Description of Prasinibacter corallicola gen. nov., sp. nov., a zeaxanthin-producing bacterium isolated from stony coral Porites lutea

Thermal stress is considered one of the main causes of mass scleractinian coral degradation; however, it is still unknown how corals can adapt to future global warming. In this study, 11 strains of coral-associated Flavobacteria were shown to produce zeaxanthin, a carotenoid antioxidant, which may help coral holobionts to alleviate thermal stress. In addition, a novel zeaxanthin-producing Flavobacterium, designated R38T, was identified using polyphasic taxonomy. Although strain R38T shared a maximum 16S rRNA gene sequence similarity of 93% with Mesoflavibacter aestuarii KYW614T, phylogenetic analyses based on whole genome and 16S rRNA gene sequences revealed that strain R38T forms a distinct branch in a robust cluster composed of strain R38T and Leptobacterium flavescens KCTC 22160T under the family Flavobacteriaceae. Strain R38T exhibited average nucleotide identities of 70.2% and 72.5% for M. aestuarii KYW614T and L. flavescens KCTC 22160T, respectively. The only detected respiratory quinone was menaquinone 6 (MK-6). The genomic DNA G + C content was 33.2 mol%. The major polar lipids were phosphatidylmethylethanolamine, phosphatidylethanolamine, one unidentified ninhydrin phospholipid, three unidentified ninhydrin-positive lipids, and three unidentified lipids. The major cellular fatty acids were iso − C15: 0, iso − C15: 0 ω6c, C16:2 DMA, and C13:1 ω3c. The distinct biochemical, chemotaxonomic, phylogenetic, and phylogenomic differences from validly published taxa suggest that strain R38T represents a new species of a new genus, for which Prasinibacter corallicola gen. nov., sp. nov. is proposed. The type strain R38T (= MCCC 1K03889T = KCTC 72444T).


Introduction
Global coral reefs have degraded dramatically in recent decades (Hughes et al. 2003;Bellwood et al. 2004) and massive corals have replaced branched corals as the dominant assemblages (Perry et al. 2015;Yu et al. 2019). Thermal stress is one of the main causes of scleractinian coral degradation (Gardner et al. 2003;Hughes et al. 2017Hughes et al. , 2018Hughes et al. , 2019. Recently, it was reported that the phycosphere bacteria Muricauda sp. GF1 protects coral endosymbionts from thermal stress by producing zeaxanthin (Motone et al. 2020). Zeaxanthin is usually synthesized in flavobacteria from terpenoids by the combinations of phytoene synthase (CrtB), phytoene dehydrogenase (CrtI), lycopene cyclase (CrtY), and β-carotene hydroxylase (CrtZ) (Zhang et al. 2018). According to the List of Prokaryotic names with Standing in Nomenclature (https:// lpsn. dsmz. de/), to date, up to 152 genera and more than 980 species have been validly published in the family Flavobacteriaceae. Members of the family are ubiquitous in terrestrial, freshwater, and marine environments, and some ones are marine animal associated (McBride, 2014). Genomic and physiological analyses have indicated that flavobacteria can degrade a diverse range of carbohydrates and proteins (Bauer et al. 2006;Qin et al. 2010;Gavriilidou et al. 2020), and a few are animal pathogens (Duchaud et al. 2007;Loch and Faisal 2015;Adamek et al. 2018). To determine the roles of flavobacteria in coral health, pure cultures were isolated from thermal tolerant corals: Porites lutea, Galaxea fascicularis, and Favia sp. The taxonomic position of a new coral-associated Flavobacterium, strain R38 T , was identified using polyphasic identification, and the results indicated that strain R38 T and other coral-associated Flavobacteria can also produce zeaxanthin. Therefore, coral-associated Flavobacteria may help alleviate host thermal stress caused by global warming.
Preliminary identification using 16S rRNA gene sequence indicated that strain R38 T (from P. lutea) may represent a new species in a new genus, therefore, polyphasic identification was performed to identify its exact taxonomic position. Type strains Leptobacterium flavescens KCTC 22160 T and Spongiivirga citrea KCTC 32990 T , obtained from the Korean Collection for Type Cultures (KCTC), were used as references. Both the new isolate and the reference type strains grew well on marine agar 2216 (BD).

Chemotaxonomic characterization
The biomass of strain R38 T and the reference strain for cellular fatty acid analysis were acquired from the third quadrant of the streaked marine agar 2216 plate incubated at 28 °C. Cellular fatty acid composition was analyzed by gas chromatography (Agilent G6890N) and identified using the Sherlock Microbial Identification System (Version 6.0) according to the manufacturer's instructions. Biomass for the analyses of quinones and polar lipids was obtained from marine broth 2216 after 3 days of incubation at 30 °C. Respiratory quinones were extracted as described by Collins (1994) and analyzed using reversed-phase high-performance liquid chromatography (HPLC) (Komagata & Suzuki, 1987). The isoprenoid quinones were eluted using a mixture of methanol/2propanol (2:1, v/v) and flow rate of 1 ml/min at room temperature and detected by UV absorbance at 270 nm. Polar lipids were extracted as described by Kamekura (1993) and identified by spraying with ethanolic molybdophosphoric acid, molybdenum blue, ninhydrin, α-naphthol/sulfuric acid, and Dragendorff's reagent after two-dimensional thin layer chromatography (TLC) (Tindall 1990).

Phylogenetic and phylogenomic analysis
The 16S rRNA gene of strain R38 T was obtained using PCR amplification with the universal primers 27F and 1492R (Lane 1991) and sequenced using the Sanger method. The 16S rRNA gene sequence similarities were determined using the EzBioCloud (Yoon et al. 2017a, b) and NCBI database. Alignment of 16S rRNA gene sequences was performed using the SINA software package (Pruesse et al. 2012) and the SILVA rRNA database. Phylogenetic trees were constructed using the maximum likelihood (Felsenstein 1981), neighbor-joining (Saitou & Nei 1987), and maximum-parsimony (Swofford 1993) algorithms in the software package MEGA version 7.0 (Kumar et al. 2016). The phylogenetic distance matrices were estimated using the Kimura two-parameter model (Kimura, 1980). The topology of the phylogenetic tree was analyzed using the bootstrap resampling method of Felsenstein (1985) with 1000 replicates. Whole-genome sequencing was performed using an Illumina HiSeq PE150 platform. Library construction was performed by PCR amplification of a 350 bp insert with A-tail ligated to paired-end adaptors, at the Beijing Novogene Bioinformatics Technology Co., Ltd. Good-quality paired reads were assembled into a number of scaffolds using the SOAPdenovo (Li et al. 2008(Li et al. , 2010. Genomic information was extracted as described by Chun et al. (Chun et al. 2018). The phylogenetic tree was reconstructed using the up-todate bacterial core gene set (UBCG v.3) according to the manual (Na et al. 2018). The average nucleotide identity (ANI) was calculated using the online ANI calculator (Yoon et al. 2017a, b). The average amino acid identity (AAI) was calculated using the EzAAI (Kim et al. 2021). Carbohydrate-active enzyme annotation was performed using the dbCAN meta server (Zhang et al. 2018). Peptidases were annotated using the Hotpep-protease method (Bush, 2020) based on the Merops database.

Zeaxanthin detection
The production of zeaxanthin by bacteria from the family Flavobacteriaceae was assessed using both genome annotation and HPLC assay. The existence of coding genes for zeaxanthin biosynthesis enzymes, namely phytoene desaturase, lycopene beta-cyclase, and beta-carotene 3-hydroxylase were checked using the Joint Genome Institute online server (https:// genome. jgi. doe. gov/ portal/) and the NCBI genome server (https:// www. ncbi. nlm. nih. gov/ genome/). Zeaxanthin was extracted from approximately 10 mg wet weight bacteria grown on an R2A plate using methanol and glass beads on a vortex for 30 s, and then analyzed using HPLC (Thermal Ultimate 3000). Samples were separated using an Agilent ZORBAX Eclipse XDB-C18 (250 mm, 5 µm particle size) at a column temperature of 35 °C, and the mobile phase comprised 90% (vol/vol) methanol containing 0.1% (v/v) formic acid at a flow rate of 1 ml/min (Motone et al. 2020). Zeaxanthin was checked using a photodiode array detector. Standard zeaxanthin was purchased from the Resource Platform of Standard Material (China).

Results and discussion
Morphological, cultural, physiological and biochemical characterization Colonies of strain R38 T on marine agar 2216 were yellow and circular. Cells of bacterial strain R38 T were gram-negative, non-spore-forming, non-motile, aerobic rods. Cells were usually 0.3-0.5 μm wide and 0.9-2.0 μm long ( Supplementary Fig.S1), being narrower than that of L. flavescens KCTC 22160 T , S. citrea KCTC 32990 T and Fulvibacter tottoriensis MTT-39 T , while being wider than that of Mesoflavibacter aestuarii KYW614 T ( Table 1). Cells of strain R38 T could reduce nitrate to nitrogen, L. flavescens KCTC 22160 T and S. citrea KCTC 32990 T could only reduce nitrate to nitrite, whereas M. aestuarii KYW614 T (Lee et al. 2014) and F. tottoriensis MTT-39 T (Khan et al. 2008) could not reduce nitrate. Enzyme characterization of strain R38 T using API ZYM strips showed a spectrum similar to that of M. aestuarii KYW614 T and F. tottoriensis MTT-39 T with the absence of β-galactosidase, β-glucuronidase, α-glucosidase, β-glucosidase, N-acetyl-β-glucosaminidase, and α-mannosidase (Table 1) (Lee et al. 2014;Yoon et al. 2013). These enzyme results differed from those of L. flavescens KCTC 22160 T and S. citrea KCTC 32990 T (Table 1). Other characteristics of strain R38 T are listed in Table 1 and the species description.

Molecular characterization and phylogenetic analysis
A nearly complete 16S rRNA gene sequence (1383 nt) of strain R38 T was obtained by Sanger sequencing and deposited in GenBank under accession number MN908337. Global alignment using the EzBioCloud database indicated that the most closely related neighbor of strain R38 T is M. aestuarii KYW614 T , with a 16S rRNA gene similarity of 93%. The next most similar members were of Bizionia, Sabulilitoribacter, Gaetbulibacter, and Algibacter genera which showed 92.7-92.9% sequence similarity. However, 16S rRNA gene phylogenetic analysis based on the maximumlikelihood algorithm indicated that strain R38 T forms a distinct branch in a stable cluster composed of strain R38 T and L. flavescens KCTC 22160 T (Fig. 1), neighbor-joining and maximum-parsimony algorithms also support this stable cluster.

Genome properties and comparison
The genome sequencing depth of strain R38 T was 333 × , and the N50 was 750,154 bp. A total of 11 contigs were obtained, the obtained genome size was 3.94 Mb, and the genomic DNA G + C content was 33.2 mol%. The genome sequencing depth of L. flavescens KCTC 22160 T was 154 × , the N50 was 1,032,064 bp, a total of 9 contigs were obtained, the obtained genome size was 4.21 Mb, and the genomic DNA G + C content was 40.9 mol%. The genome sequencing depth of S. citrea KCTC 32990 T was 155 × , the N50 was 637,254 bp, a total of 17 contigs were obtained, the obtained genome size was 4.15 Mb, and the genomic DNA G + C content was 36.3 mol%. The genomes of closely related type strains were 2.92-5.03 MB, with G + C content of 31.8-55.3 mol% (Supplementary Table S2). Thus, strain R38 T has a low genomic G + C content  Lee et al. 2014); 5, Fulvibacter tottoriensis MTT-39 T (data from Khan et al. 2008;Yoon et al. 2013 NaCl tolerance (%, w/v) 3-6 0.5-6 1-5 1-9 1-5 pH range 5-10 6-10 8-9 6-8 6-10 Nitrate reduction  Supplementary  Fig.S4).
Approximately 37 families of carbohydrate-active enzymes were detected in strain R38 T , while in closely related type strains this number was between 0.1 quantity of enzymes for carbon hydrate utilization (glycoside hydrolases, polysaccharide lyases, and carbohydrate esterases) (28 vs. 20-91) indicates that strain R38 T is weak in carbohydrate utilization (Supplementary Table S2). Approximately 71 families of peptidases were detected in strain R38 T . This quantity is higher than in most of the closely related type strains (Supplementary Table S2), indicating that strain R38 T is versatile in protein utilization. This carbohydrate and protein utilization pattern might have resulted from long-term bacteria-animal association.

Taxonomic conclusion
Based on phylogenetic analyses, strain R38 T was found to be associated with the family Flavobacteriaceae. The ANI of strain R38 T to closely related type strains (≤ 72.5%) indicates that strain R38 T belongs to a novel species , and both biochemical and chemotaxonomic characteristics (Table 1) support this species-level assignment. Furthermore, the low 16S rRNA gene similarities (≤ 93%) of strain R38 T to closely related type strains indicate that strain R38 T represents a new genus (Yarza et al. 2014), which is also supported by the differences in polar lipid profile (Table 1 & Supplementary Fig.S2). Therefore, strain R38 T represents a new species in a new genus under the family Flavobacteriaceae.

Zeaxanthin production
Coding genes for Zeaxanthin biosynthesis enzymes phytoene desaturase, lycopene beta-cyclase, and betacarotene 3-hydroxylase were examined in 115 of the 177 total genera in Flavobacteriaceae, and over 50% of these genera (62) had all three enzymes (Supplementary Table S3). Of note, all 11 strains of coralassociated flavobacteria (from approximately ten genera), including strain R38 T , were able to produce zeaxanthin according to HPLC analysis (Supplementary Table S4). These results indicate that the Flavobacteriaceae family contains important zeaxanthin producers, and corals may benefit from these symbiotic flavobacteria when confronting thermal stress (Motone et al. 2020).
The type species is Prasinibacter corallicola. Member of the family Flavobacteriaceae.
The type strain, R38 T (= MCCC 1K03889 T = KCTC 72444 T ) was isolated from stony coral Porites lutea collected from Weizhou Island in the Beibu Gulf, China. The GenBank accession number of the 16S rRNA gene sequence of the type strain is MN908337.
Author contributions GW, JL and YL isolated strains, performed experiments and wrote the manuscript. BC and ZL collected samples. HS and JL gave advices about bacteria cultivation. KY conceived and designed the experiments and approved the final manuscript.