bacteroidetes strain 32, have been isolated by us from photoautotrophic, epilithic biofilms taken from the littoral zone of Lake Constance. A.\ minutissimum (renamed from Achnanthes\ minutissima (Kützing) @czarnecki_freshwater_1994) is one of the most abundant freshwater diatoms [@patrick_diatoms_1966; @krammer_suswasserflora_1991]. It is frequently found in epilithic biofilms of Lake Constance and represents a eukaryotic pioneer during the initial processes of biofilm formation [@sekar_early_2004]. Common EPS structures of this diatom are stalks and capsules [@geitler_entwicklungsgeschichtliche_1977] which were described in the literature as "unidirectionally deposited, multilayered structures attaching cells to substrata" for stalks and "material external to the wall and often sloughed into the environment" for capsules [@hoagland_diatom_1993]. Bacteria of the Bacteroidetes phylum are frequently associated with diatoms [@amin_interactions_2012] and were also found to be prominent in xenic diatom isolates from photoautotrophic epilithic biofilms of Lake Constance [@bruckner_bacteria_2008]. Bacteroidetes strain 32 belongs to the Dyadobacter genus. Bacteria of this group were found in freshwater, soil samples, or to be associated with maize [@chelius_dyadobacter_2000; @baik_dyadobacter_2007; @zhang_dyadobacter_2010].
A.\ minutissimum (Kützing) Czarnecki was isolated from photoautotrophic epilithic biofilms of Lake Constance [@windler_purification_2012].
The diatom was either cultivated with co-isolated bacteria as "xenic culture" or additionally as "axenic culture" after removal of associated bacteria as described earlier [@windler_purification_2012].
The diatom stock cultures were cultivated in a modified liquid Bacillariophycean Medium (BM; @windler_purification_2012) in cell culture flasks with ventilation caps (Sarstedt, Newton, NC, USA), in which the cells could form a biofilm on the vessel surface.
Monthly, these cultures were scraped off and sub-cultured in new BM.
For aggregation, the xenic and axenic diatom cultures were cultivated at 100\ rpm on an orbital shaker (type 3019; GFL, Burgwedel, Germany).
The diatom cultures were exposed to a 12:12\ h light:dark cycle with a light intensity of 20-50
Bacteroidetes strain 32 was isolated by @bruckner_bacteria_2008. This bacterium was enriched from a xenic Cymbella\ microcephala Grunow strain D-32 culture, which originated from photoautotrophic, epilithic biofilms from the same sampling site as A.\ minutissimum (47°41' N; 9°11' E, Germany). The bacterium was cultivated at 22°C on agar plates containing 50% (v/v) Luria-Broth medium (diluted LB; @miller_experiments_1972), sub-cultivated monthly, and stored at 4-8°C.
In the following, A.\ minutissimum together with all co-isolated bacteria is termed a "xenic culture", whereas the purified diatom in co-culture with Bacteroidetes strain 32 is termed a "co-culture".
A.\ minutissimum was cultivated either in the presence of bacteria, the sterile supernatant of Bacteroidetes strain 32, or single fractions of the respective bacterial supernatant.
The assays were conducted in well plates (#677180, Cellstar); Greiner Bio-One, Frickenhausen, Germany).
Axenic diatom cells were washed with fresh BM and 10^5^ to
Carbohydrates (CHO) associated with cells were stained either with alcian blue or CV. For CV staining, we used a 1:100 dilution of the Gram-staining reagent described by @kaplan_biofilm_2002. Supernatants of the diatom cultures were gently removed, the dye was added to the cultures and incubated for 1\ min. Excessive dye was removed and the cultures were rinsed with water to remove non-attached cells [@izano_detachment_2007]. To quantify biofilm formation, CV was extracted from the cultures with 1\ mL ethanol and the absorption of the dye was determined photometrically at 580\ nm. The alcian blue dye solution was modified according to @staats_isolation_1999, here alcian blue (1% (w/v) in 3% acetic acid (Roth, Karlsruhe, Germany) was added and the cultures were washed once with water.
Cells were observed with an upright light microscope (Olympus BX51; Olympus Europe, Hamburg, Germany), equipped with the filter set 41020 (Chroma Technology Corp., Rockingham, VT, USA) or by using an inverted light microscope (Axiovert 40 C; Carl Zeiss MicroImaging GmbH, Göttingen, Germany). Images were taken with the Zeiss AxioCam MRm or MRc digital camera systems (Carl Zeiss MicroImaging GmbH, Oberkochen, Germany).
For the co-cultivation experiments, Bacteroidetes strain 32 was grown in diluted LB (50% (v/v)), washed three times with BM to remove the bacterial medium and the bacterial suspension was adjusted to an optical density at 600\ nm (OD600) of 0.1, of which 5\ µL were applied to the bioassay.
The xenic A.\ minutissimum cultures were inoculated with the same chl concentration as the axenic cultures and the co-cultures (for chl measurement see below).
For those experiments requiring the sterile supernatant of the bacterial culture, Bacteroidetes strain 32 was cultivated in liquid BM supplemented with 10\ mM glucose (glcBM) at 20°C and 135\ rpm.
The supernatant of the bacterial culture was harvested when the culture reached an OD600 of 0.2.
The culture was centrifuged at 5,525\ g (Allegra™ 25R centrifuge with TS-5.1-500 rotor and swinging buckets, 10\ or\ 50\ mL tubes, depending on the volume of the supernatant, Beckman Coulter, Krefeld, Germany) and the supernatant was filtered using a 0.2\ µm filter (Filtropur S; Sarstedt, Nümbrecht, Germany).
For filtration of large volumes of bacterial supernatant (>10\ mL), the bacterial supernatant was additionally prefiltered through 3.0\ µm membrane filter (Merck Millipore, Darmstadt, Germany) and 0.2\ µm filter (Whatman, Dassel, Germany).
The sterility of the filtrate was randomly tested by plating an aliquot on agar plates containing diluted LB medium (50% (v/v)) and incubated as described for Bacteroidetes strain 32.
Different volumes of the sterile bacterial supernatant and equivalent volumes of glcBM for negative control were applied to the bioassay.
For fractionation, 10\ mL of the bacterial supernatant were separated via solid-phase extraction (SPE) using C18-SPE endcapped cartridges (530\ mg; Macherey-Nagel, Düren, Germany) according to @von_elert_predator_2000. The bacterial supernatant and medium control were adjusted to pH 7 with citric acid or sodium hydroxide. Flow-through (Ft), wash fraction (Wf) and eluate were collected separately. Ft and Wf were evaporated to near dryness and diluted in 1\ mL BM. The eluate was evaporated to dryness to remove methanol and resuspended in 2\ mL BM. Aliquots of this fraction were incubated for 1\ h at 30°C and 80°C, respectively, for further dissolution of solid material. The bacterial growth medium (glcBM) was treated the same way and was used as a negative control. The fractions were sterile filtered as described above and aliquots were diluted to the original concentration (1:5 for the eluate, 1:10 for Ft and Wf). 250\ µL of the concentrated and diluted fractions were applied to the bioassay and tested for bioactivity. The unseparated bacterial supernatant was used as positive control.
For further fractionation, 910\ mL of the sterile bacterial supernatant were loaded on an endcapped C18-SPE column (10\ g; Macherey-Nagel) and were eluted stepwise with methanol at increasing concentrations (20%, 40%, 60%, 80%, and 100% methanol in ultrapure water). Each fraction was eluted with 20\ mL of the solvent, except the 100% methanol fraction. This fraction was eluted with 25\ mL. After evaporation, the residues were resuspended in 1\ mL of ultrapure water, sterile filtered, and diluted in BM. 250\ µL of the diluted fractions were tested for bioactivity.
The sterile bacterial supernatant was harvested at different growth phases of Bacteroidetes strain 32, which was cultivated in 100\ mL glcBM in triplicates.
1.2\ mL were removed at each time point.
A total of 100\ µL were used to measure OD600 of the bacterium and the supernatant of the remaining volume was sterile filtered as described above.
Half of the volume was frozen for determination of the glucose concentration via HPLC as described in @jagmann_parasitic_2010.
The other half volume was stored at 4°C until it was tested for bioactivity (50% (v/v) of the bacterial supernatants).
Growth of axenic and xenic A.\ minutissimum and of the diatom co-cultured with Bacteroidetes strain 32 was measured. The bioassay was performed as described above and the chl concentrations of suspended cells in the supernatants (non-adherent cells) and of cells embedded in the biofilm (adherent cells) were determined as described below.
Triplicates of xenic as well as axenic cultures were incubated in 100\ mL BM in cell culture flasks at standard diatom cultivation conditions.
The axenic cultures were inoculated at a cell density of 2SO4 and heated to 60-80°C when required to dissolve the pellets.
The ethanol fractions containing the oligo-and monomers were evaporated and the pellets were dried and processed as described for the polymers.
The bound CHO were extracted stepwise.
First, the cell pellets were resuspended in 1\ mL sterile-filtered tap water to keep the osmolarity in the physiological range and incubated for 1\ h at 30°C and 300\ rpm.
Cells were centrifuged (5,000\ g for 10\ min at 20°C, 5417R microcentrifuge; Eppendorf, Hamburg, Germany) and the supernatants containing the "WW" fraction were separated from the pellets.
The following centrifugation steps were conducted at 5,000\ g for 7\ min at 4°C (5417R microcentrifuge; Eppendorf).
Pellets were defatted by incubation in 1\ mL 90% ethanol at 600\ rpm for 15\ min and subsequently centrifuged [@wustman_extracellular_1997; @bahulikar_complex_2008].
This procedure was repeated 5-7\ times until the pellets were colourless or slightly green.
Cells were further washed twice with distilled water, then incubated in 1\ mL distilled water at 90°C for 1\ h and centrifuged.
The supernatants were used for determination of the "HW" and the cell pellets were incubated for 1\ h in 1\ mL 0.5\ M NaHCO3 and 0.1\ M EDTA at 95°C [@chiovitti_complex_2003] to harvest the "HB".
The remaining portion of the bound EPS was extracted by resuspending the pellets in 1\ mL 1\ M NaOH and 0.2\ M NaBH4 at 95°C for 1\ h and subsequent centrifugation of the cell fragments, resulting in "HA".
Mono-/oligo- and polysaccharides of the soluble CHO fraction as well as the total CHO content of the WW-, HW-, HB-, and HA- fractions were measured using the phenol-sulfuric acid assay [@dubois_colorimetric_1956] with glucose as standard according to @de_brouwer_physical_2002 and @bruckner_bacteria_2008 and were specified as $\frac{\mu \text{gCHO}}{\mu \text{gchl}}$.
In a follow-up experiment, we compared a standard curve for glucose with those of other CHO that were reported to be prevalent in A.\ minutissimum [@bahulikar_complex_2008].
Glucose had the steepest regression slope (see Suppl.\ Fig.\ \ref{CHO-STD},\ p.\ \pageref{CHO-STD}), indicating that other CHO might have been slightly over-estimated.
Glucuronic acid was tested as well and was found to have the lowest regression slope (see Suppl.\ Fig.\ \ref{Glu-compare},\ p.\ \pageref{Glu-compare}).
It is therefore not included in this study.
Chl was extracted by addition of a mixture of 5% methanol and 95% acetone to the diatom cell pellet. The chl of surface-adherent cells was extracted by flushing the biofilm with the extraction mixture for several times directly in the cultivation well. Chl concentrations were determined using the equation and extinction coefficients for diatoms, chrysomonads, and brown algae according to @jeffrey_new_1975, and specified as total chl concentration (a sum of chl a and c).
A.\ minutissimum is a benthic diatom growing attached to surfaces. Interestingly, we observed that xenic and axenic A.\ minutissimum cells showed very different aggregation behaviour. When cultivated on a shaker, the xenic A.\ minutissimum cells formed macroscopically visible aggregates (Fig. \ref{ax-xen}A), while the axenic diatom culture grew completely suspended and no aggregate formation was visible (Fig. \ref{ax-xen}B). Under non-shaking conditions, the xenic diatom cells were attached to the surface of the cultivation vessel, forming a biofilm. Microscopic analyses in combination with alcian blue staining revealed that the cells in the early and late stationary phase were surrounded by large capsules of bound EPS (Fig. \ref{ax-xen}, C and E). The respective diatom cells were attached to each other via the capsules, partially resulting in macroscopically visible aggregates. Cells of the axenic culture did not show any capsules, remaining freely dispersed and forming, if at all, only small aggregates of a few cells (Fig. \ref{ax-xen}, D and F). In addition to the capsules, the xenic cultures showed diffuse EPS structures that were stainable by alcian blue (Fig. \ref{ax-xen}C). This diffuse, unstructured form of EPS could also be found in the axenic culture (Fig. \ref{ax-xen}D).
Induction of capsule- and biofilm formation in the axenic A.\ minutissimum culture by co-cultivation with Bacteroidetes strain 32
A bioassay was designed to investigate the bacterial influence on capsulation/biofilm formation of A.\ minutissimum. Biofilm formation was quantified according to @izano_detachment_2007, by staining of bound CHO with CV. The dye stains the cells as well as the bound EPS and its intensity depends on the amount of attached cells and insoluble EPS and therefore is proportional to the biomass. After staining, biofilm formation is visible by eye and biofilm quantity can be determined by measurement of the alcohol-extracted CV.
We found that the biofilm formation of the xenic culture and of the co-culture with Bacteroidetes strain 32 is accompanied by much higher CV staining compared with the axenic diatom culture (Fig. \ref{bioassay-stains}A). The stable biofilm of the xenic cultures allowed rough washing steps because the cells adhered strongly to the plastic surface of the wells, resulting in high CV absorption already after 3\ d of cultivation (Fig. \ref{bioassay-stains}B). At that time, the chl concentrations were still relatively low in the xenic culture (0.38\ µg/mL; see below Fig. \ref{bioassay-chl}A). The co-culture also exhibited strong biofilm formation after 3\ d and it reached the highest values of CV adsorption after 10\ d (Fig. \ref{bioassay-stains}B). After 3\ d of cultivation the diatom cells in the xenic culture and in the co-culture exhibited stalks that apparently mediate adherence in early stages of biofilm formation (Fig. \ref{bioassay-stains}C). The cells in the xenic culture started capsule formation after 10\ d, those of the co-culture after 14\ d of cultivation, when cells were in the stationary phase. In the axenic culture, there was no biofilm and capsule formation observable even after 24\ d. Here, the cells did not adhere to the well surface and were removed during the wash steps of the staining procedure. In some cases, CV did not stain the capsules properly when the biofilm was fully developed. In these cases, the cultures were additionally stained with alcian blue for microscopic observation and visualization of capsules in the xenic culture and co-culture (Fig. \ref{bioassay-stains}C).
Capsule and biofilm induction by the sterile supernatant of Bacteroidetes strain 32 {#S32-extracts-BF}
\begin{FPfigure} \centering \caption{\textbf{Induction of capsule and biofilm formation of \emph{A.~minutissimum} by sterile bacterial supernatant of Bacteroidetes strain 32.} Biofilms were stained with CV and the absorption of the dye was determined. Cultures exhibiting capsules are marked with (Caps). (\textbf{A}) Incubation with 6%, 12.5%, 25%, 50%, and 94% (v/v) of the sterile supernatant (\emph{N} = 3). The control assays were conducted either with addition of equivalent volumes of the bacterial medium (glcBM) or the axenic diatom culture. (\textbf{B}) \emph{A.~minutissimum} incubated with different fractions of the C18-solid phase separated bacterial supernatant: eluate (Elu), eluate preheated to 30°C and 80°C, respectively (Elu 30/80°C), flow through (Ft) and wash fraction (Wf). Graphs show cultures treated with fractions which had been diluted to the original concentration. Fractions of the bacterial medium glcBM and the unseparated bacterial supernatant were used for negative and positive control (\emph{N}\ =\ 1 for the fractions, \emph{N} = 3 for the control). (\textbf{C}) \emph{A.~minutissimum} incubated with different fractions of the C18-SPE separated bacterial supernatant, eluted with increasing methanol concentrations (\emph{N} = 2). Fractions of the bacterial growth medium glcBM were used for negative controls (lower curve). Error bars indicate SD. \label{S32-extracts}} \includegraphics{biofilm-formation-figures/S32-extracts.png} \end{FPfigure}
Bacteroidetes strain 32 was cultivated in BM supplemented with 10\ mM glucose (glcBM) before harvesting the spent medium. This medium allowed bacterial growth without inhibiting diatom growth as we observed for diluted LB. Capsule and biofilm formation of A.\ minutissimum can be induced by incubation of the axenic culture with the filtered sterile supernatant of Bacteroidetes strain 32 (Fig. \ref{S32-extracts}). The intensities of CV adsorption of A.\ minutissimum cultures showed a dose-dependent pattern when treated with different volumes of the sterile bacterial supernatant and reached a maximal absorption after addition of 25% and 50% (v/v) of the bacterial supernatant (Fig. \ref{S32-extracts}A). Diatom capsules were found in cultures which had been supplemented with 12.5%, 25%, and 50% (v/v) of the bacterial supernatant. Neither the control cultures, which had been treated with the glcBM only, nor the axenic culture did induce biofilm or capsule formation. This indicates that soluble molecules released by the bacterium induce biofilm and capsules of A.\ minutissimum.
For enrichment of the biofilm-inducing substance, we further separated the bacterial supernatant by solid phase extraction (SPE) according to @von_elert_predator_2000. The fractions were diluted to the original concentration prior to the application (Fig. \ref{S32-extracts}B). The eluted fraction (Elu) showed a similar induction of capsule/biofilm formation as the unseparated bacterial spent medium (positive control). Ft and Wf of the extracted bacterial spent medium described above did not induce a stronger biofilm formation than the fractions of the bacterial growth medium, indicating that the biological activity had been completely absorbed by the nonpolar C18-sorbent and was released after extraction with methanol. Incubation of the bioactive fraction at 30°C and 80°C as well as evaporation to dryness did not diminish the biological activity. As the methanol that was used as solvent for the fractionation had been completely evaporated, we can exclude induction of capsule and biofilm formation by the alcohol itself. Furthermore, the SPE was performed with bacterial growth medium only. Here, no capsule or biofilm formation was observed in any of the fractions.
Furthermore, the bioactivity was found in fractions which were eluted with high methanol concentrations, thus increased biofilm formation was mainly found in A.\ minutissimum cultures which had been treated with the 60%-100% methanol fractions and capsules were only found in the 80%-100% methanol fractions (Fig. \ref{S32-extracts}C). These results suggest a nonpolar character of the bioactive molecule, as well as resistance to heat and dryness. The observed effect of bacterial supernatant on A.\ minutissimum is very reproducible, we were able to repeat the experiments several times independently starting from different bacterial cultures and purifying the bioactive compound(s) to varying degrees (data not shown).
To investigate the optimal growth phase of the bacterium for sufficient production of the bioactive substance, the bacterial supernatant was harvested at different time points and each supernatant was tested for bioactivity.
Figure \ref{S32-growth}A shows the growth of Bacteroidetes strain 32 in glcBM reaching an OD600 of 0.24 in the stationary phase.
Glucose consumption, as determined via HPLC analysis, resulted in a decrease of the glucose concentration from 10.3 to 8.5 mM during this time.
Biofilm formation of A.\ minutissimum, as based on an increase in extractable CV adsorption from 0.16 to 0.35, was already induced by bacterial supernatant taken in the early exponential phase of the bacterium at an OD600 of 0.016 (Fig. \ref{S32-growth}B).
Accordingly, capsule formation was only induced by supernatants taken after the bacterium reached the mid-exponential phase at an OD600 of 0.03.
Growth of A.\ minutissimum in the bioassays was determined by measuring the chl concentrations (Fig. \ref{bioassay-chl}). This method was chosen, because both co-culture and xenic culture tended to form cell aggregates, which hampered reproducible cell counting. The axenic culture reached a similar maximum chl content as the xenic culture and the co-culture (0.78±0.13\ µg/mL after 10\ d for the axenic culture, 0.86±0.05\ µg/mL and 0.81±0.06\ µg/mL after 7\ d for the xenic culture and co-culture, respectively; Fig. \ref{bioassay-chl}A). Interestingly, the chl content of the xenic culture and of the co-culture showed a plateau in the stationary phase after 17\ d of cultivation, whereas the chl concentration of the axenic culture decreased continuously. In the xenic culture and in the co-culture, the main chl concentrations were found in the biofilm fraction, indicating that these cultures consist mostly of adherent cells (Fig. \ref{bioassay-chl}, B and C). Nearly no chl was detected in the supernatant after 10 and 14\ d of cultivation, respectively, at a time when the cells exhibit capsules. In contrast, most of the chl of the axenic culture was found in the non-adherent fraction (Fig. \ref{bioassay-chl}D). The bacterium itself was not able to grow without the diatom in the diatom full medium (BM; Suppl.\ Fig.\ \ref{S32-growth-BM},\ p.\ \pageref{S32-growth-BM}).
\begin{FPfigure}
\centering
\caption{\textbf{Quantification of soluble and bound carbohydrates of axenic and xenic \emph{A.minutissimum} cultures.}
Specific carbohydrate quantities are shown for axenic and xenic cultures and specified in $\frac{\mu \text{gCHO}}{\mu \text{g~chl}}$).
Carbohydrates were extracted in different growth phases of the diatom: late-exponential growth phase (12\ d of cultivation), early stationary phase (20\ d of cultivation), and late stationary phase (33\ d of cultivation).
Soluble carbohydrates were measured as soluble polymers and monomers.
Bound carbohydrates were stepwise extracted with warm water (WW), hot water (HW), hot bicarbonate (HB), and hot alkali (HA; \emph{N} = 3; error bars indicate SD).
\label{CHOs}}
\includegraphics{biofilm-formation-figures/CHOs.png}
\end{FPfigure}
We analysed the CHO content and quality by stepwise extraction of soluble and insoluble EPS from xenic and axenic A.\ minutissimum contents as described in Materials and methods. Quantification of CHO in each fraction showed that in the late exponential growth phase (after 12\ d of cultivation), the axenic and xenic cultures exhibited similar amounts of bound CHO (Fig. \ref{CHOs}A). At that time, the xenic diatoms did not show capsules and the axenic cells secreted about 5-fold more soluble CHO compared with the xenic culture. In the early stationary phase, after 20\ d of cultivation, the amount of bound CHO in the HB fraction of the xenic culture increased (Fig. \ref{CHOs}B) and light microscopy indicated that the xenic cells started to form capsules (Fig. \ref{ax-xen}C). The capsules were fully developed in the late stationary phase, after 33\ d, when the HW and HB fractions exhibited large amounts of insoluble CHO (Figs.\ \ref{ax-xen}E and \ref{CHOs}C). Microscopic observations revealed that dissolution of capsular material correlated well with an increase in CHO content in the HW and HB fractions (Suppl.\ Fig.\ \ref{CHO-BFM},\ p.\ \pageref{CHO-BFM}). Thus, the capsular material of the xenic culture can be extracted with the hot water and hot bicarbonate treatments, leaving only very little CHO material in the hot alkali fraction. During the whole cultivation period, the amount of soluble CHO remained relatively low in the xenic culture, especially in the monomer fractions (Fig. \ref{CHOs}). The axenic A.\ minutissimum, on the other hand, showed less CHO in the insoluble fractions but produced large amounts of soluble CHO.
When cultivated in presence or absence of distinct bacteria, the diatom A.\ minutissimum showed a considerably different behaviour regarding cell aggregation, cell attachment and biofilm formation, which is most likely because of changes in the generation of soluble and insoluble EPS by the diatom. Positive as well as negative bacterial influences on aggregation formation have been described in the context of marine snow, aggregated material of cells, and organic matter in the oceans [@grossart_algaebacteria_2006; @gardes_diatom-associated_2011], and several possible mechanisms have been discussed. Bacteria may increase the amount of matrix material either by contribution of bacterial EPS [@decho_microbial_1990] or by stimulation of EPS secretion by the algae. Also bacterial modification of organic material and subsequent changes of adhesive properties has been suggested [@grossart_marine_2005]. Cellular aggregation is of ecological relevance as it increases the sinking velocity and thus plays an important role in the flux of organic matter into deeper parts of the water column [@fowler_role_1986; @gardes_diatom-associated_2011]. It also may alter the consumption by grazers and thereby the transfer of organic matter to upper trophic levels [@decho_microbial_1990].
Xenic A.\ minutissimum cells may stick to the well surface because of development of capsules of extracellular organic matter, while axenic cultures revealed no such capsules and grew suspended. Capsule and biofilm formation of axenic A.\ minutissimum, however, can be induced by addition of Bacteroidetes strain 32. Of eight other tested bacterial isolates, several led to stronger biofilm formation, but none of them was able to induce capsule formation (Suppl.\ Fig.\ \ref{CV-isolates},\ p.\ \pageref{CV-isolates}). We made similar observations in earlier experiments showing that capsulation and altered EPS structures of the freshwater biofilm diatom Cymbella\ microcephala occur in co-culture with Bacteroidetes strain 32, indicating that this might be a specific property of the bacterium [@bruckner_bacteria_2008]
A.\ minutissimum also shows capsule formation when treated with the sterile supernatant of Bacteroidetes strain 32. Apparently, the diatom is able to recognise one or more soluble molecules produced by the bacterium; a direct cell-to-cell contact between the interaction partners apparently is not required as it was shown previously for Pseudo-nitzschia multiseries [@kobayashi_direct_2009]. Interestingly, Bacteroidetes strain 32 produced the putative infochemicals even when cultivated separately, which indicates a constitutive production and secretion. As the biotic activity of this or these substances can be demonstrated already in the early exponential growth phase of Bacteroidetes strain 32, they apparently are produced by intact bacterial cells and do not constitute substances that are released after cell damage. As shown in figure\ \ref{S32-growth}B, A.\ minutissimum reacts with a sharp increase in biofilm formation even at relatively low bacterial abundance, thus either the diatom is very sensitive to the substance(s) or the production to the substance(s) is strongly increased once the bacterial density exceeds a certain threshold. The bioactive molecules were extractable via reversed phase cartridges indicating a hydrophobic character, similarly to AHLs, a major class of autoinducers produced by Gram-negative bacteria [@chhabra_extracellular_2005]. The inducibility of capsule formation using spent medium of the bacterium shows that the diatom itself is the active producer of biofilm material and the active contribution of bacterial EPS can therefore be excluded.
Former studies showed that surface materials may influence diatom attachment and biofilm formation [@gawne_role_1998; @mieszkin_effect_2012]. Polystyrene for example, represents a hydrophobic surface and was found to be preferred out of four different materials for attachment of the diatom Achnanthes\ longipes [@gawne_role_1998]. In our study, axenic A.\ minutissimum did not attach to polystyrene of the well plates and hence attachment only depends on Bacteroidetes strain 32 or its spent medium. @gawne_role_1998 and @mieszkin_effect_2012 further showed that bacterial biofilms on surfaces may influence attachment of diatoms in a positive or negative way and often precedes biofilm formation of diatoms. As biofilm of A.\ minutissimum can be also induced by the sterile supernatant of the bacterium, preceding biofilm formation of the bacterium as a sort of base layer is obviously not the only reason for attachment of the alga.
CHO quantification revealed that the axenic diatom culture contained large amounts of soluble CHO.
In the late stationary phase, we found variable amounts of soluble polymers in this culture (15.9, 11.1, and 55.6 $\frac{\mu \text{gCHO}}{\mu \text{gchl}}$; Fig. \ref{CHOs}C), resulting in a high standard deviation.
From former experiments, we know that the axenic cultures usually secrete large amounts of soluble polymeric CHO in the stationary phase (up to 10^5^ $\frac{\mu \text{gCHO}}{\mu \text{gchl}}$; Suppl.\ Fig.\ \ref{CHO-BFM},\ p.\ \pageref{CHO-BFM}).
However, in the xenic culture insoluble CHO were dominant, especially in the late-stationary phase when the capsules were fully developed.
In this growth phase, CHO of the xenic culture were primarily found in the HW and HB fractions, which correlates well with the dissolution of the capsular material (Suppl.\ Fig.\ \ref{CHO-BFM} C and D,\ p.\ \pageref{CHO-BFM}), implying that the CHO content of these fractions derives from capsular material.
Absolute values of the CHO extraction have to be interpreted with caution, as the extraction of bound CHO with warm water is controversially discussed in the literature because of possible contamination with released intracellular CHO [@chiovitti_complex_2003; @de_brouwer_does_2004].
However, we found only little amounts of CHO in this fraction (\textless 2 $\frac{\mu \text{gCHO}}{\mu \text{gchl}}$) compared to the other fractions.
We tried to minimise possible contamination with intracellular CHO previous to the extraction with warm water as we defatted the cells with ethanol, which was shown to remove intracellular material [@wustman_extracellular_1997].
We checked the reliability of the defatting step and found no CHO left after two extractions with ethanol, suggesting that all intracellular CHO were removed.
Here, axenic and xenic cells released similar amounts of CHO during the defatting steps (see Suppl.\ Fig.\ \ref{CHO-defat},\ p.\ \pageref{CHO-defat}), while chl was efficiently removed (see Suppl.\ Fig.\ \ref{Chl-defat},\ p.\ \pageref{Chl-defat}).
Glucose has been used as standard for CHO measurement as described earlier [@de_brouwer_physical_2002; @bruckner_bacteria_2008]. To evaluate the difference with the naturally occurring CHO mix of A.\ minutissimum, we tested six most abundant compounds which were found in A.\ minutissimum previously [@bahulikar_complex_2008] using the Dubois method (Suppl. Figs.\ \ref{CHO-STD},\ p.\ \pageref{CHO-STD} and \ref{Glu-compare},\ p.\ \pageref{Glu-compare}). These sugars yielded a somewhat weaker colorimetric reaction compared with glucose. Thus, the absolute values of the CHO fractions may be overestimated, however, our findings are based on a comparison of CHO contents in xenic and axenic cultures and not on absolute values.
As bacteria in the xenic culture potentially may consume the soluble CHO (especially the easily accessible monomers, whereas the bound CHO may be more resistant to bacterial degradation), we cannot estimate whether the total amounts of CHO in xenic and axenic cultures are comparable. However, it might be possible that axenic and xenic cultures produce similar amounts of CHO, but the condition of the secreted CHO changed from a dissolved to an insoluble state in the presence of bacteria. Nonetheless, in both cases it is evident that A.\ minutissimum strongly secreted CHO even when no structured EPS was visible. When nutrients become limited--thus within the stationary phase--increased EPS production and capsule formation is a known phenomenon [@lewin_capsule_1955; @bhosle_isolation_1995] and is assumed to serve also as an overflow mechanism [@staats_oxygenic_2000].
The exact function of capsules produced by A.\ minutissimum is still unclear. @geitler_entwicklungsgeschichtliche_1977 proposed that capsule formation of this alga is involved in sexual reproduction, whereas pads and stalks are regarded as common structures of vegetative cells. This proposal was based on the observation that the capsule may engulf both mating partners, however, we also found single cells to be surrounded by a capsule. According to @lewin_capsule_1955 the production of cell bound gelatinous matter may provide attachment, which is supported by our observation that cells exhibiting capsules were attached to each other or to artificial surfaces. The cells of A.\ minutissimum are motile until they form capsules. These capsules can be stained with alcian blue in acetic acid, which stains anionic polysaccharides [@staats_isolation_1999] that are thought to facilitate adhesion [@dade_effects_1990]. However, in the xenic culture and in the co-culture with Bacteroidetes strain 32, the diatom cells within a few days adhered to the surface mediated by stalks before a capsule became visible. We therefore conclude that bacterially induced capsulation may not be a prerequisite for surface-attachment, but possibly for strengthening the attachment and for irreversible binding. Diffuse EPS, also known as transparent exopolymer particles (@passow_transparent_2002 observed both in the axenic and xenic cultures, obviously did not facilitate surface adhesion, as axenic cells can be easily removed during the washing procedures. Surface adhesion of diatoms could be relevant once the cells found a favourable environment e.g., when they recognise the presence of an interaction partner. This would implicate a mutualistic character of the interaction with Bacteroidetes strain 32. Indeed, the bacterium was not able to grow in BM alone, suggesting that exudates from the diatom are utilised as carbon sources. The finding of a clearly higher abundance of bacteria in co-culture with diatoms compared with the bacterial growth in the respective diatom medium, supports that this is also the case for exudates from other diatom species [@grossart_algaebacteria_2006; @gardes_diatom-associated_2011]. Thus, a further explanation for capsulation in the presence of bacteria could be that the capsule itself may serve as a feeding ground for the bacteria and may help to keep the bacteria in close proximity to the diatom cell. It is tempting to speculate that bacteria may induce cell aggregation and biofilm formation of the diatom to keep the alga in spatial proximity and to ensure an adequate nutrient supply by the primary producer. The diatom in turn may benefit from the general properties of the biofilm, e.g., protection from toxic compounds [@ceri_calgary_1999], UV radiation [@ehling-schulz_uv-b-induced_1997], and grazer protection. They may also directly benefit from the bacterium, for example, by bacterial supply of essential nutrients (e.g., vitamins). The medium used in this study indeed contains sufficient basic nutrients for unlimited diatom growth. However, once a biofilm is established, local nutrient deficiency may occur which may be compensated by bacteria.
Microscopic analyses revealed that the bacterial cells of the xenic culture most likely cannot penetrate the capsule, which is visible by a bacteria-free area around the diatom cell (data not shown), and thus the diatom may prevent being overgrown or parasitised by bacteria. This may be much more evident when the diatom cells are stressed and thus much more vulnerable. In this context capsulation could also represent a defence mechanism. The capsule could further protect the diatom cells from toxic bacterial compounds as EPS may reduce the susceptibility of biofilm organisms to some substances [@stewart_antibiotic_2001].
The bioassay for analysis of bacterial influence on biofilm structure and quantity of A.\ minutissimum established in this study proved to be very reliable. The enhanced biofilm formation is visible by eye after staining with CV, allowing a fast screening of a large number of substances as shown here for fractions of the separated bacterial supernatant. The assay could possibly be extended to test the impact of other chemicals on the biofilm formation of this or even other diatom species. A.\ minutissimum turned out to be an excellent model organism for the investigation of biofilm formation as it produces large amounts of soluble and bound EPS and it can be cryopreserved by using a modified protocol according to @buhmann_post-cryopreservation_2013. The bioassay is based on the interaction of the alga with Bacteroidetes strain 32 as this bacterium induces the same physiological changes of capsulation and biofilm formation as shown for the satellite community of A.\ minutissimum. Up to now, little is known about the chemical interactions between diatoms and bacteria. The identification of the bioactive compound(s) of Bacteroidetes strain 32 and further physiological studies will be important to show whether the interaction between A.\ minutissimum and Bacteroidetes strain 32 has a mutualistic or an antibiotic character.
The authors gratefully acknowledge the financial support by the Konstanz Research School Chemical Biology (KoRS-CB, GSC 218) and the University of Konstanz. We also would like to thank Georg Pohnert (University of Jena) for very helpful discussions and Annette Ramsperger, Vera Bleicher, and Joana Thiel for technical support.