- Split View
-
Views
-
Cite
Cite
Nils Meyer, Arite Bigalke, Anett Kaulfuß, Georg Pohnert, Strategies and ecological roles of algicidal bacteria, FEMS Microbiology Reviews, Volume 41, Issue 6, November 2017, Pages 880–899, https://doi.org/10.1093/femsre/fux029
- Share Icon Share
Abstract
In both freshwater and marine ecosystems, phytoplankton are the most dominant primary producers, contributing substantially to aquatic food webs. Algicidal bacteria that can associate to microalgae from the phytoplankton have the capability to control the proliferation and even to lyse them. These bacteria thus play an important role in shaping species composition in pelagic environments. In this review, we discuss and categorise strategies used by algicidal bacteria for the attack on microalgae. We highlight the complex regulation of algicidal activity and defence responses that govern alga–bacteria interactions. We also discuss how algicidal bacteria impact algal physiology and metabolism and survey the existing algicidal metabolites and enzymes. The review illustrates that the ecological role of algicidal bacteria is not yet fully understood and critically discusses the challenges in obtaining ecologically relevant data.
INTRODUCTION
The key players living in the Earths’ oceans and lakes are photosynthetic unicellular microbes, and their associated bacteria, viruses and protist grazers. These organisms are found universally in freshwater and marine environments, from coastal zones to open water. Phytoplankton on average accounts for ∼50% of global primary production by photosynthetic activity (Behrenfeld et al.2006). It also acts as a sink of inorganic nutrients that are mobilised upon predation or cell death and concomitant lysis. Plankton species composition on both macro- and microscale is affected by biotic interactions as well as by temporal and spatial fluctuations of nutrients, resources and temperature (Dann et al.2016). It is estimated that up to 90% of phytoplankton-derived organic matter is recycled by heterotrophic bacteria within the microbial loop (Azam et al.1983). Such high turnover rates of phytoplankton biomass substantially impact the balance of biomass formation and respiration in aqueous systems (Bidle and Falkowski 2004). Perpetual changes in species abundance and community assemblages are shaping the plankton communities over time. These dynamic processes are often driven by intra- and interspecies interactions. Heterotrophic bacteria, for instance, can dwell on resources excreted by living phytoplankton cells or resources released after algal cell death and lysis (Bidle and Falkowski 2004). These heterotrophic bacteria can follow different trophic strategies as a response to resource availability resulting in chemical niche specialisation (Hunt et al.2008; Hansell et al.2009; Kujawinski 2011; Mayali et al.2014). During a phytoplankton bloom (a massive algal proliferation event driven by e.g. light, temperature and nutrient availability), the algal-associated microbial community can shift from being dominated by oligotrophic Alphaproteobacteria to copiotrophic Flavobacteria that grow on senescent algae and to Gammaproteobacteria that profit from products released by the degradation of algal biomass (Teeling et al.2012; Ruff et al.2014). Nutrient exchange is frequently the driver of alga–bacteria interaction, irrespective of whether the relation is symbiotic or opportunistic. Field observations suggest that phylogenetic diversity and specificity of bacteria is correlated to phytoplankton groups (Jasti et al.2005; Bunse et al.2016) or even to phytoplankton species (Pinhassi et al.2004; Amin et al.2015). In an impressive, comprehensive study of reoccurring patterns in bacterioplankton dynamics, metagenome analyses revealed recurrent patterns at the functional level, in particular with respect to algal polysaccharide degradation genes. It was concluded that despite variability between spring phytoplankton blooms, the succession of bacterial clades is driven by deterministic principles such as substrate-induced forcing (Teeling et al.2016).
Within this review, we specifically focus on the group of algicidal bacteria that inhibit microalgal growth or actively lyse unicellular algae. Other algicidal microorganisms that are not considered in this review have been the subject of intensive studies and of reviews, including work on viruses (Tomaru et al.2008; Rohwer and Thurber 2009; Martiny et al.2014) or fungi (Gerphagnon et al.2015). The investigation of the strategies of algicidal bacteria is of substantial interest for scientists from diverse fields including biogeochemistry, algal physiology, ecology and environmental sciences, and aspects from all these disciplines will be covered here. In addition, several applications of algicidal bacteria in biotechnology and environmental engineering are currently under development. These are not the subject of this contribution but are reviewed elsewhere: current and potential applications of algicidal bacteria in biofuel production are reviewed in the study by Wang et al. (2016b). For such purposes bacteria might be used to stimulate high-value product formation, to support harvesting by promoting cell aggregation or to perform cell lysis thereby reducing energy-intensive processing. The specific use of algicidal microorganisms and algicides to induce microalgal cell disruption for biomass processing is specifically addressed in a review by Demuez, Gonzalez-Fernandez and Ballesteros (2015a). Natrah et al. (2014) summarised the role of alga–bacteria interactions in aquaculture. The potential to control harmful algal blooms using algicidal bacteria is discussed in several recent reviews (Gumbo, Ross and Cloete 2008; Kim et al.2008; Jancula and Marsalek 2011; Seong and Jeong 2013; Shao et al.2013). We also want to refer here to review articles on specific organismic groups or species that include aspects on the impact of algicidal bacteria. Reviews with a focus on cyanobacteria (Dahms, Ying and Pfeiffer 2006; Leflaive and Ten-Hage 2007; Van Wichelen et al.2016), diatom–bacteria interactions (Amin, Parker and Armbrust 2012) or Pseudoalteromonas spp. (Holmstrom and Kjelleberg 1999) are noteworthy in this context. In this review, we exclude the numerous studies that claim the investigation of algicidal bacteria or algicides without any ecological rationale. Here, metabolites are introduced that either are not occurring in the environment in relevant concentrations or the interactions of species are investigated that would never encounter in nature. While these studies might be useful for biotechnological purposes, they do not contribute to the ecological and functional understanding of algicidal activity in the plankton, which is in the focus of this contribution.
A most influential review defining the field has been published in 2004 by Mayali and Azam; it summarises nicely the state of research at that time and systematically defines ways to consider ecological relevance in such plankton interactions (Mayali and Azam 2004). Examining the Mayali review, it becomes evident that substantial progress has been made, mainly driven by field and mesocosm experiments, genetic methods, elaborated bioassays and advances in chemical analytics. In this contribution, we focus on this progress with a special emphasis on emerging possibilities to address algicidal bacteria within their natural environment. We place special emphasis on algicidal metabolites, regulation of the interaction between algicidal bacteria and their target organisms and the ecological consequences of alga–bacteria interaction.
Scales of interactions
It is often suggested that algicidal bacteria have a massive impact on the productivity of aquatic systems by modulating entire communities even if, as outlined below, solid experimental prove for this hypothesis is difficult to obtain. Most yet described algicidal bacteria are lytic to phytoplankton and are frequently succeeding algal blooms of diatoms, coccolithophores, Phaeocystis spp. as well as often toxic species belonging to the groups of dinoflagellates and raphidoflagellates (Demuez, Gonzalez-Fernandez and Ballesteros 2015a). These bacteria can interact with freely floating unicellular or colonial algae, but considerations of concentrations in the open water make it likely that excretion of active algicidal metabolites would only be a feasible strategy during conditions with high cell densities (Fig. 1). It can be thus argued that interactions between freely floating organisms might not be very widely distributed.
Interactions in the phycosphere
The coupling between primary phytoplankton production and secondary production by heterotrophic bacteria creates a very heterogeneous environment (Azam and Malfatti 2007). Observed spatial heterogeneity is often driven by the formation of microbial clusters around nutrient patches and gradients. Such three-dimensional and temporal niches can originate from the influx of nutrients from terrestrial environments, from exudates of larger organisms, cell lysis and many other events, thereby creating highly productive hotspots. Independent of the nature of alga–microbe interactions, the regions around algal cells that are termed ‘phycosphere’ have to be considered (Fig. 1) (Bell and Mitchell 1972). The phycosphere is mainly enriched with fixed carbon contributed by the algae but also signalling compounds and allelochemicals can accumulate within this zone. It creates a high-nutrient, semiprotected area in which mass transport is limited to diffusion and not influenced by turbulence. The underlying reasons for creating such specific nutrient-rich surroundings by phytoplankton are diverse but not fully understood. The exudates can function as a chemical defence (Dutz, Breteler and Kramer 2005) or provide allelochemical activity (Van Donk 2007), but excretion of organic matter has also been linked to photoprotection (Cherrier et al.2015) and protection against heavy metal stress (Strmecki et al.2010). Excreted dissolved organic carbon (DOC) might also mediate mutualistic interactions with non-algicidal bacteria (Aota and Nakajima 2001) and even symbiotic trophic interaction between bacteria and phytoplankton is regulated within the phycosphere (Wagner-Döbler and Biebl 2006; Armbrust 2009; Geng and Belas 2010; Amin, Parker and Armbrust 2012; Cooper and Smith 2015; Kouzuma and Watanabe 2015; Ramanan et al.2016). The composition of algal exudates varies between phytoplankton species and even between growth phases of one species, thereby challenging the adaptive capabilities of bacteria (Barofsky, Vidoudez and Pohnert 2009; Weber et al.2013). In accordance, growth efficiency and metabolic activity of heterotrophic bacteria can be dynamically influenced by such algal exudates (Azam and Malfatti 2007; Sapp et al.2007; Natrah et al.2014). On the other hand, it has also to be considered that algae generate a chemical ‘fingerprint’ that might guide motile bacteria towards nutrient hot spots. Since the use of carbon derived from phytoplankton differs among bacterioplankton, the specific excretions will directly influence bacterial community composition (Sarmento and Gasol 2012).
Due to methodological restrictions arising from the analysis of unicellular algae, the direct demonstration of chemical gradients within the phycosphere is difficult. The monitoring of this zone and studies how it shapes the associated microbiome has mainly been done with macroalgae and biofilms as study objects (Grosser et al.2012; Saha et al.2014). Recently, a very elegant study revealed spatiotemporal patterns of bacteria around individual Chaetoceros affinis diatoms that can be clearly explained with a response to a phycosphere (Smriga et al.2016). By combining video microscopy and modelling, the role of bacterial chemotaxis in the resource exploitation around lysed algal cells could be explained. It can thus be regarded as established fact that secretions of unicellular algae provide noticeable amounts of dissolved organic matter (DOM) to the outside environment and these carbon hotspots are valuable resources for bacteria that would elsewhere have to deal with much more oligotrophic conditions (Amin, Parker and Armbrust 2012). To enter the phycosphere, bacteria are dependent on encounters. Encounter rates can be increased by directed motility (Barbara and Mitchell 2003; Xie et al.2011), and the duration of an interaction can be increased by association with the phytoplankton cells by adhesion (Grossart et al.2005; Slightom and Buchan 2009; Grossart 2010). Colonisation can lead to complex associations of both photoautotrophs and heterotrophs that are often surrounded by a joint polysaccharide dominated matrix (Fig. 1). Ultimately, the microbiome may be vertically transmitted to subsequent generations relieving the bacteria of the need of anew encounters. Bacterial successions within high-productive areas such as the phycosphere are thus generally governed by physiological and behavioural responses of the bacteria. Complex signalling chemistry involving specific chemoattractants or quorum sensing (QS) regulators can mediate these processes (Cooper and Smith 2015). Consequently, heterotrophic bacteria have to cope with diverse settings and have developed means to actively shape their environment. The diverse interaction situations in the plankton and the phycosphere have surely contributed to the evolution of a multitude of adapted algicidal strategies.
Algicidal bacteria and their lifestyle
Most known algicidal bacteria belong either to the phyla Bacteroidetes or Gammaproteobacteria (such as Alteromonas, Pseudomonas and Pseudoaltermonas) but algicidal Alphaproteobacteria are also reported (Kirchman 2002; Goecke et al.2013). Only a minority are Gram positive such as members of the phylum Actinobacteria and the genus Bacillus (Firmicutes) (Mayali and Azam 2004; Zhou et al.2015). Isolates of algicidal bacteria have been obtained from diverse regions of the oceans; in fact, surface waters (e.g. Jung et al.2008), algal bloom samples (Park et al.2010) and sediments (Lenneman, Wang and Barney 2014) are rich reservoirs. These bacteria have adapted their strategies to the diverse aqueous environments. Based on the available data, it is however hard to predict under which situations or in which environments the prevalence of algicidal bacteria is highest. DeLong, Franks and Alldredge (1993) observed that the bacteria in macroaggregates and marine snow are dominated by algicidal Cytophaga (Bacteroidetes) and Gammaproteobacteria. A systematic screening, however, is still lacking but first studies already point towards an increased hit rate in particle-associated compared to free-living bacteria and a generally low host specificity (Park et al.2010). This is in agreement with the observation that nutrient shuttling between algae and bacteria is the basis for these microbial interactions. A high nutrient availability as it is found in the phycosphere is often a prerequisite for the production of algicides. Under laboratory conditions, algicides are often released under optimised culturing conditions at high cell densities (Mayali and Doucette 2002; Roth et al.2008). This mirrors an environmental situation with increased algicide expression rates at later phytoplankton bloom phases or in nutrient hotspots. The observed preferred algicide production at high cell densities might be additionally explained with the involvement of bacterial QS mechanisms controlling algicide production in a density-dependent way (Williams et al.2007). Interestingly, a change in lifestyle depending on the nutrient availability or cell density has been observed for bacteria that can switch from an algicidal to a symbiotic strategy (Amaro et al.2005; see also section Regulation).
The variety of microbial strategies to take advantage of algal primary production is remarkable. Besides the use of algal exudates as a resource, microbes can also actively modulate algal activity and performance to increase their supply of nutrients. Two fundamentally different lines of attack have been identified. One is only effective in situations where direct contact or close proximity of the algicidal bacteria with the host organism is found (Fig. 1). The other relies on the indirect interaction via excreted diffusive algicides. The contact-based algicidal activity has been shown for numerous bacteria (see for example Wang et al.2005; Roth et al.2008), whereas examples for non-contact interactions are less frequently reported (Paul and Pohnert 2011; Li, Geng and, Yang 2015). Even unusual modes of action have been observed; for example, a freshwater Saprospira sp. (Bacteroidetes) forms bundle-like structures that are suspected to be involved in catching cyanobacterial prey before lysis (Shi et al.2006). The most drastic way how algicidal activity is manifested is clearly the lysis of algal cells. Other activities that are also summarised under the concept of algicidity include the induction of morphological changes, growth inhibition modulated by excreted factors or nutrient competition as well as the immobilisation of previously mobile algae (Wang et al.2005; Mu et al.2007; Bai et al.2011; Lenneman, Wang and Barney 2014; Demuez et al.2015b; Li et al.2016).
Specificity of algicidal bacteria
The specificity of algicidal activity is a central topic if environmental engineering by the addition of bacteria is envisioned. Investigation of specificity also offers opportunities to understand the mode of action of algicides and their targets. As a consequence, many studies survey for algicidal activity using a broad selection of algal test organisms. Specificity is commonly characterised by toxic threshold concentrations but unfortunately, most of these experiments are not comparable since the species tested are arbitrarily selected and the assays are not standardised. Many algicidal bacteria show a broad host range and inhibit or lyse microalgae from different phyla (Skerratt et al.2002; Roth et al.2008; Park et al.2010). Others are only active against specific groups or certain species. However, even within species selectivity was observed with an algicidal bacterium from the phylum Bacteroidetes. It was active against three Karenia brevis isolates while six other isolates of the same species were unaffected. The same bacterium was also active against two out of five Alexandrium isolates indicating a highly random, unexplained specificity (Roth et al.2008). Park et al. (2010) surveyed patterns of algicidal specificity and observed that 231 of 487 particle-associated bacterial strains were algicidal while only 55 of 249 free-living bacterial isolates showed activity. Interestingly, 80% of particle-associated strains could kill multiple algal species, whereas 75% of the isolates of free-living bacteria killed only one of the six tested dinoflagellates and raphidophytes. If this pattern universally holds true, this finding might support strategies for the isolation of host-specific bacteria.
In the case of Cytophaga sp., results on algicidal specificity in lab experiments could indirectly be linked to field observations. The Cytophaga sp. was initially isolated during the bloom of the harmful red tide alga Chattonella antiqua. In laboratory co-incubation experiments, it was proven to be rather unspecific and lysed diatoms, dinoflagellates and raphidophytes (Imai, Ishida and Hata 1993). This lack of specificity was also reflected in the bacterial abundance during field surveys. Cytophaga concentrations reached highest values after the peak of C. antiqua. However, the abundance of this algicidal bacterium also followed the total amount of microalgal biomass even when C. antiqua was absent, indicating that in field situations, it can also profit from resources provided by a wider selection of algae (Imai et al.2001). Several promising results have been obtained from studies on mixed species assemblies in micro- and mesocosms. An important set of experiments demonstrating nicely the effects on entire communities was performed with an algicidal bacterium Shewanella sp. (Gammaproteobacteria). Extracts of this bacterium are specifically algicidal against dinoflagellates, while a chlorophyte, a cryptophyte and a diatom tested were not affected (Pokrzywinski et al.2012). The active extract exhibited photosystem II inhibition and induced loss of cell membrane integrity but no conclusive cellular targets could be identified (Tilney et al.2014b). In mixed laboratory microcosms initiated from three natural dinoflagellate blooms, the Shewanella extract resulted in decreased effects in communities compared to unicellular cultures. The entire eukaryotic community composition was, however, shifted to increasing proportions of heterotrophic protists. Diatom abundance also increased at low algicide concentrations. In addition, ciliate abundance, as well as a bactivorous chrysophytes, increased (Tilney et al.2014a). This shift toward communities dominated by heterotrophic and bacterivorous organisms is often observed in treatments using algicidal bacteria. Remarkably, in the study cited the effect was also observed in the absence of algicidal bacteria after treatment with the algicidal extract. This observation points towards a determining effect of the organic substances released upon cell lysis. Another example of the effect of specific algicides was reported by Jung et al. who identified a Pseudomonas fluorescens strain that was introduced as a candidate for the control of the freshwater diatom Stephanodiscus hantzschii. In indoor mesocosms with natural species assemblies collected during an S. hatzschii bloom, they confirmed the reduction of this alga by 88% while the amount of other phytoplankton remained unaffected (Jung et al.2008). A remarkable specificity was also observed during microcosm incubations with a not further characterised algicide extracted from the actinobacterial Streptomyces alboflavus (Cai et al.2016). This algicide is active against Phaeocystis globosa and could remove this species from microcosms without affecting others. Exposure to the algicide led to the instant increase of the plasma membrane permeability and disruption of the photosynthetic system, but these activities could not explain the underlying mode of specificity.
Regulation of algicidal interactions
It can be hypothesised that pronounced regulation of activity will be required for bacteria to maintain efficient algicidal strategies in an environment with spatial and temporal heterogeneity. On the other hand, algae might respond with induced resistance or even defence mechanisms against the attacker in situations with fluctuating pathogen pressure. Only a few pioneering studies have reported on ecologically relevant regulated interactions between algicidal bacteria and host algae. Besides dilution effects, a complicating factor in environmentally realistic studies is the complexity of consortia with multiple species that might interact in a complex network. In the following, concepts and, if available, first examples of regulation during interactions will be introduced.
Chemotaxis
Finding of prey species might be supported by the chemotactic behaviour of some bacteria, directing them along chemical gradients towards microalgae (Fig. 2a). In most reported cases, a positive chemotaxis could be observed, leading the organism to higher concentrations of the signal (Sonnenschein et al.2012; Smriga et al.2016). Thus, for example, Li et al. (2016) could demonstrate in a comprehensive study that Chitinimonas prasina (Betaproteobacteria) behaves chemotactically towards the diatom Thalassiosira pseudonana. Chemotaxis of the flagellated cells is directed towards the algae via a yet unknown cue and upon encounters the cells adhere on the algae with their flagella. The bacterium uses chitinases to metabolise chitin from the cell walls of diatoms. Chitin digestion leads to loss of cell integrity in the diatoms and supplies the bacteria with nutrients and energy. Contact between bacteria and the algae is a prerequisite for algicidal activity. Interestingly, all tested diatoms besides Phaeodactylum tricornutum were lysed, which is in accordance with the observation that this is also the only diatom that lacks chitin in its cell walls.
Nutrient cues
Nutrient availability could also be a possible cue to trigger algicide production in bacteria. However, evidence for such mechanisms is very limited and rather indirect. This lack of knowledge might be attributed to the difficulty to keep bacteria under fully nutrient-deprived conditions, which would be a pre-requisite for further experiments on the topic. This experimental limitation might explain that such mechanisms have only been reported for the cyanobacterium Oscillatoria laetevirens, where algicide production is regulated by nutrient availability (Ray and Bagchi 2001). Under conditions that limit growth (e.g. ammonia as N-source or phosphate limitation), stimulated algicide production was observed while in optimum growth conditions (e.g. organic N-source) the algicide production was abolished.
Quorum sensing
The cell-to-cell communication at both intra- and interspecies level is another factor to be considered if the dynamics of algicide production is concerned. The mechanism of concentration-dependent cell-to-cell-communication called QS was first studied with the marine Gammaproteobacterium Vibrio fischeri (Aliivibrio fischeri) nearly 40 years ago by Nealson and Hastings (1979). They showed that the bacterial enzyme luciferase responsible for bioluminescence occurred only at high cell densities and is under the control of secreted autoinducer signalling molecules. Bacterial behaviour regulation under the concentration-dependent regulation by autoinducers was in detail described in reviews by Bassler (Bassler 1999; Waters and Bassler 2005). QS is used by bacteria to initiate processes with high energy or resource costs only when they are beneficial to the population (Keller and Surette 2006). Gram-negative bacteria use not only acylated homoserine lactones (AHLs) as autoinducers (Fuqua and Greenberg 2002) but also 4-quinolon amide (Pesci et al.1999; Mashburn-Warren et al.2008), fatty acids and fatty acid methyl ester with similar function were reported (Deng et al.2010). In planktonic Gram-negative and Gram-positive bacteria, furanone complexes (AI-2) also induce genetic expression (Skerratt et al.2002; Williams et al.2007). Due to their lifestyle in a highly diluted changing environment, it has early been postulated that algicidal bacteria regulate their activity by means of QS. However, until now, only a few studies were able to unambiguously prove the connection of the two processes (Fig. 2b). An indirect evidence for QS-type regulation of algicidal activity was provided by Nakashima et al. (2006b). They show that a prodigiosin-type pigment PG-L-1 (see section Alkaloids), isolated from a Gammaproteobacterium, has algicidal activity against various red tide phytoplankton species. The production of the pigment PG-L-1 is completely inhibited in the presence of erythromycin or chloramphenicol but can be recovered by the addition of AHLs. This induced production and the fact that the pigment production was inhibited by beta-cyclodextrin, an inhibitor of QS activities, are in fact pointing towards a control of prodigiosin production by QS. Other marine Gammaproteobacteria such as Serratia sp., Pseudoalteromonas bacteriolytica and V. ruber also produce prodigiosin-like pigments (Lewis and Corpe 1964; Sawabe et al.1998; Shieh et al.2003). For Serratia sp., production is also controlled by AHLs (Thomson et al.2000). The investigation of the exometabolome of the marine P. piscicida revealed the bacterial signalling molecule to be 2-heptyl-4-quinolone (see section Alkaloids). This metabolite is not only a QS mediator, but also induces slow lysis in three strains of Emiliania huxleyi at biologically relevant nanomolar concentrations (Harvey et al.2016). The quinolone thus mediates cross-kingdom interactions that can shift the phytoplankton population dynamics and trigger the intraspecies cell-to-cell communication (Diggle et al.2007; Dubern and Diggle 2008). Paul and Pohnert (2011) studied the lytic activity of Kordia algicida (Bacteroidetes) on the diatom Skeletonema costatum and found that a concentration-dependent growth-inhibiting or lytic effect of the bacteria was caused by diffusible substances. Further tests showed that the algicidal agent is a protease (see section Enzymes) released by the bacteria. Induction with the spent medium of dense K. algicida cultures led to an increased protease release, which is a first evidence for a QS-regulated process. Further studies are however necessary to identify the mode of regulation in this system.
Quorum quenching
In general, the use of signal molecules for the control of vital functions makes the producer susceptible to development of resistance or to interference strategies targeting the signal molecules (reviewed in Grandclement et al.2016). There is much evidence from marine bacteria that they employ so-called quorum quenching (QQ) strategies in which QS signalling is suppressed. Identified mechanisms include the inactivation of N-acyl homoserine lactones by lactonases or by cleavage of the amide bond through the action of acylases (Yates et al.2002; Romero et al.2008; Safari et al.2014). Some bacteria and microalgae are also able to locally raise the pH which can affect the stability of QS molecules in their surroundings (Yates et al.2002). There are several examples for QQ in the marine environment (Romero, Martin-Cuadrado and Otero 2012). Especially in dense marine microbial communities, such as sediments, biofilms and on surfaces of algae, a large number of marine bacteria interfering with AHL signalling could be found (Romero et al.2011). A broader survey revealed that 84 isolates out of 464 from estuarine and oceanic water (0 and 10 m depths) are able to interfere with C6- and C10-AHL signalling. Prevalence of QQ mechanisms is lower in estuarine isolates (2% of the investigated samples) than in oceanic samples (28%) (Romero, Martin-Cuadrado and Otero 2012). Given this wide distribution, it will be of interest to specifically address the effect of QQ on algicidal mechanisms. Especially in the close contact situations within the phycosphere, biofilms or marine snow such mechanisms might become relevant. Several microalgae have also developed strategies for QQ or QS inhibition, but up to now no direct link to interference mechanisms with algicidal activity is reported. The diatom Nitzschia cf. pellucida, for example, produces a haloperoxidase which deactivates bacterial β-keto-AHL QS molecules. The disruption of these signal molecules is H2O2 dependent (Syrpas et al.2014). The alga also produces the secondary metabolite cyanogen bromide that is catalysing the cleavage of peptide bonds as they occur in AHLs (Vanelslander et al.2012). The unicellular green alga Chlamydomonas reinhardtii can secrete lumichrome that mimics AHLs from associated bacteria (Teplitski et al.2004; Rajamani et al.2008, 2011). The effects of 19 microalgal strains commonly used in aquaculture on QS-regulated gene expression of three reporter strains were studied. Two out of 19 tested algae inhibited the AHL-dependent pigment production in Chromobacterium violaceum (Betaproteobacteria). Reduced green fluorescent protein production in a QS-controlled reporter strain Escherichia coli was observed with six algae (Natrah et al.2011). Again, this broad distribution of complex signalling event prompts further investigation in the field of algicidal bacteria.
Evasive strategies
Evasive strategies are another possible form of resistance against algicidal activity. Algae have means to form morphologically distinct resting stages as a survival strategy that could potentially be employed to avoid contact with algicidal bacteria (Fig. 2c) (Anderson and Rengefors 2006; Mertens et al.2012). Once the conditions ameliorate, these resting stages can fuel a novel generation of proliferating algae. However, this interaction mechanism is purely theoretical since to our knowledge, no experimental evidence for the mechanism has been provided till to date. In general, marine bacteria can however induce such resting stages. Roseobacter sp. (Alphaproteobacteria), for example, induces cyst formation in the dinoflagellate Alexandrium tamarense that is suspected to play an important role in bloom dynamics (Adachi et al.1999). Whether this Roseobacter sp. exhibits algicidal activity is however not proven.
Induced resistance
A first example for an induced resistance against algicidal bacteria has been found in the diatom Chaetocerous didymus. This alga was the only resistant diatom in a screening performed with the algicidal bacterium K. algicida (Fig. 2d) (Paul and Pohnert 2011). Interestingly, the resistance of the algae goes ahead with an induced production of own proteases that are suspected to counteract the lytic enzymes of the bacteria (Paul and Pohnert 2013). These proteases are substantially more effective compared to the bacterial counterparts and can not only be triggered by the bacteria but also by signals contained in the >30 kDa fraction of a bacterial culture filtrate.
Complex interactions
A more complex interaction, potentially involving multiple regulators, has been identified in the Alphaproteobacterium Phaeobacter gallaeciensis (Seyedsayamdost et al.2011b). The ‘Jekyll-and-Hyde’ chemistry of this bacterium illustrates impressively that assigned functions of key organisms might change dramatically in a context-dependent way. P. gallaeciensis is a bacterium that interacts with the bloom-forming coccolithophore E. huxleyi. In situations where the E. huxleyi bloom is growing, P. gallaeciensis produces the auxin phenylacetic acid and the broad-spectrum antibiotic tropodithietic acid that support the development of the alga. However, once the algal bloom decays, the bacterium switches its lifestyle to an algicidal type. It releases the potent and selective roseobacticides that lyse the aging algae. The master switch triggering this dramatic change is p-coumaric acid, a lignin degradation product released by aging algae (Seyedsayamdost et al.2011b). p-Coumaric acid also triggers the production of a putative 226 g mol−1 roseobacticide that causes loss of motility and cell enlargement in phytoplankton (Sule and Belas 2013). Other highly complex interactions take place in the phycosphere of the dinoflagellate A. tamarense. In the presence of associated Alpha-, Beta- and Gammaproteobacteria, the growth of the alga is suppressed. The dinoflagellate, however, associates closely with the small flagellate alga Ochromonas sp. that ingests bacteria. This interaction, entirely taking place in the phycosphere, relieves bacterial pressure from the dinoflagellate and restores growth (Hu et al.2015). Cascading effects are also to be considered since alga-associated microbial communities might modulate the potency of algicides (Mayali and Doucette 2002; Roth et al.2008).
Algae are also supposed to be able to fend off algicidal bacteria by means of antibiotics. Most antibacterial metabolites were, however, found from marine microalgae by using cell lysates or extracts (Desbois, Mearns-Spragg and Smith 2009; Michalak and Chojnacka 2015) and thus the evaluation of their true ecological function is difficult. A compound class that has been highly investigated in this context is the oxylipin family from diatoms that are predominantly released upon cellular disruption but also within defined events during regular growth in cultures and the field (Vidoudez and Pohnert 2008; Vidoudez et al.2011a, b). Especially the polyunsaturated aldehydes produced by many diatoms have antibacterial activity (Adolph et al.2004; Ribalet et al.2008; Balestra et al.2011), comparable to similar metabolites found in higher plants (Blée 2002). These data suggest a possible defence mechanism of diatoms against microorganism including algicidal bacteria but the connection has not been shown in field studies till to date.
Bacteria–bacteria interactions are widely studied for members of the Roseobacter clade, which produce antibacterial substances potentially providing a competitive advantage e.g. during surface colonisation of macroalgae (Manefield et al.2002; Rao, Webb and Kjelleberg 2006; Nissimov, Rosenberg and Munn 2009; Goecke et al.2010; Berger et al.2011; Cude et al.2012; Cude and Buchan 2013; Dang and Lovell 2016). Principles within bacterial competition are reviewed by Hibbing et al. (2010). Long et al. (2003) linked production of antibiotic compounds by a particle-associated Alteromonas sp. to a multitude of functions as the metabolites shape not only the species composition of the bacterial community in a model particle system but they also inhibit growth of one cyanobacterium and three diatoms.
The complexity of the observed interactions and the close association of the involved organisms suggest that some algicides, especially those with known antibiotic activity, might also affect the microbial community and consequently also algal symbionts, resulting in the amplification of algicidal effects. On the other hand, alga-associated bacteria might also provide protection from algicidal bacteria either by the production of antibiotics or by metabolization of algicides. Since marine microbial communities are part of tightly connected networks, common investigations of bilateral interactions are most likely only scratching the surface of the hidden regulation of network dynamics (see also Amin et al.2015).
Methods for the investigation of algicidal activity
The sections above make it clear that a truly ecologically relevant investigation of algicidal principles is highly challenging. In order to distinguish between algicidal activity mediated by contact interactions and by diffusible metabolites, bioassays can be performed by either infecting algae directly with bacteria or by treating them with bacterial exudates in cell-free spent medium (Fig. 3a and b). In order to circumvent overlaying effects of remaining medium, bacteria can be grown on agar plates and resuspended. However, care has to be taken since algicidal activity is often overlooked if it is under the control of signalling chemicals. In these cases, induction of the production of algicidal metabolites or QS-based upregulation of algicidal activity has to be performed before the generation of extract or cultures for bioassays (Fig. 3d and e). Non-contact co-culturing is an approach that covers both the above-mentioned mechanisms of regulation (Fig. 3c). It is also a reliable method to determine if an observed activity is dependent on cell–cell contact (Kim et al.2009b; Paul, Mausz and Pohnert 2013). If algicidal activity in more complex communities is investigated, a potential additional overlaying effect of algal metabolites has to be considered. Conspecifics or other species in the experiment might be affected by allelochemicals released during bacteria-induced lysis. Furthermore, the concept of strictly separating direct and indirect modes of action is challenged by the existence of a mixed mode mechanism that includes both the requirement for direct contact and the excretion of a chitinase (Li et al.2016).
Quantitative aspects
The above-mentioned methods are suitable for the identification of compounds with algicidal activity, and many studies are limited to this aspect. If the compounds, however, have a true ecological function can only be judged after their quantification in nature or in laboratory settings that provide close to natural conditions. Only if the active concentrations of isolated algicides fall in the range of the concentrations determined analytically, a true ecological function can be postulated. However, many of the below-mentioned metabolites are active in the micromolar range, thus in concentrations that rarely occur in nature, or, as it might be argued based on hydrodynamic considerations, that can only be reached in close contact situations where metabolites accumulate in the diffusion limited zone (the phycosphere) around cells of the algae (Breckels et al.2011). Especially this phycosphere concept has attracted much attention since in fact most interactions of algae and their associated microbial consortia are occurring in direct proximity to the algal cells (Rolland et al.2016). Lysis of Alexandrium sp., for example, is caused by bacteria from its phycosphere that is highly diverse in natural communities (Hasegawa et al.2007; Wang et al.2010). It is however extremely difficult to determine the actual concentrations of metabolites in the immediate vicinity of algal cells and mostly modelling approaches allow their estimation (Breckels et al.2010). A first direct monitoring of the pigment fucoxanthin in the vicinity of macroalgal cells using Raman microscopy revealed that bacteria in the immediate vicinity of the algae will be exposed to millimolar concentrations of active metabolites, even if concentrations of the compound in the water are close to the detection limit (Grosser et al.2012). Truly ecologically relevant investigations would thus have to consider not only bulk concentrations, but also the influence of gradients and the dimensions of the receiving organisms. Given the high concentrations that might accumulate within the phycosphere, some metabolites with moderate activity might play an important role locally. In addition, complexity is even increased since metabolites might change their function in a concentration-dependent manner, as it has been shown for indole-3-acetic acid. The bacterially produced phytohormone can promote both algal growth and algal death in a concentration-dependent manner thereby dynamically influencing performance of the microalga E. huxleyi (Segev et al.2016).
Linking lab and field data
In the first view, the setup of most lab experiments with dense cultures contradicts the average situation in the plankton, where the typical abundance of the entire microbiome is comparably low (roughly 105–106 cells/mL and up to 107 cells/mL during a bloom). However, such cell densities are only average values and since microscale patches with locally increased biomass play an important role in plankton ecology, such hotspots might boost bacterial abundance even in the natural environment (Doubell, Prairie and Yamazaki 2014). A hitherto poorly addressed problem in the connection of field and laboratory work lies in the often purely correlative interpretation. Prevalence of algicidal bacteria is often high during a declining phytoplankton bloom phase but it is not clear if this is indeed linked to the lytic activity of bacteria (Skerratt et al.2002). During declining blooms, the predisposition of algicidal bacteria for the efficient use of released nutrients resulting from cell lysis might already cause sufficient advantage leading to their dominance (Biddanda and Benner 1997; Azam and Malfatti 2007). Lab experiments often lead to the claim that algicidal bacteria can control phytoplankton blooms; however, this assumption is till to date not fully supported by field data. Chemical analysis of algicides in the field or transcriptomic monitoring of genes relevant for algicide production might help in the future to solve this issue.
Another still unresolved problem is the role of how to evaluate the influence of signalling chemistry in naturally occurring plankton communities. Such specific interactions often have been observed for mutualistic interactions, but even if there is evidence from lab data that algicidal bacteria use (secondary) metabolites from their prey as cues field data are still not available (Seymour, Ahmed and Stocker 2009; Seyedsayamdost et al.2011b; Li et al.2016) (see section Regulation).
In field studies, it also remains difficult to distinguish if the dominance of an algicidal bacterium is truly due to its algicidal activity or rather due to its successful competition for nutrients or simple opportunistic effects. Often it remains unsolved if the occurrence of an algicidal bacterium is not rather coinciding with the decline of a bloom. The assignment of the emitter and target organism of a natural product in the field is as well very difficult. In fact, even the presence of algicidal compounds in the field has to our knowledge not consistently been demonstrated. A possible way to solve this lack of knowledge could involve transcriptomics data that record the upregulation of genes required for algicide production or a chemical analysis of the respective metabolites in the field. If the molecular targets of the algicides are known, transcriptomics or proteomics of the target algae could also indicate its effectiveness during the interaction. Such protocols are however still not routinely available for field samples even if progress in method development has been made (McCarren et al.2010; Moustafa et al.2010; Keeling et al.2014; Wang et al.2014, 2015; Amin et al.2015; Balzano et al.2015; Cooper and Smith 2015; Guo et al.2016a).
Active components
The algicides produced by bacteria are very diverse in structure and mode of action. The metabolites span a wide range of polarity from polar amino acid derivatives to less polar fatty acids and their size spans several orders of magnitude. Algicides comprise small metabolites with molecular weights below 500 g mol−1, intermediate molecular weight compounds like peptides and even functional enzymes several kDa in size. This diversity is a major challenge for the structure elucidation of algicides. There are no universal extraction or separation methods that can be recommended. Algicide purification and structure elucidation has rather to start with optimised extraction procedures for the respective compound classes. In many cases, preliminary experiments that characterise the compound class of interest are required before extraction and purification are pursued. These can include heat or pH stress that denatures enzymes. Also, different extractions, e.g. liquid/liquid or solid-phase extraction can be undertaken to enrich algicides from the medium. Bioassay-guided fractionations with separation techniques that are optimised for the algicide classes are often the basis for the full characterisation (Prince and Pohnert 2010).
The structural diversity is also reflected in the multifaceted mode of action of algicides. Small metabolites generally have a defined target, e.g. the inhibition of an enzyme or the interaction with membranes, and therefore inactivate or disturb vital cellular functions of the host leading to cell death. On the other hand, enzymes catalyse biochemical transformations that lead to the degradation of the target organism. Mixed modes of action are also observed. The following sections will give an overview of the variety of algicidal metabolites and enzymes. We classified the active principles according to their biosynthetic origin or, when biosynthetic evidence is not obvious, based on common structural features. We exclude the numerous studies where claims of algicidal metabolites were made without any ecological rationale. Especially in the field of aquaculture and industrial applications, several structures from bacteria that do not occur in the plankton were brought forward that might act against microalgae. These metabolites can represent useful tools for biomass processing but are not ecologically relevant. We also do not list activities assigned to not fully characterised fractions or to primary metabolites administered in unrealistically high concentrations. However, as outlined above, the ecologically relevant concentrations are often not available for interacting plankton species and we thus can often not judge if algicidal metabolites are truly ecologically relevant. Algicides exclusively produced by cyanobacteria are excluded since they can rather be categorised as allelopathic. However, interesting parallels between bacterial and cyanobacterial algicides are highlighted. The following sections list identified principles but we do not validate the potential ecological role of the identified compounds.
Alkaloids
Alkaloids comprise a very diverse class of algicidal compounds (Fig. 4). The scaffolds include indoles, quinolones and others. The mechanism of algicidal activity of these nitrogen-containing metabolites with different biosynthetic origin is equally diverse.
Many algicides are derived from tryptophan and therefore belong to the class of indole alkaloids. Indole itself, isolated from a periphyton biofilm, inhibits the growth of cyanobacteria. This inhibition seems to be caused by impaired thylakoid membrane integrity, interrupted electron transport in photosystem II and reduction of the effective quantum yield (Wu et al.2011). Another representative of the indole alkaloids is isatin, which was isolated from a marine Pseudomonas sp. as well as from a freshwater Shewanella sp. (Gammaproteobacteria) (Sakata, Yoshikawa and Nishitarumizu 2011; Li et al.2014b). The growth of the diatom Chaetoceros ceratosporum is inhibited by isatin but not by indole (Sakata, Yoshikawa and Nishitarumizu 2011). Furthermore, isatin inhibits cyanobacterial growth (Li et al.2014b). Cyanobacteria are also negatively affected by 3-methylindole produced by a freshwater Aeromonas species (Guo et al.2016b). Production of 3-methylindole is regulated by QS (see section quorum sensing). Another anti-cyanobacterial indole derivative is trans-3-indoleacrylic acid produced by a Rhodococcus species (Actinobacteria) isolated from soil. Interestingly, a synergistic effect of trans-3-indoleacrylic acid and the co-isolated algicide L-pyroglutamic acid was observed (Wang et al.2013). The indole moiety is also a structural element of other, more complex algicides like the indole-thiazole bacillamide. Bacillamide was isolated from a marine Bacillus species and is toxic against dinoflagellates and raphidophytes without affecting the tested diatoms, green algae or cyanobacteria (Jeong et al.2003).
An important group of algicidal indole alkaloids are the tricyclic β-carbolines. Kodani and co-workers isolated harmane from an algicidal freshwater Pseudomonas species. Harmane as well as the structurally related norharmane inhibited the growth of various cyanobacteria, but not of two green algae species tested (Kodani et al.2002). This specificity makes them potential control agents for harmful algal blooms. Harmine and norharmane, isolated from the freshwater Bacillus flexus, showed algicidal activity against cyanobacteria but not against green algae (Alamri and Mohamed 2013). A marine Brachybacterium species (Actinobacteria) is the source of 1-acetyl-β-carboline that inhibited growth of naked dinoflagellates and raphidophytes as well as of thecate dinoflagellates (Kim, Son and Jeong 2015). It is noteworthy that β-carbolines are also allelopathic toxins since they can be produced by cyanobacteria as well. The cyanobacterium Nodularia harveyana from brackish water produces norharmane and norharmalane with toxicity against several other cyanobacteria (Volk 2005, 2006). Comparing threshold activities of norharmane and norharmalane revealed that saturation of the double bond at position 3–4 leads to a decrease of cytotoxicity. Furthermore, comparison of these β-carbolines with harmane revealed lower minimal toxic quantities against some of the species tested due to the additional methyl group. The hypothesis that increased hydrophobicity might facilitate penetration through the plasma membrane is strengthened by the fact that in many cases the hydrophobic alkaloid base shows higher toxicity than the more hydrophilic hydrochloride (Volk 2006).
An example for a quinoline alkaloid algicide is 2-n-pentyl-4-quinolinol, which is produced by a marine Pseudoalteromonas species and a marine Alteromonas species. The antimicrobial activity of this compound also spans diatoms and cyanobacteria affecting not only growth, but also mobility and adhesive properties (Long et al.2003; Wigglesworlth-Cooksey, Cooksey and Long 2007; Sakata et al.2008). The structurally related compound 2-heptyl-4-quinolone was isolated from the marine P. piscicida and negatively affected growth rates of a coccolithophore, but not of a chlorophyte or diatom (Harvey et al.2016). Another marine bacterium from the genus Alteromonas produces two algicidal pyrones (see section Others) as well as the quinolone alkaloids 2-undecen-1΄-yl-4-quinolone and 2-undecyl-4-quinolone. Both compounds were more toxic to dinoflagellates and raphidophytes than to the green algae or haptophytes tested (Cho 2012). Compared to 2-undecyl-4-quinolone, the unsaturated double bond in 2-undecen-1΄-yl-4-quinolone leads to an increased toxicity and the authors suggest more toxic oxidised hydroxyl compounds as a reason.
An example for an alkaloid algicide based on peptide precursors is the tripyrrole prodigiosin that has been detected in the marine Gammaproteobacterium Hahella chejuensis (Jeong et al.2005; Kim et al.2006; Zhang et al.2016). Prodigiosin inhibits the growth of some dinoflagellates, a raphidophyte and a haptophyte. Another red pigment that was assigned to the prodigiosin family is produced by a marine Gammaproteobacterium. In contrast to many other algicides, this pigment reduces production of reactive oxygen species with a negative effect to the raphidophyte Chattonella marina, for which ROS production is essential (Nakashima et al.2006a).
Amino acid derivatives, peptides and proteins
Numerous algicidal metabolites are structurally related to amino acids, although biosynthetic origin has not been studied in most cases (Fig. 5). A marine Vibrio sp. produces β-cyano-L-alanine inhibiting growth of several cyanobacteria, but not of green algae, diatoms or a dinoflagellate tested (Yoshikawa et al.2000). A different mode of action has been observed in the freshwater Stenotrophomonas maltophilia (Gammaproteobacteria) that actively reduces the availability of nutrients that are essential for the target algae. The bacterium produces a hydroxamate-type siderophore that negatively affects cyanobacterial growth by limiting iron availability (Liu et al.2014). It turns out that interactions via siderophore production and the accompanied carbon-for-iron exchange are a common type of algal–bacterial interactions (Kurth et al.2016).
Peptides and proteins can also have algicidal activity. A marine Vibrio shiloi strain, for example, produces a linear proline-rich dodecapeptide termed toxin P that inhibits photosynthesis in zooxanthellae isolated from corals only in the presence of ammonium ions (Banin et al.2001). The authors conclude that the peptide might facilitate uptake of ammonia, which in turn uncouples photosynthesis. Photosynthesis also seems to be the target of argimicin A, a pentapeptide isolated from a freshwater Sphingomonas sp. (Alphaproteobacteria) (Imamura et al.2000; Hibayashi and Imamura 2003). Argimicin A and the structurally related argimicins B and C inhibit growth of cyanobacteria without affecting green algae (Hibayashi and Imamura 2003; Yamaguchi et al.2003). This is in accordance with the hypothesis that argimicin A inhibits photo energy transfer from phycobilisome, an accessory pigment complex of photosystem II in cyanobacteria (Hibayashi and Imamura 2003). So far uncharacterised is a potentially cyclic algicidal peptide produced by an Alteromonas sp. with diatom-lytic activity (Wang et al.2016a).
Diketopiperazines are also a major class of bactericidal compounds. Apart from the indole alkaloid isatin (see section Alkaloids), the freshwater Shewanella sp. isolated by Li et al. (2014b) is also a source for hexahydropyrrolo[1,2-a]pyrazine-1,4-dione, an algicidal compound from the group of the diketopiperazines. Structurally this compound can be described as a cyclic dipeptide derived from glycine and proline (cyclo[Gly-Pro]). In contrast to isatin, which inhibits growth of the cyanobacteria Microcystis aeruginosa and Synechococcus sp., the diketopiperazine was only algicidal to M. aeruginosa, affecting cell wall integrity and intracellular structuring (Li et al.2014b). Cyclo[Gly-Pro] was also isolated from a freshwater Stenotrophomonas sp. together with the algicide hydroquinone. Hydroquinone is also toxic to M. aeruginosa as well as Synechococcus sp. (Lin et al.2016). This utilisation of structurally different algicides with at least overlapping host range is striking and might be a strategy to minimise resistance. Affected target organisms of the algicidal cyclo[Gly-Pro] were further studied by Li and co-workers after its isolation from a freshwater Bacillus species. Only four out of six cyanobacterial species tested were affected by this diketopiperazine (Li, Geng and Yang 2015). The co-isolated structurally related compound cyclo[Pro-Val] showed the same profile of algicidal activity as cyclo[Gly-Pro]. Recently, cyclo[Gly-Pro] has also been isolated from a marine Bacillus species and mechanistic studies showed that it inhibits photosynthetic activity and causes oxidative stress in the haptophyte Phaeocystis globosa (Tan et al.2016). Other anti-cyanobacterial diketopiperazines include cyclo[Gly-Phe] isolated from a freshwater Aeromonas sp. (Guo et al.2016b) as well as cyclo[Pro-Leu] and cyclo[4-OH-Pro-Leu] isolated from a freshwater Flavobacterium Chryseobacterium sp. (Guo et al.2015). Despite their structural similarity, the latter two have distinct characteristics and act synergistically. Both algicides cause oxidative stress but cyclo[Pro-Leu] mainly inhibits antioxidases, while cyclo[4-OH-Pro-Leu] interrupts photosynthetic electron flux.
Amino-acid derivatives are among the most frequently identified algicides. Many studies use complex media based on peptone and yeast extract for bacterial cultivation. However, such media are not well suited for the purpose due to high background levels of amino-acid-derived metabolites. The surplus of amino-acid derivatives in the medium can mask low-abundant algicides due to insufficient chromatographic separation leading to false identifications. In addition, amino-acid derivatives might correctly be identified as algicidal but are not a product of the bacterium under study but rather a medium component. Consequently, special efforts have to be made to prove bacterial origin whenever complex media are used for cultivation.
Enzymes
The release of enzymes is closely associated with an algicidal lifestyle of bacteria (Smith et al.1992; Martinez et al.1996; Paul and Pohnert 2011; Lei et al.2015; Li et al.2016). Many algicidal enzymes belong to the hydrolases and might serve a dual function being algicidal as well as facilitating utilisation of host polymers by extracellular degradation. The hydrolases primarily target polymers of algal cell walls, which are easily accessible and vital for algal cellular integrity. An example is a marine Pseudoalteromonas species that produces two proteases that are algicidal to a diatom. One of the proteases is a serine protease and the other that has a 6-fold lower algicidal activity belongs to the metal proteases (Lee et al.2000, 2002; Kohno et al.2007). The activity is specific since another serine protease produced by the bacterium as well as several commercially available proteases did not exhibit algicidal activity. Substrate specificity and affinity to the target may be determining factors for the algicidal activity of proteases. The marine K. algicida also produces lytic proteases that can target diatoms. While four tested diatom species were lysed by the bacteria, one showed resistance in infection experiments. Closer inspection revealed an induced protease production by the algae that might be linked to a defence mechanism counteracting the enzymes from bacteria. In the resistant diatom, proteases are substantially upregulated in the presence of the lytic bacteria triggered by an as of yet unknown mechanism (Paul and Pohnert 2011, 2013). Apart from lytic activity, bacterial proteases can also affect cell motility. Excreted proteins from three Flavobacteria reduce motility of a dinoflagellate, and inhibition experiments indicated that these proteins are proteases (Mayali et al.2008). However, the authors doubt that motility is the primary target of the protease but rather interpret the activity as a secondary effect of general toxicity. Algicidity has also been associated with the production of β-glucosidases by a marine P. haloplanktis strain with specific toxicity to the dinoflagellate Prorocentrum spp. without affecting other dinoflagellates (Kim et al.2009a). Naturally associated bacteria in the phycosphere of the dinoflagellate A. tamarense also release carbohydrate-degrading enzymes directly causing the lysis of the alga (Wang et al.2010). The potential target of β-glucosidases and other glycosidases is the cell wall. Activity of amylases, cellulases and xylanases in the marine bacterium Flammeovirga yaeyamensis (Bacteroidetes) has been linked to the degradation of algal cell walls as well (Chen, Bai and Chang 2013). Similarly, several cellulolytic marine bacteria isolated from mollusc gut negatively affect algal cell wall integrity (Munoz et al.2014). A complex toxic mechanism has been elucidated for freshwater C. prasina. As discussed in the section Regulation, this mechanism involves chemotaxis to diatoms, attachment to the cells via flagella and subsequent enzymatic degradation of the cell wall assisted by chitinases (Li et al.2016). An indirect algicidal mechanism is used by Rheinheimera sp. (Gammaproteobacteria) and the Flavobacterium Aquimarina sp. which produce L-amino acid oxidases (Chen, Lin and Sheu 2010; Chen, Sheu and Sheu 2011). The hydrogen peroxide production that happens concomitantly with the amino acid oxidation is responsible for the algicidal effect on green algae as well as on cyanobacteria.
Polyketides
Despite the great variety of polyketides known, only one algicidal polyketides has been identified to date (Fig. 6). A polyether algicide was isolated from a mangrove sediment St. malaysiensis (Zheng et al.2013). This polyether NIG355 is a stereoisomer of the known polyketide nigericin (Harvey et al.2007). The isolated polyether inhibited growth of a haptophyte and a dinoflagellate but was less active against a green alga (Zheng et al.2013). Although the authors did not comment on the mode of action, it can be hypothesised that it functions as an ionophore like nigericin does (Kevin, Meujo and Hamann 2009).
Terpenes
An algicidal terpene has been isolated from a freshwater Deinococcus sp. (Li et al.2015). The algicide was identified as the red pigment deinoxanthin, which is acting on specific target organisms (Fig. 7). Only 5 out of 23 tested species of microalgae were susceptible and these belonged to different taxonomic groups, namely dinoflagellates, diatoms and haptophytes. The mechanism of action is not yet fully understood but the authors suggest the chromophore to facilitate ROS formation. This is in agreement with a measurable increase in ROS and subsequent loss of cellular integrity upon deinoxanthin treatment. Damage by ROS also seems to be involved in the anti-cyanobacterial activity of a triterpenoid saponin produced by a Streptomyces sp. isolated from soil (Luo et al.2013). Another anti-cyanobacterial terpene derivative is a 3-oxo-ionone, which was isolated from a periphyton biofilm tentatively assigned by comparison with GC/MS libraries (Wu et al.2011). Similar to the co-isolated indole, the terpene also affects photosynthesis.
Fatty acids and their derivatives
This small group of algicides covers simple fatty acids as well as more complex derivatives (Fig. 8). Although fatty acid derivatives are well known in algal allelopathic interactions (Legrand et al.2003), they are rarely reported as bacterial algicides. A marine Vibrio sp. produces palmitoleic acid that inhibits growth of dinoflagellates (Li et al.2014a) and to a minor extend of diatoms and a raphidophyte. Upon treatment with palmitoleic acid, impaired integrity of cell wall and membranes were visible prior to total cell lysis. A more complex algicide build from a fatty acid and an amino acid is N-9-hexadecenoylalanine methyl ester. This algicide was isolated from the two marine species Loktanella koreensis and Roseovarius lutimaris (Alphaproteobacteria), and inhibited growth of a diatom (Ziesche et al.2015). Furthermore, N-acyl-homoserine-lactones produced by several marine bacteria from the Roseobacter clade have been identified as algicides (Ziesche et al.2015).
Others
The two 2-pyrones 3-hexyl-4-hydroxyl-6-pentyl-2H-pyran-2-one and 3-hexyl-4-hydroxyl-6-heptyl-2H-pyran-2-one were isolated from a marine Alteromonas sp. (Cho 2012) (Fig. 9). Similar to the quinolones produced by this organism (see section Alkaloids), these pyrones are toxic to dinoflagellates and raphidophytes but less toxic to the green algae or haptophytes tested. The more lipophilic 3-hexyl-4-hydroxyl-6-heptyl-2H-pyran-2-one possesses higher algicidal activity a trend also described for other algicides, such as harmane. Benzoic acid isolated from a marine Thalassospira sp. (Alphaproteobacteria) is algicidal against the harmful dinoflagellate Karenia mikimotoi (Lu et al.2016). As a potential mechanism of action, acidification of the cell leading to enzyme inhibition is discussed. Furthermore, a marine actinobacterial Brevibacterium sp. produces (2-isobutoxyphenyl)amine with toxicity to a dinoflagellate via degradation of membranes (An et al.2015). A larger group of algicides produced by marine P. gallaeciensis are the roseobacticides (Seyedsayamdost et al.2011a,b, 2014). The production of these algicides is triggered by algal breakdown products (see section Regulation). Potent algicidal activity of the roseobacticides with loss of cellular integrity and chloroplasts has been shown for a haptophyte; a cryptophyte and a diatom were also affected.
Structures of algicides—general considerations
In conclusion, algicides are very versatile and possess a wide range of molecular targets. There is until now no way to predict activities since structurally similar algicides can address distinct targets while structurally diverse compounds result in similar phenotypic effects. Three algicidal modes of action have been shown for several algicides from different natural product classes and represent common algicidal principles: (i) the destruction of cell membranes and cell walls as structural features leading to loss of cellular integrity and lysis; (ii) the production of ROS leading to disturbance in redox homeostasis, radical reactions and consequently to loss of cellular functions; (iii) the impairment of photosynthesis depriving the cells of energy. These modes of action are not independent of each other since lipid peroxidation due to ROS disturbs membranes and ROS may damage photosynthetic pigments. Blocked photosynthesis, on the other hand, also facilitates the generation of ROS. Consequently, separating the true primary target from secondary effects is very challenging. Furthermore, there might be a bias since parameters such as ROS or photosynthetic activity are easily accessible with standard toolkits. The lack of other targets might be caused by a lack of suitable methods for their detection. Many studies thus remain in a descriptive state and do not report the primary targets of the algicides but merely secondary effects caused by cellular stress or dysregulation. Future research should go into detail with the identification of algicidal mechanisms using, for example, structure-activity tests, assays to identify molecular targets using standard biochemical methods or recent tools, such as activity-based protein profiling.
Ecological role of algicidal bacteria—the microbial loop
Based on the above considerations, we now want to summarise the general role of algicidal bacteria within the marine environment. Mostly in spring when nutrient levels are high, specific microalgae can become dominant in the surface ocean. Such algal blooms develop mainly under the influence of nutrient availability (nitrogen, phosphorus, silicate and iron) but microbial as well as grazing activity can also promote bloom termination. Phytoplankton blooms are highly productive, releasing up to 20% of the primary production as DOC into the ocean. Phytoplankton-derived organic matter is highly diverse and the complexity is even increased by bacterial metabolism (Kujawinski 2011). DOC bioavailability is tightly linked with bacterial growth and community composition (Wear et al.2015). Bloom-associated bacteria can be assigned to the three major heterotrophic lineages Flavobacteria, Alphaproteobacteria (including Roseobacter spp.) and Gammaproteobacteria (Buchan et al.2014), and distinct populations of bacteria are observed during the process of decomposition of the algal bloom (Teeling et al.2012). A recent spatial-temporal study examining interactions of phytoplankton blooms and bacterioplankton community composition revealed that members of Bacteroidetes and Alphaproteobacteria dominated the bacterial community composition without being correlated to bloom-related variables. Less abundant bacterial phyla, on the other hand, were strongly associated with phytoplankton biomass or diatom:dinoflagellate ratio and showed high niche specificities (Bunse et al.2016). DOM as sum parameter of all dissolved organic matter is thus not appropriately reflecting the influence of resource availability for bacteria. Uptake experiments using the specific substrates glucose, amino acids and ATP revealed preferences of Alpha- and Gammaproteobacteria as well as Bacteroidetes resulting in variable performance depending on the DOM fraction present. In contrast, Roseobacter as a dominant group that includes algicidal members were overrepresented on all substrates (Alonso-Sáez and Gasol 2007,). In general, the microbial loop is dominated by bacteria utilising organic matter released by exudation, lysis and hydrolysis of phytoplankton (Azam et al.1983; Bidle and Falkowski 2004). The entity of DOM generated by these processes is one of the largest pools of organic carbon globally. It might be recycled within the microbial loop or move to higher (grazing) or lower trophic levels (viruses) (Wilhelm and Suttle 2012). However, dominant processes also lead to its escape from the food web in form of inorganic gaseous carbon (CO2, dimethylsulfide, etc.) or that it contributes to the ocean storage of non-degradable DOC in the deep sea known as the biological pump (Jiao et al.2010; Buchan et al.2014). Algicidal bacteria can actively modulate the carbon shuttling described above by arresting or inhibiting phytoplankton growth. More commonly, high lysis rates of phytoplankton populations triggered by algicidal bacteria will increase the abundance of DOM and POM (particulate organic matter) in the ocean (Mayali and Azam 2004; Roy and Chattopadhyay 2007). However, algicidal activity can also be very specific. Lysis of only one species or taxon can cause species shifts in the oceans’ plankton without significantly affecting the overall photosynthetic activity (Kang et al.2011). After lysis, resistant phytoplankton species and those creating temporary or resting stages might be responsible for the post-blooming shift in short-time phytoplankton diversity and also for longer lasting changes of dominance in the plankton (von Dassow and Montresor 2011). Especially specific bloom-terminating events by bacterial activity might substantially affect plankton composition (Sohn et al.2004; Mayali, Franks and Azam 2008; Meyer et al.2014). It can be argued that the frequency of such events might be substantial since especially high productive situations, as observed in algal blooms can nurture bacteria with algicidal effects (Bowen, Stolzenbach and Chisholm 1993). Solid data on the prevalence of algicidal effects in the field are however not available today—especially since methodological limitations of surveillance capacities as well as of analytical possibilities might lead to the bias towards contact or within-bloom interactions.
Conclusions and open questions
It becomes clear that algicidal bacteria and their lifestyles are highly diverse. This is also the case for the underlying regulative principles of alga–bacterium interactions. Multiple different mechanisms causing algicide activity and specificity have been identified in the last years, but often the true molecular targets or modes of action are not entirely understood. Some algicides seem to be rather species or taxa specific while others have a broader target spectrum, but no common patterns emerge, which again reflects the diversity. Mainly motivated by potential industrial applications many algicidal principles from small molecular weight secondary metabolites to enzymes have been identified. In most cases, however, the ecological function and concentrations of the algicides in the natural environment remain elusive. These active principles will now pave the way towards more elaborate studies involving in situ analytics as well as knock-out strategies to evaluate their true function in the interactions. It is fascinating to see that novel approaches in field work are already contributing to a better understanding of the role of algicidal bacteria within plankton communities. Especially the multiple regulations of alga–bacteria interactions contribute to our emerging understanding of interaction dynamics. Often novel insights also reveal additional layers of complexity. These become evident in micro- and mesocosm experiments, and it will be a major challenge to transfer the knowledge to field studies. Emerging possibilities in chemical analysis as well as transcriptomics will surely help to address the task. Even then, some more fundamental questions are still open, including the mechanisms of host specificity, or the identification of the true molecular targets of most algicides. In this context, the establishment of standard panels of microalgae for testing could be the task of an urgently needed initiative for standardisation. This is especially important since the emerging applications of algicidal bacteria in ecosystem engineering and biotechnology call for a reliable characterization of their activities before large-scale manipulations are envisaged.
Acknowledgments
We thank the German Research Foundation for funding within the framework of the CRC 1127 ChemBioSys.
Conflict of interest. None declared.