Diatoms are unicellular, phototrophic eukaryotes of the class Bacillariophyceae. Their defining features are their "spectacularly" designed cell walls [@kroger_diatom_1998], and their four-membraned plastids, which originate from secondary endocytobiosis [@kroth_diatom_1999]. The cell walls of diatoms are scaffolded by biomineralised silica, to which polysaccharides and proteins are attached [@kroger_diatom_1998]. This so called frustule is split along a "girdle" region into two parts called thecae, which explains the etymology of this algae group's name: "dia" and "temnein" together mean "cut in half" in Greek. Several diatom genera possess one or two major slits (called raphes) on one or either frustule surface (called valves). Raphes and valve pores can secrete extracellular polymeric substances (EPS). By definition, polymeric substances are characterised as extracellular, if they occur outside of the plasma membrane [@hoagland_diatom_1993]. EPS are relevant for the attachment of cells to a surface, their motility, as well as for biofilm formation.
Cell division in diatoms is coupled to the transfer of one of the thecae to each daughter cell. The daughters only synthesise the smaller hypotheca, so that one lineage is continually shrinking, which leads to a sexual reproduction cycle [@geitler_formwechsel_1932]. Thus, the long-term cultivation of diatoms, maintenance of those cultures, as well as life cycle studies encounter unique problems (see @chepurnov_experimental_2004 for a review). Besides decreasing cell size, another such problem can be the decrease of the specific surface area and pore size of frustules, due to lower salt concentrations [@vrieling_salinity-dependent_2007]. This results in a denser biosilica packing, and illustrates that diatom cells can modulate their frustule morphology through metabolic processes, which in turn may be influenced by genetic modification. It is envisioned to utilise such modified diatom frustules to engineer nanomaterials, which may be useful for chemical catalysis, particle separation and other applications [@kroger_prescribing_2007].
The frustules also help to trace diatoms through the geological record. Non-marine diatoms appear in ca. 70 million years old strata from the Late Cretaceous [@chacon-baca_70_2002], and likely radiated from marine species that evolved earlier in that period [@harwood_cretaceous_1995]. Genetic evidence points to an even earlier origin, up to 266\ million years ago [@kooistra_evolution_1996]. Thus, the radiation of diatoms into almost all moist habitats likely started in the Late Permian, Trias or Jurrasic. Since then, diatoms have occupied a diverse range of ecological niches and follow various life styles: planktonic [@kooistra_origin_2007] and benthic [@round_benthic_1971] in sea- and fresh water, epibiotically on both plants and animals [@tiffany_epizoic_2011; @majewska_diatoms_2015], and also terrestrial [@souffreau_molecular_2013].
Their diversity and abundance make diatoms a major biogeochemical and ecological force.
Besides driving the silica cycle
by incorporating orthosilicic acid into their cell walls,
diatom photosynthesis contributes approx. 20%
to the global net primary production and oxygen production
[@mann_species_1999; @field_primary_1998].
Additionally, diatoms often dominate the initial phase of phytoplankton blooms.
Such blooms occur naturally due to the upwelling of nitrate-, phosphate- and iron-rich deep waters, as well as due to the influx of these nutrients from land [@capone_microbial_2013].
Some phytoplankton blooms are harmful to the higher trophic levels due to the exhaustion of oxygen by heterotrophs that degrade the remains of the primary producers.
Blooms can also be harmful to humans directly, due to the production of toxin by the microorganisms [@smayda_what_1997].
Nonetheless, they are highly productive events in terms of CO2 fixation, and export much of that as particulate organic carbon to deeper ocean layers [@buesseler_decoupling_1998], thus feeding the benthic heterotrophs.
Organic exudates of marine diatoms can also become aerosolised via sea spray,
and have recently been suggested as an important source for ice-nucleating particles, which play a role in cloud formation [@wilson_marine_2015].
For these reasons, diatoms need to be considered when discussing climate-change, ocean acidification and other global environmental phenomena.
Just as diatoms, biofilms occur naturally in many different habitats with at least a small availability of water [@kolter_microbial_2006]. Biofilms are aggregated microbial cells, which are often embedded in an EPS matrix and attached to a surface [@vert_terminology_2012]. Many bacterial, archaeal and eukaryotic organisms besides the diatoms possess the capacity to form biofilms. The EPS matrix contains carbohydrates, proteins, proteoglycans, extracellular DNA, and modulates abiotic factors, such as moisture, electrochemistry, mechanical stability, and others [@flemming_eps_2007]. For example, it may adsorb nutrients from the surrounding medium, so that their availability within the biofilm is increased. Conversely, toxicants may be excluded, or sequestered to insoluble end stages [@hullebusch_role_2015], so that their toxicity to cells within the biofilm is reduced. Because of these properties, biofilms offer favourable conditions for many species, including diatoms.
Biofilms are not only beneficial for the cells embedded in them, but also for outside organisms, as well as for the ecosystems as a whole.
Primitive animals for example may have co-evolved with the O2-producing cyanobacterial biofilms
before the oxygenation of Earth's oceans [@gingras_possible_2011].
More obvious and present is the utility of biofilms as a food source
for grazing invertebrates [@poff_herbivory_1995], as well as fish [@carpentier_feeding_2014] and even birds [@kuwae_biofilm_2008].
Humans utilise various types of biofilms not as food itself, but for food production, such as aquacultural shrimp rearing [@thompson_importance_2002] or abalone larvae settlement [@stott_testing_2004].
On the ecosystem level, phototrophic biofilms contribute to the mechanical stability of sediment due to their secretion of adhesive organic compounds (see @widdows_impact_2002 for a review).
Because biofilms fulfil such crucial ecosystem functions,
their increased understanding, protection and appropriate application
is a crucial part of environmental management by us humans.
Diatoms and other phototrophic organisms, of course depend on the availability of photosynthetically active radiation (PAR) in their habitat. Thus, diatom biofilms are most often found in riverine, lacustrine [@kwandrans_diversity_2007] and intertidal [@sahan_community_2007] ecosystems. Incidently, these are the aquatic habitats that are also populated with anthropogenic structures. The undesired colonisation of man-made structures is called biofouling and negatively impacts shipping and other machinery. This impact occurs as increased hull drag, as higher maintenance costs due to clogging, as well as the costs of biofilm removal [@molino_biology_2008]. Biofouling research has been conducted on the most prevalent biofouler species, and has achieved important insights into the biofilm formation processes. For example, the pre-conditioning of surfaces by heterotrophs speeds up the development of phototrophic biofilms [@roeselers_heterotrophic_2007], which in turn were found to mediate the adherence of Bryozoa larvae [@dahms_effect_2004]. It is unclear how well the models of prokaryotic biofilm formation [@stoodley_biofilms_2002; @monds_developmental_2009] can be applied to eukaryotes such as diatoms and other algae, but succession clearly occurs in underwater habitats just as it does in terrestrial ecosystems. There are anti-fouling strategies to inhibit the undesired formation of biofilms, but these are often accompanied by toxic side-effects on the environment [@karlsson_practical_2006]. Therefore, biofilm control strategies need to carefully balance economic interests with ecological consequences. Of particular interest are the chemical cues that foster the formation and resilience of biofilms, so that they may be countered in non-cytotoxic ways.
In the soil, the surface of plant roots was recognised as the interface of complex biochemical interactions between plants and bacteria in the early 20^th^ century, and summarily termed the "rhizosphere" [@hiltner_uber_1904; @hartmann_lorenz_2007]. There, both fungi and bacteria interact with the plant rhizome and each other, in both symbiotic and parasitic manners. In particular, the exchange of nutrients was understood to mutually influence the interaction partners and the microenvironment, in which they interact with each other. In the algae communities, this understanding was summarised by @bell_chemotactic_1972 in the "phycosphere" concept. Here too, the immediate surrounding of the cells is understood as the space of their interaction, which influences both their own co-evolution as well as their common impact on the ecosystem [@amin_interactions_2012].
In particular, metabolites, toxins and other chemical cues mediate these interactions (see review by @ianora_new_2006 and references therein). These chemical cues are often called "signalling" or "messenger" molecules, which implies evolution of the production and secretion mechanisms on the side of their producers, as well as sensing mechanisms on the side of the recipients. This surely is the case in many interactions, but in a broader ecological context, a more useful term may be "infochemical". Infochemicals are understood from the point of effect on the recipient [@dicke_infochemical_1988], regardless of the specific nature or source of a chemical cue. The recipient has evolved to detect infochemicals, and to react to them. The information is in the olfactory pathways of the beholder, so to speak. Although infochemical production may not have evolved to fulfil an information transfer function, this possibility is of course not excluded from the understanding of chemical cues as infochemicals. For example, environmental phosphate availability has been linked to diatom-specific cyclin responses [@huysman_genome-wide_2010]. Regardless of their source, phosphates can in this context be considered as infochemicals, to which diatoms can adjust the regulation of their cell cycle. In summary, infochemicals are naturally occurring substances that trigger a physiological or behavioural response in an organism to some aspect of its biotic or abiotic environment. However, non-natural chemicals can fulfil such functions as well (e.g. act as pheromones), and unintentionally influence the natural chemical communication. In ecology and ecotoxicology, this understanding is summarised under the term "infochemical effect" [@klaschka_infochemical_2008].
Both nutrients and toxins can be understood as infochemicals, because they both affect intracellular processes of the recipient. Examples for beneficial effects of infochemicals are the increased proliferation of certain bacteria when incubated together with diatoms [@grossart_interactions_1999], and algae that depend on bacteria for essential vitamins [@croft_algae_2005]. Conversely, antibiotic infochemicals are for example algicidal chemicals produced by bacteria [@lovejoy_algicidal_1998], bacterial compounds that inhibit the swarming of other bacteria [@bottcher_chimeric_2014] and algal products that inhibit bacterial biofilm formation [@ren_inhibition_2002]. The types of the chemical mediators of these interactions are diverse: small molecules (above citations) as well as proteases have been found [@paul_interactions_2011].
In the more specific context of diatoms, some chemical cues are released only upon cell death and then act as infochemicals to other organisms. For example, diatom cells that are damaged by grazing copepods release unsaturated aldehydes. These reduce the reproductive success of the copepods, and thus reduce the grazing load on the diatom culture in the long-term [@pohnert_diatom/copepod_2005]. Such aldehydes have also been found to affect heat-shock protein expression in sea urchin embryos [@romano_nitric_2011] and the transition from vegetative to reproductive stages in the development of sea squirt larva [@castellano_diatom-derived_2015]. The proposed mode of action of unsaturated aldehydes in such animals involves their intracellular nitric oxide (NO) messaging pathways to the effect that the aldehydes may reduce NO levels, which in turn may modify gene regulation downstream. @bruckner_growth_2011 reported that the secretion of extracellular organic chemicals by several freshwater diatoms is triggered by bacterial infochemicals. Their chemical identification however, is ongoing. Recently, @amin_interaction_2015 showed that Pseudo-nitzschia multiseries cell division is promoted by the Sulfitobacter-derived indole-3-acetic acid (IAA) in marine settings. Thus, specific chemical relationships between bacterial sources and diatom recipients are a current topic with high interests in the identification of chemical cues.
\sectionmark{\emph{A. minutissimum} as a model organism}
The diatom A.\ minutissimum (Kützing) @czarnecki_freshwater_1994 is a cosmopolitan species complex of early colonisers in freshwater habitats [@round_four_1996; @johnson_changes_1997; @potapova_morphological_2007]. A.\ minutissimum is found in the littoral zone along the shore of Lake Constance, where it forms photoautotrophic–heterotrophic biofilm communities with satellite bacteria [@bahulikar_diatoms_2006]. Such bacteria have previously been shown to modulate A.\ minutissimum's organic secretions such as EPS and amino acids [@bruckner_bacteria_2008; @bruckner_growth_2011]. Because these modulations occurred also upon treatment with spent bacterial medium, A.\ minutissimum apparently reacts to soluble chemical substances produced by the bacteria. This diatom's fast colonisation of new substrates is facilitated by the quick formation of adherence-providing EPS structures such as stalks. These structures consist of aggregated EPS at the apical part of a cell, which grow into a shaft [@wang_extracellular_1997]. Besides stalks, A.\ minutissimum also produces EPS capsules, which envelope the whole diatom cell (Fig.\ \ref{frustule-vs-capsule}). In contrast to stalks, capsules appear later in the culture's growth phase, but their function is less apparent and may be manifold [@lewin_capsule_1955; @geitler_entwicklungsgeschichtliche_1977]. Capsules have been suggested to play a role in substrate attachment, nutrient capture, reproduction, and grazer defence. Additionally, they appear to provide A.\ minutissimum with a mechanical barrier against bacteria [@windler_purification_2012].
A.\ minutissimum has been established as a laboratory model organism in the form of xenic cultures (i.e.\ associated with naturally co-occurring bacteria) and axenified suspension cultures [@myklestad_rate_1989; @windler_purification_2012]. Moreover, A.\ minutissimum has been used as an in situ biomonitor for heavy metals in the environment. Frustules react to increased heavy metal levels by deformations, which can be quantified microscopically [@morin_scanning_2008;@falasco_morphological_2009; @cantonati_diatom_2014]. A.\ minutissimum can also be useful to archaeologists, because its abundance was found to correlate with anthropogenic increases in turbidity and eutrophication [@rodriguez-ramirez_climatic_2015]. Although preliminary, such results may confirm earlier findings that A.\ minutissimum is comparatively more resilient than other diatoms; In particular against mechanical stress [@peterson_resistance_1992]. In summary, A.\ minutissimum is an ecologically relevant model organism with versatile application options in the laboratory and in the field. However, it is underrepresented in biofilm research, which often focusses on marine species due to their economical impact on shipping by biofouling of hulls.
Light microscopy (LM) and electron microscopy (EM) have been developed up to their respective physical limits largely independently. Utilising stimulated fluorescence emission and depletion, LM has even breached the previously dogmatised Abbe diffraction limit [@klar_subdiffraction_1999; @hell_microscopy_2009]. Correlation techniques between these two microscopy methods enable a two-step approach to many imaging projects: observe many events or structures of interest using LM, and subsequently investigate structural details by EM (see @mironov_correlative_2009 for a review). CLEM techniques have been applied to a wide variety of sample types: from crystal grains [@wilding_correlative_1973] to small model animals [@kolotuev_precise_2010]. The correlation results from the ability to find the same locations within a sample with both microscope types. Several techniques are able to facilitate the correlation (reviewed by @sosinsky_markers_2007), such as utilising landmarks within the sample or the direct labelling of the locations of interest. Using landmarks, one has to take into account that samples structures may deform due to the harsh EM preparation procedures of fixating, freezing and drying [@hoagland_diatom_1993]. Direct labelling requires staining of the structures of interest with for example green fluorescent protein (GFP) and other fluorophores. These can double as catalysts for the photo-oxidation of 3',3'-diaminobenzidine into a polymeric precipitate, which can in turn be stained with the electron-dense osmium tetroxid [@maranto_neuronal_1982; @grabenbauer_correlative_2012]. Because proteins can be tagged with GFP, many intra- and subcellular structures become observable by CLEM. In the present thesis, observations were focussed on complete diatom cells, and a minimally invasive correlation technique for their in situ marking in biofilms was tested.
This thesis approaches the chemical communication of diatoms and bacteria from a methodological standpoint. It is the opinion of the author that the quality of scientific work is determined largely by the quality of the available techniques and tools. It therefore became a central theme of this thesis to develop and improve methods of investigating biofilms, and to make them available to the research community.
Chapter \ref{BF-form} will present a biofilm model system based on A.\ minutissimum and a Bacteroidetes strain. The model system was developed to advance the screening of a variety of sample types for biofilm-inducing effects. We additionally asked, which influence the bacterium has on the EPS production of the diatom.
Chapter \ref{signal-extraction} will present work on the up-scaling and fractionation of bacterial supernatants, guided by the afore-mentioned bioassay. It was hypothesised that a multi-step liquid-liquid extraction combined with a solid-phase extraction could extract biofilm-inducing fractions from bacterial supernatant.
Chapter \ref{assay-opt} will explain workflow automations and optimisations. How to remove bottlenecks in the measurement and data processing? And how to increase sample throughputs and replicate numbers? Our answers to these questions enhance the applicability of the bioassay for large sample sets, such as bacterial mutant strains.
Chapter \ref{capsule-microstructure} will take an electron microscopic view into the A.\ minutissimum biofilms, and elucidate the microstructure of this diatom's EPS capsule. We solved the problem of finding the exact same cells in both light and electron microscopy with a simple, biofilm-compatible technique. Moreover, a novel microstructure type in freshwater diatoms will be described, and a model for the capsule formation will be proposed.