Twenty-seven type strains of Marinobacter and 5 strains isolated from deep sea hydrothermal vents in our lab were selected for analysis (Table 1), which all are deposited in the MCCC (https://mccc.org.cn/).

Growth in the autotrophic sulfur-oxidizing medium was observed, which was modified with sodium thiosulfate as the sole electron donor and sodium bicarbonate as the sole carbon source according to He et al. (He et al., 2023). The artificial seawater in the autotrophic thiosulfate oxidization medium included 30.00 g NaCl, 0.25 g NH₄Cl, 0.33 g KCl, 0.14 g CaCl₂·2H₂O, 4.18 g MgCl₂·6H₂O, 0.14 g K₂HPO₄, 0.50 mg NiCl₂·6H₂O, 0.50 mg Na₂SeO₃·5H₂O and 1000 mL distilled water. After autoclaving, the medium was supplemented by sterilized 10 mL trace mineral solution (https://www.atcc.org/products/md-tms), 1 mL vitamin solution (https://www.atcc.org/products/md-vs), 5 mmol/L NaHCO_3, and 10 mmol/L Na₂S₂O₃. Before inoculation, the cells grown in marine 2216 medium (BD Difco, San Diego, CA, USA) were washed three times with sterilized seawater as inoculum to avoid interference from organic carbon sources and then transferred to the autotrophic media five times at a ratio of 1:20 with 0.5 mL of the inoculum to 10 mL autotrophic thiosulfate oxidization medium. The strains were inoculated into the aerobic medium and cultured at 28℃ Biochemical Incubator (ZXSD-1270) for 15 d and the cells number, thiosulfate, and sulfate concentrations were measured (refer to Section 2.4). In the autotrophic thiosulfate oxidization solid medium, 1.5% agar and 0.5 ppm phenol red were added (Ruby et al., 1981). pH change was used as an indicator of thiosulfate oxidation, the values of pH were detected by adding phenol red in the agar plates. During autotrophic sulfur-oxidizing incubation, 10 μL of the culture was sampled on days 0, 0.5, 1, 1.5, 2, 3, 4, 5 and 7 for cell counting by under microscopy (Nikon 80i, Japan). Each sample was counted in five grids (containing 80 subgrids) of a hemocytometer under microscopy and subsequently, the results were extrapolated to calculate the cell concentration (Randolph, 1944). All the counts and tests below were done in duplicates.

The heterotrophic thiosulfate oxidization medium was detected and investigated in the following medium (0.1% yeast extract, 0.5% peptone, 10 mmol/L Na₂S₂O₃), prepared with artificial seawater. Before inoculation, the cells were washed three times with sterilized seawater and then transferred to the heterotrophic media at a ratio of 1:20. The strains were inoculated into the medium and cultured at 28℃ for 15 d and the cells’ number, pH, and sulfate concentrations were measured. In heterotrophic growth, 200 μL of the culture was determined the absorbance at OD600 nm.

Thiosulfate and sulfate concentrations in the supernatant from autotrophic thiosulfate oxidization tests were measured by an ICS-2100 ion chromatography (Dionex, USA). The amount of sulfate (${{\rm {SO}}_4^{2-}} $) from heterotrophic thiosulfate oxidization tests was also determined spectrophotometrically by barium chloride colorimetric assay. Sulfate was measured by adding 1:1 barium chloride solution (10% w/v) with bacterial culture supernatant followed by mixing the suspensions vigorously. The white turbidity due to barium sulfate formation was measured at 450 nm with a Varioskan LUX (Thermo Scientific, Waltham, MA, USA). The amount of turbidity formed is proportional to the sulfate concentration standard sulfate solutions were made by dissolving Na₂SO₄ in deionized water to known concentrations in the range 0 mmol/L to 3 mmol/L (Behera et al., 2014).

In this study, 53 genomes of Marinobacter genus were downloaded from National Center for Biotechnology Information database, NCBI (https://www.ncbi.nlm.nih.gov/). Up-to-date bacterial core gene (UBCG) was used to construct a phylogenomic tree and infer the phylogenomic relationship of Marinobacter (Na et al., 2018). The phylogenomic tree was visualized, modified, and annotated by One Table (tvBOT) (Xie et al., 2023). Genome annotation was performed using the software BLASTP 2.13.0+, the e value was set to 10⁻⁵ (Camacho et al., 2009), and using the Rapid Annotation Subsystems Technology (RAST) server (Aziz et al., 2008). The e value indicates the probability that other sequences are more similar to the target sequence than this displayed sequence in a random situation. In general, if e < 10⁻⁵⁰, the database match should be the result of a homology relationship with a very high confidence level; and an e value less than 10⁻⁵ is the result of a more sexually acceptable S-value.

The proteins for sulfur oxidation were searched in the 53 genomes of Marinobacter by locally blastP with reference sequences, including soxB form Allochromatium vinosum DSM 180 (accessions: Q1W3E6) (Hensen et al., 2006), Tsd from Allochromatium vinosum DSM 180 (accessions: ADC61061) (Denkmann et al., 2012), Sqr from Aquifex aeolicus VF5 (accessions: O67931) (Griesbeck et al., 2002), AprA from Megalodesulfovibrio gigas DSM 1382 (accessions: T2G6Z9) (Fritz et al., 2002), DsrA from Archaeoglobus fulgidus DSM 4304 (accessions: Q59109) (Pott and Dahl, 1998), and Fcc from Allochromatium vinosum DSM 180 (accessions: Q06530) (Chen et al., 1994).

Sox gene clusters including key gene soxB is essential for thiosulfate oxidation. The phylogenetic relationship of the soxB protein of Marinobacter with other 90 sulfur oxidizer soxB proteins of Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, Chlorobi, and Campylobacteria were aligned with MUSCLE (Edgar, 2004) in the phylogenetic software MEGAX (Kumar et al., 2018), and constructed the maximum-likelihood tree with 1000 bootstraps using the “LG + G + I” model (Felsenstein, 1981). Then the Linux version of IQ-tree 2.0 (Trifinopoulos et al., 2016) was used to construct a phylogenetic tree, using the One Table (tvBOT) (Xie et al., 2023) to visualize the phylogenetic tree. The complete Sox multi-enzyme complex encoding genes including SoxRSVWXYZABCDEFGH were mapped for sox gene clusters by online tool chiplot (https://www.chiplot.online/tvbot.html).

In the study, 27 type strains and 5 deep sea hydrothermal vents strains were chosen for physiologic characterization (Table 1). Among them, only 5 type strains and 1 deep sea hydrothermal vent isolate showed autotrophic sulfur oxidizing ability with thiosulfate as electron donor, including type strains M. guineae M3B^T, M. aromaticivorans D15-8P^T，M. vulgaris F01^T, M. profundi PWS21^T, M. denitrificans JB02H27^T and deep sea hydrothermal vents strain Marinobacter sp. ST-1M (99.93% similarity to Marinobacter salsuginis SD-14B^T) (Table 1, Fig. S1). Five type strains had growth in the autotrophic thiosulfate oxidization liquid medium from initial 1 × 10⁵ cells/mL to 8.88 × 10⁷−9.72 × 10⁷ cells/mL in 5 d (Fig. 1a). These strains grew fast in the first 24 h, and reached a peak of cell density 10⁷ cells/mL. Among the five positive type strains, M. profundi PWS21^T had the fastest growth rate and the highest number of cells in the stable phase.

Ion chromatography was used to detect the concentration changes of thiosulfate and sulfate during autotrophic sulfur oxidation growth. These 5 type strains produced sulfate ranging from 0.25 mmol/L to 1.44 mmol/L (non-bacterial abiotic control was 0.17 mmol/L) with the consumption of thiosulfate from 3.60 mmol/L to 9.10 mmol/L (the beginning was 12.97 mmol/L) in 5 d, but without ${{\rm {SO}}_3^{2-}} $ detected, under autotrophic growth conditions (Fig. 1b). Marinobacter guineae M3B^T had the highest ability with the production of sulfate to 138.23 mg/L.

On the solid medium of autotrophic thiosulfate oxidization containing phenol red, five Marinobacter strains turned the color of the media from purple to yellow (Fig. S2), indicating the pH was reduced by sulfur oxidation. They were M. guineae M3B^T，M. aromaticivorans D15-8P^T, M. vulgaris F01^T, M. denitrificans JB02H27^T and Marinobacter sp. ST-1M. Marinobacter profundi PWS21^T couldn’t grow on the autotrophic thiosulfate oxidization agar medium.

BaCl₂ colorimetric assay was used to test sulfate production in heterotrophic thiosulfate oxidation medium of 27 type strains and 5 strains from deep sea hydrothermal vents (Table 1, Fig. 2). The t-test results revealed that 6 Marinobacter strains showed significant sulfate producing ability (p < 0.05), especially M. antarcticus ZS2-30^T and M. vulgaris F01^T with highly significant differences (p = 0) of 3.10 mmol/L and 9.45 mmol/L sulfate production. In addition, the other four were M. algicola DG893^T, M. daepoensis SW-156^T, Marinobacter sp. ST-1M and Marinoabcter sp. ST-43, which produced sulfate by 2.20 mmol/L, 1.72 mmol/L, 1.72 mmol/L and 1.73 mmol/L, heterotrophic thiosulfate oxidation growth. The pH obviously decreased from 7.0 to below 6.0 in the late phase growth culture of M. guineae M3B^T and M. vulgaris F01^T due to acid production (Fig. 2).

As representative strains, M. guineae M3B^T and M. vulgaris F01^T were chosen for further mixotrophic ability identification. Marinobacter guineae M3B^T was isolated from marine sediment collected from Antarctica (Montes et al., 2008), and M. vulgaris F01^T was isolated from the solar saltern of Weihai, China (Zhang et al., 2020).

Under autotrophic growth conditions with ${{\rm {HCO}_3^-} } $/CO₂ as the sole carbon source, M. guineae M3B^T oxidized thiosulfate from 12.97 mmol/L to 8.62 mmol/L with the sulfate production up to 1.44 mmol/L in 7 d (Fig. 3a); and under heterotrophic growth conditions, it also oxidized thiosulfate by reducing the concentration from 11.53 mmol/L to 4.71 mmol/L with the production of sulfate to 0.47 mmol/L in 3 d (Fig. 3b).

Under autotrophic growth conditions, M. vulgaris F01^T could oxidize thiosulfate from 12.96 mmol/L to 7.10 mmol/L, with the production of sulfate to 0.81 mmol/L in 7 d; and under heterotrophic growth conditions, it oxidized thiosulfate from 11.53 mmol/L to 4.55 mmol/L with the production of sulfate to 1.78 mmol/L. In the late growth phase of M. vulgaris F01^T culture, elemental sulfur, as well as sulfur-containing complexes were observed. Morphological observations under scanning electron microscope (SEM) showed that elemental sulfur was mainly shaped as spherical (Fig. 4a) and the sulfur-containing complexes were irregular or elliptical in morphology (Fig. 4b), and energy dispersive spectrometer (EDS) analyses revealed that the atomic percentage of sulfur in the particles ranged from −2.76% to22.84%.

Thiosulfate can be oxidized to sulfate by the Sox system (SoxABCDXY) in the periplasm. Four protein components, soxYZ, soxXA, soxB, and soxCD are required for the complete oxidation of thiosulfate to sulfate (Wasmund and Mussmann, 2017). Six Marinobacter strains encode a complete core set of Sox genes (soxC-soxD-soxY-soxZ-soxA-soxB) (Fig. 5), including M. antarcticus ZS2-30^T, M. guineae M3B^T, M. orientalis W62^T, M. pelagius HS225^T, M. salinus Hb8^T, and M. vulgaris F01^T, which are similar with Amphritea japonica and Neptunomonas marina. Fourteen Marinobacter genomes contain SoxB (six strains with an e value of 0), which may involve in converting thiosulfate to sulfate.

The phylogenetic tree based on 90 SoxB protein sequences (Fig. 6) can be divided into seven clusters. From bottom to top in the tree : Cluster Ⅰ, Alphaproteobacteria containing genera Rhodovulum and Roseinatronobatcer; Cluster Ⅱ, the genera Acidihalobacter and Thiohalorhabdus in Gammaproteobacteria; Cluster Ⅲ, Gammaproteobacteria, mainly containing the genera Thioalkalivibrio, Ectothiorhodospira, Leucothrix and Thiothrix; Cluster Ⅳ, mainly Thiobacillus, Sulfuriferula, Thiomonas, Pandoraea, Advenella in Betaproteobacteria, Azospirillum in Alphaproteobacteria, and Acidiferrobacter, Sulfuricaulis in Gammaproteobacteria; Cluster Ⅴ, Gammaproteobacteria; Cluster Ⅵ, Marinobacter in Gammaproteobacteria and Chlorobaculum belonged to Chlorobi, and Cluster Ⅶ, Sulfurimonas, Sulfurovum, and Nitratiruptor in Campylobacteria. Marinobacter formed a separate lineage within Gammaproteobacteria and close to Thiomicrorhabdus arctica and Aequoribacter fuscus. Marinobacter spp. might obtain the soxB gene through lateral gene transfer from sulfur oxidizing Gammaproteobacteria in the environment.

Other sulfur oxidizing related proteins including Tsd, Sqr, AprA, DsrA, and Fcc were also searched in all 53 Marinobacter genomes with reference proteins in this study. Thirty strains contain Tsd (3 strains with e value less than 10⁻⁵⁰, M. aromaticivorans D15-8P^T, M. changyiensis CLL7-20^T, M. salicampi ISL-40^T), which catalyzes thiosulfate to tetrathionate; 30 strains encode Sqr (2 strains with e value less than 10⁻⁵⁰, M. oulmenensis Set74^T, M. persicus M9B^T) and 29 strains encode Fcc (no strain with e value less than 10⁻⁵⁰), which oxidize sulfide (HS^–/S^2–) to ZVS; 6 strains were annotated AprA (no strain with e value less than 10⁻⁵⁰) for sulfite oxidation; no Marinobacter strain contain DsrA, which oxidizes sulfide to sulfite (Fig. 7). It indicated that sulfur oxidizing potential exists in Marinobacter spp. According to their physiologic tests, the results showed that sulfur oxidizing Marinobacter strains probably oxidize thiosulfate by Sox pathway.

Kyoto Encyclopedia of Genes and Genomes (KEGG) annotation analysis revealed that the presence of the two essential enzymes of the Calvin cycle (CBB cycle), glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and phosphoribulokinase (PRK) in autotrophic thiosulphate-oxidizing Marinobacter strains, including 5 type Marinobacter, M. guineae M3B^T, M. aromaticivorans D15-8P^T, M. vulgaris F01^T, M. profundi PWS21^T, and M. denitrificans JB02H27^T, suggesting their autotrophic potential.