Interspecific interactions of heterotrophic bacteria during chitin degradation

(1)

Interspecific interactions of

heterotrophic bacteria during chitin degradation

Dissertation

zur Erlangung des akademischen Grades des Doktors der Naturwissenschaften (Dr. rer. nat.)

vorgelegt von Nina Jagmann

an der Universität Konstanz

Mathematisch-Naturwissenschaftliche Sektion Fachbereich Biologie

Konstanz, Mai 2012

Tag der mündlichen Prüfung: 20.07.2012

1. Referent: Prof. Dr. Bernhard Schink, Universität Konstanz 2. Referent: Prof. Dr. Bodo Philipp, Universität Münster 3. Referent: Prof. Dr. Christof Hauck, Universität Konstanz

(2)

SUMMARY

In their natural habitats, bacteria live in multi-species microbial communities and are, thus, constantly interacting with bacteria of other phylogenetic groups. In order to prevail in these interspecific interactions, such as the competition for nutrients, bacteria have developed numerous strategies. During the degradation of polymers such interspecific interactions are likely to occur, because degradation starts as an extracellular process. In one possible interaction scenario, investor bacteria, which invest energy in the production of extracellular enzymes, face the danger of being exploited by opportunistic bacteria that compete for degradation products. To investigate such a scenario and to characterize the strategies employed by the bacteria involved, we established two co-culture model-systems consisting of bacteria that co-exist in aquatic environments and with the polymer chitin as carbon, nitrogen, and energy source. Aeromonas hydrophila strain AH-1N, which releases extracellular chitinases, was employed as investor bacterium in both co-cultures.

In the first co-culture, chitin embedded in agarose served as substrate.

Flavobacterium sp. strain 4D9, which cannot degrade embedded chitin due to its cell- associated chitinases, was employed as opportunistic bacterium. The strategies applied by strain 4D9 in order to acquire nutrients included active integration into the biofilm formed by strain AH-1N on the chitin beads and interception of the chitin monomer N-acetylglucosamine (GlcNAc), leading to overgrowth of strain AH-1N by strain 4D9 in the biofilm.

In the second co-culture, suspended chitin served as substrate. Pseudomonas aeruginosa strain PAO1, which is unable to degrade chitin, was employed as opportunistic bacterium. In the first phase of the co-culture, strain PAO1 grew with ammonium, acetate, and possibly GlcNAc and other compounds, which were released by strain AH-1N. In the second phase, strain PAO1 produced quorum sensing (QS)-controlled secondary metabolites, among them the redox active pigment pyocyanin. Pyocyanin inhibited the enzyme aconitase of strain AH-1N through the production of reactive oxygen species causing a block of the citric acid cycle. This led to a massive acetate release by strain AH-1N, which supported substantial growth of strain PAO1. Strain AH-1N was finally inactivated by pyocyanin and presumably other secondary metabolites. Further investigation of this parasitic growth strategy of strain PAO1 revealed that, while catabolite repression of GlcNAc

(5)

metabolism by acetate did not play a role, the action of isocitrate lyase was a key metabolic requirement for the transition into the second phase. Besides its role in acetate utilization, this enzyme was crucial for the utilization of GlcNAc. The ability to synthesize amino acids was a metabolic requirement of strain PAO1 as well. The overexpression of the QS effector protein PqsE regulating pyocyanin production could not restore formation of pyocyanin in auxotrophic mutants. P. aeruginosa possesses three QS systems. For the QS response of strain PAO1 in the co-culture, both the rhl and the 2-alkyl-4(1H)-quinolone system were crucial, whereas the las system was dispensable. Lack of the rhl signal synthase could be complemented by cross-talk with signals of strain AH-1N. By applying transposon mutagenesis and screening for mutants of strain PAO1 with defects in QS-regulated processes, we could identify several genes that were involved in the regulation of pyocyanin production. Among them were members of the gene cluster PA1415-PA1421, mutations of which led to a decreased production of pyocyanin and accelerated growth with the polyamine spermin.

By employing our co-culture model systems to study interspecific interactions, we could identify strategies of bacteria that are likely to be important in their natural habitats. With regard to P. aeruginosa, our model system offers the possibility to study QS under conditions that are more ecologically relevant than in single culture.

(6)

ZUSAMMENFASSUNG

Bakterien sind in ihren natürlichen Habitaten Bestandteil mikrobieller Gemeinschaften und befinden sich somit in ständiger Interaktion mit Bakterien verschiedener phylogenetischer Gruppen. Um in diesen Konkurrenzsituationen zu bestehen, haben Bakterien eine Vielzahl verschiedener Strategien entwickelt. Solch interspezifische Interaktionen sind während des Abbaus von Polymeren sehr wahrscheinlich, da dieser als extrazellulärer Prozess initiiert wird. In einem möglichen Interaktionsszenario werden Bakterien (Investoren), die Energie in die Produktion von extrazellulären Enzymen investieren, von opportunistischen Bakterien ausgenutzt, welche mit ihnen um die Abbauprodukte konkurrieren. Um ein solches Szenario zu untersuchen und die von den beteiligten Bakterien angewandten Strategien zu charakterisieren, entwickelten wir zwei Modellsysteme, die aus jeweils einer Co-Kultur bestanden. Diese Co-Kulturen enthielten Bakterien, die in denselben aquatischen Habitaten vorkommen, sowie das Polymer Chitin als Kohlenstoff-, Stickstoff-, und Energiequelle. Als Investor in beiden Co-Kulturen wurde Aeromonas hydrophila Stamm AH-1N eingesetzt, welcher extrazelluläre Chitinasen sezerniert.

In der ersten Co-Kultur wurde in Agarose eingebettetes Chitin als Substrat verwendet. Als opportunistisches Bakterium wurde Flavobacterium sp. Stamm 4D9 eingesetzt, der aufgrund seiner zellassoziierten Chitinasen kein eingebettetes Chitin abbauen kann. Die Strategien, die Stamm 4D9 anwandte, um an Nährstoffe zu gelangen, beinhalteten die aktive Integration in den von Stamm AH-1N auf den Chitinkugeln gebildeten Biofilm und das Abfangen des Chitinmonomers N- Acetylglucosamin (GlcNAc). Dies hatte zur Folge, dass Stamm AH-1N durch Stamm 4D9 im Biofilm überwachsen wurde.

In der zweiten Co-Kultur wurde suspendiertes Chitin als Substrat verwendet. Als opportunistisches Bakterium wurde Pseudomonas aeruginosa Stamm PAO1 eingesetzt, welcher Chitin nicht abbauen kann. In der ersten Phase der Co-Kultur nutzte Stamm PAO1 Ammonium, Acetat und wahrscheinlich GlcNAc und andere Verbindungen als Wachstumssubstrate, die von Stamm AH-1N freigesetzt wurden. In der zweiten Phase bildete Stamm PAO1 quorum sensing (QS)-regulierte Sekundärmetabolite, darunter den redoxaktiven Farbstoff Pyocyanin. Pyocyanin inhibierte durch die Bildung von reaktiven Sauerstoffspezies das Enzym Aconitase des Stammes AH-1N, was eine Blockade des Citratzyklus zur Folge hatte. Dadurch

(7)

kam es zur Freisetzung einer großen Menge von Acetat durch Stamm AH-1N, welches von Stamm PAO1 als Substrat genutzt werden konnte. Stamm AH-1N wurde schließlich durch Pyocyanin und vermutlich auch durch andere Sekundärmetabolite inaktiviert. Weitere Untersuchungen dieser parasitischen Wachstumsstrategie ergaben, dass die Aktivität der Isocitratlyase eine wichtige metabolische Voraussetzung für den Übergang von der ersten in die zweite Phase durch Stamm PAO1 darstellte. Eine mögliche Katabolitrepression des GlcNAc- Stoffwechsels durch Acetat spielte hierbei keine Rolle. Neben ihrer Beteiligung an der Assimilation von Acetat war die Isocitratlyase essentiell für das Wachstum mit GlcNAc. Eine weitere metabolische Voraussetzung für Stamm PAO1 stellte die Fähigkeit, Aminosäuren zu synthetisieren, dar. Die Bildung von Pyocyanin durch Mutanten, die für Aminosäuren auxotroph waren, konnte selbst durch Überexpression des QS-Effektorproteins PqsE, das diese Bildung reguliert, nicht wiederhergestellt werden. P. aeruginosa besitzt drei QS-Systeme. Für die QS- Antwort des Stammes PAO1 in der Co-Kultur waren das rhl- und das 2-alkyl-4(1H)- Chinolon-System essentiell, während das las-System entbehrlich war. Die Mutation der Signalsynthase des rhl-Systems konnte durch Signale von Stamm AH-1N komplementiert werden. Die Durchführung einer Transposonmutagenese und die darauffolgende Suche nach Mutanten des Stammes PAO1 mit Defekten in QS- regulierten Prozessen führten zur Identifizierung von Genen, die an der Regulation der Pyocyaninbildung beteiligt waren. Darunter befanden sich Gene des Genclusters PA1415-PA1421, deren Mutation zu einer verringerten Bildung von Pyocyanin und einem beschleunigtem Wachstum mit dem Polyamin Spermin führten.

Die Nutzung unserer Modellsysteme für die Analyse interspezifischer Interaktionen führte zur Identifizierung von Strategien der beteiligten Bakterien, die auch in ihren natürlichen Habitaten wirksam sein könnten. In Bezug auf P. aeruginosa ermöglicht es unsere Co-Kultur, QS unter Bedingungen zu analysieren, die ökologisch relevanter sind als die einer Reinkultur.

(8)

LIST OF PUBLICATIONS

This thesis is based on the following publications and manuscripts:

CHAPTER 1 Jagmann, N., Styp von Rekowski, K., and Philipp, B. (2012) Interactions of bacteria with different mechanisms for chitin degradation result in the formation of a mixed-species biofilm.

FEMS Microbiol Lett 326(1): 69-75.

CHAPTER 2 Jagmann, N., Brachvogel, HP., and Philipp, B. (2010) Parasitic growth of Pseudomonas aeruginosa in co-culture with the chitinolytic bacterium Aeromonas hydrophila. Environ Microbiol 12(6): 1787-1802.

CHAPTER 3 Jagmann, N., Hupfeld, M., and Philipp, B. Metabolic requirements for the parasitic growth strategy of Pseudomonas aeruginosa in co-culture with the chitinolytic bacterium Aeromonas hydrophila. (Manuscript)

CHAPTER 4 Jagmann, N., Bleicher, V., and Philipp, B. The role of quorum sensing of Pseudomonas aeruginosa in co-culture with Aeromonas hydrophila and the identification of novel genes involved in quorum sensing-regulated processes. (Manuscript)

(9)

GENERAL INTRODUCTION

Interspecific and interkingdom interactions of bacteria

In their natural environment, bacteria do not exist as independently acting cell populations, but they are part of multispecies environments, in which bacteria of different phylogenetic groups are constantly interacting with each other (Keller and Surette, 2006; Haruta et al., 2009). These bacterial communities are integral components of most biological systems (Straight and Kolter, 2009). Interspecific interactions of bacteria are ranging from competition, during which for example virulence factors are employed, to cooperation, commensalism, parasitism, and predator-prey interactions. Cooperation between bacteria involves for example symbiotic interactions like syntrophy (McInerny et al., 2008), whereas commensalistic interactions can be mediated for example by the release of waste products by bacteria that serve as growth promoting substances for other bacteria (Ohno et al., 1999; Ueda et al., 2004). In a parasitic interaction, Pseudomonas aeruginosa lyses cells of Staphylococcus aureus in order to get access to iron in low-iron environments (Mashburn et al., 2005). During a predator-prey interaction, Bdellovibrio species predate Escherichia coli by attaching to and subsequently replicating within the cells (Koval and Bayer, 1997).

Microbial communities are often found in close association with eukaryotes as well.

Consequently, interkingdom interactions of bacteria with plants, animals, and fungi exist. These interactions comprise for example cooperation like the symbiosis between rhizobia and legumes (Robertson et al., 1985) and between mammals and their intestinal microbiota (Hooper et al., 1998) or parasitism involving pathogenic bacteria.

The focal point of interspecific and interkingdom interactions is the acquisition of nutrients (Hibbing et al., 2010). To be able to persist in such interactions and to react to changes in their environment, bacteria have developed multiple strategies such as antibiotic production, motility, or coordinated activities, which are often mediated by producing, sensing, and responding to chemical information.

(10)

Intraspecific interactions of bacteria

Interspecific and interkingdom interactions of bacteria have been acknowledged for a long time, but the view on the complexity of these interactions has changed dramatically over the last 25 years, after it became obvious that intraspecific interactions of bacteria are widespread (Keller and Surette, 2006; West et al., 2006;

Williams 2007; West et al., 2007; Straight and Kolter, 2009). These intraspecific interactions are mediated by communication and cooperation between single cells of one population. Communication occurs when one or several individuals produce a signal that can be perceived by other individuals, which alter their behaviour in response to this signal (Keller and Surette, 2006). Cells of some Myxococcus species, for example, have long been known to communicate via chemical signals to form fruiting bodies as multicellular behaviour (Shimkets, 1990). By now, it is clear that various bacteria communicate and cooperate in order to perform multicellular behaviours, and the sociobiology and so-called social lives of bacteria have become major research areas (West et al., 2007; Dunny et al., 2008). These intraspecific interactions of bacteria have great influence on interspecific interactions as competitive strategies that require cooperative behaviour can be carried out (Hibbing et al., 2010). In addition to chemical communication (see below), numerous signalling pathways, e.g. two-component systems, have been identified in bacteria, which enables them to respond to a variety of environmental cues (Williams et al., 2007).

Thus, bacteria cannot simply be viewed as passive nutritional sinks taking up nutrients and reproducing without taking note of their environment, but they have developed numerous strategies to ensure their acquisition of resources (Straight and Kolter, 2009; Hibbing et al., 2010).

Quorum sensing

During intraspecific interactions, bacteria release diffusible signal molecules and respond to them by changes in gene expression. These bacterial cell-to-cell communication mechanisms were termed quorum sensing (QS) (Fuqua et al., 1996).

According to the general model of quorum sensing, bacteria constantly produce and release signal molecules into the environment. When a certain threshold concentration of signals is reached, which often coincides with high cell densities, the signal binds to its cognate signal receptor protein, and this complex leads to the activation or repression of the transcription of quorum sensing target genes, such as

(11)

genes encoding for enzymes for the biosynthesis of secondary metabolites or virulence factors. Additionally, this complex activates transcription of the signal synthase gene leading to a strong increase in signal molecules and, in consequence, to an enhanced expression of target genes, thus creating a positive feedback loop (Fig. 1).

luxR luxI

Fig. 1. The basic quorum sensing module. A LuxI-type synthase produces the signal molecule, which binds to a LuxR-type receptor protein, when it has reached a certain threshold concentration. The complex of receptor protein and signal molecule regulates the expression of certain target genes and enhances expression of luxI (positive feedback loop).

QS has been originally defined as cell-density dependent gene regulation, but it has become obvious that QS can be dependent on other environmental factors as well.

Nevertheless, QS is used as a general term for bacterial cell-to-cell communication based on diffusible signal molecules (Williams, 2007; Williams and Camara, 2009).

The phenomenon of cell-to-cell communication was initially discovered in the Gram- negative bioluminescent bacterium Vibrio fischeri and was originally termed autoinduction (Nealson et al., 1970). Bacteria of the genera Vibrio and Photobacterium live as symbionts in the light organs of sepiolid squids, which use bacterial bioluminesce for reducing their silhouette by counterillumination thus evading predators or disguising during search for prey (Nealson and Hastings, 1991).

V. fischeri, the model organism for the molecular background of QS, possesses a

(12)

signal synthase, LuxI, which produces N-(3-oxo-hexanoyl)-homoserine lactone (3- oxo-C6-HSL) as signal molecule (Eberhard et al., 1981). After binding 3-oxo-C6-HSL the signal receptor LuxR activates transcription of the lux operon, which is responsible for production of bioluminescence, and enhances the expression of luxI as positive feedback loop (Engebrecht et al., 1983). This phenomenon was thought to be limited to marine bioluminescent bacteria, until homologues of LuxR and LuxI were discovered in Erwinia carotovora and Pseudomonas aeruginosa (Gambello and Iglewski, 1991; Bainton et al., 1992). Since then, different LuxR/LuxI/N-acyl-HSL dependent-QS systems have been discovered in a variety of Gram-negative bacteria, some of which employing several interrelated systems (Williams, 2007).

So far, N-acyl-HSL-mediated QS has been found only in Gram-negative bacteria. QS in Gram-positive bacteria differs with regard to signal molecules and signal transduction (Jayaraman and Wood, 2008). These bacteria employ so-called autoinducer peptides (AIPs) as QS signalling molecules, which are actively exported from the cell. When a certain threshold of AIPs is reached, they bind to a histidine kinase of a two-component system, and the corresponding response regulator leads to activation or repression of target genes.

Both Gram-negative and Gram-positive bacteria share a QS system that employs a group of borate diester-containing furanones, collectively termed autoinducer 2 (AI- 2), as diffusible signal molecules (Chen et al., 2002). The gene encoding AI-2 synthase (LuxS) is widespread among bacteria and is found for example in Proteobacteria, Firmicutes, Actinobacteridae and in genera of the Cytophaga group (Vendeville et al., 2005). Therefore, it was suggested that AI-2 is used for interspecies communication (Xavier and Bassler, 2003). However, it has also been suggested that LuxS has an important metabolic function in recycling S- adenosylhomocysteine, a precursor of AI-2, and that AI-2 represents a metabolite rather than a signal molecule (Winzer et al., 2002).

As described above, intraspecific cell-to-cell communication and the resulting coordination of behaviour increases the complexity of interspecific and interkingdom interactions of bacteria. Additionally, QS can also have a direct influence on other bacteria during interspecies interactions. Burkholderia cepacia, for example, is able to sense and respond to N-acyl-HSLs synthesized by P. aeruginosa but not vice versa (Riedel et al., 2001). Escherichia coli and Salmonella species do not synthesize N-acyl-HSLs but possess an LuxR homologue, which enables them to

(13)

intercept signal molecules from other bacteria leading for example to reduced biofilm formation in E. coli (Ahmer, 2004).

Equally, QS has a direct influence on interkingdom interactions, and several eukaryotes are able to react to bacterial signal molecules. Spores of the green microalga Ulva intestinalis, for example, preferentially attach to N-acyl-HSL producing bacterial biofilms (Tait et al., 2005). In response to QS by P. aeruginosa, Candida albicans switches from growth as filamentous cells to growth as yeast-form cells, which are resistant to P. aeruginosa attack (Hogan et al., 2004). And the red alga Delisea pulchra is able to inhibit QS of Vibrio harveyi by producing brominated furanones that act on the signal receptor protein (Defoirdt et al., 2007).

Quorum sensing in Pseudomonas aeruginosa

One of the best studied and most complex QS mechanism is found in P. aeruginosa (Williams, 2007). This bacterium possesses three QS systems (see also Chapter 4), which control about 10 % of its genome (Schuster and Greenberg, 2006) (Fig. 2).

Two QS systems, las and rhl, employ N-acyl-HSLs as signal molecules. The las system consists of the signal synthase LasI, which produces N-3-oxo-dodecanoyl- homoserine lactone (3-oxo-C12-HSL), and the signal receptor LasR. The las system exerts both transcriptional and translation control over the rhl system (Latifi et al., 1996), which consists of RhlI, producing N-butanoyl-homoserine lactone (C4-HSL), and the signal receptor RhlR. However, it has been shown that the hierarchical relationship of both systems is conditional and dependent on the growth environment (Duan and Surette, 2007). The third QS system of P. aeruginosa employs 2-alkyl- 4(1H)-quinolones (AQs) as signal molecules (Dubern and Diggle, 2008). For the synthesis of more than 50 different AQs from anthranilate and α-keto fatty acids, transcription of pqsABCD is required. One important signal molecule of this QS system is 2-heptyl-4-quinolone (HHQ), which is the precursor of the second important signal molecule 2-heptyl-3-hydroxy-4-quinolone, which has been termed the Pseudomonas Quinolone Signal (PQS) and is produced from HHQ by PqsH. Both PQS and HHQ bind to their cognate receptor PqsR, which activates the expression of the pqsABCDE and phnAB operons. The latter is responsible for provision of the AQ-precursor anthranilate from chorismate. PqsE, which belongs to the family of metallo-hydrolases, is the effector protein of the AQ quorum sensing system and

(14)

NH O

pqsA

pqsH

pqsB pqsC pqsD pqsE phnA phnB pqsR

lasR lasI

r hlR r hlI

HN

O O O

O

HN

O O O

NH O

OH 12

4

Fig. 2. The N-acyl-HSL-dependent and AQ-dependent quorum sensing network in P.

aeruginosa that, partially together with other signalling pathways, regulates numerous target genes and controls the expression of multiple virulence determinants, some of which are listed above. The las system (in green), consisting of the signal synthase LasI producing 3- oxo-C12-HSL and the signal receptor LasR, exerts control over the rhl system (in red), consisting of RhlI producing C₄-HSL and RhlR. AQs (AQ system in blue) are synthesized from anthranilate (provided via PhnAB) by the action of PqsABCD, and PQS is produced from HHQ via PqsH. Both PQS and HHQ bind to the signal receptor PqsR, which drives expression of the pqsABCDE operon. PqsE is the effector protein of the AQ system controlling target gene expression. The AQ system is positively regulated by the las system and negatively regulated by the rhl system, which itself is positively regulated by the PqsR/PQS-complex of the AQ system. Our study indicated that the PqsR/HHQ complex activated the rhl system as well (dashed blue arrow; see Chapter 4). In the absence of LasR, RhlR/C4-HSL positively regulates lasI and pqsH expression (dashed red arrows; Dekimpe and Déziel, 2009). The QS systems are subject to a number of additional regulators, which further modulate the response of QS to a variety of environmental cues. Solid arrows indicate positive regulation; solid T-bars indicate negative regulation.

(15)

drives the expression of PQS/HHQ-dependent genes. The biochemical function of PqsE the associated downstream signal transduction is still unknown. The AQ system is linked to both the las and rhl system, with LasR/3-oxo-C12-HSL positively regulating transcription of pqsH, pqsR, and pqsA (Déziel et al., 2004; McGrath et al., 2004; Wade et al., 2005) and with RhlR/C4-HSL negatively regulating pqsR and pqsA (Wade et al., 2005; Xiao et al., 2006). The AQ system in turn positively regulates the rhl system (Heeb et al., 2011). All three QS systems have their own regulons that partially overlap. Target genes of QS in P. aeruginosa include extracellular enzymes such as elastase and alkaline protease, secondary metabolites, such as pyocyanin, siderophores, rhamnolipids, and hydrogen cyanide, and are involved in biofilm formation (Dubern and Diggle, 2008). For many QS-regulated genes, co-regulation by other factors is additionally required, and P. aeruginosa possesses various other signalling pathways and global regulators with partially overlapping regulons (Schuster and Greenberg, 2007). Additionally, the QS systems themselves are subject to a number of additional regulators further modulating the QS response (Juhas et al., 2005; Schuster and Greenberg, 2006). QS together with these mechanisms shape the behaviour of this bacterium in response to changing environmental conditions.

Biofilm formation as QS-mediated interaction

Biofilms are considered to be the predominant lifestyle of bacteria in the environment (Costerton et al., 1995). Biofilm formation as coordinated activity can be influenced by QS, and there is a difference in gene expression in cells living in a biofilm compared to planktonic cells (Hentzer et al., 2003; Lazazzera, 2005; Dötsch et al., 2012). Biofilms are defined as matrix-enclosed bacterial populations or communities that adhere to each other and are either attached to a surface or freely floating (Costerton et al., 1995). The development of a biofilm is described as a five-stage process (Stoodley et al., 2002). In the beginning of biofilm formation planktonic cells reversibly attach to a surface (stage 1), before they produce extracellular polymeric substances (EPS), which are composed of proteins, polysaccharides, and nucleic acids, leading to irreversible attachment (stage 2) and the formation of microcolonies (stage 3). These subpopulations start to interact with each other forming macrocolonies, and the biofilm structure develops (stage 4), before single cells actively disperse from the biofilm (stage 5). Biofilms are highly structured containing

(16)

physiologically specialized. One important characteristic of cells within biofilms is their increased resistance to environmental stress and antimicrobial agents.

Co-culture model systems for interspecific interactions employed in this thesis In nutrient-limited environments like the oligotrophic Lake Constance, interspecific interactions during competition for nutrients are likely to occur. In these environments the concentration of dissolved organic matter is low, and organic particles are therefore considered as hot spots for microbial metabolic processes (Simon et al., 2002; Azam and Malfatti, 2007). In aquatic systems polymeric organic compounds constitute a major portion of total organic matter, and are, thus, a major food source for heterotrophic bacteria (Unanue et al., 1999). As polymers are too large to be directly taken up by bacterial cells, their degradation starts as an extracellular process mediated by extracellular hydrolytic enzymes. The resulting oligo- and monomers are subsequently taken up by the cells. The degradation of polymers may therefore lead to different interspecific interactions of bacteria. In particular, bacteria that invest energy in enzyme production (investor bacteria) face the danger of being exploited by bacteria that scavenge the degradation products (opportunistic bacteria).

To investigate this interspecific interaction scenario during polymer degradation we set up two different co-culture model systems with chitin as substrate. Chitin is the second most abundant polysaccharide on earth after cellulose and the most abundant in aquatic systems (Gooday, 1990; Pruzzo et al., 2008), as it is part of the cell wall of molds and certain green algae, and a major constituent of the cuticles and exoskeletons of worms, molluscs, and arthropods (Keyhani and Roseman, 1999). It has been estimated that more than 10¹¹ tons of chitin are produced annually in the aquatic biosphere (Keyhani and Roseman, 1999). Chitin is composed of linear strains of β-1,4-linked N-acetylglucosamine (GlcNAc) residues that are cross-linked by hydrogen bonds.

As investor bacterium in both co-culture model systems, we employed the Gram- negative Gammaproteobacterium Aeromonas hydrophila. Aeromonads are abundant in aquatic environments, and A. hydrophila, which is an important fish pathogen and can also infect humans (von Graevenitz, 1987), was isolated from Lake Constance (Styp von Rekowski et al., 2008). A. hydrophila possesses a QS system, the ahy system, employing C4-HSL as signal molecule (Swift et al., 1997). For the degradation of chitin, this bacterium releases extracellular chitinases, which

(17)

hydrolyse the glycosidic bonds of chitin, before chitin oligomers are taken up into the cells (Li et al., 2007; Lan et al., 2008). Thus, A. hydrophila invests energy into enzyme production and faces the danger of being exploited by opportunistic bacteria matching the criteria of an investor bacterium.

As opportunistic bacterium in the first co-culture model system, we employed Flavobacterium sp. strain 4D9, which had been isolated from Lake Constance as well (Styp von Rekowski et al., 2008). Flavobacteria, some species of which are fish pathogens (Duchaud et al., 2007), form a group together with Cytophaga species and are members of the diverse phylum of Gram-negative bacteria known as the Bacteroidetes (McBride et al., 2009). Members of the Cytophaga-Flavobacterium group are abundant in aquatic systems and are known polymer degraders (Kirchman, 2002; Alonso et al., 2007). N-acyl-HSL-mediated QS was described for a member of the marine Bacteroidetes (Romero et al., 2010), but until now, QS has not been reported for members of the Cytophaga-Flavobacterium group (Bruhn et al., 2005). Genome analysis of members of this group of bacteria suggests that chitin degradation proceeds via cell-bound chitinases, which would create a close linkage between chitin hydrolysis and oligomer uptake (McBride et al., 2009). In aquatic habitats, however, polymers are usually entangled into larger aggregates (Simon et al., 2002; Azam and Malfatti, 2007), making it difficult for bacteria with cell-associated enzymes to get into contact with the polymers, and presumably forcing them to scavenge oligo- and monomers produced by others. To account for this situation in this co-culture, chitin was embedded into agarose.

As opportunistic bacterium in the second co-culture model system, we employed the Gram-negative Gammaproteobacterium P. aeruginosa. P. aeruginosa is a metabolically versatile bacterium and a primary agent of opportunistic human infections (Driscoll et al., 2007), which is especially dangerous for individuals suffering from cystic fibrosis or being immunocompromised (Gang et al., 1999; Jones et al., 2010). Apart from that, P. aeruginosa can be found in various environmental habitats (Ringen et al., 1952; Pellet et al., 1983; Hardalo and Edberg, 1997) and has also been detected in Lake Constance (Hans Güde, ISF Langenargen, personal communication). As described above, P. aeruginosa possesses three QS and various signalling systems that control production of several virulence factors in response to environmental changes. Even though both a chitinase and a chitin- binding protein are encoded in the genome of P. aeruginosa, this bacterium was

(18)

reported not to grow with chitin (Folders et al., 2000; Folders et al., 2001). In this co- culture, suspended chitin was used as substrate.

Thus, both co-culture model systems consisted of bacteria that co-exist in the environment and a polymer that occurs in the natural habitats of these bacteria.

Aims of this thesis

The general interest underlying this thesis was to study interspecific interactions of bacteria during competition for nutrients. For this, co-culture model systems for interspecific interactions during polymer degradation should be established. The aim of this thesis was to apply these model systems to identify strategies, which are employed by the bacteria involved in order to prevail in the interactions. Furthermore, once these strategies had been identified and characterized, we aimed at investigating genes and regulatory pathways underlying these strategies.

(19)

Interactions of bacteria with different mechanisms for chitin degradation result in the formation of a mixed-species biofilm

Nina Jagmann, Katharina Styp von Rekowski, Bodo Philipp

FEMS Microbiology Letters (2012) 326(1): 69-75

Abstract

In this study, interactions between bacteria possessing either released or cell- associated enzymes for polymer degradation were investigated. For this, a co-culture of Aeromonas hydrophila strain AH-1N as an enzyme-releasing bacterium and of Flavobacterium sp. strain 4D9 as a bacterium with cell-associated enzymes was set up with chitin embedded into agarose beads to account for natural conditions, under which polymers are usually embedded in organic aggregates. In single cultures strain AH-1N grew with embedded chitin, while strain 4D9 did not. In co-cultures, strain 4D9 grew and outcompeted strain AH-1N in the biofilm fraction. Experiments with cell-free culture supernatants containing the chitinolytic enzymes of strain AH-1N revealed that growth of strain 4D9 in the co-culture was based on intercepting N- acetylglucosamine from chitin degradation. For this, strain 4D9 had to actively integrate into the biofilm of strain AH-1N. This study shows that bacteria using different chitin degradation mechanisms can co-exist by formation of a mixed-species biofilm.

Introduction

Degradation of polymers by heterotrophic bacteria has to be initiated as an extracellular process. For this, bacteria produce extracellular hydrolytic enzymes that degrade the polymer into oligomers and monomers that can be taken up by the cells.

Extracellular hydrolytic enzymes can either be released into the environment, or they

(20)

can remain associated to the cells (Wetzel, 1991; Vetter and Deming, 1999). Both degradation mechanisms have contrasting advantages and disadvantages.

Enzyme-releasing bacteria bear a risk of not being rewarded by their energetic investment, because the polymer degradation products may be lost by diffusion or by scavenging by opportunistic bacteria (also called cheaters), which do not release extracellular enzymes (Allison, 2005). Bacteria with cell-associated enzymes minimize that risk by achieving a tight coupling between the hydrolysis of polymers and the uptake of oligo- and monomers. However, polymeric substrates in the open water do not usually occur as free compounds but are embedded into larger organic aggregates or assembled to complex organic gels (Simon et al., 2002; Verdugo et al., 2004; Azam and Malfatti, 2007). While bacteria with cell-associated enzymes have only limited access to polymers embedded within such networks, enzyme- releasing bacteria are able to hydrolyze these polymers. Bacteria with these contrasting mechanisms for polymer degradation co-exist in aquatic environments and are, consequently, interacting with each other during competition for the respective polymer. Thus, both bacteria must have strategies to compensate for the respective disadvantages of their degradation mechanisms during these interactions.

Chitin, a polymer of β-1,4-linked N-acetyl–D-glucosamine (GlcNAc) is the most abundant polymer in aquatic environments (Gooday, 1990; Pruzzo et al., 2008).

Chitin degradation via released chitinases has been well described for marine bacteria of the genera Vibrio and Pseudoalteromonas (Keyhani and Roseman, 1999;

Baty et al., 2000; Meibom et al., 2004) and for freshwater bacteria of the genus Aeromonas (Janda, 1985; von Graevenitz, 1987; Lan et al., 2008). On the contrary, chitin degradation via cell-associated chitinases is largely unexplored. It has been described that many chitinolytic bacteria of the Cytophaga/Flavobacterium-group of the Bacteroidetes, which are abundant inhabitants of marine and freshwater environments and contribute significantly to polymer degradation in the open water (Cottrell and Kirchman, 2000; Kirchman, 2002; Lemarchand et al., 2006; Alonso et al., 2007, Beier and Bertilsson, 2011), do not release chitinases (Sundarraj and Bath, 1972; Gooday, 1990). Recent genome analyses of several Bacteroidetes such as Flavobacterium johnsoniae suggest that chitin degradation in this group of bacteria proceeds via surface-bound chitinolytic enzymes that are very similar to the well- described starch utilization system (sus) of Bacteroides thetaiotaomicron (Bauer et al., 2006; Xie et al., 2007; Martens et al., 2009; McBride et al., 2009).

(21)

The goal of our study was to investigate interactions of bacteria with contrasting mechanisms for chitin degradation to identify the strategies they apply for overcoming their respective disadvantages. As this is difficult to study within natural communities, we set up a reductionistic laboratory model system with a defined co- culture of aquatic bacteria, Aeromonas hydrophila strain AH-1N and Flavobacterium sp. strain 4D9. Previously, we reported that strains of Aeromonas and of the Cytophaga/Flavobacterium group were dominant in the same enrichment cultures, in which the microflora of the littoral zone of the oligotrophic Lake Constance had been supplied with artificial organic particles as substrate (Styp von Rekowski et al., 2008).

Thus, members of these bacterial groups co-exist in the same environment. As described above for polymers in general, naturally occurring chitin is usually linked to other organic components such as proteins or glucans (Gooday, 1990). To account for this in our study, we embedded chitin into agarose beads.

Material and Methods

Cultivation of bacteria

Aeromonas hydrophila strain AH-1N (Lynch et al., 2002) and Flavobacterium sp.

strain 4D9, a Lake Constance isolate formerly called Cytophaga sp. strain 4D9 (Styp von Rekowski et al., 2008; gene bank accession number EF395377), were cultivated in the mineral medium B (Jagmann et al., 2010). When acetate (5 mM) and tryptone (0.1 %) were used as carbon and energy sources, 5 mM NH4Cl was present in the medium. When suspended chitin (0.5 % (w/v)), embedded chitin (2 chitin-containing agarose beads per test tube) or GlcNAc (5 mM) served as carbon, energy and nitrogen source, ammonium was omitted from the medium. Both strains were maintained on solid (1.5 % w/v agar) medium B plates containing 1 % tryptone.

Preparation of suspended and embedded chitin

Suspended chitin was prepared as described previously (Jagmann et al., 2010). For preparation of embedded chitin, medium B was supplied with suspended chitin and with agarose (GenAgarose, LE; Genaxxon) both to final concentrations of 1 %. After autoclaving 25 ml of the suspension were poured into a Petri dish (diameter 8.5 cm).

Agarose beads were punched out with a truncated 1 ml pipette tip. Each bead had a

(22)

volume of about 100 µl and contained chitin with a GlcNAc content of approximately 5 µmoles.

Growth experiments

All growth experiments were carried out in a volume of 4 ml in 15 ml test tubes. Pre- cultures of strains AH-1N and 4D9 were incubated in medium B containing tryptone on an orbital shaker (SI50 Orbital Incubator; Stuart Scientific) at 200 r.p.m. for 13- 16 h at 21° C. Growth of pre-cultures was measured as optic al density at 600 nm (OD600) with a spectrophotometer. Pre-cultures were harvested by centrifugation at 6000 x g for 3 min, washed with medium B, and were used to inoculate main cultures with suspended or embedded chitin at OD600=0.001 for strain AH-1N and at OD600=0.0005 for strain 4D9, which equals 10⁶ cells ml^-1 in both cases. Main cultures with GlcNAc or acetate were inoculated at OD600=0.01 for both strains. Main cultures were incubated on a rotary mixer (scientific workshop; University of Konstanz) at 120 r.p.m. at 16° C.

Cell-free culture supernatant of strain AH-1N was prepared by incubating the main cultures with suspended chitin in 100 ml of medium B in a 500 ml Erlenmeyer flask without baffles on an orbital shaker (Innova 4000 incubator shaker; New Brunswick) at 200 r.p.m. for 4 days at 30° C. At this point of time, chitinolytic enzyme activ ities were maximal, and the culture supernatant was processed by two centrifugation steps at 16,100 x g for 15 min at 15° C and filter-sterilization (pore size 0.2 µm ).

Before use for growth experiments the supernatant was supplemented in the same way as medium B (Jagmann et al., 2010).

Growth of bacteria with acetate or GlcNAc as substrates was measured as OD600

with a spectrophotometer (M107 with test-tube holder; Camspec). Growth of bacteria with suspended or embedded chitin was measured by determination of colony forming units (CFUs) as described previously (Jagmann et al., 2010). Growth of bacteria with embedded chitin was daily inspected for the disappearence of chitin from the agarose beads. When chitin had completely disappeared from the agarose beads, CFUs of the suspended and the biofilm fraction were determined subsequently. To determine CFUs of the biofilm fraction single agarose beads were washed in 500 µl of potassium phosphate buffer (50 mM, pH 6) and processed as described previously (Styp von Rekowski et al., 2008) Colonies of the individual strains in co-cultures could unambiguously be differentiated, because strain AH-1N

(23)

formed smooth whitish colonies, while strain 4D9 formed structured orange colonies.

Colonies of both strains did not show any inhibiting effect on each other.

Quantification of substrates and degradation products

Suspended chitin in test tubes was quantified by measuring its filling level as described previously (Jagmann et al., 2010). Samples for measuring chitin degradation products were centrifuged in 1.5 ml plastic tubes at 16,100 x g for 15 min at room temperature, and supernatants were stored at -20° C until further analysis.

To determine chitin degradation products during incubation in cell-free supernatant of strain AH-1N, samples were centrifuged as described above. Supernatants were subsequently incubated at 100° C for 5 min to inhibit ch itinolytic enzymes. After a further centrifugation step, supernatants were transferred into new plastic tubes and stored at -20° C until further analysis. Acetate, the mono mer, dimer (N,N’- diacetylchitobiose (Sigma)) and trimer (N,N’,N’’-triacetylchitotriose (Sigma)) of GlcNAc were determined by ion-exclusion HPLC as described previously (Klebensberger et al., 2006). Ammonium was determined as described previously (GDCH, 1996).

Determination of chitinolytic enzyme activities and protein determination

Chitinolytic enzyme activities during growth of strains AH-1N and 4D9 with suspended or embedded chitin were determined indirectly with 4-methyl- umbelliferone (4-MU) derivatized substrates (Colussi et al., 2005). Assays were performed in 96-well black microtiter plates (Nunc) and contained 10 µl of the respective sample and 90 µl of McIlvaine buffer (pH 7). Cell-free culture supernatant was obtained by centrifugation at 16,100 x g for 15 min. To measure chitinolytic enzyme activity in the biofilm fraction single agarose beads were washed in 500 µl medium B and homogenized with a plastic pestle in 100 µl of the same medium.

Assays were started by adding 25 µM of either 4-MU-N’-acetyl-β-D-glucosaminide (4- MU-GlcNAc, Sigma) for measuring chitobiase activities or 4-MU-N’,N’’-diacetyl-β-D- chitobioside (4-MU-(GlcNAc)2, Sigma) for measuring chitinase activities. Enzyme activities were determined at room temperature by measuring the fluorescence of released 4-MU at 465 nm after exciting at 340 nm in a microplate reader (Genios, Tecan) over a time period of 4 min. Activities were calculated using a 4-MU standard fluorescence curve in the range of 0 to 20 µM. Protein concentrations in culture

(24)

supernatants were determined using the Pierce BCA Protein Assay Kit (Thermo Scientific).

Results and Discussion

Growth in single culture with suspended and embedded chitin

To confirm that A. hydrophila strain AH-1N and Flavobacterium sp. strain 4D9 employed different mechanisms of chitin degradation both strains were incubated with suspended and embedded chitin, respectively, as the sole source of carbon, nitrogen, and energy.

With suspended chitin strain AH-1N grew concomitant with chitin degradation and reached numbers of 1.5 x 10⁹ CFUs ml^-1 within 120 hours (Fig. 1). Cleavage of 4- MU-(GlcNAc)2 was detected in cell free culture supernatants with a specific activity of 120 mU (mg protein)^-1 indicating the presence of a released chitinase. With

Fig. 1. Growth of A. hydrophila strain AH-1N and of Flavobacterium sp. strain 4D9 with suspended chitin in single cultures. CFUs of strain AH-1N (filled squares), CFUs of strain 4D9 (filled circles), decrease of chitin in cultures of strain AH-1N (open squares) and of strain 4D9 (open circles). Error bars represent standard deviation (n=3).

embedded chitin strain AH-1N grew in the suspended and in the biofilm fraction attached to the agarose beads. During growth, chitin disappeared from the agarose beads, while the agarose itself was not utilized. Chitin had completely disappeared

10⁵ 10⁶ 10⁷ 10⁸ 10⁹

10¹⁰ 0

20

40

60

80

100 100

50 150 200 250

0 300 350

time [h]

cfu ml-1 chitin decrease [%]

10⁵ 10⁶ 10⁷ 10⁸ 10⁹

10¹⁰ 0

20

40

60

80

100 100

50 150 200 250

0 300 350

time [h]

cfu ml-1 chitin decrease [%]

(25)

from the agarose beads after 15 days of incubation. At this point of time, strain AH- 1N had reached a final number of 3 x 10⁸ CFUs ml^-1 in the suspended fraction and 2.2 x 10⁸ CFUs ml^-1 in the biofilm fraction (Fig. 2A). Cleavage of 4-MU-(GlcNAc)2

(0.032 mU ml^-1) and of 4-MU-GlcNAc (0.013 mU ml^-1) indicating the presence of a released chitinase and chitobiase, respectively, could only be detected in the biofilm fraction, while it was below the detection limit in the culture supernatant. When cell- free culture supernatant of strain AH-1N containing chitinolytic enzymes was incubated with embedded chitin only about 40 % of the activity disappeared from the culture supernatant within short time (Fig. 3A). This activity was recovered from the agarose beads at the end of the incubation (not shown). These results indicate that physicochemical interactions alone are not sufficient to cause the strong accumulation of enzymes at the agarose beads in cultures of strain AH-1N. Rather, biofilm formation by strain AH-1N could serve as a strategy for minimizing diffusive loss of released enzymes and degradation products and for preventing exploitation by opportunistic bacteria.

Flavobacterium sp. strain 4D9 grew similar to strain AH-1N with suspended chitin and reached numbers of about 1.1 x 10⁹ CFUs ml^-1 within 170 hours concomitant with chitin degradation (Fig. 1). In cell-free supernatants of strain 4D9 no chitinolytic activities could be detected. A low 4-MU-GlcNAc-cleaving activity of 7 mU (mg protein)^-1 was detectable when cells of strain 4D9 and chitin were centrifuged and resuspended in fresh medium with 0.1 % of the detergent Triton X- 100 for solubilising particle-associated enzymes (Rath and Herndl, 1994). This result indicates that chitinolytic enzymes of strain 4D9 are either cell- or chitin-associated.

With embedded chitin CFUs of strain 4D9 had increased only slightly in the suspended and the biofilm fraction after 32 days of incubation (Fig. 2A) and chitin did not disappear from the agarose beads. Apparently, strain 4D9 was not able to grow with embedded chitin. If strain 4D9 released chitinases, these enzymes would certainly have reached chitin within the agarose beads (Svitil and Kirchman, 1998).

Thus, these results indicated that the chitinolytic enzymes of strain 4D9 were associated to the cells, which is in agreement with genome analyses of F. johnsoniae and other Bacteroidetes. The fact that strain 4D9 could not access embedded chitin clearly illustrated a disadvantage of this chitin degradation mechanism.

(26)

Growth in co-culture with embedded chitin

To investigate whether strain 4D9 had strategies to overcome this disadvantage in co-culture with enzyme-releasing bacteria strains AH-1N and 4D9 were incubated in co-culture with embedded chitin. In these cultures chitin had disappeared from the agarose beads after 32 days of incubation indicating a strong delay in chitin degradation compared to the single culture of strain AH-1N. At this point of time strain AH-1N had reached 5-fold and 8.7-fold lower CFU numbers in the suspended and in the biofilm fraction, respectively, compared to the single culture (Fig. 2A).

Fig. 2. A) Growth of A. hydrophila strain AH-1N and of Flavobacterium sp. strain 4D9 with embedded chitin in single and co-cultures. Single cultures of strain AH-1N were incubated for 15 days and single cultures of strain 4D9 as well as the co-cultures were incubated for 32 days. CFUs of the suspended fractions were taken at t=0 and t=end and CFUs of the biofilm fractions were taken at t=end. CFUs of strains AH-1N in single culture (black bars), 4D9 in single culture (dark grey bars), AH-1N in co-culture (light grey bars), and 4D9 in co- culture (open bars). Error bars represent standard deviation (n=3); B) Photograph of an agarose bead from a co-culture of A. hydrophila strain AH-1N and of Flavobacterium sp.

strain 4D9 with embedded chitin after chitin had been depleted to a large extent; bar equals 1.5 mm.

In contrast, strain 4D9 reached 34-fold higher CFU numbers in the suspended and 13,700-fold higher CFU numbers in the biofilm fraction compared to its single culture (Fig. 2A). Growth of strain 4D9 in the biofilm fraction of the co-culture was visible by

t_endbiofilm

co- culture

single cultures

10⁵ 10⁶ 10⁷ 10⁸ 10⁹ 10¹⁰

10⁴

cfu ml-1

t₀suspension t_endsuspension AH-1N

AH-1N 4D9

4D9

t_endbiofilm

co- culture

single cultures

10⁵ 10⁶ 10⁷ 10⁸ 10⁹ 10¹⁰

10⁴

cfu ml-1

AH-1N 4D9

4D9

co- culture

single cultures

10⁵ 10⁶ 10⁷ 10⁸ 10⁹ 10¹⁰

10⁴

cfu ml-1

AH-1N 4D9

4D9

A

B

(27)

the formation of its characteristic orange colonies on the surface of the agarose beads (Fig. 2B). These colonies turned red upon treatment with KOH, indicating the presence of the pigment flexirubin, which is characteristic for bacteria of the Cytophaga/Flavobacterium group (Reichenbach et al., 1980).

Apparently, strain 4D9 was able to grow especially in the biofilm fraction of the co- culture, even though it could not degrade embedded chitin itself, and it even overgrew strain AH-1N. The strong growth stimulation of strain 4D9 in the biofilm fraction could be the outcome of different strategies. First, strain 4D9 might have been able to access chitin within the agarose bead by penetrating into cavities within the agarose that had resulted from chitin degradation. However, as strain 4D9 only grew on the periphery of the agarose beads (Fig. 2B) this was rather unlikely.

Second, strain 4D9 might have grown with organic substrates that were released by strain AH-1N. These could have been either chitin degradation products or other substrates.

Identification of growth substrates for strain 4D9 in co-culture with embedded chitin To identify the substrates causing the strong growth stimulation of strain 4D9 in the biofilm fraction of the co-culture, it was first analyzed which compounds were released during growth of strain AH-1N with embedded chitin in single cultures.

These analyses revealed that acetate and ammonium were transiently released, while GlcNAc and its oligomers could not be detected (not shown). However, strain 4D9 grew very poorly with acetate (Fig. 4) ruling out this compound as a substrate.

Second, it was analyzed which products are formed by chitinolytic enzymes of strain AH-1N by incubating embedded chitin in cell-free supernatant of this strain. During this incubation chitin largely disappeared from the agarose beads, and HPLC analysis showed that up to 2 mM of GlcNAc accumulated (Fig. 3B). As strain 4D9 could grow with GlcNAc (Fig. 4), growth of strain 4D9 in the co-culture might be based on GlcNAc. To investigate this possibility, strain 4D9 was incubated with embedded chitin in cell-free supernatant of strain AH-1N. In these cultures GlcNAc did not accumulate and strain 4D9 reached about 1.400-fold higher CFU numbers in the suspended fraction (Fig. 5A) and about 64-fold higher CFU numbers in the biofilm fraction (Fig. 5B) compared to the control, in which strain 4D9 was incubated with embedded chitin in medium B. If embedded chitin was omitted from cell-free culture supernatant strain 4D9 reached only 48-fold higher CFU numbers in the suspended

(28)

fraction compared to the control (Fig. 5A). This relatively small growth must have been due to organic compounds in the culture supernatant of strain AH-1N, which have not been identified so far. These results indicated that GlcNAc released from chitin by the chitinolytic enzymes of strain AH-1N was most likely the main growth substrate for strain 4D9 in the co-culture.

Fig. 3. Characterization of chitinolytic enzyme activities in cell-free culture supernatants of A.

hydrophila strain AH-1N incubated with embedded chitin (four chitin-containing agarose beads). A) Activity of chitinases (cleavage of 4-MU-(GlcNAc)2) for determining absorption of chitinolytic enzymes to the agarose beads over time. Values are expressed as percentages of control measurements (set to 100 %) from supernatants incubated without agarose beads;

B) Release of GlcNAc from embedded chitin by chitinolytic enzymes. Error bars represent standard deviation (n=3).

As GlcNAc could not be detected in the supernatant of single cultures of strain AH- 1N with embedded chitin, this bacterium apparently exhibited a tight coupling of polymer hydrolysis and GlcNAc uptake. To interfere with this tight coupling strain 4D9 had to actively integrate into the biofilm for establishing a close contact to zones of chitin hydrolysis and GlcNAc release. This was supported by the fact that in the presence of strain AH-1N strain 4D9 grew mainly in the biofilm fraction (Fig. 2A), while it grew mainly in the suspended fraction when incubated in cell-free supernatant only (Fig. 5AB) indicating that there was no selective pressure for biofilm formation in the absence of strain AH-1N. As the growth rate with GlcNAc of strain AH-1N (µ = 0.133 h^-1) was about 3 times higher than the growth rate of

60

20 80 100

0

time [h]

40 120 140

0 20 40 60 80 100

[ %] of control

60

20 80 100

0

time [h]

40 120 140

0 20 40 60 80 100

[ %] of control

time [d]

2

1 3 4

0 5 6

0

GlcNAc [mM]

2.5

2

1.5

1

0.5

time [d]

2

1 3 4

0 5 6

0

GlcNAc [mM]

2.5

2

1.5

1

0.5

A B

Interspecific interactions of heterotrophic bacteria during chitin degradation