A Micrarchaeon Isolate Is Covered by a Proteinaceous S-Layer

ABSTRACT

In previous publications, it was hypothesized that Micrarchaeota cells are covered by two individual membrane systems. This study proves that at least the recently cultivated “Candidatus Micrarchaeum harzensis A_DKE” possesses an S-layer covering its cytoplasmic membrane. The potential S-layer protein was found to be among the proteins with the highest abundance in “Ca. Micrarchaeum harzensis A_DKE,” and in silico characterization of its primary structure indicated homologies to other known S-layer proteins. Homologues of this protein were found in other Micrarchaeota genomes, which raises the question of whether the ability to form an S-layer is a common trait within this phylum. The S-layer protein seems to be glycosylated, and the micrarchaeon expresses genes for N-glycosylation under cultivation conditions, despite not being able to synthesize carbohydrates. Electron micrographs of freeze-etched samples of a previously described coculture, containing “Ca. Micrarchaeum harzensis A_DKE” and a Thermoplasmatales member as its host organism, verified the hypothesis of an S-layer on the surface of “Ca. Micrarchaeum harzensis A_DKE.” Both organisms are clearly distinguishable by cell size, shape, and surface structure.

IMPORTANCE Our knowledge about the DPANN superphylum, which comprises several archaeal phyla with limited metabolic capacities, is mostly based on genomic data derived from cultivation-independent approaches. This study examined the surface structure of a recently cultivated member “Candidatus Micrarchaeum harzensis A_DKE,” an archaeal symbiont dependent on an interaction with a host organism for growth. The interaction requires direct cell contact between interaction partners, a mechanism which is also described for other DPANN archaea. Investigating the surface structure of “Ca. Micrarchaeum harzensis A_DKE” is an important step toward understanding the interaction between Micrarchaeota and their host organisms and living with limited metabolic capabilities, a trait shared by several DPANN archaea.

INTRODUCTION

Archaea exhibit a variety of cell surfaces. The most common cell surface type is a proteinaceous surface layer (S-layer), which is often glycosylated (1–3). S-layers consist of one or two proteins, which can self-assemble and form a two-dimensional (2D) paracrystalline layer, spanning the whole cell (4, 5). The S-layer is often the only cell wall component, but in some methanogenic species, it occurs in combination with an additional layer of pseudomurein or methanochondroitin. The proteins forming this layer are directly or indirectly anchored to the underlying cytoplasmic membrane (1). The lattice symmetry is either oblique (p1, p2), square (p4) or hexagonal (p3, p6), depending on the number of identical S-layer proteins, which form one morphological subunit (6, 7). The fine structure of the protein subunits, arranged on the lattice, results in regularly spaced pores, which can be observed by electron-microscopic (EM) techniques (8–10). After the first discovery of a glycosylated S-layer protein in Halobacterium salinarum (3, 11), more and more glycosylated S-layers were discovered in Archaea and Bacteria (12, 13). The comparison of several S-layer proteins in different Archaea revealed in some cases a poorly conserved amino acid sequence but often no sequence similarity at all; in contrast, the lattice type is often shared by prokaryotic families or genera (1).

S-layers are found—or at least postulated to be encoded in the genomes—in all four archaeal superphyla, including the DPANN superphylum. Besides the eponymous phyla (Diapherotrites, Parvarchaeota, Aenigmarchaeota, Nanoarchaeota, and Nanohaloarchaeota) (70), this superphylum comprises organisms of the Woesearchaeota and Pacearchaeota (14), Huberarchaeota (15), Micrarchaeota (16), Altiarchaeota (17), and Undinarchaeota (18), as well as several so-far-undefined phyla (19, 20). To date, the obligate symbiont Nanoarchaeum equitans is the only known DPANN member with an S-layer (21). While the cell surface type of other DPANN members is still unknown, Micrarchaeota and Altiarchaeota were postulated to have an inner and outer membrane (16, 17, 22). Micrarchaeota are acidophilic organisms, occurring in various habitats around the world (23). It is speculated that these organisms are dependent on other microorganisms, because of their reduced genomes and lack of several essential genes or even metabolic pathways (16). In natural habitats, they are often associated with members of the Thermoplasmatales (24). In fact, the reduced genome in combination with a small cell size is a common characteristic of members of the DPANN superphylum, which is postulated to contain several archaeal symbionts. Just a few DPANN members have been cultivated under laboratory conditions so far (21, 25–29). Recently, we described the cultivation of the Micrarchaeota member “Candidatus Micrarchaeum harzensis A_DKE” in coculture with “Candidatus Scheffleriplasma hospitalis B_DKE,” a member of the Thermoplasmatales, as a potential host (30). For better readability here, we refer to these two archaeal strains as Scheffleriplasma hospitalis and Micrarchaeum harzensis, although they still have “Candidatus” status.

The hypothesis that the cell surface of Micrarchaeota is composed of two membranes was based on cryo-electron microscopy and tomographic reconstructions of environmental samples containing Micrarchaeota (22). The latter were distinguished from other community members by cell size, shape, and presence of “two surrounding layers” of unknown composition. In our ongoing experiments of the above-mentioned coculture of S. hospitalis and M. harzensis, we aimed at characterizing the surface type of the micrarchaeon using a variety of approaches. We conducted transcriptomic and proteomic analyses in order to study a putative S-layer protein encoded in the genome of M. harzensis. Finally, we aimed at showing the presence of an S-layer, by analyzing the cell surface structure on freeze-etched samples or on frozen, fully hydrated cells of this culture.

RESULTS AND DISCUSSION

Transcriptomic and proteomic evidence for the production of a putative S-layer protein by M. harzensis.

The examination of the previously published genome of M. harzensis (30) revealed a gene (Micr_00292/slp_Mh) proposed to encode a protein with similarities to S-layer domain PF05123 (see Table S3 in the supplemental material). The genome of the possible host organism S. hospitalis does not contain such a gene. This is in line with previous results revealing that S-layers are rare among members of the Thermoplasmatales and occur only on cells of the family Picrophilaceae (31). The putative S-layer-encoding gene of M. harzensis was found to be the third most highly expressed gene within the transcriptome (Table S3; transcriptomic data have already been published [30]). A proteomic analysis of the above-mentioned coculture of M. harzensis and S. hospitalis corroborated this result, as the potential S-layer protein was the M. harzensis protein with the highest abundance within the proteome (Table S4).

In silico analysis of SLP_Mh reveals homology to S-layer proteins.

S-layer proteins share only a few traits among their primary structure, which renders their identification on a bioinformatic level difficult. S-layer proteins are described to have a molecular mass between 40 and 200 kDa, an isoelectric point between 4 and 6, and a content of 40% to 60% hydrophobic as well as only a few sulfur-containing amino acids (7, 32). With a predicted molecular mass and an isoelectric point of 101 kDa and 5.55, respectively, as well as being composed of about 50% hydrophobic and almost no sulfur-containing amino acids (Table S3), the S-layer protein of M. harzensis (SLP_Mh) matches these characteristics quite well. In addition, numerous putative N- and O-glycosylation sites located throughout the whole sequence (Fig. 1) indicate possible glycosylation of SLP_Mh, which is a common trait of many known archaeal and bacterial S-layer proteins (3, 33). Lastly, conserved S-layer domains at the N and C termini of the protein, as well as a single domain in between, with strong homology to S-layer proteins from several Archaea, such as Pyrococcus horikoshii, Methanococcus jannaschii (both TIGR01564), and Methanosarcina species (TIGR01567), could be detected. In contrast to SLP_Mh, this specific domain is tandemly duplicated in Methanosarcina species (34). Overall, the presented data are consistent with the hypothesis that SLP_Mh is indeed an S-layer protein.

FIG 1.

Putative domain architecture of SLP_Mh. Sequence motifs such as the N-terminal region containing a putative transmembrane domain or signal peptide are shown in yellow. Putative N- and O-glycosylation sites are highlighted by pink circles and purple diamonds, respectively. The peptide antigen of the SLP_{Mh 133–147} IgG is highlighted in teal. Putative domains with homology to S-layer proteins (N/C-SLP; SLP), copper binding proteins (CoB), and glycoside hydrolases (GH) are depicted as arrows and are colored according to their respective prediction probability: blue, >90%; green, ∼50%; orange, <30%. The respective database accession numbers of the putative domains are given as well. For more detailed information on the predicted domains, see Table S5.

Further conserved motifs in the SLP_Mh primary structure allow speculation about how exactly the protein could be attached to the M. harzensis cell membrane. Unfortunately, there are conflicting predictions regarding whether there is a signal peptide (Phobius and SignalP) or transmembrane domain (TMHMM) located at the N terminus of SLP_Mh, and the compiled proteomic data do not suffice to verify the cleavage of the putative signal peptide in the mature protein (data not shown). In case of a signal peptide, which is predicted to be cleaved between Ala26 and Gly27 upon secretion, resulting in the loss of the transmembrane domain, SLP_Mh anchoring might depend on a second SLP with a transmembrane domain acting as a stalk, as described for several Sulfolobales species (5). Other than transmembrane domains, archaeal S-layer proteins can be anchored by either a lipid modification or interaction with cell wall polymers (reviewed in reference 7). As a corresponding signal motif could not be detected, SLP_Mh does not seem to comprise a glycosylphosphatidylinositol anchor. Still, other sorts of lipid modifications to anchor the protein in the cell membrane are possible, i.e., as described for Haloferax volcanii (35). Further experiments with isolated SLP_Mh are necessary to answer this question, however. Since there is no indication of synthesis of archaeal cell wall polymers in the M. harzensis genome (30), anchoring of the S-layer via cell wall interaction as described for Methanothermus species (2) is not likely. At its very N and C termini, SLP_Mh shows weak homology to glycoside hydrolases (Fig. 1). These domains could potentially allow glycan binding and therefore might facilitate membrane anchoring in a lipopolysaccharide matrix or physical interaction with S. hospitalis cells. Since the former has so far been exclusively described in Gram-negative Bacteria (see examples in references 7 and 34), such as Caulobacter crescentus (36), and M. harzensis also lacks crucial enzymes for the synthesis of lipopolysaccharides (see below), the latter seems to be more likely. Due to homology to a copper binding protein, the N-terminal domain could also allow binding of divalent cations (i.e., Ca²⁺), which are known to stabilize S-layer structures in many bacterial and archaeal cells, i.e., in Haloferax volcanii (7, 37, 38).

S-layers might be a common characteristic of Micrarchaeota members.

A BLASTp database search for similarities to SLP_Mh revealed 52 homologues in 38 of 49 examined Micrarchaeota genomes. Whereas some of these homologues were already annotated as S-layer proteins, others are so far annotated as hypothetical or uncharacterized proteins. An alignment of these proteins revealed conserved regions throughout the whole sequence, but especially at the N-terminal end containing the S-layer domain. Figure 2 shows the first 150 positions of this alignment which cover the S-layer domain of SLP_Mh, revealing a high conservation at the amino acid level. Cross-referencing the BLAST hits with the PFAM protein domain database revealed that 38 of the proteins show homologies to N-terminal S-layer domains and one to a C-terminal S-layer domain (Table S2). This finding suggests that other Micrarchaeota can also form an S-layer, which would make the proteinaceous surface layer a common characteristic of the phylum. In fact, so far, we could not find a nearly complete genome of a Micrarchaeota member that does not seem to possess a corresponding gene for an S-layer protein. Whether these genes are expressed under natural conditions cannot be determined in all cases, due to missing data. Still, the homologue of SLP_Mh in “Ca. Micrarchaeum acidiphilum ARMAN-2” (UNLARM2_0493; GenBank no. EET90051.1) was found to be the most abundant protein assigned to ARMAN-2 in proteomic data of acid mine drainage (AMD) biofilms (16).

FIG 2.

MUSCLE alignment of SLP_Mh homologues at the N terminus. Depicted are the first 150 alignment positions of each sequence, with the NCBI accession numbers on the left and a sequence logo at the bottom. Amino acids shown in the corresponding sequence logo are colored according to their polarity (black, hydrophobic; green, hydrophilic; red, anionic; blue, cationic), and conservation of amino acids at each position is indicated through their height. The blue background color highlights predicted S-layer domains.

Immunostaining reveals potential glycosylation of the mature protein.

To detect the S-layer protein SLP_Mh, we produced polyclonal rabbit antibodies against a 14-amino-acid peptide of the protein (Fig. 1). The functionality of the antibody was verified via SDS-PAGE and Western blotting, followed by immunodetection (Fig. 3b and c). Lysed cells of the coculture containing M. harzensis and S. hospitalis were compared with the S. hospitalis pure culture. Whereas no signal was obtained with the cell lysate of the pure culture, the cell lysate of the coculture showed one signal corresponding to a protein with a molecular mass higher than 130 kDa. Interestingly, this is roughly 30 kDa more than what was expected based on the molecular mass deduced from the protein sequence. The 130-kDa signal was also distinctly visible in Coomassie blue-stained coculture cell lysates (Fig. 3b), which suggests a high abundance of this protein in the culture.

FIG 3.

Detection of S-layer protein and glycosylation. SDS-PAGE (8% T resolving gel) of cell lysate samples from a coculture containing “Ca. Micrarchaeum harzensis A_DKE” and “Ca. Scheffleriplasma hospitalis B_DKE,” a pure culture containing only “Ca. Scheffleriplasma hospitalis B_DKE,” and a culture of E. coli Rosetta pRARE pBAD202_slp_Mh-6× His. Gels were either stained with PAS (a) and subsequently with Coomassie (b) or subjected to Western blotting followed by immunodetection, using anti-SLP_{Mh 133–147} primary antibodies (c). The bands corresponding to SLP_Mh and recombinant SLP_Mh (*) are highlighted. M, molecular weight standard.

Possible explanations for a higher apparent molecular mass are posttranslational modifications. Many S-layer proteins are described in the literature as glycosylated (3, 33). Moreover, the bioinformatic analysis of SLP_Mh also revealed numerous putative glycosylation sites (Fig. 1). Hence, this potential reason for the higher molecular mass was analyzed in further experiments.

SLP_Mh is a glycosylated protein.

To investigate the hypothesis of a glycosylated protein, we conducted periodic acid-Schiff (PAS) staining of the cell lysate from coculture and pure culture and correlated it to the immunodetection with the developed antibody (Fig. 3a). PAS staining revealed a corresponding band in the coculture, which correlated with the SLP_Mh signal after immunodetection. This point together with the absence of the signal in the pure culture strongly suggests that the higher molecular mass is due to glycosylation. The additional signals stained in pure and coculture seem to be glycosylated proteins of the Thermoplasmatales member.

In order to provide further evidence for the posttranslational modification of SLP_Mh, we conducted heterologous expression in Escherichia coli. We hypothesized that the lacking posttranslational modification should lead to a protein matching the theoretical molecular mass of 100 kDa. As expected, immunostaining showed a corresponding band at around 100 kDa, while the PAS stain was negative (Fig. 3). Thus, the higher apparent molecular mass of SLP_Mh is very likely a result of posttranslational glycosylation.

Interestingly, bioinformatic analyses could not provide evidence for the synthesis of activated carbohydrates by M. harzensis. Still, the organism seems to possess several parts of a canonical N-glycosylation machinery. Table S6 summarizes the results of the bioinformatic search for corresponding genes needed for glycosylation reactions. Detected glycosyltransferases typically transfer carbohydrates from UDP-glucose, UDP–N-acetyl-galactosamine, GDP-mannose, or CDP-abequose. Besides the latter, all other carbohydrates can potentially be produced by the host organism S. hospitalis. Further analyses will be needed to reveal how these molecules are exchanged by the two organisms. In this regard, another study (30) provided evidence that the two organisms might interact similarly to Ignicoccus hospitalis and N. equitans. For these cells, a cytoplasmic bridge was proven to exist by electron tomography (39), which would render the exchange of even complex metabolites possible.

Electron micrographs help to visualize the surface structure of M. harzensis.

As a final proof for the existence of an S-layer covering M. harzensis cells, coculture samples containing M. harzensis and S. hospitalis were examined via electron microscopy. As a control, we also examined the pure culture of S. hospitalis.

Cryo-EM images of coculture samples show cells with a larger diameter (∼0.8 to 1.5 μm) and one surrounding layer, as well as cells with a smaller diameter (∼0.4 μm) and two layers (Fig. 4). The latter resembled the cells identified as ARMAN (archaeal Richmond Mine acidophilic nanoorganism) cells in a previous study (22). Single cells as well as cells maintaining cell contact could be observed. For this study, we focused on single cells. Deduced from published information on the surfaces of almost all members of the Thermoplasmatales, including ultrathin sections (31, 40–42), we identified the larger cells as the Thermoplasmatales relative, rendering the observed layer a cytoplasmic membrane. We hypothesized that the smaller cells were in fact M. harzensis cells and suggest based on our previous findings that the inner layer is a cytoplasmic membrane and the outer one is the S-layer. We are aware that these interpretations must be made with caution. A projection through a sample like this—a 3D volume with a thickness of 400 nm or more—does not provide clear evidence of the structure of cell walls. Analysis of such samples need a (cryo)tomographic approach for enabling the interpretation of structural details.

FIG 4.

Cryo-EM micrograph of fully hydrated coculture cells, frozen in liquid ethane. The image shows a single “Ca. Scheffleriplasma hospitalis B_DKE” cell (Sh) next to a few “Ca. Micrarchaeum harzensis A_DKE” cells (Mh). Arrows indicate cytoplasmic membranes (CM) and S-layer (SL). Bar, 200 nm.

In the data obtained so far, we were able to quantify the dimensions of ∼20 M. harzensis cells and calculated an average inner diameter (cytoplasmic membrane) of ∼400 nm and an outer diameter (including S-layer) of ∼440 to 450 nm, leaving a pseudoperiplasmic space of 20 to 25 nm (Table S7).

A fast, elegant, and direct way to unequivocally visualize the 2D paracrystalline layer formed by S-layer proteins on the surface of cells is the electron-microscopic imaging of metal-shadowed, freeze-etched cultures (43–45). Hence, we conducted this experiment with a sample of a coculture containing the DPANN member and its host S. hospitalis. In our setting, these experiments were greatly facilitated by using current software (EM-MENU5; SerialEM) for image acquisition and in parallel controlling the microscope. Both packages enable the user to get a live presentation of the power spectrum of each image that is fetched by the camera. In these power spectra, regular features—for example, crystalline layers—in the micrographs and hence from the sample can be observed instantaneously.

Comparable to the information obtained by cryo-electron microscopy, all electron micrographs of the freeze-etched coculture showed two cell types, distinguishable by cell diameter and surface structure (Fig. 5 and Fig. S1). The diameters of the observed cells matched those determined from cryo-electron micrographs. The cells with a larger diameter, a pleomorphic shape, and a rough surface (lacking any regular or crystalline structure as displayed in Fig. 5c) are most likely the host organism S. hospitalis. This is corroborated by images of the pure S. hospitalis culture, as the cells in these images have the same characteristics (Fig. S1). M. harzensis cells have a smaller diameter and are evenly round. All these cells clearly displayed a crystalline lattice on their surface, an S-layer (Fig. 5d and Fig. S1).

FIG 5.

Electron micrographs of freeze-etched, platinum-carbon-shadowed cells and corresponding power spectra. The images show surface and cell morphology of dividing “Ca. Scheffleriplasma hospitalis B_DKE” (Sh) and “Ca. Micrarchaeum harzensis A_DKE” (Mh) cells (a), a closeup of a single “Ca. Micrarchaeum harzensis” cell (b), and the corresponding power spectra of “Ca. Scheffleriplasma hospitalis B_DKE” (c) and “Ca. Micrarchaeum harzensis A_DKE” (d) cells in panel a. Bars, 200 nm.

The quantitative determination of the lattice, i.e., the distance of the molecular complexes, is notoriously difficult on small round cells. The small diameter inevitably results in a highly bent lattice, the regularity in projection images is reduced, and nice reflexes in the power spectrum are hardly visible (Fig. S1). Nevertheless, we conducted measurements on 10 cells with a visible S-layer pattern and determined the distance of molecular complexes to be around 16 ± 1 nm (Fig. S1). Most likely, the complexes are arranged on a hexagonal lattice; at the limited resolution achieved so far, we cannot determine whether the symmetry of the crystalline lattice is p6, p3, or even p2. Note that the width of the periplasm (∼20 nm) and center-to-center distance of S-layer complexes (∼16 nm) in M. harzensis correspond to characteristics of N. equitans cells (46). While there is no significant sequence conservation between the N. equitans S-layer protein (NEQ300) and SLP_Mh, the putative domain architectures of both proteins determined via HHpred are quite similar (Table S8).

Hence, we concluded that cells assigned to S. hospitalis possess one membrane only, with no additional surface polymer (as in Thermoplasma acidophilum [41]), while M. harzensis cells exhibit one membrane covered by an S-layer. From our results, we postulate that the published structural data in reference 22 potentially show Micrarchaeum-like cells with a surface layer, similar to M. harzensis and N. equitans (22). A potential route to obtain final proof for this would be to perform comparable freeze-etching experiments. An alternative route is cryo-electron tomography in conjunction with subtomogram averaging, which is clearly more demanding and time-consuming but has the potential to yield molecular details of the S-layer under investigation (36).

In coculture, M. harzensis cells are often observed in direct contact with S. hospitalis cells (29, 30). Their cell surface types could play an important role during the possible interaction between these two Archaea, since other studies showed that S-layers can have a role in recognition and interaction between prokaryotic cells in general and especially between different Haloferax species (47). S-layer proteins have been discussed to stabilize the cell as a kind of exoskeleton (37) and influence or stabilize the proteins in the underlying membrane, like for example the proteins of the archaellum complex (48). At the site of direct interaction, however, between N. equitans and I. hospitalis, the S-layer of N. equitans is seen to be decomposed specifically at the attachment site (39).

Conclusion.

During the last decades, cultivation-independent approaches expanded our knowledge about the diversity and evolution of microorganisms. However, laboratory cultures remain essential for detailed characterization of an organism’s genomic potential. In this study, cultivation of the micrarchaeon M. harzensis proved to be a critical point in order to investigate its cell morphology. Via electron micrographs of metal-shadowed, freeze-etched cells of the coculture containing M. harzensis and its putative host S. hospitalis, we were able to identify the DPANN member as round cells with one membrane and a proteinaceous S-layer. This discovery was supported by the detection of an S-layer-protein-encoding gene in the genome of M. harzensis. The identification of homologous proteins in other Micrarchaeota genomes suggests a distribution in the whole phylum.

Therefore, we propose a surface type similar to that of Nanoarchaeum equitans, with one membrane and a surface-spanning S-layer for members of the phylum Micrarchaeota. Glycosylation of the M. harzensis S-layer protein seems to be dependent on a cooperation of M. harzensis and S. hospitalis. While glycosylation of the S-layer protein could be catalyzed by M. harzensis alone, which expresses a functional N-glycosylation system, activated carbohydrates for chain elongation must be provided by S. hospitalis. Further advanced examination techniques like subtomogram averaging can reveal details of the S-layer ultrastructure at near-atomic resolution, including the anchoring mechanism of SLP_Mh in the membrane (36, 49).

Information about the surface of M. harzensis and S. hospitalis will most probably be essential to understand the fundamental basis of the direct physical contact between these two Archaea (30, 50, 51). Interaction studies with other membrane proteins of M. harzensis and S. hospitalis, as well as transplantation of the S-layer into other organisms to observe its potential physical interaction, might help to reveal a putative function of SLP_Mh during the interaction.

MATERIALS AND METHODS

Culturing conditions.

Cultures containing “Candidatus Micrarchaeum harzensis A_DKE” and “Candidatus Scheffleriplasma hospitalis B_DKE,” as well as the pure culture of the latter, were grown as described in reference 30.

E. coli Rosetta pRARE cells carrying plasmid pBAD202_slp_Mh-6×His for recombinant expression of M. harzensis S-layer protein were cultivated in a shaking flask containing Terrific Broth medium (1.2% [wt/vol] tryptone, 2.4% [wt/vol] yeast extract, 0.5% [wt/vol] glycerol, 17 mM KH₂PO₄, 72 mM K₂HPO₄) supplemented with 50 μg mL⁻¹ kanamycin and 30 μg mL⁻¹ chloramphenicol at 37°C and 180 rpm.

MS-based proteomic analyses.

Two 100 mL replicates of an M. harzensis-S. hospitalis coculture as well as a S. hospitalis pure culture were centrifuged for 2 min at 15,500 × g and 4°C. The pellets were resuspended in 500 μL Tris-HCl buffer (pH 6,8) and lysed using a Branson digital sonifier, model 102C (Branson Ultrasonics Co. Ltd., Shanghai, China), for 2 min (0.5-s pulse, 20-s pause) with 60% intensity. Cell debris was pelleted via centrifugation for 10 min at 9,000 × g and 4°C. The protein concentration in the supernatant was determined using the Bradford assay (52). Samples were mixed with Laemmli buffer (53) to a final protein concentration of 0.1 μg μL⁻¹, incubated at 95°C for 10 min, frozen in liquid nitrogen, and stored at −80°C until further analyses.

Proteins were stacked in the top of a 4 to 12% NuPAGE gel (Invitrogen, Carlsbad, CA, USA) and stained with R-250 Coomassie blue before being in-gel digested using trypsin (sequencing grade; Promega GmbH, Charbonnières-les-Bains, France) as previously described (54). The resulting peptides were analyzed by online nanoliquid chromatography coupled to tandem mass spectrometry (MS/MS) (Ultimate 3000 RSLCnano and Q-Exactive HF; Thermo Fisher Scientific, Waltham, MA, USA). Peptides were sampled on a 300-μm by 5-mm PepMap C₁₈ precolumn (Thermo Fisher Scientific, Waltham, MA, USA) and separated on a 75-μm by 250-mm C₁₈ column (Reprosil-Pur 120 C₁₈-AQ, 1.9 μm; Dr. A. Maisch HPLC GmbH, Ammerbuch-Entringen, Germany) using a 200-min gradient. MS and MS/MS data were acquired using the Xcalibur software (Thermo Fisher Scientific, Waltham, MA, USA). Peptides and proteins were identified using Mascot (version 2.6.0; Matrix Science Ltd., London, UK) through concomitant searches against M. harzensis and S. hospitalis databases, a homemade classical contaminant database, and the corresponding reversed databases. Trypsin P was chosen as the enzyme, and two missed cleavages were allowed. Precursor and fragment mass error tolerances were set at 10 and 25 milli-mass units (mmu), respectively. Peptide modifications allowed during the search were carbamidomethyl (C, fixed), acetyl (Protein N-term, variable) and oxidation (M, variable). The Proline software (55) was used to filter the results: conservation of rank 1 peptides, peptide score of ≥25, peptide length of ≥6 amino acids, false discovery rate of peptide-spectrum-match identifications of <1% as calculated on peptide-spectrum-match scores by employing the reverse database strategy, and minimum of 1 specific peptide per identified protein group. Proline was then used to perform a compilation, grouping, and MS1 quantification of the protein groups on the basis of razor and specific peptides. For each replicate, intensity-based absolute quantification (iBAQ) (71) values were normalized by the sum of iBAQ values in the analyzed sample. The normalized iBAQ values were then summed to provide the mean iBAQ value of each quantified protein.

In silico analysis of protein characteristics.

The theoretical molecular mass and isoelectric point, as well as the relative amino acid content of SLP_Mh, were calculated using the CLC Main Workbench 20.0.1 (Qiagen, Aarhus, Denmark). Conserved domains in the amino acid sequence of SLP_Mh were identified based on hidden Markov models via the HHpred server (56–58) using Pfam-A_v34 (59), TIGRFAMs_v15.0 (60), and PDB_mmCIF70_3_Mar (61) databases as references. Putative N- and O-linked glycosylation sites, as well as GPI anchor motifs, were determined using the web servers NetNGlyc 1.0 (62), NetOGlyc 4.0 (63), and GPI-SOM (64), respectively.

Identification of SLP_Mh homologues.

Homologues of SLP_Mh in other Micrarchaeota genomes were detected via BLASTp search with the following parameters: number of threads, 4; expect, 0.05; word size, 6; matrix, BLOSUM62; gap cost, existence 11, extension 1. A total of 51 available Micrarchaeota genomes (Table S1) were used in this study. BLASTp hits in Micrarchaeota (Table S2) were analyzed for known protein domains using the HMMER 3.1b1 algorithm (65) in combination with the Pfam database (72) (release 33.1). For visualization, an alignment of SLP_Mh with all 52 protein sequences detected via BLASTp search was created using MUSCLE v3.8.425 (66) implementation of CLC Main Workbench 20.0.1 (Qiagen, Aarhus, Denmark). The alignment was visualized using the CLC Main Workbench.

Antibody production.

Polyclonal rabbit antibodies against the potential S-layer protein of M. harzensis were generated by GenScript Biotech B.V. (Leiden, Netherlands). A suitable antigen peptide was designed and evaluated with the proprietary OptimumAntigen design tool. The chosen antigen was the peptide region between amino acid positions 133 and 147 (NRGVKTDQYGATKT). During peptide synthesis, a cysteine was added to the C terminus of the peptide and conjugated to keyhole limpet hemocyanin (KLH). The conjugated peptide antigen was used for rabbit immunization.

Cloning and recombinant expression of slp_Mh-6×His.

The slp_Mh gene (Micr_00292) was PCR amplified from genomic DNA isolated from an M. harzensis-S. hospitalis coculture using the oligonucleotide primers 1 and 2 (Table 1), which introduced a 6× His tag-encoding sequence to the 3′ end, as well as complementary overlaps to the target vector pBAD202 (Invitrogen, Carlsbad, CA, USA). The plasmid pBAD202 was linearized via inverse PCR using primers 3 and 4 (Table 1). The linearized vector and the slp_Mh-6×His fragment were gel purified using the Wizard SV gel and PCR cleanup system (Promega, Mannheim, Germany) and assembled via isothermal in vitro ligation according to reference 73. The resulting plasmid (pBAD202_slp_Mh-6×His) was transferred into E. coli cells. To avoid difficulties during protein production due to different codon usage of E. coli and the micrarchaeon, E. coli Rosetta pRARE (Merck, Darmstadt, Germany) cells were used.

TABLE 1.

Primers used for cloning of plasmid pBAD202_slp_Mh-6×His

No.	Orientation	Sequence (5′ → 3′)a
1	Forward	GTTTAACTTTAAGAAGGAGATATACATACCATGAAAAGTCTAAACCCAAAGAGGATAGC
2	Reverse	CCGCCAAAACAGCCAAGCTGGAGACCGTTTTTAGTGATGGTGGTGATGATGGCTCGTGCTTGCCGATG
3	Forward	AAACGGTCTCCAGCTTG
4	Reverse	GGTATGTATATCTCCTTCTTAAAGTTAAAC

^aSequence overlaps with pBAD202 are underlined; the 6× His tag-encoding sequence is in bold.

E. coli Rosetta pRARE pBAD202_slp_Mh-6×His was cultivated as described above. Upon reaching an optical density at 600 nm (OD₆₀₀) of 0.8, expression of slp_Mh-6×His was induced by addition of 1 mM l-(+)-arabinose. After incubation for 18 h at 30°C, the OD₆₀₀ of the culture was determined using a Genesys 20 spectrophotometer (Thermo Fisher Scientific, Schwerte, Germany), and a sample (1 mL) was centrifuged at 16,000 × g for 2 min. The pellet was resuspended in 75 μL of 2× SDS loading dye per OD₆₀₀ of 0.2, heated for 10 min at 95°C, and centrifuged for 5 min at 16,000 × g prior to loading on the gel.

Preparation of cell lysate from archaeal cells.

Samples (50 mL each) of a dense M. harzensis-S. hospitalis coculture and a S. hospitalis pure culture were centrifuged at 15,500 × g and 4°C for 15 min. After resuspension of the pellets in 1 mL protein buffer (50 mM HEPES, 150 mM NaCl [pH 8.0]), OD₆₀₀ was measured using a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, Schwerte, Germany) and was adjusted to 0.65 in both samples. Four volumes of each sample were mixed with 1 volume of 5× SDS loading dye (600 mM Tris-HCl [pH 6.8], 50% [vol/vol] glycerol, 5% [wt/vol] sodium dodecyl sulfate, 0.25% [wt/vol] orange G, 250 mM dithiothreitol), heated for 10 min at 95°C, and centrifuged at 16,000 × g for 5 min, prior to loading on the gel.

SDS-PAGE, Western blotting, and protein staining methods.

Cell lysate samples were separated via denaturing SDS-PAGE according to reference 53 in hand-cast 8% Tris-glycine gels. Separated proteins were transferred from the acrylamide gel to a nitrocellulose membrane (Roth, Karlsruhe, Germany) via a semidry blot with a Trans-Blot Turbo device (Bio-Rad, Munich, Germany) at 1.3 A for 12 min using a continuous blotting buffer system (330 mM Tris, 267 mM glycine, 15% [vol/vol] ethanol, 5% [vol/vol] methanol [pH 8.8]).

PAS staining of glycosylated proteins in acrylamide gels following SDS-PAGE was performed according to reference 67. The gels were subsequently stained with InstantBlue Coomassie protein stain (Abcam, Cambridge, UK) according to the manufacturer’s instructions.

For immunostaining, the membrane was blocked overnight at room temperature with TBST (20 mM Tris, 500 mM NaCl, 0.05% [vol/vol] Tween 20 [pH 7.5]) containing 3% (wt/vol) skim milk powder and incubated with a rabbit anti-SLP_{Mh 133–147} primary antibody (GenScript, Leiden, Netherlands), diluted 1:200 in TBS (10 mM Tris, 150 mM NaCl [pH 7.5]) containing 3% (wt/vol) bovine serum albumin for 1 h. The blot was washed with TBST (four times for 5 min each) and incubated with a goat anti-rabbit alkaline phosphatase secondary antibody (Sigma-Aldrich, Steinheim, Germany) diluted 1:30,000 in TBST containing 3% (wt/vol) skim milk powder for 45 min. After washing with TBST (four times for 5 min each) and several brief rinses with distilled water (dH₂O), colorimetric band visualization was achieved using the AP conjugate substrate kit (Bio-Rad, Munich, Germany) according to the manufacturer’s instructions.

Electron-microscopic samples.

For electron-microscopic imaging, 3 mL of a dense culture were centrifuged at 10,000 × g for 10 min. The cell pellet was resuspended in the remaining supernatant (∼15 μL), transferred onto a gold carrier, and rapidly frozen by being plunged into liquid nitrogen. After transfer into a high-vacuum freeze-etch device (pressure, <1 mPa; CFE-50, Cressington, Watford, UK), samples were freeze fractured at a sample temperature of −97°C. The surface water was removed by sublimation at this temperature for 4 min (freeze-etching), and the resulting sample surfaces were coated with 1.5 nm Pt/C (shadowing angle, 45°) and 15 nm C (shadowing angle, 90°). The resulting replicas were cleaned by floating on freshly prepared sulfuric acid (70% [vol/vol]) for 17 h and then washed twice with double-distilled water (ddH₂O) before being picked up onto hydrophilized copper grids (600 mesh, hex) for imaging.

For cryo-electron microscopy of suspensions of the Archaea, two ∼1-mL portions of a grown cell culture were gently concentrated by centrifugation (5,000 × g, 5 min) and resuspended in a minimum amount of supernatant (ca. 10 μL). Three microliters was applied onto a Cu grid with a holey carbon film (Quantifoil R 2/2; Quantifoil, Jena, Germany), blotted, and cryoimmobilized in liquid ethane using a Leica EM-GP2 plunge freezer (Leica, Wetzlar, Germany). Grids were stored in liquid nitrogen before they were inserted into a cryoARM 200 (Z200FSC; JEOL GmbH, Freising, Germany). Samples were kept at <96 K throughout all imaging steps. The grid was screened at low magnification (×30), and maps were acquired at medium magnification (×8,000) when searching for suitable sample areas. Images were taken under strict low-dose conditions using a Rio16 CMOS camera (Ametek-Gatan GmbH, Munich, Germany) at a nominal magnification of ×20,000 (pixel size, 0.39 nm), using SerialEM version 3.8.6 (68).

Electron micrographs of freeze-etched samples were recorded on a JEM-2100F transmission electron microscope (JEOL GmbH, Freising, Germany) using a CMOS camera (F416; TVIPS GmbH, Gauting, Germany), using pixel sizes of about 0.25 to 1 nm (relative magnification, ×10,000 to ×40,000) and the software packages SerialEM 3.8.6 (68) or EM-MENU 5 (TVIPS GmbH, Gauting, Germany). The corresponding power spectra of selected image areas were determined with built-in algorithms using the software package IMOD (University of Colorado, Boulder, CO, USA) (69).