A novel fold for acyltransferase-3 (AT3) proteins provides a framework for transmembrane acyl-group transfer

Abstract

Acylation of diverse carbohydrates occurs across all domains of life and can be catalysed by proteins with a membrane bound acyltransferase-3 (AT3) domain (PF01757). In bacteria, these proteins are essential in processes including symbiosis, resistance to viruses and antimicrobials, and biosynthesis of antibiotics, yet their structure and mechanism are largely unknown. In this study, evolutionary co-variance analysis was used to build a computational model of the structure of a bacterial O-antigen modifying acetyltransferase, OafB. The resulting structure exhibited a novel fold for the AT3 domain, which molecular dynamics simulations demonstrated is stable in the membrane. The AT3 domain contains 10 transmembrane helices arranged to form a large cytoplasmic cavity lined by residues known to be essential for function. Further molecular dynamics simulations support a model where the acyl-coA donor spans the membrane through accessing a pore created by movement of an important loop capping the inner cavity, enabling OafB to present the acetyl group close to the likely catalytic resides on the extracytoplasmic surface. Limited but important interactions with the fused SGNH domain in OafB are identified, and modelling suggests this domain is mobile and can both accept acyl-groups from the AT3 and then reach beyond the membrane to reach acceptor substrates. Together this new general model of AT3 function provides a framework for the development of inhibitors that could abrogate critical functions of bacterial pathogens.

Keywords: Salmonella enterica; acetylation; bacteria; biochemistry; chemical biology; computational biology; lipopolysaccharide; membrane protein; surface polysaccharides; systems biology.

Figure 1.. Reaction catalysed by OafB, and its predicted topology and structure.

(A) Schematic of the acetylation reaction carried out by OafB. C2 of rhamnose in the repeating O-antigen unit (shown partially, see text) is O-acetylated, likely using acetyl-CoA as acetyl group donor. The acetyl group can migrate (dotted arrow) to the C3 (indicated by *). R = CoA; Man = mannose; Gal = Galactose (Pearson et al., 2020; Micoli et al., 2014) (B) TOPCONS topology prediction of OafB with N- and C-termini of each transmembrane helix (TMH) indicated. Consistent with the RaptorX structure, TOPCONS predicts short cytoplasmic loops and a long periplasmic loop between TMH 3 and 4. (C) RaptorX predicted structure of OafB with AT3 domain (TMH1-10) coloured light blue, loop between TMH3–4 coloured blue, TMH11 coloured orange, and the SGNH domain coloured grey (D) Topology schematic of OafB based on the RaptorX predicted structure. The RaptorX structure has 11 TMH, with a long periplasmic loop between TMH 3 and 4 consisting of a short helix followed by an unstructured region. TMH1–10 form the AT3 domain and are coloured blue, TMH11 forms the linking region between the AT3 and SGNH domains and is coloured orange.

Figure 1—figure supplement 1.. OafB structure as predicted by RaptorX (yellow) and AlphaFold (lilac).

Domains (according to the topology in Figure 1A of Pearson et al., 2020) were isolated from each model and directly compared using a least-squares fit: (A) AT3 domain (Pfam PF01757) (residues 1–338); (B) SGNH domain (residues 422–640); (C) 11th TMH and the periplasmic linking region (residues 328–421). All three domains are largely similar across the two models. (D) The RaptorX and AlphaFold models were superimposed. Variation in linker orientation between the two models results in the SGNH domains further apart than one might expect. This supports the idea of a flexible, mobile linker between domains.

Figure 2.. RaptorX predicted structures of AT3 domain containing proteins.

(A) Fused AT3-SGNH proteins with OafB for comparison. Left to right: OafB from Salmonella enterica subsp. enterica ser. Paratyphi A (OafB); PglI from Neisseria gonorrhoeae (PglI-NG); OatA from Staphylococcus aureus (OatA-SA); OatA from Listeria monocytogenes (OatA-LM). (B) Standalone AT3 proteins with OafB for comparison. Left to right: OafB-SPA; Oac from Shigella flexneri (Oac-SF); WecH from Escherichia coli (WecH-EC); IcaC from S. aureus (IcaC-SA). The structure of the AT3 domain is largely conserved between these proteins.

Figure 3.. Structure of OafB, an AT3-SGNH protein (left, panels A and C), compared to DltB from Streptococcus thermophilus, an MBOAT (membrane-bound O-acyltransferase) protein (right, panels B and D).

Both OafB and DltB have 11 TMH (C, D); however, the arrangements differ considerably. (C) OafB and (D) DltB viewed from periplasm, coloured blue to red, N- to C-termini with transmembrane helices numbered.

Figure 3—figure supplement 1.. Analysis of the RaptorX model for the transmembrane domain of OafB, embedded in a model Escherichia coli membrane and simulated under equilibrium conditions.

(A) Root-mean-square deviation (RMSD) of the backbone atoms. RMSD values remain below 1 nm for the duration of each simulation, even at an elevated temperature (320 K), indicating that there are no major structural changes occurring. (B) Root-mean-square fluctuation (RMSF) by residue (calculated for the backbone atoms). Lilac-shaded bars indicate residues identified by TOPCONS topology analysis as being in the transmembrane α-helices. (C) STRIDE secondary structure analysis of the transmembrane domain simulated at 303 K. Alpha-helices, turns, 3–10 helices, and coils are indicated in pink, teal, blue, and white, respectively. Secondary structure is maintained throughout the 100 ns simulation (1001 frames). 13 helices are maintained across the simulation, 11 of which are transmembrane.

Figure 4.. The loop between transmembrane helix (TMH) 3 and 4 controls the formation of a transmembrane channel lined with essential residues.

(A) Initial equilibrated structure of the transmembrane domain (residues 1–376) of OafB. Residues 1–94 and 136–376 in pale blue, and the novel re-entrant loop residues 95–135 in blue. The two largest cavities identified by the ProPores2 server are shown as red surfaces. Initially, the loop between TMH 3 and 4 occludes the central pore in the AT3 domain. (B) Structure of the transmembrane domain after 50 ns equilibrium molecular dynamics at 320 K. The loop between TMH 3 and 4 is dynamic – this snapshot from the end of the simulation shows the loop to have moved away from the centre of the cavity to allow a channel to form. This channel could be occupied by the acetyl group donor. (C) Same snapshot as (B), but pore shown as a transparent red surface. Important conserved residues shown as red spheres: these residues line the pore in the centre of the AT3 domain.

Figure 4—figure supplement 1.. Multiple sequence alignment of selected AT3 domain-containing proteins.

Multiple sequence alignments using the MafftWS tool of Jalview of the amino acid sequences of the AT3 domains of the proteins: OafB from Salmonella enterica subsp. enterica ser. Paratyphi A (OafB-SPA); PglI from Neisseria gonorrhoeae (PglI-NG); OatA from Staphylococcus aureus (OatA-SA); OatA from Listeria monocytogenes (OatA-LM); Oac from Shigella flexneri (Oac-SF); WecH from Escherichia coli (WecH-EC); IcaC from S. aureus (IcaC-SA). Where these proteins were AT3-SGNH fusions, the sequence was truncated before the start of the SGNH domain to only include the AT3 domain. Regions of conservation are boxed and totally conserved amino acids in red boxes.

Figure 4—figure supplement 2.. AT3 domain-containing proteins, with conserved essential residues (identified via sequence alignment) mapped onto their respective RaptorX structure predictions (red spheres).

In all cases, these residues (excluding those in the catalytic triad of the SGNH domain) line a channel between helices within the AT3 domain. This suggests that this family of proteins may have a common acyl donor and mechanism of action.

Figure 5.. Interactions between the transmembrane domain of OafB and the putative acetyl donor molecule, acetyl coezyme-A.

Note these structures are taken from simulations in which we use the full OafB protein, but only the transmembrane domain (residues 1–377) is shown here for clarity. (A) Left: Initial structure of the OafB transmembrane domain. Essential residues H25 and R14 shown as spheres, the loop between helices 3 and 4 is shown in dark blue. Phosphate headgroups of the phospholipids shown as tan spheres. The loop initially occludes the pore in the AT3 domain. Right: Structure from the end of the steered MD simulation in which acetyl coenzyme-A was pulled into the central channel within the AT3 domain. The loop between helices 3 and 4 has moved away from the centre of the pore towards transmembrane helix 2 (to the left) sufficiently to allow passage of acetyl-coenzyme A into the transmembrane domain. (B) A pocket of basic residues is observed in the AT3 domain, complementary to the 3’-phosphate of acetyl coenzyme-A. Several high occupancy hydrogen bonds are observed between the 3’-phosphate and transmembrane domain residues R14, R74, R338, and K279. (C) Charge transfer interactions between acetyl coenzyme-A and the transmembrane domain of OafB were identified via ONETEP Energy Decomposition Analysis. Loss of electron density is depicted as a purple surface and gain of electron density in a green surface. Significant charge transfer interactions were identified between the 3’-phosphate of acetyl coenzyme-A (losing electron density) and surrounding basic residues R14, R74, R338, and K279 (gaining electron density) (D) Electrostatic potential of the OafB transmembrane domain. Protein in black New Cartoon representation; calculated electrostatic potential overlayed in Surface representation. Scale from –10 V (blue) to +10 V (red). (E) Electrostatic potential of acetyl coenzyme-A. Molecule in Licorice representation; calculated electrostatic potential overlayed in Surface representation. Using the same scale as the AT3 domain, it is clear that acetyl coenzyme-A and the AT3 domain are complementary.

Figure 6.. Interactions between the transmembrane and periplasmic domains in the closed state of OafB.

(A) Closed OafB structure in New Cartoon representation (TM AT3 domain in lilac, SGNH-ext in orange, extra helix in SGNH domain (α8) in teal, and remainder of SGNH domain in grey). Catalytic triad (D618, H621, and S430) as red spheres. Residues identified in high-occupancy or important hydrogen bonding interactions between the SGNH and AT3 domains in Licorice representation. (B) Hydrogen bonding observed between SGNH-ext and AT3 domains in all equilibrium simulations of the closed OafB structure. High-occupancy hydrogen bonding between side chain: carboxylate of E387 and hydroxyl of S129; amine of K132; carboxylate of E98 and hydroxyl of T386. (C) Hydrogen bonding between the SGNH and AT3 domains observed in the OafB₁₀₀ simulations. E189 can hydrogen bond to S474 via its backbone carbonyl (green dashed line), or to L408 via its carboxylate side chain (blue dashed line) but cannot form these interactions simultaneously. Time series of the separation of these residues shown in Figure 6—figure supplement 1.

Figure 6—figure supplement 1.. Separation between the carboxylate of E189 and backbone NH of L408 (teal), and backbone O of E189 and backbone NH of S474 (indigo).

Cut-off distance (3 Å) for hydrogen bonding shown as a dashed grey line. For the first 40 ns, the backbone carbonyl of E189 can form a hydrogen bond to the backbone amide NH of L474. After this point, the flexible loop in which E189 resides (between TMH 5 and 6) shifts, breaking this interaction and allowing the formation of a hydrogen bond between the sidechain carboxylate of E189 and the backbone amide NH of L408 instead.

Figure 7.. Largest cavity in the closed state of OafB.

(A) Largest cavity identified by the ProPores2 Webserver in the closed OafB structure shown as a red surface. The cavity extends down into the transmembrane domain. (B) Closed OafB in Cartoon representation: transmembrane domain coloured light blue, periplasmic loop between TMH 3 and 4 coloured dark blue, SGNH domain coloured grey with the periplasmic linking region in orange, and additional helix in the SGNH domain in teal. The SGNH catalytic triad (Ser430, Asp618, and His621) is shown as yellow spheres. Largest pore identified by the ProPores2 Webserver shown as a transparent pink surface. A single O-antigen unit is shown inside the pore in van der Waals representation, coloured by atom name (cyan for carbon; red for oxygen; white for hydrogen). The single O-antigen unit can be comfortably accommodated in this space in several orientations.

Figure 8.. Two proposed models for the O-antigen rhamnose moiety acetylation by OafB.

In both panels, OafB is shown in a cartoon representation with the AT3 domain coloured light blue, periplasmic loop between TMH 3 and 4 coloured blue, SGNH domain coloured grey, the periplasmic linking region in orange, and additional helix in the SGNH domain in teal. Catalytic triad of the SGNH domain (Ser430, Asp618, and His621) shown as red spheres. 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoglycerol phosphate headgroups shown as transparent tan spheres. (A) Undecaprenyl phosphate carrier lipid with a single O-antigen unit in Licorice representation. O3 of the rhamnose moiety (marked by a red sphere) has a vertical separation from the periplasmic leaflet surface of ~9 Å. The OH of S430 has a vertical separation from the bulk membrane of ~12 Å. (B) OafB-lipopolysaccharide (LPS) system after equilibration steps - one LPS molecule with a single O-antigen unit shown in Licorice representation, and its rhamnose moiety shown in van der Waals representation. OafB remains in an open conformation. After relaxation, the oligosaccharide of LPS has kinked, with the distance between the lipid headgroups of the periplasmic leaflet surface and the acetylation site reduced to ~29 Å (from ~42 Å, as shown in Figure 8—figure supplement 1). This is comparable to the ~35 Å separation between the OH of S430 and the periplasmic leaflet surface.

Figure 8—figure supplement 1.. One molecule of lipopolysaccharide (LPS) with a single O-antigen unit from Salmonella enterica subsp. enterica ser.

Paratyphi A, shown in the Licorice representation. The acetylation site (O3 of rhamnose) and one phosphorus of the Lipid A are shown as spheres. This model was generated using the CHARMM-GUI LPS Modeller and shows the LPS unit before it has been embedded in the membrane and equilibrated. The separation between the lipid A phosphate groups (which would be level with the phospholipid headgroups of the periplasmic leaflet of the inner membrane) is done vertically from the acetylation site by a distance of 42 Å.

Appendix 1—figure 1.. The SGNH domain of OafB exhibits a wide range of orientations relative to the transmembrane domain.

The trajectory has been fitted (rotation and translation) to the initial position of the transmembrane domain, and the positions of the SGNH domain from every fifth frame of a single trajectory (OafB₁₀₀ without elastic network) overlaid. SGNH domain coloured by timestamp, with red = 0 ns, and blue = 250 ns.

Appendix 1—figure 2.. Porcupine plots of the first two major motions of OafB in unrestrained equilibrium molecular dynamics simulations, identified via principal component analysis.

Magnitude and direction of the motion shown with the coloured spikes: blue indicates smaller motions, red indicates larger motions. Values on colour bars indicate the magnitude of the motions in Angstroms. (A) A pedal bin lid-type motion is observed. The protein opens and closes, hinged at the linker (residues 370–380). (B) The periplasmic domain is able to rotate freely about the linker region.