The limits to growth – energetic burden of the endogenous antibiotic tropodithietic acid in Phaeobacter inhibens DSM 17395

Phaeobacter inhibens DSM 17395, a model organism for marine Roseobacter group, was studied for its response to its own antimicrobial compound tropodithietic acid (TDA). TDA biosynthesis is encoded on the largest extrachromosomal element of P. inhibens, the 262 kb plasmid, whose curation leads to an increased growth and biomass yield. In this study, the plasmid-cured strain was compared to the wild-type strain and to transposon mutants lacking single genes of the TDA biosynthesis. The data show that the growth inhibition of the wild-type strain can be mainly attributed to the TDA produced by P. inhibens itself. Oxygen uptake rates remained constant in all strains but the growth rate dropped in the wild-type which supports the recently proposed mode of TDA action. Metabolome analysis showed no metabolic alterations that could be attributed directly to TDA. Taken together, the growth of P. inhibens is limited by its own antibacterial compound due to a partial destruction of the proton gradient which leads to a higher energetic demand. The universal presence of TDA biosynthesis in genome-sequenced isolates of the genus Phaeobacter shows that there must be a high benefit of TDA for P. inhibens in its ecological niche despite the drawback on its metabolism.


Introduction
Due to its nutritional versatility Phaeobacter inhibens DSM 17395, originally isolated from seawater at the Spanish Atlantic coast in Galicia [1,2] is commonly used as a model system to investigate the physiology and metabolic capabilities of marine bacteria living in nutrient-rich environments [3][4][5]. In addition to its 3.2 Mb chromosome, P. inhibens DSM 17395 carries three extrachromosomal elements of 65, 78 and 262 kb size, which were classified as chromids based on a relatively synonymous codon usage [6,7] but will be further on designated as "plasmids" for reasons of simplicity. The largest plasmid encodes, in addition to proteins involved in metabolic pathways and exopolysaccharide formation, proteins involved in the formation of the antibacterial compound tropodithietic acid (TDA) [8][9][10]. In a recent study by Zhao et al. (2016) [11], the probiotic effect of P. inhibens in aquacultures is attributed to TDA in killing unwanted marine pathogens [12][13][14].
TDA is supposed to act by disturbing the proton gradient due to an exchange of extracellular protons for cytoplasmic cations [15]. Furthermore, Wilson et al. (2016) [15] proposed a counteracting resistance mechanism consisting of an active proton export in P. inhibens via the γ-glutamyl cycle. They claimed that three genes of P. inhibens DSM 17395 (tdaR1 to tdaR3) are necessary for conferring an increased TDA tolerance (S2 File). However, experiments to select TDA-resistant or -tolerant strains of Escherichia coli, Pseudomonas aeruginosa or Staphylococcus aureus were not successful and furthermore, Salmonella typhimurium mutants with nonfunctional efflux pump and porin genes showed the same TDA susceptibility as wild-type strains [16]. Porsby et al. (2011) [16] explained an observed inhibition of the TDA-producing Phaeobacter strain 24-7 by a low TDA concentration with the prediction that TDA-producing strains express their resistance mechanism only during active TDA-production, which was not active in their experimental setup. In process-controlled bioreactors, P. inhibens DSM 17395 showed a faster growth (3.2-fold higher growth rate), a~40% higher growth yield and a significant higher carbon usage efficiency when the 262 kb plasmid encoding the TDA biosynthesis genes was missing [8]. The TDA biosynthetic gene cluster (S2 File) consists of 6 genes coding for proteins involved in the biosynthesis, 1 transcriptional regulator and 3 genes which were described as resistance genes by Wilson et al. (2016) [15]. The biosynthesis of TDA and catalyzed steps are proposed by Brock et al. (2014) [10]. They proposed the biosynthesis to start from phenyl-acetate. The TDA production is induced by N-3-hydroxydecanoylhomoserine lactone (R-3OHC 10 -HSL) and also by TDA itself [9].
In this study, we cultivated different transposon mutants of genes that are essential for the TDA biosynthesis and compared them to the 262 kb plasmid-cured mutant and the wild-type of P. inhibens DSM 17395 to analyze the impact of endogenously produced TDA on P. inhibens itself. The data clearly show that the absence of TDA is the major reason for the higher biomass yield of mutant strains lacking a functional TDA biosynthesis. Furthermore, our analysis documents that with increasing TDA concentration the wild-type strain uses a higher share of oxygen to keep up its proton gradient. Whereas the mutants deplete the whole of the available sources the wild-type decreases its growth rate to 35% compared to ½ OD max despite the fact that substrate is still available. We could not detect any increase of respiratory activity as a defensive counteraction against TDA on metabolic level.

Strains and growth conditions
Strains used in this work are listed in Table 1. Transposon mutagenesis was performed and mutants were checked for homogeneity as previously described [17] (S1 File).
All cultivations were performed in a defined salt water medium [18] with 1% of casamino acids as carbon source. Casamino acids were obtained from Merck (Darmstadt, Germany), prepared in a 5x stock solution and adjusted to pH 6.5 prior to sterile filtration. All cultivations were conducted at 150 rpm and 28˚C in an orbital incubator (Certomat BS-1, orbit 50 mm; Sartorius, Göttingen, Germany). Growth was followed spectrophotometrically (OD 600nm ) and gravimetrically. TDA formation was followed optically by the brownish complex in the medium [19]. The absorption of the TDA-iron complex was measured at 398 nm [9]. Additionally, we determined absorption at 302 nm which is the main absorption maximum for TDA (S3 File, S3 Fig).
For growth curves and metabolome analysis, 500 mL Erlenmeyer flasks with 3 baffles were used.
For determination of the oxygen uptake rate, the cultures were also grown in 500 mL baffled Erlenmeyer flasks with 100 mL medium. On the bottom inside each flask was an optical oxygen sensor spot (SP-PSt3-YAU-D7-YOP; PreSens, Regensburg, Germany) attached. The incubator was equipped with an online oxygen monitoring device (Shake Flask Reader; Pre-Sens, Regensburg, Germany). The measurement of the dissolved oxygen (DO) in the culture medium is based on fluorescence quenching [20] and the amount of fluorescence signal decrease is correlated to the oxygen concentration in the solution. Data acquisition and analysis was carried out with Shake Flask Reader Software v2.0.0 (PreSens, Regensburg, Germany) and the oxygen uptake rate (OUR) was automatically calculated. The specific OUR was calculated with the growth rate during the linear growth phase.

Medium exchange experiment
Medium exchange experiments were performed in 250 mL Erlenmeyer flasks with 3 baffles. Cells were grown to ½ OD max and harvested by centrifugation. The culture supernatants were sterile filtered and the cells were resuspended in either their own or the culture supernatant of another strain before the resulting growth rate was determined. For a detailed workflow description of the medium exchange experiment see results and discussion.

Amino acid quantification and metabolome analysis
Amino acids were quantified in samples taken at half-maximal and maximal growth of 4 biological cultivations. Samples were prepared and analyzed by a 1260 Infinity HPLC system equipped with a fluorescence detector (Agilent Technologies, Waldbronn, Germany) and a Poroshell HPH-C18 separation column (4.6 x 100 mm, particle size 2.7 mm; Agilent Technologies) as described previously [21]. Samples for metabolome analysis were taken of 4 biological cultivations at half-maximal growth and each cultivation was analyzed in triplicates. Cell lysis, preparation and metabolome analysis was performed as described previously [18,22] with some minor modifications: ethanol was replaced by methanol and 500 μL of the polar phase were dried prior to derivatization.

Results and discussion
Growth behavior of mutants with a missing or incomplete TDA biosynthesis Four transposon mutants (tdaA, tdaB, tdaC, tdaE), the plasmid-cured mutant (Δ262-kb) and the corresponding wild-type P. inhibens DSM 17395 were cultivated with casamino acids as the carbon source. In absence of a functional TDA biosynthesis, the maximal biomass concentration was reached after approximately 20 h of growth due to the depletion of any available carbon source while the wild-type showed a drastically reduced growth rate, lower final biomass yield and entered the stationary growth phase after 30 h despite a remaining high concentration of amino acids (total amount 6.04 ± 1.44 mM, corresponding to 16% of the initial concentration) in the medium (Fig 1, S1 Table). Overall the different transposon mutant strains showed an approximately 11% lower biomass yield than the Δ262-kb mutant strain while the growth rates during the linear growth phase remained similar. In contrast the biomass yield of the wild-type was reduced by approximately 35% compared to the transposon mutants and by approximately 41% compared to the Δ262-kb mutant strain. The difference in amino acid depletion for plasmid-cured mutant and wild-type was already observed by Trautwein et al.
(2016) [8]. The lower growth yield for the wild-type was explained by its slow replication rate or the metabolic burden for plasmid carriage and production of exopolysaccharides. However, our growth experiment showed that the growth inhibition of the wild-type is directly connected to the TDA biosynthesis as its biosynthesis is encoded by the 262 kb plasmid. The data suggest that only a minor effect can be attributed to other metabolic pathways encoded by the 262 kb plasmid or the metabolic costs of plasmid carriage.

Medium exchange between wild-type and TDA-free mutants
To exclude the possible costs of TDA biosynthesis itself as the source for the decrease of growth yield, a selected transposon mutant (tdaE) and the plasmid-cured mutant (Δ262-kb) were applied in medium exchange experiments with the wild-type (Fig 2). We selected the transposon mutant of the tdaE gene encoding an acyl-CoA dehydrogenase which catalyzes the first step of the tda operon in the TDA biosynthesis [10] to avoid side effects of precursors on the metabolism. The presence of TDA in the supernatant of the wild-type under the given growth conditions was verified by measurement of absorption at 302 and 398 nm (S3 File and S3 Fig). A growth inhibition was observed for both mutants after transferring the cells into the culture supernatant of the wild-type. Furthermore, the wild-type showed an increased growth rate when transferred to the TDA-free supernatant of the mutants. As a control experiment, we performed an exchange of the culture supernatant between the two TDA-free mutants to check whether the process of changing the medium or any difference in the amino acid concentration has an influence on the growth rate. Growth rates after this medium exchange were similar to the growth rates before the change (S1 Fig). The growth inhibition caused by the culture supernatant of the wild-type showed that a produced compound of the wild type is the cause for the inhibition effect. The control experiment with an exchange between transposon and plasmid-cured mutant approved that the growth inhibition is directly connected to TDA.

Constant oxygen uptake rates allow only lower growth rates in presence of TDA
The resistance mechanism proposed by Wilson et al. (2016) [15] assumes that glutamate is available in the medium and γ-aminobutanoic acid (GABA) would be excreted. However, glutamate is one of the preferred amino acids and is rapidly consumed by P. inhibens [5]. In contrast to the proposed resistance mechanism of Wilson et al. (2016) [15], we could not detect any accumulation of γ-aminobutanoic acid in the culture supernatant. P. inhibens shows only a decreased growth rate pointing to a high energy demand. A partial destruction of the proton gradient means that more respiratory activity is required to obtain the necessary ATP regenerated by ATP hydrolase. In previous studies, an almost 2-fold higher oxygen consumption per cellular dry weight was observed for the wild-type compared to the plasmid-cured mutant in processed-controlled bioreactors [8]. Thus, we measured the oxygen saturation in Erlenmeyer flasks and determined the oxygen uptake rate (OUR) in relation to the biomass at different time points of growth (Table 2, S2 Table, S2 Fig). All mutants showed a constant growth rate until CDW max while the wild-type strain showed a decreasing growth rate with proposed accumulation of TDA in the culture supernatant, correlating with the brown color. In contrast, the OUR remained comparable at approximately 1000 L h -1 g -1 for the wild-type and the mutants until the transition phase. In the context of the mode of action of TDA [15], this indicates that at constant oxygen uptake rates per cell dry weight, the presence of TDA leads to a slower growth rate due to lower ATP yield without any decrease of respiratory activity. Metabolome analysis of wild-type and TDA-free mutant strains As we could not detect a higher respiratory activity, we performed a metabolome analysis of the wild-type and the two mutant strains (tdaE and Δ262-kb) to analyze possible differences in their metabolism (Fig 3, S3 and S4 Tables). Only few changes in metabolites could be observed for the tdaE mutant compared to the wild-type strain (Fig 3A) in cells at half-maximal biomass yield. N-acetyl-serine was only detectable in the wild-type strain and may be involved in regulatory functions. Increased levels of valine and tyrosine in the wild-type in comparison to the tdaE mutant may be attributed to the different amino acid utilization at half-maximal growth as both amino acids were less or not used by the wild-type strain at this stage of growth (S5 Table). From half-maximal growth on, isoleucine and leucine are less consumed by the wild-type which may explain the observed increased intracellular levels at half-maximal growth (S1 Table). 2-Isopropylmalate is directly linked to increased levels of valine. These experiments show that the inhibition by TDA has only minor direct effects on metabolic processes except slower biosynthesis rates. Furthermore, we compared the metabolome of the tdaE transposon mutant with the plasmid-cured strain ( Fig 3B). Again, we detected differences in isoleucine, valine and its degradation product 3-aminoisobutanoate.
Other metabolites which are altered between the strains can be attributed to genes encoded by the 262 kb plasmid involved into the proposed lysine degradation pathway via pipecolate in P. inhibens [23]. One gene (locus tag: PGA1_262p02210) shows sequence similarity to the pipecolate oxidase in Pseudomonas putida KT 2440 (40% identity, E-value 4E-97) [24]. Thus, concentrations of several metabolites of the different lysine degradation pathways are altered: cadaverine, pipecolate, 5-aminopentanoate and 2-aminoadipate. The unidentified compound D200302 shows the typical mass spectrum and retention index of a modified sugar and can be attributed to alterations due to the absence of the exopolysaccharide biosynthesis proteins encoded by a gene cluster on the 262 kb plasmid (locus tags: PGA1_262p00050 to PGA1_262p00130) with sequence similarities to the succinoglycan biosynthesis of Rhizobium meliloti 1021 [25] (S6 Table). Overall, none of the changes in the metabolome can be directly attributed to the influence of TDA.

Conclusion
In this work we demonstrate that the presence of the antibacterial compound TDA in P. inhibens DSM 17395 is associated with a dramatic impact on the growth of the bacterium. The TDA-producing wild-type strain shows a reduced growth rate and-more importantly-a significantly lower biomass yield at the same oxygen uptake rate as the mutants, we could not observe any counteraction neither on metabolic level nor on the respiratory activity. Accordingly, our data support the effect of TDA on the proton motive force proposed by Wilson et al. (2016) [15]. Moreover, we show that TDA is the major reason for the observed growth restriction by the 262 kb plasmid previously reported by Trautwein et al. (2016) [8]. Only a minor part can be attributed to other genes encoded by the plasmid. As TDA biosynthesis genes encoding plasmids can be found in all Phaeobacter species [6,26,27], TDA must provide a significant advantage to P. inhibens in a competition for nutrients which compensates the high energetic costs.