Structure of the surface layer of the methanogenic archaean Methanosarcina acetivorans

Abstract

Archaea have a self-assembling proteinaceous surface (S-) layer as the primary and outermost boundary of their cell envelopes. The S-layer maintains structural rigidity, protects the organism from adverse environmental elements, and yet provides access to all essential nutrients. We have determined the crystal structure of one of the two “homologous” tandem polypeptide repeats that comprise the Methanosarcina acetivorans S-layer protein and propose a high-resolution model for a microbial S-layer. The molecular features of our hexameric S-layer model recapitulate those visualized by medium resolution electron microscopy studies of microbial S-layers and greatly expand our molecular view of S-layer dimensions, porosity, and symmetry. The S-layer model reveals a negatively charged molecular sieve that presents both a charge and size barrier to restrict access to the cell periplasmic-like space. The β-sandwich folds of the S-layer protein are structurally homologous to eukaryotic virus envelope proteins, suggesting that Archaea and viruses have arrived at a common solution for protective envelope structures. These results provide insight into the evolutionary origins of primitive cell envelope structures, of which the S-layer is considered to be among the most primitive: it also provides a platform for the development of self-assembling nanomaterials with diverse functional and structural properties.

The cell envelopes of archaeal and bacterial cells contain, at a minimum, a cytoplasmic membrane (CM) that sequesters its genetic, cellular, and metabolic potential within a confined, concentrated space. External to this membrane the Archaea and bacteria have evolved distinct molecular structures to provide for protection from the environment, to ensure structural rigidity, and to define cell shape that, together, mediate interactions with the external milieu. Most Archaea use a proteinaceous surface (S-) layer, which encloses the CM (Fig. 1), as the outermost cell envelope structure (1). In contrast, bacteria generally possess one or more peptidoglycan layers surrounding the CM and where the Gram-negative types are further encased by an outer membrane (OM). The archaeal S-layer has been postulated to be the most primitive component of the cell envelope structure (2) and may have originated before the evolution of the murein-containing cell envelope/sacculus. Despite their essential biological function and with many potential applications in nanotechnology (3), we still have limited knowledge of any microbial S-layer structure.

Fig. 1.

The M. acetivorans cell envelope. (A) Electron micrograph of the cell envelope of Methanosarcina barkeri, a representative species of the order Methanosarcinales. Cytoplasmic membrane (CM), S-layer (SL), and metachondroitin (MC) are indicated. Reprinted from ref. 10. Copyright American Society for Microbiology. (B) Cartoon of the cell envelope indicating relative positions of cell envelope structures including the periplasmic-like space (PLS) contained between the S-layer and cytoplasmic membrane. (C) Linear cartoon representation of the domain organization of the M. acetivorans MA0829 S-layer protein including: N-terminal signal sequence (S), N- and C-terminal DUF1608 domains (NTR and CTR), tether region (T), and membrane anchor (M).

Electron microscopy studies have provided medium-resolution information on the architecture of S-layers of several Archaea and bacteria. The protein subunits making up the S-layers are arranged in 2D crystalline lattices that have oblique, tetragonal, or hexagonal symmetry with hexagonal symmetry predominating for archaeal S-layers (reviewed in refs. 3, 4). Several high-resolution crystal structures of S-layer–associated proteins have been determined; however, these structures represent but a single component of a multicomponent S-layer (5), isolated domains of multidomain proteins (6, 7), or are computationally annotated proteins not shown to be localized on the cell surface (8, 9).

The Methanosarcina acetivorans cell envelope is typical of archaeal species belonging to the order Methanosarcinales, whereby the cytoplasmic membrane is enclosed by an S-layer (9). It may also be decorated with an external layer of heteropolysaccharide, termed methanochondroitin, depending on the environmental conditions of cell culture (10). We previously identified (9) the S-layer protein of M. acetivorans as MA0829 (671 aa). It has an N-terminal signal peptide, tandem-duplicated DUF1608 domains (11), followed by a negatively charged tether of ∼60 aa and a predicted C-terminal transmembrane helix. The latter presumably anchors the S-layer to the cytoplasmic membrane (Fig. 1C). Protein domains of the domain of unknown function (DUF)1608 protein family have average amino acid lengths of 250–300 aa, are typically found in contiguous pairs within a polypeptide chain, and, with the exception of two extreme halophilic Archaea, are exclusively found in the methanogenic Euryarchaeota (12).

Since the discovery of microbial S-layers over fifty years ago (13), knowledge of their molecular properties that govern cell integrity and permeability has remained limited. To elucidate the structural features of the M. acetivorans S-layer that provide for these protective and transport functions, we have determined the structure of the C-terminal DUF1608 domain of MA0829 in two crystal forms by X-ray crystallography. The structures allow a model of the tandem DUF1608 MA0829 protein and for its assembly into the 2D S-layer lattice. Our results provide high-resolution insight into the functional and structural properties of an S-layer required to maintain cell metabolism.

Results and Discussion

Crystal Structure of the MA0829 C-Terminal DUF1608 Domain.

The ability to determine the M. acetivorans MA0829 tandem DUF1608 S-layer structure has been limited by availability of suitable diffracting crystals. However, the second DUF1608 domain or C-terminal repeat (CTR) (i.e., the C-terminal DUF1608 repeat) containing amino acids 294–570 of the mature MA0829 amino acid sequence was successfully crystallized in two different crystal forms. The structure (Fig. 2A) was determined by selenomethionine (SeMet) single-wavelength anomalous diffraction (SeMet-SAD) to a resolution of 2.3 Å using SeMet-labeled protein, and the second crystal form obtained with unlabeled protein was solved by molecular replacement at a resolution of 2.36 Å. Both structures have excellent geometry (Table S1) and are virtually identical (Fig. S1). The SeMet-labeled structure contains the complete CTR (Thr-295 through Thr-568), whereas the native protein structure comprises amino acids Glu-297 through Glu-566 with two short disordered loop regions (Ser-447 to Ala-455 and Gly-484 to Ser-488).

Fig. 2.

The crystal structure of the S-layer protein of M. acetivorans. (A) The MA0829 CTR shown in ribbon representation. Domains I and II are colored red and blue, respectively, and the connector subdomain is colored yellow. (B) Crystallographic CTR dimer that likely represents the physiological MA0829 NTR-CTR protein. Domains I and II of the symmetry-related CTR molecule are colored orange and cyan, respectively, and the connector subdomain is colored dark green. (C) Sequence alignment of the sequences for the MA0829 NTR and CTR DUF1608 domains with identical residues in gray shading. Secondary structure elements above the alignment are based on the MA0829 CTR structure and are colored according to the CTR domains in A. Residues predicted to be involved in the NTR-CTR interface based on the CTR dimer structure are indicated by green circles. Numbering is according to the mature processed sequence upon removal of the signal peptide.

The MA0829 CTR structure reveals two structurally related domains composed primarily of β-sheet (Fig. 2 and Fig. S2). Both domains have a β-sandwich fold with two opposed antiparallel β-sheets as their structural core. The N and C termini of the polypeptide are located in close proximity in the first domain (domain I); thus, the chain passes from domain I to domain II and back to complete the bipartite CTR structure. As the chain passes from the first domain to the second, it folds back on itself to form a small ∼40-aa connector subdomain consisting of an elongated three-stranded β-sheet (Fig. 2A, yellow). There is significant structural similarity between domains I and II (3.1 Å rmsd for the alignment of 66 residues with 3% sequence identity), although each has differences in both topology and the number of β-strands (Fig. S3): domain II has an additional three-stranded β-sheet appended to one of the β-sheets. The connector subdomain is not more tightly associated with one domain over the other and appears to limit the range of motion between the two β-sandwiches without completely rigidifying the overall structure.

Crystallographic CTR Dimer Structure.

The MA0829 DUF1608 CTR is a dimer in solution; however, the asymmetric unit of both crystal lattices contains a single DUF1608 CTR molecule. The Protein Interfaces, Surfaces and Assemblies (PISA) program (14), which identifies possible protein interfaces and quaternary assemblies, was used to determine a possible CTR dimer structure. The top PISA prediction of a MA0829 CTR dimer was the same for both the P622 and C2 space groups and structural superposition of the predicted dimers (Fig. S1) reveals that they are virtually identical (DALI Z score of 49.6 and 1.6 Å rmsd for the alignment of 511 residues with 100% sequence identity). The intermolecular interface, mediated by residues solely from domain I, buries ∼925 Ų and 875 Ų of each CTR molecule, or 7.1% and 7.2% of the surface of each CTR molecule, in the interface for the P622 and C2 space groups, respectively. The interfaces of the CTR dimers were further compared using the program SC (15) to calculate the geometric shape complementarity (Sc) of the interfaces. Perfectly correlated and uncorrelated surfaces have theoretical Sc values of 1 and 0, respectively, with a range of 0.70–0.76 seen for biologically relevant protein–protein interfaces. The Sc values for the two predicted interfaces are 0.73 and 0.82 for the P622 and C2 dimers, respectively, revealing that the interfaces are highly complementary. In total, 28 aa contribute to the intermolecular interface with an extensive network of 16 hydrogen bonds formed between eight residues from each subunit (see Fig. S5).

DUF1608 N-Terminal Repeat and CTR Domains Are Highly Homologous.

A high degree of sequence identity between the MA0829 N- and C-terminal DUF1608 repeats (79% identical and 87% similar; Fig. 2C) suggests that the DUF1608 crystallographic CTR structure is representative of the N-terminal repeat (NTR) structure. A sequence alignment of the NTR and CTR DUF1608 domains (Fig. 2C) and mapping of sequence variability unto the MA0829 CTR surface (Fig. S4) using the program Consurf (16) reveals conservation of the residues that form the CTR dimer interface. Of the 28 aa contributing to the intermolecular interface of the CTR dimer, 21 are identical between the NTR and CTR domains, 2 residues are conservatively substituted, 1 residue is semiconservatively substituted, and 4 residues are nonconserved substitutions. Of the four nonconserved residues in two instances CTR residues Met-574 and Ile-575), the intersubunit interactions are mediated by the backbone atoms of the nonconserved residues with conserved residues in the symmetry-related chain, and, thus, the interactions mediated by these two residues would be conserved in the NTR-CTR interface. A third nonconserved interaction occurs at the N terminus of the CTR chain (residue Tyr-296) but is not involved in the primary interaction between the two surfaces. The greatest degree of variability between the NTR and CTR sequences occurs primarily on the extracellular face of MA0829 CTR.

The crystallographic CTR dimer structure would also be representative of the tandem DUF1608 NTR-CTR domain structure of the full-length MA0829 polypeptide because all major structural components are preserved in the two domains. To compare the CTR-CTR interface with that of a predicted NTR-CTR interface, a model of the MA0829 NTR-CTR tandem DUF1608 protein was generated. The NTR DUF1608 domain was homology-modeled using the Phyre server (17) with the CTR domain as a template; the NTR model was superimposed on one DUF1608 domain of the CTR dimer to generate the NTR-CTR tandem DUF1608 protein (Fig. 3). The interfaces of the CTR dimer and the predicted NTR-CTR tandem DUF1608 protein are virtually identical (Fig. 3 C and D): the minor differences in sequence between the NTR and CTR DUF1608 domains, including three short amino acid insertions, are primarily in surface-exposed positions that do not contribute to the dimer interface. The termini of the two CTR molecules in the dimer in both crystal forms are in close proximity (∼9 Å) and easily spanned by the five amino acid linker that connects the NTR and CTR in the tandem DUF1608 MA0829 protein.

Fig. 3.

Comparison of the interfaces of the crystallographic CTR dimer and the predicted NTR-CTR protein. (A) Crystallographic CTR dimer with CTR domains labeled and colored as in Fig. 2A. (B) Model of the NTR-CTR tandem DUF1608 protein. The DUF1608 domains are labeled and colored as in Fig. 2B. The position of the 12-aa residues inserted near the N terminus of the NTR model is indicated by an arrow. (C) Crystallographic CTR dimer with residues that contribute to the intermolecular interface shown in space-filling representation. Interfacing amino acid residues from one CTR domain are colored dark gray, whereas those from the other CTR domain are colored light gray. (D) Predicted NTR-CTR tandem DUF1608 protein with residues that contribute to the predicted interdomain interface shown in space-filling representation. Interfacing amino acid residues from the different domains are colored as in C.

The properties of the interfaces, including shape complementarity and buried surface area, and the equivalent dimer structures from two different CTR crystal forms strongly suggests that the crystallographic CTR dimer is representative of the CTR dimer in solution. Moreover, there is strong sequence conservation of structurally important residues that contribute to the intermolecular interactions in the CTR dimer and by extension to the intramolecular interactions in the tandem DUF1608 MA0829 protein. It infers similar structural properties across a large family of paralogous and orthologous tandem DUF1608 S-layer proteins of the Methanosarcinaceae (Fig. S5).

Model of the M. acetivorans S-Layer Hexagonal Tile.

The CTR crystal form I contains a hexagonal lattice (crystallographic space group P622) and examination of symmetry related molecules reveals a highly plausible model for the self-assembly of MA0829 into the M. acetivorans S-layer. Three crystallographic CTR dimers are arranged around a threefold crystallographic axis and serve as the basic repeating unit of the S-layer (Fig. 4 and Fig. S6). Translation of this trimeric unit in two dimensions throughout the crystal lattice generates a flat sheet with sixfold symmetry elements consistent with the molecular features of hexagonal archaeal S-layers visualized by electron microscopy (EM) (2, 18, 19).

Fig. 4.

The putative M. acetivorans S-layer structure inferred from crystal packing in the hexagonal crystal lattice. (A) Five S-layer tiles from the 2D sheet in the hexagonal crystal form as viewed from the extracellular side. The six CTR crystallographic dimers that contribute to the primary pore are shown in colors of the spectrum. CTR dimers that span adjacent tiles are shown in gray. The positions of primary (P), trimer (T), and asymmetric (A) pores are indicated by arrows. The trimeric building block that is the basic repeating unit of protomers in the S-layer is comprised of the gray, green, and yellow crystallographic dimers surrounding the indicated trimer pore. (B) Cutaway side view of the S-layer with the extracellular surface at the top. The distance between the arrowheads is ∼240 Å. (C) A single tile as seen from the extracellular surface (Upper) and from the periplasmic-like space (Lower). A smoothed molecular surface is shown and, on the top, the tile is tilted 40° away from the viewer from the view seen in A. (Lower) The tile is rotated 180°. (D) A single tile colored by electrostatic surface potential with the extracellular surface (Upper) and the view from the periplasmic-like space (Lower). (E) Cutaway views of the primary (P), asymmetric (A), and trimer (T) pores, respectively.

The intermolecular interactions in the trimeric MA0829 CTR assembly occur at the threefold axis of symmetry within the trimer and by interactions at the sixfold axis of symmetry that coincides with the primary pore of the S-layer sheet (discussed below). The interactions at the threefold axis are primarily between residues of domain I of each DUF1608 tandem repeat closest to the threefold axis. The intermolecular interactions at the sixfold axis occur between domain II of two MA0829 protomers. Forty-nine amino acid residues contribute to the interface between adjacent protomers with nine hydrogen bonds and two salt bridges stabilizing the interface. In total, ∼695 Ų of an MA0829 protomer is buried in the interface with the adjacent protomer in the trimeric assembly. The shape complementarity of the interface is high with an average value of 0.70 for the three interprotomer interfaces.

Sequence alignment of the MA0829 NTR and CTR sequences reveals residues making hydrogen bonds and salt bridges in the trimeric assembly are 100% conserved between the MA0829 NTR and CTR tandem DUF1608 repeats (Fig. S5). In addition to these interactions, an ammonium citrate molecule from the crystallization solution is bound at the intermolecular interface between individual MA0829 molecules at the threefold axis. Two residues from each protomer, Lys-387 and Lys-390 from one CTR protomer and Thr-536 and Glu-541 from the adjacent CTR protomer, make hydrogen bonds to the citrate (Fig. S6B). These residues are conserved in analogous positions (NTR residues: Lys-110, Lys-113, Thr-259, Glu-264) in the N-terminal DUF1608 repeat, suggesting that citrate, or a related molecule, is likely to be bound in this interface and involved in stabilizing interacting proteins in the S-layer.

Two-Dimensional S-Layer Lattice.

The overall appearance of the S-layer sheet is made up of a series of hexagonal tiles with each tile containing six complete CTR dimers and halves of an additional six CTR dimers that connect the tile with six surrounding tiles (Fig. 4). The extensive network of intermolecular interactions between S-layer subunits includes hydrogen bonds, salt bridges, and van der Waals interactions (Fig. S5) but no covalent bonds, in agreement with prior observations (2). In total, the interface between a CTR dimer and the surrounding molecules buries 2,700 Ų (11.4%) of the solvent accessible surface of a CTR dimer in the interface with 17% of the surface exposed residues of a CTR dimer participating in the intermolecular interfaces. The orientation of the S-layer relative to the CM is obvious as the N and C termini of all chains are located on the same face of the sheet. An equally striking feature of the 2D sheet is a funnel-like structure on the sixfold axis of symmetry that points toward the CM. The overall morphology of the S-layer, including center-to-center spacing between “primary” pores, 120 Å, and S-layer height, 45 Å, is consistent with the data from EM studies of other microbial S-layers (2, 18–21). The sequences of putative S-layer proteins from related Methanosarcinales also have a similar architecture with tandem DUF1608 domains; a high degree of conservation of amino acid residues in functionally and structurally important positions suggests that these Archaea will have S-layers with properties similar to that of M. acetivorans (Fig. S5).

The inner and outer surfaces of the S-layer are overwhelmingly negative (Fig. 4D). The structure of the NTR-CTR sheet would be essentially identical to the CTR-CTR sheet; however, minor differences occur in sequence between the NTR and CTR DUF1608 domains (Fig. 2C) and would result in a modest increase in negative charge for the S-layer composed of NTR-CTR tandem DUF1608 proteins versus the S-layer model based on the CTR-CTR DUF1608 structure alone.

S-Layer Pores.

Porosity of the S-layer sheet is evident by the presence of three distinct pore types: P or “primary pores” on the sixfold axis, T or “trimer pores” on the threefold axis at the center of the trimeric building block, and A or “asymmetric pores” at the interface of two adjacent trimeric building blocks (Fig. 4). The P pore is the most prominent structural feature of the S-layer and has the overall appearance of a funnel when viewed from the extracellular side of the S-layer. The primary pore constricts to ∼13 Å before opening slightly at the cytoplasmic face of the pore (Fig. 4 and Fig. S7). The T pore on the threefold axis is roughly triangular in shape and constricts to an approximate diameter of ∼8 Å at the narrowest point. Unlike the P pore, the dimensions of the T pore remain relatively unchanged throughout its vertical profile (Fig. S7). The A pore is an irregular shaped gap that exists at the interface of the trimeric building blocks (Fig. 4). At its narrowest point, the A pore has dimensions of 5 × 14 Å. The surfaces of the P and T pores is overwhelmingly negative (Fig. 4D), whereas the A pore is slightly less negative because of the presence of several positively charged residues (CTR residues: Lys-387, Lys-401, and Arg-537; corresponding NTR residues: Lys-110, Lys-124, Arg-260) at the periphery of the pore. The dimensions of all three pores could vary considerably depending on the positioning of amino acid side chains, but their limited size and strong negative charge suggest that the S-layer is a formidable charge and size barrier that protects the organism from noxious or chaotropic compounds and/or hydrolytic enzymes. Thus, the pores exhibit distinct morphological differences in shape and dimension.

All three S-layer pore types would allow passage of small molecular weight nutrients (e.g., amino acids, purines, pyrimidines, and vitamins) needed for cell growth. Additionally, the P pore is sufficiently large to allow passage of siderophores and conceivably, oligopeptides, oligonucleotides, and lipids. The loops that form the primary pore adopt slightly different conformations in the P622 and C2 crystal forms, suggesting that the size of the primary pore may vary slightly (Fig. S8) and could allow passage of larger compounds. The intense negative charge of the pore surfaces, and of both the internal and external surfaces of the S-layer, would impede the movement of negatively charged molecules which presumably cross the S-layer as salts. As a primitive precursor of more advanced bacterial envelope structures, the molecular properties of which are easily tailored to adapt to changing environmental conditions (e.g., through incorporation of substrate-specific porins and high-affinity receptors), the large pores of the S-layer are a necessary feature. The incorporation of other proteins that contain DUF1608 domains, nine of which have been identified in the M. acetivorans genome and three of which have been shown to be surface-exposed (9), could allow modification of the sieve-like properties of the S-layer in response to environmental cues.

Sequence Identity Suggests a Common S-Layer Architecture for Methanosarcinaceae.

To determine whether the S-layer of related Methanosarcinales species would form a similar structure as that seen for M. acetivorans, we compared the primary amino acid sequences of the M. acetivorans MA0829 S-layer protein with the S-layer proteins of Methanosarcina mazei Gö1 (MM_1976) and M. barkeri strain Fusaro (Mbar_A1758) identified by our previous studies (9, 12), and Methanococcoides burtonii DSM 6242 (Mbur_1690), identified by BLAST searches. Secretion of Mbur_1690 by M. burtonii has been demonstrated by experiments that found this protein in M. burtonii culture supernatants (22). All four S-layer sequences have a similar domain organization (Fig. S5), with an N-terminal signal sequence followed by tandem DUF1608 domains, a tether region, and C-terminal transmembrane anchor. There is also significant sequence conservation in the DUF1608 amino acid residues that govern the intermolecular interactions between the M. acetivorans S-layer protomers (Fig. S5), suggesting that related Methanosarcinales species plus two haloarchaeal species, Haloarcula marismortui NP and Natronomonas pharaonis DSM2160 (12), have S-layers with structures similar to that proposed for M. acetivorans by this study.

More distantly related Archaea also appear to use β-sandwich structures in their S-layers. Sulfolobales S-layers are composed of stalk (SlaB) and canopy (SlaA) proteins, both of which are predicted to contain multiple β-sandwich domains (23). Although the Sulfolobales S-layers exhibit three and sixfold symmetry elements, the overall architecture and pore types are distinct from those of our model of the M. acetivorans S-layer. A lack of detectable sequence homology between the Methanosarcinales and Sulfolobales S-layer proteins suggests different solutions for protective structures have been met through the use of structurally related β-sandwich domains.

Structural Homology Between the M. acetivorans S-Layer Protein and Viral Envelope Proteins.

A search of known protein structures deposited in the Protein Data Bank was used to find proteins structurally similar to the M. acetivorans S-layer NTR and CTR domains (Table S2). Most striking is the structural homology between the S-layer protein and envelope proteins of eukaryotic RNA viruses of the Flaviviridae and Togaviridae families. These two virus families use structurally homologous envelope proteins in their capsids (24, 25), although they differ in the number and organization of these proteins in the mature virion (24, 26). Similarity between the S-layer protein and the viral envelope proteins is strongest between MA0829 CTR domain II and domain I of the viral envelope proteins (Fig. 5). In viral envelope proteins, domain I is a central structural element flanked by viral domains II and III, and it mediates a variety of protein–protein interactions with adjacent molecules in the virus envelope (25–27). Flexibility in the hinge regions linking domain I with domains II and III is believed to allow the large conformational changes required during virion maturation and fusion with host cell membranes (28). This suggests that conformational changes in the M. acetivorans S-layer protein may also allow limited bending of the S-layer to fit around the considerably larger archaeal cell membrane surface. A lack of fivefold symmetry in the M. acetivorans S-layer hexagonal tile structure would preclude formation of smaller virus-shaped bodies and, thus, contribute to cell envelope stability. Although both the M. acetivorans S-layer protein and viral envelope proteins form large protein assemblies, the lack of significant primary amino acid sequence similarity and limited chance for lateral gene exchange between these phylogenetically diverse organisms suggest that Methanosarcinales and viruses may have solved the common requirement for protective barriers through convergent protein evolution.

Fig. 5.

Structural homology between the M. acetivorans S-layer protein and eukaryotic virus envelope proteins. Superposition of domain II of the MA0829 CTR (colored as in Fig. 2A) with domain I of the envelope proteins of West Nile virus (PDB ID code 3I50; colored orange), a member of the Flaviviridae family (A), and Sindbis virus (PDB ID code 1Z8Y; colored green), a member of the Togaviridae family, in the alphavirus subfamily (B). Domain II of the MA0829 CTR is rotated ∼210° on the y axis relative to the orientation shown in Fig. 2A, and the N and C termini of the domains are labeled. Multiple breaks are visible in domain I of the envelope proteins, as there are multiple connections between domain I and domains II and III.

Conclusions.

The structure of the M. acetivorans S-layer protein DUF1608 domain and the working model for the 2D S-layer lattice are instances in which a high-resolution crystal structure has recapitulated, in high resolution, the symmetry and structural features of microbial S-layers visualized by medium-resolution EM studies. As a representative of a large family of homologous archaeal Methanosarcinaceae proteins, this previously undescribed S-layer protein structure allows insight into the properties, evolution, and adaptation of the cell envelope structures of key methanogenic species involved in anaerobic carbon cycling. The atomic level view of the archaeal S-layer reveals the molecular architecture of a protective, yet porous, barrier: it also has potential to facilitate the design of nanomaterials requiring periodic spacing and/or well-defined pore structures.

Materials and Methods

Cloning and Purification of M. acetivorans MA0829 C-Terminal DUF1608.

The DUF1608 CTR (amino acid residues 294–570 of the mature amino acid sequence) from M. acetivorans C2A genomic DNA was cloned into the NdeI and XhoI sites of pET22b (Novagen). The resulting construct has a methionine residue added to the N terminus of the CTR sequence and a C-terminal hexahistidine tag.

The expression plasmid was transformed into E. coli Rosetta (DE3) cells (Novagen). Cells were grown in LB media at 37 °C, and growth continued for 4 h after protein expression was induced with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) at an OD₆₀₀ of 0.6. SeMet-labeled protein was produced at the same temperature using the autoinduction method (29). For protein purification, cell pellets were resuspended in lysis buffer [20 mM Na₂HPO₄ (pH 7.4), 0.5 M NaCl, 15 mM imidazole] containing protease inhibitor mixture (Sigma) and PMSF (100 μM). The cell suspension was broken using a French press, and the clarified supernatant was incubated with Ni- nitrilotriacetic acid (Ni-NTA) agarose beads (Qiagen) for 60 min at 4 °C. The beads were washed extensively with lysis buffer, and bound protein was eluted using buffer containing increasing concentrations (50 mM, 100 mM, 300 mM) of imidazole. The proteins were further purified by gel-filtration chromatography using a HiPrep Sephacryl S-300 size exclusion column (GE Life Sciences) equilibrated in 20 mM Na₂HPO₄ (pH 7.4), 0.5 M NaCl. Fractions containing the MA0829 CTR were concentrated to 10 mg/mL (native protein) or 5 mg/mL (SeMet-labeled protein) for crystallization screening.

Protein Crystallization.

Native and SeMet-labeled protein crystals were grown at 18 °C using the hanging drop vapor diffusion method. Crystals of the native protein grew in 0.2 M sodium sulfate, 20% (wt/vol) polyethylene glycol (PEG)3350 using a 1:1 ratio of protein to well solution and were flash frozen for data collection without any additional manipulation. SeMet-labeled protein crystals grew in 0.4 M ammonium citrate, 20% (wt/vol) PEG3350 using a 1:2 ratio of protein to reservoir solution and were cryoprotected in reservoir solution containing 20% (vol/vol) glycerol.

Structure Determination.

Single-wavelength datasets for both native and SeMet-labeled MA0829 CTR protein crystals were collected on beam line 24-ID-C at the Advanced Photon Source at Argonne National Laboratory. Datasets were processed and scaled with DENZO and Scalepack (30). The structure of the MA0829 CTR in crystal form I, crystallographic space group P622, was determined by SeMet-SAD at a resolution of 2.30 Å using the program HKL2MAP (31), which uses the SHELX suite of programs (32). Model building was performed with Arp/Warp (33) and Coot (34). The model was refined with Refmac5 (35), using translation, libration, screw (TLS) parameters (36) in the later stages of refinement, which is part of the CCP4 suite (37). A second crystal form of unlabeled MA0829 CTR, in crystallographic space group C2, was solved by molecular replacement using Phaser (38) at a resolution of 2.36 Å using the model from the hexagonal crystal form. The native protein structure was refined as per the SeMet-labeled protein. Refinement statistics are summarized in Table S1. Structural homology searches were performed using DALI (39), and figures were prepared with PyMol. Electrostatic calculations were performed with Adaptive Poisson-Boltzmann Solver (APBS) software (40) using the PDB2PQR server (41) with the following parameters: parameters for solvation energy (PARSE) force field, N and C termini with neutral charge, and ionic strength of 100 mM.

Pore Dimension Calculations.

The program HOLE (42) was used to calculate the dimensions of the symmetrical primary and trimer pores. The dimensions of the asymmetric pore were calculated by visual inspection of the pore, and the distance between the closest atoms on either side of the pore was measured and the van der Waals radii (3.4 Å) of the atoms subtracted.