A Whiff of Taxonomy – Atribacter laminatus

by Christoph

The image is way too large to show it here, but when you join me and Hug et al. (2016) for "A new view of the tree of life" here in STC (or here in Nature, or in Wikipedia), you can spot a bunch of names of bacterial phyla in the top left corner that sound unfamiliar. One of those, Atribacteria, is at the top of the column with Aquificae, Calescamantes, ... , Fusobacteria. You can also see the phyla Dictyoglomi (now Dictyoglomota) and Thermotogae (now Thermotogota, the "-ota" suffix indicating the taxonomic phylum level) in this column, and Elio had a closer look at the enigmatic "coated" Thermotoga maritima years ago here in STC. "Enigmatic" not so much because of their posh "coat," but it appears as a genomic hybrid in which about a quarter of its genes originate from Archaea. Now, without further ado, enter Atribacter laminatus, the first cultivated representative of the phylum Atribacteria and the main subject of this episode of our sporadic A Whiff of Taxonomy series.

Figure 1. Scanning electron microscopy of A. laminatus RT761 cells. Scale bar, 1.5 µm. Frontispiece: Screen capture from Suppl. Movie 4 of a 3D-rendered reconstruction from cryo-ET of a A. laminatus RT761 cell. Red: O‑LML; blue: M‑LML; yellow: I‑LML; green: ribosome‑sized particles. Source

Atribacteria (now Atribacterota) were previously grouped as a 'Candidatus phylum' as there was no cultivated representative until recently. They eked out a nameless existence in silico, labeled only with the "license plates" OP9 and JS1, in the (almost) unmanageably vast collections of metagenomic data of microbial dark matter. OP9 and JS1 sequences are globally distributed, and in some cases abundant, in samples from anaerobic marine sediments, deep geothermal environments, sediments associated with methane hydrates and hydrocarbon seeps, hypersaline microbial mats, landfill leachates, and anaerobic digesters/reactors and petroleum reservoirs.

Dodsworth et al. (2013) figured that the provenance of the OP9 bacteria from "dark matter" should be reflected in the name and chose the Latin adjective āter (black, dark-colored, gloomy/murky) as prefix. They came up with Atribacteria (pronounced Ā.tri.bac.te'ri.a) and Nobu et al. (2016) concluded later that: "...the most parsimonious analysis of the available data would suggest that the 'Atribacteria', inclusive of OP9 and JS1, is a single candidate phylum within the Bacteria." To come full circle, and perhaps so as not to disappoint the name patrons in hindsight, the cultivated Atribacter laminatus presented itself to Katayama, Nobu et al. (2020) as colonies that are "...light brown circular and convex disks on the deep agar slant."

Figure 2. Morphology and membrane structure in A. laminatus RT761 cells showing the presence of three lipid membrane-like layers (LMLs) with the innermost layer enclosing the nucleoid. A Screen capture from Suppl. Movie 1. B Screen capture from Suppl. Movie 2. A zoom of the white‑boxed area is shown in the bottom right-hand corner. Outer (O‑LML), middle (M‑LML), and inner (I‑LML) lipid membrane-like layers are marked. White circle in B: invagination in the I‑LML. Scale bar ~100 nm. Modified from Source

Under the scanning electron microscope (SEM), cells from these light-brown colonies "are a rod or ovoid shape tapered with pointed ends with 0.6–0.8 μm wide and 1.3–1.8 μm long," as the authors say (Figure 1). The cells were happily growing when taken for SEM as you can observe elongated cells, some with incipient constrictions in the middle and one in the center of the image that is in the process of binary fission. "...tapered with pointed ends" is reminiscent of Borrelia, but these cells are considerably slimmer (see here in STC). They clearly do not have the typical dome-shaped poles of rod-shaped bacterial cells such as E. coli or B. subtilis. To me, the cells of A. laminatus have the shape of a lemon (since lemons come in different shapes, here is a picture of what I imagine), and this is even more evident in the cryo-electron micrographs (Figure 2).

So far, so normal. However, the researchers were in for a big surprise when they looked at A. laminatus by cryo-electron microscopy (Figure 2). The Gram‑negative cells have a third membrane. They carefully avoid their assignement as Outer/Inner membrane (OM, IM) and talk of outer (O‑LML), middle (M‑LML), and inner (I‑LML) lipid membrane-like layers instead. The I‑LML completely encloses the nucleoid but not as tightly as apparently an intra‑cytoplasmic membrane in Thiomargarita magnifica (see here in STC).

In addition to the nucleoid, the I‑LML encloses approximately half of the cytoplasmic volume. The other half is confined between the M‑LML and I‑LML. The 3D reconstruction of a cell cryo-tomogram (see frontispiece) indicates that ribosome‑sized particles (green dots) are present in comparable numbers in both the "inner" and the "outer" cytoplasm (the innermost region of the cell is almost devoid of ribosomes due to the volume occupied by the nucleoid).

At present, one can only speculate why the "outer" cytoplasm is notably less electron-dense than the "inner" (see Figure 2), but there is no indication that the I‑LML could be a eukaryote-like nuclear envelope. In the past, this had led to intense debates about the intra-cytoplasmic membrane system in Gemmata obscuriglobus (Planctomycetota) (see here in STC). Invaginations of the I‑LML can be occasionally observed (white circle in Figure 2B), but they are less frequent and less extended than the invaginations of the IM in Gemmata.

Figure 3. Morphology and membrane structure in RT761 cells showing the presence of three lipid membrane-like layers (LMLs) with the innermost layer enclosing the nucleoid. Original slice picture is shown in Suppl. Fig. 4. Black arrowheads indicate the outer (1), middle (2), and inner (3) LMLs. White arrowheads indicate the 2.2 nm thick layer (1) and faint layers (2). Scale bar, 200 nm. Source

After examining numerous longitudinal cell sections, the authors concluded that the O‑LML and the M‑LML are continuously separated by a ~30 nm wide space containing an almost completely continuous electron-dense layer of ~2 nm thickness that could well be peptidogylcan (Figure 3). While the O‑LML and the M‑LML appear to be tightly connected, this is certainly not the case with the L‑LML, which does not follow the curvature of the M‑LML with regard to shape and distance. They observe discontinuous, stacked "thin layers" of unknown nature below the M‑LML in the "outer" cytoplasm, but they do not mention that these are embedded in a ~100 nm thick layer of less electron-dense material that lies below the entire M‑LML (most clearly seen here).

A. laminatus reproduces by binary fission (see Figure 1), but its compartmentalization by three membranes with two cytoplasms raises the intriguing question of how and where the cells organize their divisome. A few initial clues – and more open questions – can be obtained from Supplementary Movie 2, which shows a cell at the very end of division, shortly before the daughter cells completely separate (one frame is shown in Figure 2B). The "inner" cytoplasm (between I‑LML and M‑LML), which contains the two already fully replicated and separated daughter chromosomes in the to‑be daughter cells, remains connected by a thin plasma bridge until the very last step. This plasma bridge appears as so thin, less than membrane-width, that it is conceivable that the membranes can spontaneously flip over. But what separates the "outer" cytoplasm? The A. laminatus genome encodes a homolog of the ring‑forming division protein FtsZ (see here⛓️‍💥 in STC) with an unusual N‑terminal extension, likely a signal sequence (for export to the "outer" cytoplasm?). The A. laminatus genome also encodes a homolog of FtsA without a detectable signal sequence. Usually, FtsA tethers growing FtsZ filaments to the inner membrane (IM), so that the formation of FtsZ rings is only possible at midcell where the nucleoid is thin, since the bulk nucleoid excludes ring formation and functional analogs of the MinCDE system prevent ring formation at the cell poles. Do you see where the open questions are?

Figure 4. Cell structures of select species are shown for "Ca. Atribacteria", Thermotogae, and Dictyoglomi. The illustrations indicate the outer membrane (black), cytoplasmic membrane (blue), intracytoplasmic membrane (purple), and nucleoid (yellow). *For RT761, the shown schematic requires further investigation to conclude the identity/role of each layer. Modified from Source

Katayama, Nobu et al. (2020) point out that the genome of A. laminatus includes a significantly higher proportion of genes coding for proteins with transmembrane helices and atypical signal sequences (for export) than genomes of typical Gram-negative "diderm" bacteria (see Elio's take on monoderms/diderms here in STC). This is also reported for other bacteria with anatomies that deviate from the usual blueprint (Figure 4). While in A. laminatus the cytoplasm is divided into two compartments while the periplasm is as thin as usual, an oversized toga envelops the cytoplasm of Thermotoga maritima, which results in a significantly expanded periplasm. In Dictyoglomus thermophilum, several cells even share a periplasm and the outer membrane. Doesn't it look as if the classical distinction between monoderms and diderms is not really suited to adequately describe the whole zoo of variations on the theme observed in nature?

Addendum It's a tidbit only for die-hard microbiologists, though, I cannot resist quoting here a whole paragraph from the Materials & Methods section of the Katayama, Nobu et al. (2020) paper. Pay particular attention to the time taken to cultivate by enrichment and finally isolate A. laminatus RT761. Three years! And note that DNA sequencing was used here as a handy analytical tool to verify the purity of the culture(s). They obtained the complete genome sequence as an encore, so to speak. The time scale turned on its head: when the genome of E. coli was first sequenced and published in 1997, the preparation of chromosomal DNA libraries took a few weeks, but sequencing and genome assembly took, well, years. But then, that is now pre-history.

Sediment samples were mixed at a 1:2 volume ratio with a formation water (= water from the sample site) to make slurry samples in an anaerobic chamber. The slurry samples were dispensed as 20 ml-aliquots into 70-ml serum vials and were then sealed using butyl rubber stoppers and aluminum crimps in an anaerobic chamber. The slurries were incubated without the addition of any nutrients under an atmosphere of N2/CO2 (80:20) at a temperature higher than the water temperatures of original environments (45°C rather than 25°C). After 90 days, 2 ml of the methane-producing culture of slurry sample was inoculated into a saline mineral medium supplemented with 1 g·l⁻¹ glucose, 1 g·l⁻¹ Bacto peptone (BD), 0.1 g·l⁻¹ yeast extracts (BD), 5 mM coenzyme M, and 0.1 mM titanium (III) citrate as reducing agent. The saline mineral medium contained 350 mM NaCl, 30 mM NaHCO3, 15mM MgCl2⋅6H2O, 10 mM NH4Cl, 1 mM KH2PO4, 1 mM CaCl2⋅2H2O, 1 ml·l⁻¹ trace elements solution, 1 ml·l⁻¹ vitamin solution, and 1 ml·l⁻¹ resazurin solution. Cultivation was performed in 75-ml serum vials containing 20 ml of medium under an atmosphere of N2/CO2 (80:20). Enrichment cultures were grown, followed by successive transfer six times at intervals of approximately 80 days. Individual cells were isolated in a pure culture using the deep agar slant method combined with dilution-to-extinction method (10-fold dilutions from 10^–1 to 10^–8) with a saline mineral medium supplemented with 1 g·l⁻¹ glucose, 0.1 g·l⁻¹ yeast extracts, 0.1 mM titanium (III) citrate, and 8 g·l⁻¹ agar. After 40 days of incubation, a single colony was picked from 10⁶-dilution culture and transferred to fresh liquid medium. This procedure was repeated three times. Purity of the culture was verified by microscopy and further confirmed by no contaminant sequences in DNA sequencing data of genomic DNA. The pure culture of strain RT761 was incubated at 45°C in saline mineral medium amended with 16 mM glucose, 0.2 g·l⁻¹ yeast extracts and 0.5 g·l⁻¹ cysteine hydrochloride.