QQ bacteria isolation and screen
Coral samples for bacteria isolation were collected from the Luhuitou fringing reef in Sanya, China (109°28′27″ E, 18°12′19″ N) on December, 2019. Coral fragments of Pocillopora damicornis, Acropora millepora, Galaxea fascicularis, Acopora formosa and Acropora digitifera were collected, flushed with sterilized seawater and homogenized by grinding and stirring. The homogenates were then serially diluted (10− 1 to 10− 9 times) in sterilized seawater, and 100 µL of the dilutions were spread onto marine agar 2216E and incubated at 28°C for 2–7 days. Single bacterial colonies were picked and streaked onto new plates to obtain pure strains.
To screen the obtained strains with QQ activity, a disc diffusion assay (Bauer et al. 1966) was performed using the biosensor strain Chromobacterium violaceum CV026 (McLean et al. 2004). Briefly, C. violaceum CV026 was cultured in Luria-Bertani broth (LB: consisting of 10 g L− 1 tryptone, 5 g L− 1 yeast extract, 10 g L− 1 NaCl, sterilized at 121°C) supplemented with 25 µg·mL− 1 kanamycin. N-hexanoyl-L-homoserine lactone (C6-HSL, Sigma-Aldrich, USA) and 2,5-dimethyl-4-hydroxy-3[2H] furanone (Aladdin, China) were prepared by dissolution in methanol. Each 6 mL of overnight reporter strain culture and 0.5 µM C6-HSL were added to 54 mL of LB media containing 0.8% agar until the temperature of the media was about 45°C. Sterile filter paper circles (6 mm diameter) were then placed on the solidified plates for sample loading. Bacterial suspension (10 µL) in 2216E broth was pipetted onto the filter paper for testing. In addition, 10 µL of furanone (0.1 M), methanol solvent and 2216E broth were used as positive, negative and blank controls in this plate-based bioassay. After incubation for 24 h at 28°C, the inhibition of pigment production around the disc (a colorless ring) was checked. Visible colorless haloes around the filter paper circles represent potential positive QQ activity. To ensure reliability, QQ activity screening experiments were repeated three times independently.
Genomic DNA of potential QQ bacterial strain was extracted using the Bacterial DNA Kit (Omega Bio-tek, USA) following the manufacturer's instructions. The 16S rRNA gene was amplified by PCR using the primers 27F (5′AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-GGTTACCTTGTTACGACTT-3′) (Weisburg et al. 1991), and sent for sequencing (Tianyi Huiyuan Biotechnology, China). The 16S rRNA gene of the QQ bacterial strain and reference sequences were aligned using Clustal W (Larkin et al. 2007). The phylogenetic tree was constructed by the neighbor-joining method using MEGA software (Tamura et al. 2021).
Extraction and Identification of QSI
The QQ bacterial strain Pseudoalteromonas SCSIO 43740 was fermented in10 L of 2216E broth for 72 h (28°C, 180 rpm). The fermentation broth was then extracted twice with an equal volumes of ethyl acetate, by vigorous shaking for 10 minutes. The aqueous fractions were discarded and the organic extract fractions were combined and evaporated in a rotary evaporator at 45°C. The organic residues were dissolved in methanol (Nithya et al. 2010), filtered through 0.22µm filters and concentrated by nitrogen flow. The extracts were then loaded onto a silica gel column chromatography (10g; 100–200 mesh) and eluted with a stepwise system of petroleum ether/ethyl acetate = 1:0, 50:1, 25:1, 9:1, 6:1and 3:1 (v/v, 1 L each). The fractions were grouped according to their thin-layer chromatography (TLC) profile. These single band fractions were evaporated to dryness and tested for QSI activity. After drying, the fractions were dissolved in methanol and tested for QSI activity at 16.4 mg/ml to better discriminate the most active fraction.
As previously described (Cho et al. 2016), nuclear magnetic resonance (NMR, Bruker Biospin, Fällanden, Switzerland) analysis was carried out after reconstitution of the bioactive fraction in deuterated methanol (CD3OD, Cambridge Isotope Laboratories, Inc), containing tetramethyl silane (TMS) as analytical solvents. NMR spectra were recorded on a Bruker Avance III spectrometer (Bruker Biospin) (500 and 126 MHz for 1H and 13C NMR, respectively) equipped with a 1.7 mm MicroCryoprobe (Bruker Biospin), using the signal of the residual solvent as internal reference (δH 3.31 and δC 49.15 ppm for CD3OD). All mass spectra were acquired on a hybrid iontrap time-of-flight mass spectrometer (Bruker Biospin, Fällanden, Switzerland) equipped with an electrospray ionization source (ESI-MS).
Coral culture and experimental design
Ten healthy coral individuals (diameter ~ 25cm) were collected from the Luhuitou fringing reef (Sanya, China). Prior to use, the coral P. damicornis was divided into 120 coral fragments (5 cm each) and transferred to a sand-filtered seawater tank. The corals were cultured in the laboratory for 30 days under the same initial environmental conditions (temperature, 29 ± 0.5°C; pH, 8.1 ± 0.05; and salinity, 31.0 ± 0.5‰). All tanks were maintained at 200 µmol/m2 s− 1 light in a 12-h light/dark cycle. Figure 1 shows the schematic diagram of this study, which included a control group (Con: 29°C) and a high temperature group (HT: 31°C), and a high temperature with QQ bacteria inoculation group (HTQQ: 31°C + QQ bacteria) (Fig. 1). Coral fragments were randomly assigned to each group.
As previously described (Rosado et al. 2019), after the initial 5-day coral acclimation period at 29°C, in two of the treatments the temperature was gradually increased from day 1 to day 5 until it reached 31°C. The controlled experiment started when the maximum temperature was reached. In each treatment, 5 coral samples were taken from tanks on days 0, 10 and 30. The timing of the addition of experimental strains is shown in Fig. 1. QQ bacteria were added every 10 days, and the final concentration of inoculated QQ bacteria in the tank was 1 × 107 CFU mL− 1.
Following live measurements of physiological performance, fragments were snap frozen in liquid nitrogen and stored at -80°C until further processing, when they were airbrushed with 0.22 µm filtered seawater, homogenized to obtain tissue slurry (Diax 900, Heidolph Instruments) and aliquoted into individual 1 mL samples for measurements of host protein, superoxide dismutase and catalase activity, and symbiont density.
Measurement of coral physiological parameters
A total of five physiological parameters of coral holobionts were assessed at each time point, including photosynthetic efficiency (Fv/Fm), symbiont algal density, total protein content, superoxide dismutase (SOD) activity and catalase (CAT) activity. Symbiont algal density was counted by light microscope (Olympus Biomicroscope, Olympus Japan Ltd). The counts were repeated three times for each sample. Coral photosynthetic efficiency (Fv/Fm) was assessed using the diving-PAM system (Walz GmbH, Effeltrich, Germany), with the glass fiber-optic probe positioned over the oral disc (Ralph et al. 2005; Rosado et al. 2019). To avoid interference from photoinhibition, measurements were taken after 30 min of darkness to ensure full recovery of photosynthetic reaction centers. Coral protein concentration was determined in triplicate per sample using a BCA protein concentration kit (Beyotime Biotechnology, China). SOD concentration was determined in triplicate per sample using a superoxide dismutase activity colorimetric assay kit (Beijing Sollerbauer Technology Co, China). Results are expressed in ‘standard cytochrome c’ SOD units (U), by measuring the ratio of auto-oxidation rates in the presence and absence of the sample (Gardner et al. 2017). CAT concentrations in coral holobionts were determined using the CAT activity assay kit (Beijing Sollerbauer Technology Co, China) following the manufacturer’s instructions. SOD and CAT activities were normalized to protein concentrations.
16S rRNA gene sequencing
The homogenized coral fragment subsamples (~ 1 g) were used for microbial DNA extraction using HiPure Soil DNA Kits (mBio, USA). The quality and quantity of extracted DNA was checked on a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA). Prokaryotic 16S rRNA genes were amplified by the primer pair F341 (5′-CCT ACG GGN GGC WGC AG-3′) and R806 (5′-GGA CTA CHV GGG TAT CTA AT-3′) as described in previous study (Cole et al. 2009). A 6-mer-barcode sequence was linked on the primer for distinguishing different samples. Briefly, the reactions were carried out in a 50 µL volume using a Bio-Rad PCR cycler: pre-denaturation at 94°C for 2 min; 30 cycles of denaturation at 95°C for 10 s, annealing at 62°C for 30 s, and elongation at 68°C for 30 s; with a final extension for 10 min at 68°C. To generate the sequencing libraries, PCR products for each sample were pooled and purified using the AMPure XP Beads (Qiagen, USA) and quantified using Qubit 3.0. The sequencing was performed on a NovaSeq platform (Illumina, USA) using a paired-end 250-bp sequence read run (Gene Denovo, Guangzhou).
Sequencing data were processed using the QIIME 2 pipeline (Bolyen et al. 2019). Sequences were trimmed and filtered according to previously described methods (Callahan et al. 2016). DADA2 is used for quality control, noise reduction and filtering to obtain a preliminary amplicon variation sequence (ASV) table. The original ASV table was rarified to a depth of 7,000 sequences per sample for further diversity calculation. The ASV sequence taxonomy annotation was conducted by blast against the EzBioCloud 16S database (Yoon et al. 2017).
Microbial network analysis
To evaluate the co-occurrence patterns of coral microbes, network analysis was performed using the Molecular Ecological Network Analyses Pipeline (MENAP) platform based on random matrix theory (http://ieg4.rccc.ou.edu/mena). To reduce network complexity, only the top-class nodes with strong positive (r > 0.6) and strong negative (r < -0.6) relationships were shown in the network diagrams. The networks were visualized in Cytoscape v3.9.1 (Shannon et al. 2003), with the Network-Analyser tool used to calculate each network’s complexity, namely its number of nodes, clustering coefficients, link number, and average connectivity (Berry and Widder 2014). The robustness test is a powerful method to measure network stability (Montesinos-Navarro et al. 2017). We calculated microbial network stability employing natural connectivity (Wu et al. 2019) in R software using the ggClusterNet package.
Statistical analysis
The parameters are presented as mean ± standard error. The data were subjected to a normality test and a variance homogeneity test. One-way analysis of variance (ANOVA) was used to statistically analyze the differences, was marked using the LSD alphabet notation method. The ASV counts were normalized by the metagenomeSeq method (Paulson et al. 2013). Based on the ASV abundance information, principal component analysis (PCoA) was performed to investigate the microbial structure relationship between samples, and the weighted unifrac distance was selected for calculation. The significance of β-diversity difference was determined using ANOSIM tests. The beta-dispersion was represented by the average inner group distance in the weighted unifrac dissimilarity matrix. Statistically significant differences in taxa abundance were determined using Welch’s t-test in the STAMP program. The above statistical tests were all performed using the R vegan package, and the R ggplot2 package was used for plotting (Dixon 2003; Ginestet 2011).