The MTB-bearing sediment samples were collected from the estuary of the Shihe River in Qinhuangdao city, eastern China. The sampling site was located inside the estuary of Shihe River into the Bohai Sea (39°57′55.6″N, 119°47′9.4″E). The sediment comes from a brackish lacustrine environment with a salinity of ~23.9 ppt, a pH of 7.5 and a temperature of 19°C measured in July 2015. Surface sediments were collected near the shore at water depths of ~1–2 meters and transferred to 1-l plastic bottles with a sediments: water ratio of ~2:1. The bottles were shipped back to the laboratory and stored at room temperature (~20°C) in dim light to set up microcosms. Magnetotacic bacteria in the microcosms were routinely checked with the hanging-drop technique (Schüler, 2002) using an Olympus microscope BX51 equipped with phase-contrast, fluorescence and a DP70 digital camera system (Olympus Corp., Tokyo, Japan). Although they were collected from the same sample site, MTB populations differed among individual microcosms after several weeks of incubation in the laboratory, and even varied within the same microcosm with time (Li et al., 2017). Here, we focused on one microcosm dominated by one group of magnetococci (i.e., SHHC-1) that swam toward the south pole of a bar magnet (north-seeking MTB) (Figure 1A).

Approximately 200 ml of slurry from this microcosm was used to magnetically concentrate the SHHC-1 cells by using a modified magnetic separation apparatus previously described in Jogler et al. (2009). The collected MTB cells were washed three times with Milli-Q water and then re-suspended into ~100 μl of Milli-Q water for additional experiments: ~5 μl of MTB cells were used for TEM analyses, ~20 μl were boiled for 10 min and stored at −20°C for PCR amplification of 16S rRNA genes, and the remaining ~75 μl were prepared for coupled FISH-SEM analyses as previously described in Li et al. (2017). TEM samples were stored in a pure N₂ atmosphere prior to TEM observations.

SixteenS rRNA genes of the MTB cells were amplified using the universal bacterial primers 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-GGTTACCTTGTTACGACTT-3′) (Lane, 1991), and sequenced using protocol described in Li et al. (2017). Briefly, each 50 μl PCR mixture contained 1 μl of template, 25 μl of DreamTaq PCR Master Mix (MBI Fermentas), 2 μl of each primer (10 μM), and 20 μl of Milli-Q water. The PCR conditions were 95°C for 3 min, 30 cycles at 95°C for 1 min, 55°C for 1.5 min, 72°C for 1.5 min, and a final 10 min extension at 72°C. To avoid potential amplification biases, triplicate PCR products from each sample were pooled and purified using 0.8% (w/v) agarose gel electrophoresis (described below). All PCR controls without added template were negative. PCR products were purified using an E.Z.N.A.® Gel Extraction Kit (Omega Bio-tek, Inc. USA). They were ligated with the pMD19-T vector (TaKaRa, Japan), then cloned in Escherichia coli (strain DH5α) competent cells (Tiangen, Beijing, China) according to the manufacturer instructions. Thirty clones were randomly picked and sequenced using the vector primers M13-47 (5′-CGCCAGGGTTTTCCCAGTCACGAC-3′) and RV-M (5′-GAGCGGATAACAATTTCACACAGG-3′) at the Huada Genome Center (Beijing, China). After discarding sequences of insufficient length (<1,000 bp), the remaining sequences were then aligned with their close relatives using the ClustalW algorithm for manual correction, and a phylogenetic tree was subsequently constructed using the neighbor-joining method (Saitou and Nei, 1987) in the MEGA software package (v. 7.0) (Kumar et al., 2016). Bootstrap values were calculated with 1,000 replicates.

To phylogenetically and morphologically identify the SHHC-1 cells, the probe SHHC1228 (5′-CTCCAGGTCACCCATTCGCCGCTCT-3′, positions 1,228–1,252) was designed to target specifically all 16S rRNA gene sequences identified in this study. Probe specificity was evaluated by using the online probe evaluation tools probeCheck and probeBase (Loy et al., 2008; Greuter et al., 2016). No other sequence in the SILVA111 database has a matching complementary sequence (the minimum number of mismatches is 2). The universal bacterial probe EUB338 (5′-GCTGCCTCCCGTAGGAGT-3′) was used as a positive control probe of bacteria for FISH (Amann et al., 1990). Probe EUB338 was synthesized and fluorescently labeled with fluorescein phosphoramidite FAM at the 5′ end, while the SHHC1228 probe was synthesized and fluorescently labeled with the hydrophilic sulfoindocyanine dye Cy3 at the 5′ end. E. coli cells were added in approximately same amount to the SHHC-1 cells and were used as inner control cells of non-magnetotactic bacteria.

Coordinated FISH-SEM analysis was carried out using the procedure described in Li et al. (2017). Briefly, ~1 μl of cell mixtures (E. coli + SHHC-1) were placed directly on a high-precision cover glass (Paul Marienfeld GmbH & Co. KG, Germany) and dried in air at ambient temperature, followed by gradual dehydration in 50, 80, and 100% ethanol baths (3 min in each bath). In situ hybridization was performed at 46°C for 3 h in 9 μl of hybridization buffer (0.9 M NaCl, 20 mM Tris-HCl [pH 7.5], 0.02% [wt/vol] sodium dodecyl sulfate [SDS], 35% [vol/vol] formamide). One microliter of the EUB338 probe (50 ng/μl) and 1 μl of the SHHC1228 probe of interest (50 ng/μl) was then added to the mixture. After 3 h of hybridization the samples were incubated in a washing buffer (0.08 M NaCl, 20 mM Tris-HCl [pH 7.5], 5 mM EDTA [pH 8.0], 0.01% [wt/vol] SDS) at 48°C for 30 min, then washed in Milli-Q water three times, dried in air at ambient temperature and finally observed using an Olympus epifluorescent microscope BX51. After fluorescence microscopy observations, the cells mounted on cover glasses were coated with carbon using a Leica ACE200 Low Vacuum Sputter Coater (Leica Microsystems, Wetzlar, Germany), and observed using a Zeiss Ultra-55 field-emission gun scanning electron microscope (Carl Zeiss, Germany) operating at 5 kV.

Conventional TEM observations were performed on a JEM2100 microscope (JEOL Ltd., Tokyo, Japan) operating at 200 kV at the IGG-CAS (Beijing, China). High resolution TEM (HRTEM) observations were carried out on a JEM-2100F microscope (JEOL Ltd., Tokyo, Japan) operating at 200 kV at the IMPMC (Paris, France). This microscope was equipped with a field emission gun, an ultra-high resolution (UHR) pole piece, a JEOL detector with an ultrathin window, and a scanning TEM (STEM) device. STEM Z-contrast images were acquired in high angle annular dark field (HAADF) mode. Chemical composition analyses were performed by energy dispersive X-ray spectrometry (EDXS) elemental mapping in HAADF-STEM mode. HAADF-STEM tomography was performed on a FEI Tecnai F20 microscope at 200 kV at the ICMPE (Thiais, France). Acquisition was made using the Digital MicrographTM (Gatan) STEM tomography module, and 3-D reconstruction was performed using the DigiECT software (available at http://www.digisens3d.com/fr/logiciel-tomographie/soft/23D_Electron_Tomography_Software.html).

The crystal length (along the long axis) and width (perpendicular to the long axis) of the magnetosomes were measured from the TEM images. The shape factor was defined as width/length. Crystal habits of magnetosomes were determined by a combination of Fourier analyses of HRTEM images of individual particles and crystallographic investigations, as previously described in Faivre et al. (2008) and Li et al. (2015). For each particle, the zone axis and in plane crystallographic directions were determined from the 2-D Fast Fourier transform (FFT) of the HRTEM image, enabling the the stereographic projection to be determined. Idealized shapes modeled using the KrystalShaper software package (available at http://www.jcrystal.com/) were then compared to the observed ones.

TEM observations showed that the magnetically concentrated cells contained predominantly one group of magnetococci, called SHHC-1, with two bundles of magnetosome chains, and a few (< 5%) other magnetotactic cocci forming single magnetosome chain or non-MTB cells (Figure 1B). The SHHC-1 cells were coccoid in shape with an average diameter of 2.1 ± 0.4 μm (n = 150). They contained two bundles of flagella, and with an average magnetosome number of 38 ± 17 per cell (n = 150) (Figure 1C). HAADF-STEM tomography observations clearly evidenced that each bundle was composed of two parallel closely assembled magnetosome chains (Figures 1D–F and Supplementary movie). Smaller magnetosomes were often observed at the ends of the chains. Within the same SHHC-1 cell, the two bundles of magnetosome chains were commonly distributed on opposite sides of the cell body in close proximity to the cell membrane. Such an arrangement of the magnetosomes in multiple chains supports the previous assumption that repulsion forces between parallel magnetic dipoles may drive the chains or bundle of chains apart from one another. Thus, the magnetic torque acting on them by the external applied or the geomagnetic field can be effectively transferred to the entire cell body (Hanzlik et al., 1996). In the case of SHHC-1, each bundle of chains may behave as one enlarged SD particle because all the individual magnetosomes may have their magnetic moments uniformly aligned along the chain axis, as previously shown by electron holography observations performed on another uncultured, phylogenetically unknown, magnetococcus strain having two bundles of magnetosome chains (Simpson et al., 2005).

STEM-EDXS mapping (in HAADF mode) clearly shows that the SHHC-1 magnetosomes are rich in Fe and O, consistently with magnetite. Sulfur-rich globules with various sizes, ranging from several to hundreds of nanometers were also often observed within the SHHC-1 cells (Figure 2). These features were consistent with previous observations of many uncultured magnetococci that can form sulfur globules as well as magnetite-type magnetosomes, even when sulfide is absent in the sediment sample (Moench, 1988; Cox et al., 2002; Araujo et al., 2016). This confirmed that most of magnetococci cells may have an autotrophic or mixotrophic metabolism based on the oxidization of reduced sulfur compounds (Araujo et al., 2016).

To phylogenetically and structurally identify the strain SHHC-1, 16S rRNA genes of the magnetically enriched cells were amplified with a total of 30 clones being sequenced. Twenty-three 16S rRNA gene sequences with full-length had more than 99% of sequence identity. Coordinated FISH-SEM analysis using the SHHC1228 FISH probe designed to match the 23 identical 16S rRNA gene sequences was conducted on the magnetically enriched cells. As shown in Figures 3A–D, all bacterial cells, including the inner control E. coli cells, were positively labeled (green) by the 5′-FAM-labeled universal bacterial probe EUB338, while only the group of magnetococci with two bundle of magnetosome chains were positively labeled (red) by the 5′-Cy3-labeled SHHC-1-specific probe SHHC1228. This confirmed that the twenty-three 16S rRNA gene sequences corresponded to strain SHHC-1. Phylogenetic analyses based on the 16S rRNA gene shows that the strain of SHHC-1 affiliates with the Alphaproteobacteria class of the Proteobacteria phylum (Figure 3E). Moreover, it was phylogenetically close to (1) the uncultured magnetotactic coccus sp. clone WHI-8 (97% sequence identity), detected in intertidal sediments of the Xiaoshi Island in the North Yellow Sea (Weihai, China) (Chen et al., 2015), and (2) the uncultured magnetotactic coccus sp. CF24 (AJ863157) (95% sequence identity), detected in marine sediments in Germany (Flies et al., 2005). However, the bacterial and magnetosome morphologies of both WHI-8 and CF24 clones have not been described due to that both studies did not link their 16S rRNA gene sequences to the bacterial cells (Flies et al., 2005; Chen et al., 2015). Furthermore, to our knowledge, the morphological features and spatial arrangement of SHHC-1 magnetosomes have not been reported in any cultured and uncultured magnetococci. Therefore, SHHC-1 represents a novel magnetotactic coccus morphotype which is detected from the estuary of the Shihe River in Qinhuangdao City, eastern China.

Magnetite crystals of SHHC-1 had an average length of 83.3 ± 19.2 nm and an average width of 69.9 ± 17.3 nm with a shape factor of 0.84 ± 0.07 (n = 416) (Figures 4A–C). Both the crystal length and width distributions were negatively skewed and the width/length ratio was nearly constant. These are typical features for prismatic and cuboctahedral magnetite magnetosomes (Devouard et al., 1998; Isambert et al., 2007). This indicates that magnetosomes within SHHC-1 grew homothetically within magnetosome membranes, a mechanism similar to that identified in known magnetotactic Alphaproteobacteria (Li et al., 2009; Jandacka et al., 2013). However, unlike what is generally observed in other cultured and uncultured magnetotactic Alphaproteobacteria where prismatic and cuboctahedral magnetosomes have a relatively consistent morphology (Pósfai et al., 2013a), there is a large diversity of magnetosome shapes in SHHC-1 even within the same cell (Figures 4D,E). Most particles had rectangular (~53%, n = 286) and rhomboidal (~26%) projections, whilst others displayed triangular (~6%) or irregular (~15%) shapes (Figure 4E). HRTEM observations performed on dozens of individual magnetosomes coupled with morphological modeling demonstrates that most particles with larger sizes (i.e., mature magnetosomes) are consistent with elongated octahedrons composed of only {111} faces (Figures 5E–L). Variations in the elongation ratios occur among individual magnetosomes (Supplementary HRTEM images). A few particles generally with smaller sizes (i.e., immature magnetosomes) have a cuboctahedral-like shape based on {111}, {110}, and {100} faces (Figures 5A–D). In addition, crystal defects including twining, multiply twining, and stack dislocations were very often observed within SHHC-1 magnetosomes (~15% of defect frequency, n = 286) (Figures 4E, 6).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.