Pyomelanin from Pseudoalteromonas lipolytica reduces biofouling

Summary Members of the marine bacterial genus Pseudoalteromonas are efficient producers of antifouling agents that exert inhibitory effects on the settlement of invertebrate larvae. The production of pigmented secondary metabolites by Pseudoalteromonas has been suggested to play a role in surface colonization. However, the physiological characteristics of the pigments produced by Pseudoalteromonas remain largely unknown. In this study, we identified and characterized a genetic variant that hyperproduces a dark‐brown pigment and was generated during Pseudoalteromonas lipolytica biofilm formation. Through whole‐genome resequencing combined with targeted gene deletion and complementation, we found that a point mutation within the hmgA gene, which encodes homogentisate 1,2‐dioxygenase, is solely responsible for the overproduction of the dark‐brown pigment pyomelanin. In P. lipolytica, inactivation of the hmgA gene led to the formation of extracellular pyomelanin and greatly reduced larval settlement and metamorphosis of the mussel Mytilus coruscus. Additionally, the extracted pyomelanin from the hmgA deletion mutant and the in vitro‐synthesized pyomelanin also reduced larval settlement and metamorphosis of M. coruscus, suggesting that extracellular pyomelanin released from marine Pseudoalteromonas biofilm can inhibit the settlement of fouling organisms.


Introduction
Marine biofouling communities are surface-dwelling communities composed of all types of marine bacteria, invertebrates and diatoms, and interactions among these organisms govern the nature of biofouling communities (Mieszkin et al., 2013;Lee et al., 2014;Dang and Lovell, 2016). Microfouling organisms located on the surface of a marine substratum create serious issues worldwide, including an increase in drag force and metal corrosion, as well as a reduction in the efficiency of energy transfer (Dobretsov et al., 2013;Karpov et al., 2014b). Tributyltin-based antifouling technologies have been considered the most effective approach for preventing the accumulation of fouling organisms (Omae, 2003;Chambers et al., 2006). In September 2008, the use of organotin compounds was banned due to their adverse effects on nontarget marine organisms and ecological environments. The discovery of natural antifouling compounds has received growing attention due to the global efforts being made to develop more environmentally friendly antifouling technologies (Yebra et al., 2004;Qian et al., 2010Qian et al., , 2015. Pseudoalteromonas is a genus of Gammaproteobacteria that is commonly found in various marine environments (Gauthier et al., 1995;Wietz et al., 2010;Salazar et al., 2016). Pseudoalteromonas spp. possess important features that have drawn increasing interest from researchers in the ecological and pharmaceutical sciences: (i) they are often found in association with eukaryotic hosts, including marine phytoplankton, sponges, mussels, scallops, tunicates and corals (Holmstrom and Kjelleberg, 1999); (ii) they produce a variety of biologically active extracellular compounds, including pigmented secondary metabolites, extracellular enzymes and extracellular polysaccharide (EPS) (Kalinovskaya et al., 2004;Bowman, 2007). The colonization of marine surfaces by marine bacteria often involves biofilm formation, which aids survival in competitive environments or serves as a cue to settlement by invertebrate larvae (Qian et al., 2007;Egan et al., 2008). Biofilms formed by Pseudoalteromonas produce bioactive compounds that inhibit the settling of various fouling invertebrates and algae (Holmstrom and Kjelleberg, 1999;Egan et al., 2002;Bowman, 2007). We recently reported the occurrence of morphological diversification in EPS production during the biofilm formation of various Pseudoalteromonas spp. (Zeng et al., 2015). In particular, we found that genetic variants generated in Pseudoalteromonas lipolytica biofilms with alteration in capsular polysaccharide or cellulose production exhibit strong antifouling activities, which indicates that biofilm variants might produce extracellular compounds with enhanced antifouling activities (Zeng et al., 2015).
The genus of Pseudoalteromonas has been divided into pigmented and non-pigmented clades (Bowman, 2007). A correlation between the antifouling activities of Pseudoalteromonas and pigment expression has been observed by several research groups Bowman, 2007;Huang et al., 2011). However, the molecular mechanisms underlying the secretion of pigmented secondary metabolites by Pseudoalteromonas and the regulation of pigment production remain largely unknown. In this study, we isolated a pigmented variant during P. lipolytica biofilm formation and found that the release of dark-brown pigments was caused by a mutation in the hmgA gene, which encodes homogentisate 1,2-dioxygenase and that this mutation led to the hyperproduction of pyomelanin. Deletion of the hmgA gene in P. lipolytica reduced the larval settlement and metamorphosis of the mussel Mytilus coruscus, a significant biofouling organism found in China, and this finding suggests that extracellular pyomelanin released from marine bacteria can inhibit the settlement of fouling organisms.

Pigmentation during P. lipolytica biofilm formation
Like other marine Gammaproteobacterial strains, P. lipolytica is able to form pellicle-type biofilm due to the establishment of oxygen and nutrient gradients at the interface of the air in the nutrient-rich medium SW-LB (sea water LB) or nutrient-poor medium 2216E marine broth (Armitano et al., 2014;Zeng et al., 2015). Two types of genetic variants with relatively high proportions were recently characterized during P. lipolytica biofilm formation, and the results demonstrated that each type of variant leads to changes in the colony morphology from smooth to wrinkled and opaque to translucent (Zeng et al., 2015). In addition to these changes in the colony morphology of P. lipolytica, we observed an increased amount of brownish pigments of a later-stage biofilm compared with that of planktonic cultures (Fig. 1A). We then quantified the level of brownish Fig. 1. Pigmentation occurring during P. lipolytica biofilm formation. A. A substantial amount of pigments are produced during biofilm formation for 10 days at different growth temperatures. B. Quantification of pigment production in 2216E medium. C. Quantification of pigment production in SW-LB medium. Data are from three independent cultures. One SD is shown in B and C, and representative images are shown in A. Biotechnology, 10, 1718-1731 pigment in the supernatant by measuring the absorbance at 400 nm. When cultured in 2216E medium, P. lipolytica produced a eightfold higher amount of this brownish pigment after prolonged static incubation (day 10) at 25°C (Fig. 1B), and similar trends were also observed when cultured in nutrient-rich SW-LB medium (Fig. 1C). In addition, the release of this brownish pigment was further enhanced by culturing at a higher temperature during biofilm formation (Fig. 1). Compared with that observed at 15°C, the brownish pigment production was increased 39-fold after prolonged static incubation (day 10) at 37°C in SW-LB medium (Fig. 1C), suggesting that the production of brownish pigment by P. lipolytica is induced during biofilm formation and in response to the upshift of temperature.
Pigmented variants isolated from P. lipolytica biofilms Genetic variants generated during biofilm formation have been shown to account for the phenotypic changes observed in biofilm cells compared with planktonic cells (Stewart and Franklin, 2008;Zeng et al., 2015). Thus, we aimed to determine whether pigmented variants are also generated during P. lipolytica biofilm formation. A single colony of wild-type P. lipolytica was propagated in SW-LB medium at 25°C for 10 days without shaking to allow formation of a biofilm. The biofilm cells were then destructively sampled, and randomly selected cells were plated onto SW-LB agar for morphological examination. As reported earlier, the majority of phenotypic variants showed a wrinkled or translucent morphology compared with the wild-type ancestral cells, which presented a smooth and opaque morphology (Zeng et al., 2015). Of the 5000 randomly selected cells, three cells presented a change from a non-pigmented morphology to brownish pigment morphology ( Fig. 2A). These pigmented variants (named as P1, P2 and P3) were re-inoculated into fresh medium for three rounds of overnight passaging, and the production of brownish pigment remained constant, suggesting that genetic changes had occurred. We then quantified the production of brownish pigment by these variants by measuring the absorbance of the supernatants of the liquid cultures after prolonged incubation at 25°C. Although the P1 variant appeared brownish on the agar plate, it did not show increased pigment production in the supernatant of the liquid culture compared with the wild-type strain, indicating that P1 mostly produces intracellular pigments ( Fig. 2A). In contrast, the P2 and P3 variants showed significantly higher production in the supernatants after shaking culture compared with the wild-type strain, and the P3 variant produced a 2.8 AE 0.7-fold higher amount of pigment than the P2 variant ( Fig. 2B). Additionally, an extended incubation of the P3 variant resulted in a visible release of the brownish pigment into solid agar, but this visible release was not observed with the wild-type strain; however, the colours of the colonies of the two strains were comparable, suggesting that the P3 variant mostly produces extracellular pigments (Fig. 2C). When cultured statically in liquid culture, the P3 variant showed the formation of a more brownish and thicker pellicle along with an increased release of brownish pigment into the supernatant (Fig. 2D). Collectively, these results demonstrate that genetic variants that exhibit hyperproduction of this brownish pigment are naturally generated during the formation of P. lipolytica biofilms.
Point mutation in hmgA leads to pyomelanin overproduction We subsequently explored the genomic changes that had occurred in the genome of the P3 variant through whole-genome resequencing. The P3 variant was resequenced using the Illumina HiSeq 2000 platform, and 454 Mb of sequencing data was produced. In general, the depth of the whole-genome resequencing exceeded 98%, and the coverage exceeded 99.82%, suggesting that the sequence data were of high quality. In summary, 12 point mutations (designated M1 to M12) were identified in the P3 strain, and these included eight nonsynonymous mutations and one nonsense mutation (Table S1). Among the non-synonymous mutations, we found that M11 was located in the coding region of the hmgA gene, which encodes homogentisate 1,2-dioxygenase (HmgA). HmgA catalyses the conversion of homogentisic acid (HGA) into 4-maleylacetoacetate (Arias- Barrau et al., 2004;Wang et al., 2015a,b). The accumulation of HGA can lead to the formation of pyomelanin due to autopolymerization in closely related marine bacteria, such as Vibrio, Pseudomonas and Shewanella (Coon et al., 1994;Wang et al., 2013;Hocquet et al., 2016). Thus, we reasoned that the brownish pigment produced by the P3 variant could be due to the inactivation of HmgA. To test this hypothesis, we first constructed an in-frame deletion of the hmgA gene in P. lipolytica (Wang et al., 2015a,b). As expected, the deletion mutant DhmgA produced a significantly higher amount of extracellular brownish pigment than that of the wild-type strain when grown planktonically (Fig. 3). Expression of hmgA in trans via the pBBR1MCS-hmgA plasmid under the control of its own promoter restored the morphology of the DhmgA mutant and the P3 variant to that of the non-pigmented wild type (Fig. 3), suggesting that the brownish pigment produced by the P3 variant is due to the inactivation of HmgA. Moreover, the pigment produced by the DhmgA mutant exhibited the highest absorbance at 400 nm in visible light spectrum, which was the same as the pigments generated by the biofilm (Fig. S1), suggesting that the brownish pigment should contain same or similar compounds (Ruzafa et al., 1995;Lagunas-Munoz et al., 2006;Chatfield and Cianciotto, 2007).
To further confirm the brownish pigment was pyomelanin, pigment produced by DhmgA mutant and the synthetic HGA-based pyomelanin were extracted and analysed by Fourier transform infrared spectrometer (FTIR) spectroscopy which is regarded as the most informative method for the structural analysis of melanins (Bilinska, 1996;Schmaler-Ripcke et al., 2009). The dried extracted pigment resulted in a black powder with the following solubility properties indicative of melanin: insoluble in organic solvents (ethanol, chloroform and acetone) and soluble in NaOH solutions with a pH of 10 (data not shown) (Turick et al., 2002). The FTIR spectra of the brownish pigment extracted from the P. lipolytica DhmgA mutant shared definite signature peaks as previously reported for Shewanella algae pyomelanin and synthetic pyomelanin produced by auto-oxidation of HGA (Turick et al., 2002) (Fig. S2). The peaks at the following wave numbers and their corresponding structures include the bands at 3300 cm À1 , 2900 cm À1 and 1560 cm À1 . Moreover, the fingerprint regions between 500 cm À1 and 1000 cm À1 of these samples resemble each other (Turick et al., 2002;Schmaler-Ripcke et al., 2009) (Fig. S2). Collectively, these results demonstrate that the pigment produced by DhmgA mutant shared key characteristics of known melanin and also shared a high degree of similarity to the synthesized pyomelanin.

L-tyrosine metabolism is involved with production of pyomelanin
HmgA is an important enzyme that participates in the Ltyrosine catabolic pathway (Arias- Barrau et al., 2004). Four genes participating in the L-tyrosine catabolic pathway are present in P. lipolytica (Table S2) and other Pseudoalteromonas species that have been isolated from diverse marine environments, including extreme environments, such as SM9913, suggesting that this pathway is conserved in Pseudoalteromonas (Table S3). The distribution of the four enzymes belonging to the Ltyrosine catabolism pathway on two different chromosomes is also a common feature in these Pseudoalteromonas strains. The melA gene (AT00_02290 on chromosome I) encodes 4-hydroxyphenylpyruvate dioxygenase (4-HPPD), which is responsible for the synthesis of HGA. The hmgA gene (AT00_15690 on chromosome II) is responsible for the degradation of HGA to 4-maleylacetoacetate. The two other genes, fahA (AT00_02285 on chromosome I) and maiA (AT00_15695 on chromosome II), encode enzymes that convert 4-maleylacetoacetate to acetoacetate, and fumarate substrates that are subsequently utilized in the tricarboxylic acid (TCA) cycle ( Fig. 4) (Schmaler-Ripcke et al., 2009). We observed that the pyomelanin production in the biofilm of P. lypolytica was increased under elevated temperature ( Fig. 1). Thus, we measured the expression of two key genes involved in the production of HGA under higher temperature (37°C) and found that the production of the pyomelanin in P. lypolytica is neither due to a reduced expression of hmgA gene which is responsible for HGA degradation nor due to an increased expression of the melA gene which is responsible for the HGA synthesis (Table S5). Thus, we reasoned that the increased production of pigments in biofilms and under high temperature might involve with the generation of the genetic variants with increased pigment production in P. lypolytica. To further test whether pyomelanin is produced through the L-tyrosine catabolic pathway in P. lipolytica, we tested the pyomelanin production of the P. lipolytica by culturing statically through the addition of 5 mM L-tyrosine. In sea water where no L-tyrosine is present, the addition of L-tyrosine not only induced pyomelanin production ( Fig. 5A and B) but also promoted cell growth (Fig. 5C). In SW-M9 medium where 0.2% casamino acid was added into sea water or in SW-LB medium, the addition of L-tyrosine induced pyomelanin production ( Fig. 5A and B) without affecting cell growth (Fig. 5C). These results suggest that the production of pyomelanin is induced by the presence of more L-tyrosine.
Deletion of hmgA reduces plantigrade settlement of M. coruscus Because the antifouling activities of Pseudoalteromonas are often associated with pigmentation, we then explored whether pyomelanin overproduction is associated with the antifouling activities of P. lipolytica. Hence, we first tested whether biofilms formed by the DhmgA mutant affect the plantigrade settlement of M. coruscus. Biofilms of the wild-type and DhmgA strains were prepared using the same concentration of bacterial suspensions, and attached biofilms that formed on glass slips after 2 days of incubation were used for the settlement assay. The results showed that at initial concentrations of 5 9 10 6 CFU ml À1 and 1 9 10 7 CFU ml À1 , the plantigrade settlement-inducing activities of the attached biofilms formed by DhmgA were reduced 42% and 62% compared with that of the wild-type strain respectively (Fig. 6A, upper panel, Kruskal-Wallis test, P < 0.001). Moreover, no significant difference in biofilm density was observed between the DhmgA mutant and the wild-type strain at the initial concentrations of 5 9 10 6 CFU ml À1 and 1 9 10 7 CFU ml À1 respectively (Fig. 6A, lower panel, Kruskal-Wallis test, P > 0.05).
In addition, we also tested the antifouling activity of the P3 variant. Similar results in terms of plantigrade settlement were found for the comparison of the P3 variant with the wild-type strain: the inducing activities of the P3 variant were reduced 34% and 42% at the initial concentrations of 5 9 10 6 CFU ml À1 and 1 9 10 7 CFU ml À1 respectively (Fig. 6B, upper panel, Kruskal-Wallis test, P < 0.00).
We also quantified the cell densities of the attached biofilms from the two strains, and the results showed that the density of attached cells in the biofilm of the P3 variant was slightly higher than that found for the wild type (Fig. 6B, lower panel, Kruskal-Wallis test, P < 0.001), suggesting that the decreased plantigrade settlement percentage obtained with the P3 variant was not due to a decrease in the cell density of the attached biofilm. A. Pigment production by the wild-type strain after static incubation in sea water, SW-M9 and SW-LB medium with and without L-tyrosine for 7 days at 37°C. B. Quantification of pigment production by the wild-type strain after static incubation with and without L-tyrosine. C. The cell density (CFU ml À1 ) at initial inoculation and after incubation for 7 days. Data are from three independent cultures. One SD is shown in B and C, and representative images are shown in A. Collectively, these results indicate that pyomelanin production leads to increased activities against M. coruscus settlement.
Deletion of hmgA reduces larval settlement and metamorphosis of M. coruscus We next tested whether biofilms formed by the DhmgA mutant and the pyomelanin hyperproduction variant impact larval settlement and metamorphosis of M. coruscus. In the case of initial concentrations of 5 9 10 6 CFU ml À1 , the percentages of larval settlement and metamorphosis in biofilms developed by DhmgA and pyomelanin hyperproduction variant were reduced 73% and 73% compared with that of the wild-type strain respectively (Fig. 7A, Kruskal-Wallis test, P < 0.001). In the case of the initial concentrations of 1 9 10 7 CFU ml À1 , the percentages of larval settlement and metamorphosis in biofilms developed by pyomelanin hyperproduction variant and DhmgA were reduced 84% and 90% compared with that of the wild-type strain respectively (Fig. 7A, Kruskal-Wallis test, P < 0.001). Moreover, higher biofilm density in the pyomelanin hyperproduction variant and lower biofilm density in the DhmgA mutant were observed when compared with that of the wild-type strain at the initial concentrations of 1 9 10 6 CFU ml À1 (Fig. 7B, Kruskal-Wallis test, P < 0.05). In the case of the initial concentrations of 5 9 10 6 CFU ml À1 and 1 9 10 7 CFU ml À1 , higher biofilm density in the pyomelanin hyperproduction variant was observed (Kruskal-Wallis test, P < 0.05), while no significant difference was observed between the DhmgA mutant and the wild-type strain (Kruskal-Wallis test, P > 0.05). It suggests that the decreased larval settlement and metamorphosis was not due to a change in the cell density of the attached biofilm. Collectively, these results indicate that pyomelanin production leads to increased activities against M. coruscus settlement.
Inhibiting effect of pyomelanin in larval settlement and metamorphosis of M. coruscus To further confirm the inhibiting effect of pyomelanin in larval settlement and metamorphosis, we checked whether the biosynthesis pyomelanin extracted from the P. lipolytica DhmgA mutant and synthesized pyomelanin from the auto-oxidation of HGA inhibit larval settlement and metamorphosis of M. coruscus. Our results showed that pyomelanin extracted from the P. lipolytica DhmgA mutant had a significant inhibiting effect on larval settlement and metamorphosis at all concentrations (Fig. 8A, Kruskal-Wallis test, P < 0.001). In the case of pyomelanin synthesized from the auto-oxidation of HGA, significant inhibiting effect was also observed in all the concentrations tested (Fig. 8B, Kruskal-Wallis test, P < 0.001). No larval motility was observed when exposed the larvae to the extracted pyomelanin or in vitro-synthesized pyomelanin at all the concentrations tested (data not shown). Taken together, these results demonstrate that the production of pyomelanin reduces the M. coruscus larval settlement and metamorphosis.

Discussion
Pigments are organic compounds that confer characteristic colours to living tissues and are usually involved in vital processes (Liu and Nizet, 2009;Soliev et al., 2011). The Pseudoalteromonas genus has been divided into two clades based on differences in pigment production (Table S4). Pigmented species of this genus, such as P. rubra, P. denitrificans, P. tunicata, have been shown to exhibit strong antifouling activities Holmstrom et al., 2002;Bowman, 2007). However, few evidences have been shown that the antifouling activities were directly induced by the pigment production. In this study, we first demonstrated that the pyomelanin overproduction mutant greatly reduced the M. coruscus settlement and metamorphosis. Furthermore, the purified pyomelanin from the DhmgA mutant which hyperproduces pyomelanin and the synthesized pyomelanin produced by the in vitro auto-oxidation of HGA reduced the M. coruscus settlement and metamorphosis, suggesting an important ecological role of pyomelanin secreted by marine bacteria. Many marine bacteria are natural producers of melanin-like compounds. Pyomelanin is one class of melanin derived from tyrosine via the intermediate HGA, and the formation of HGA is catalysed by 4-HDDP; HGA is excreted, auto-oxidized and polymerized to form pyomelanin. The present study demonstrated that P. lipolytica DhmgA mutant produces pyomelanin, similar to Vibrio cholerae and Shewanella colwelliana (Kotob et al., 1995;Valeru et al., 2009).
The physiological function of pyomelanin has been previously explored in the clinically important pathogens. For example, in Legionella pneumophila, the secreted pyomelanin conferred a ferric reductase activity which played an important role in iron uptake, thus enhancing growth under iron-limiting conditions (Chatfield and Cianciotto, 2007;Zheng et al., 2013). In Vibrio cholera, pyomelanin production results in increased toxin-coregulated pilus and cholera toxin expression and increases the ability of the bacteria to colonize the intestine of infant mice (Valeru et al., 2009). Although pyomelanin is related to the virulence and pathogenicity of pathogenic bacteria, pyomelanin production is also associated with numerous survival advantages. The primary advantage is thought to be protection against ultraviolet  radiation. In Burkholderia cenocepacia and Pseudomonas aeruginosa, pyomelanin has been reported to scavenge free radicals and is thus able to protect cells from in vitro and in vivo sources of oxidative stress (Keith et al., 2007;Rodriguez-Rojas et al., 2009). In Bacillus anthracis, the loss of hmgA can lead to the accumulation of pyomelanin, which protects B. anthracis cells from UV damage (Han et al., 2015). In Shewanella oneidensis, the release of pyomelanin into the surrounding environments could contribute to the survival in oxygen limited environment by increasing the anaerobic respiration capacity (Turick et al., 2009). Here we demonstrated that the secreted pyomelanin from marine bacteria also plays an important role in mediating the settlement and metamorphosis of M. coruscus and thus has its potential in control biofouling. The hmgA variants that produce pyomelanin have been isolated from pathogenic bacteria during biofilm growth in a laboratory or from clinical isolates from infected patients (Boles et al., 2004;Keith et al., 2007;Valeru et al., 2009). For example, naturally occurring pyomelanin hyperproducing Vibrio cholerae variants have been found during growth under commonly used experimental culture conditions (Valeru et al., 2009;Wang et al., 2011). Variants of Pseudomonas aeruginosa that overproduced pyomelanin have been previously isolated from patients with cystic fibrosis and bronchiectasis (Drenkard and Ausubel, 2002), and they were also identified in biofilm communities grown in the laboratory (Boles et al., 2004;Rodriguez-Rojas et al., 2009). The hyperproduction of pyomelanin in these variants is due to the loss or reduced activity of the key member of the L-tyrosine catabolic pathway, HmgA. It has been reported that higher temperature could increase the mutation rate and contribute to bacteria evolution (Puurtinen et al., 2016). Moreover, variant subpopulations often emerge in bacteria during biofilm mode of growth (Stewart and Franklin, 2008). In this study, we found that the increased pyomelanin production under elevated temperature or during biofilm formation was neither due to a reduced expression of hmgA gene nor due to an increased expression of the melA gene. Thus, the increased pyomelanin production might be induced by the generation of the pyomelanin hyperproduction variants, such as hmgA mutant, in the biofilm of P. lipolytica under elevated temperature.
The Pseudoalteromonas genus is often found in association with eukaryotic hosts and adopted an attached lifestyle in the native marine environment (Holmstrom and Kjelleberg, 1999). Surface colonization of marine bacteria often involves biofilm formation and plays important role in numerous critical marine processes, including nutrient regeneration and organic matter remineralization (Dang and Lovell, 2016). Once forming biofilms, the establishment of concentration gradients of oxygen and nutrient could lead to the generation of genetic variants through mutation or recombination (Stewart and Franklin, 2008). In this study, hmgA variant of P. lipolytica was generated from laboratory-cultured biofilms in nutrientrich SW-LB medium; we hypothesized that similar mutations might also occur in the marine habitats where marine bacteria constantly face with the availability of the oxygen and nutrient. Indeed, variants isolated from laboratory-cultured biofilms formed by pathogenic bacteria share similar mutations to those found in clinical isolates during infection (Drenkard and Ausubel, 2002;D'Argenio et al., 2007;Woo et al., 2012). Pseudoalteromonas strains that produce pigments were previously isolated in various marine environments. Of the 24 pigmented Pseudoalteromonas strains identified to date, five, namely P. nigrifaciens (Baumann et al., 1984), P. aliena , P. distinct (Ivanova et al., 2000), P. carrageenovora (Ivanova et al., 2000) and P. haloplanktis (Ivanova et al., 2000), have been characterized as melanin-like pigment-producing strains (Table S4). Enzymes responsible for pyomelanin production are generally found in Pseudoalteromonas strains according to the genome annotation in NCBI database (data not shown). Thus, it remains to be determined whether these melanin-producing Pseudoalteromonas strains carry genetic variations in the hmgA gene or the production of the pyomelanin is induced at the transcriptional level.
Marine-derived natural antifouling products are good potential antifoulants. The settlement of marine invertebrate larvae can be influenced by positive or negative cues produced by different marine organisms (Walters et al., 1996;Dobretsov et al., 2013). As previously reported, natural antifouling agents could be classified into six main groups according to their chemical structures: fatty acid-related compounds, polyketide-related compounds, terpenoid-related compounds, steroidrelated compounds, alkaloid-related compounds and non-ribosomal polypeptide-related compounds (Dobretsov et al., 2013;Qian et al., 2015). In addition, extracellular polymeric substances isolated from biofilms formed by Pseudoalteromonas ulvae and Oceanobacillus iheyensis have been shown to exhibit antifouling activities (Kavita et al., 2014;Brian-Jaisson et al., 2016). In this study, we showed that pyomelanin exhibits strong antifouling activities. Pyomelanin is composed of complex polyphenolic heteropolymers and might represent a new group of chemical compounds that could serve as antifouling agents. The molecular mechanisms underlying the antifouling activity of pyomelanin as well as the physiological and ecological roles of pyomelanin production by marine Pseudoalteromonas in diverse marine ecosystems need to be further investigated.

Experimental procedures
Strains and growth conditions Escherichia coli strain WM3064 strain was grown in Luria-Bertani (LB) medium containing 0.3 mM DAP (2,6-diamino-pimelic acid) at 37°C (Dehio and Meyer, 1997). Pseudoalteromonas lipolytica SCSIO_04301 was isolated from sediment at 63 m deep in the South China Sea (Zeng et al., 2014). P. lipolytica DhmgA mutant strain used in this study has been constructed as we previously reported (Wang et al., 2015a,b). P. lipolytica was grown in 2216E medium (0.5% peptone, 0.1% yeast extract and different concentrations of the trace elements) (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) or SW-LB medium (1% tryptone and 0.5% yeast extract dissolved in sea water) at 25°C unless indicated. Chloramphenicol (30 lg ml À1 ) was used for the maintenance of pBBR1MCS-based plasmids (Zeng et al., 2015).

Pigment production quantification
Wild-type P. lipolytica strain was cultured to OD 600 of 1.0 as planktonic stage or cultured by statically as biofilm formation for different days at different temperature. The cultures were centrifuged at the 17 000 g for 15 min after indicated days to remove cell pellet. The production of pigment in the supernatant was estimated by measuring the absorbance of the supernatant at 400 nm as previous reports have shown that pigment production can more easily be distinguished by measuring the absorbance at a lower wavelength (Ruzafa et al., 1995;Lagunas-Munoz et al., 2006;Chatfield and Cianciotto, 2007). To monitor the secreted pigments, sea water, SW-M9 (0.2% casamino acid dissolved in sea water) and SW-LB medium supplemented L-tyrosine or not were used to culture P. lipolytica by shaking or statically for different days, and the pigments were accessed by measuring the absorbance of the supernatant at 400 nm.

Isolation of biofilm variants
For the isolation and quantification of biofilm variants, biofilms were harvested, uniformly homogenized and were plated onto SW-LB agar plates as previously described (Zeng et al., 2015). A total of 5000 colonies were examined to screen for variants with altered colour morphology. Colour morphology of these variants was further examined after three rounds of overnight passaging.

Whole-genome resequencing
The P3 variant genome was sequenced by BGI Co., Ltd. (Shenzhen, Guangdong, China) using the whole-genome shotgun method with the Illumina HiSeq 2000 sequencing platform as previously reported (Zeng et al., 2015). SNPs and InDels were detected based on an alignment of the assembly sequence and the wild-type reference (Zeng et al., 2014).

Complementation analysis
The broad-host-range vector pBBR1MCS was used to express the target genes in P. lipolytica. The coding region of the hmgA gene (GenBank accession number JDVB01000008.1) and its promoter region were amplified by PCR using the primer pair hmgA-F (5 0 -CGGAATTCTCGACCATATCGCTGCCATCACG-3 0 ) and hmgA-R (5 0 -CCCAAGCTTTTATACGTCCTGTTTACCTG TA-3 0 ) and ligated to pBBR1MCS-Cm after enzymatic digestion with EcoRI and HindIII. The resulting recombinant plasmid was sequenced using the primers pBBR1MCS-f and pBBR1MCS-r and conjugated into P. lipolytica according to a previously described method (Wang et al., 2015a,b).

FTIR analysis
The pigment of DhmgA mutant was extracted according to a previously published method with minor modifications (Schmaler-Ripcke et al., 2009). Briefly, a 200 ml culture of DhmgA mutant was grown for 3 days in SW-LB supplemented with 2.5 mM L-tyrosine. Culture supernatant was collected by centrifugation and filter sterilization. The supernatant was acidified with 6 M HCl to a pH of 2.0 and allowed to precipitate overnight at room temperature. After centrifugation (16 500 g, 30 min), the pellet was resuspended in 2.5 ml of water at pH of 12 and dialysed in 3.5 kDa dialysis tubing for 24 hr against water. The dialysed pigment was then dried using the vacuum centrifugal concentrator (EYELA Tokyo Rikakikai Co Ltd CVE-3110, Tokyo, Japan). Synthetic pyomelanin was used as a positive control compound in FTIR analysis. Synthetic pyomelanin was produced by autooxidation of a 10 mM HGA (Tokyo Chemical Industry. Co. Ltd, Tokyo, Japan) at pH 10 with constant stirring for 3 days. Polymerization was stopped, and precipitation was started by adjusting the pH to 2 with 6 M HCl and further treated as described above. The pigment extracted from the DhmgA mutant and synthetic pyomelanin was used for FTIR analysis (Shimadzu Corporation, IR Affinity-1, Kyoto, Japan).
Culturing of the mussels Plantigrades and pediveliger larvae of the mussel Mytilus coruscus were supplied by the Institute of Marine Science and Technology of Shengsi, Zhoushan, China.
The plantigrades and pediveliger larvae of M. coruscus were cultured following a previously described method with some modifications (Wang et al., 2012;Yang et al., 2013Yang et al., , 2014a. Briefly, the plantigrades were transferred and cultured in aerated glass beakers with filtered sea water (acetate fibre filter, 1.2 lm pore size) at an initial density of 10 plantigrades ml À1 and fed a diet consisting of the algae Isochrysis galbana at a concentration of 10 5 cells ml À1 day À1 . Plantigrades with a shell length of 2.01 AE 0.04 mm and a shell height of 1.10 AE 0.03 mm were used for the settlement bioassays. Pediveliger larvae were transferred and cultured in aerated glass beakers with filtered sea water at an initial density of 5 larvae ml À1 and fed a diet consisting of the algae I. galbana at a concentration of 5 9 10 4 cells ml À1 day À1 . The plantigrades and pediveliger larvae were cultured for more than 1 week and then were used for the settlement bioassays.

Mussel settlement bioassay
Biofilms of P. lipolytica and pigment production variants were prepared following a modified method (Zeng et al., 2015). Briefly, each strain was cultured in 2216E at 25°C for 48 h and then cells were harvested. Cell pellet was washed three times by autoclaved filtered sea water (AFSW, GF/C filter, pore size = 1.2 lm; Whatman International Ltd., Maidstone, UK), and final cell density was adjusted to colony-forming unit (CFU) of 10 6 to 10 7 per ml. Cell suspension was transferred to 24-well plate and sterile glass Petri dishes with one piece of sterile glass slip. Bacteria that remained firmly attached on surfaces of plates and the glass slips were viewed as irreversible attached bacterial biofilms. For each strain, six replicates were conducted. For the plantigrade settlement bioassay, ten plantigrades were transferred into individual glass Petri dishes containing 20 ml of AFSW and a glass slip with monospecific bacterial biofilm (Yang et al., 2014a,b). Six replicates were conducted. The inducing activity was determined by the percentage of settled plantigrades after 12 h. The settled plantigrades were verified after attachment to the surface with byssal threads. A glass slip without biofilm was included as a negative control in all assays. For the pediveliger larval settlement bioassay, ten pediveliger larvae were added into each 24-well plate containing bacterial biofilms and 1.5 ml of pyomelanin solution of different concentrations (Yang et al., 2011(Yang et al., , 2014a. A group of pediveliger larvae were exposed to bacterial biofilms as a positive control. Six replicates were conducted. In addition, pediveliger larvae were also exposed to pyomelanin from biosynthesis and chemical synthesis without the addition of bacterial biofilms as a control. The inducing activity was evaluated by the percentage of metamorphosed individuals (post-larvae) after 72 h following our previously developed method (Wang et al., 2012). Post-larvae were examined at 1009 magnification under an Olympus stereoscopic microscope and the images of the pediveliger larvae and post-larvae of M. coruscus are shown in Fig. S3. A negative control was included with twenty pediveliger larvae and 1.5 ml of AFSW in the absence of attached biofilm. Assays were conducted at 18°C in darkness with six replicates for each condition.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Fig. S1. The visible light spectra of the pigments collected from the superanatnat of the P. lipolytica DhmgA mutant grown for 2 days and wild-type P. lipolytica cultured statically in SW-LB medium at 25°C for 10 days. Fig. S2. P. lipolytica DhmgA mutant strain produced pyomelanin. Fig. S3. The settlement and metamorphosis of the pediveliger larvae of Mytilus coruscus. Table S1. Point mutations revealed by whole-genome resequencing of the pyomelanin hyper-production variant P3 isolated from P. lipolytica biofilms. Table S2. Tyrosine metabolic pathway in P. lipolytica predicted by compared with Pseudomonas putida NBRC 14164. Table S3. Identification of four L-tyrosine catabolic pathway related enzymes in the genome of the represented Pseudoalteromonas strains. Table S4. List of pigmented Pseudoalteromonas strains isolated from various marine or other habitats. Table S5. The fold change of the hmgA and melA transcription level in P. lipolytica under stress conditions compared to planktonic stage at 25°C.