The three-dimensional mesh-like structure of peptidoglycan (PGN) creates robust physical properties for bacterial survival. The linear backbone chains of alternating N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) are interconnected by stem peptides (Koraimann, 2003). PGN can be expanded and turned over due to muramidase cutting the backbone sugar chains. The immune systems of mammals and invertebrates have PGN-degrading enzymes, such as lysozyme, that efficiently kill invading bacteria as front-line defense weapons.

Gram-negative bacteria have six protein secretion systems, spanning the inner membrane, periplasm, and outer membrane in the typical cell envelope structure of Gram-negative bacteria (Alvarez-Martinez and Christie, 2009; Hueck, 1998; Kanonenberg et al., 2013; Korotkov et al., 2012; Lasica et al., 2017; Pukatzki et al., 2009). The physical barrier of the PGN layer hinders the assembly of multicomponent secretory systems. Most secretion systems have a lytic transglycosylase component, called autolysin, to create a pore in the PGN layer, allowing the large multicomponent complex to pass through (Dijkstra and Keck, 1996). The local degradation of the PGN layer by autolysin is also involved in creating pores for the insertion of flagella (Scheurwater et al., 2008). However, unregulated autolysin could damage bacteria since the PGN layer is critical for the mechanical support of bacterial cells.

Brucella abortus is a facultative intracellular pathogenic bacterium causing brucellosis in both humans and cattle (Pappas et al., 2006). Humans suffer from recurrent fever and debilitating musculoskeletal, cardiac, and neurological complications during the chronic stage of brucellosis (Archambaud et al., 2010). B. abortus produces and secretes several effector proteins to survive in host immune cells (Myeni et al., 2013). A recent study discovered secretion activator gene A (SagA) in B. abortus, together with its proteinaceous inhibitor peptidoglycan hydrolase inhibitor A (PhiA) (Del Giudice et al., 2013; 2019). SagA is the pore-forming autolysin in the type IV VirB system and responsible for preventing the fusion of Brucella-containing vacuoles (BCVs) with lysosomes of immune cells. SagA is homologous to typhoid toxin secretion A (TtsA), which was recently determined an autolysin in the toxin translocation system of Salmonella (Hodak and Galan, 2013). Both SagA and TtsA are dissimilar from mammalian lysozymes exhibiting the same muramidase activity (Del Giudice et al., 2013). Since SagA is involved in the early stages of intracellular replication of B. abortus, SagA may be a target for the treatment of brucellosis.

The crystal structure of TtsA from Salmonella enterica serovar Typhimurium (S. Typhimurium, StTtsA) was determined in complex with meso-diaminopimelic acid (DAP), revealing the overall clamp-like shape. DAP is found in the stem peptide in the PGN for linking the stem peptides in most Gram-negative bacterial PGN and Bacillus sp. in Gram-positive bacteria. Instead of DAP, Lys is found in most Gram-positive bacteria. A previous study proposed reaction and substrate recognition mechanisms highlighting the bound DAP molecule (Geiger et al., 2020). However, the catalytic mechanism and substrate-binding mode remained elucidated since the structure did not contain the substrate PGN. Here, we report the crystal structure of SagA in complex with a PGN repeating unit demonstrating the substrate-binding site, providing functional implications in the SagA system. The crystal structure and functional implications of PhiA are also studied.

Protein expression and purification of SagA and PhiA

The sagA gene sequence without the transmembrane region (residues 1-174, SagAΔTM) and phiA gene was synthesized after codon optimization for Escherichia coli expression (Bioneer, Korea) (Puigbo et al., 2007). Then, the DNA molecule encoding SagA∆TM or PhiA was inserted into the expression vector pET28a using the restriction enzyme sites NcoI and XhoI (Hyun et al., 2020). The resulting pET28a-SagA∆TM and pET28a-PhiA FL vectors contained the hexa-histidine tag at the C-terminus of the protein. The procedures for the protein expression and purification of SagA∆TM were previously reported (Hyun et al., 2020). Briefly, E. coli BL21 (DE3) cells harboring the resulting plasmid were cultured in LB broth and disrupted by sonication. For selenomethionine (SeMet)-labeled SagA∆TM, E. coli B834 (DE3) was cultured in M9 medium supplemented with L-selenomethionine. The cell lysate was mixed with Ni-NTA agarose resin (1 ml; Qiagen, Germany) in lysis buffer containing 20 mM Tris (pH 8.0), 150 mM NaCl, and 2 mM β-mercaptoethanol. After washing with buffer containing 20 mM imidazole, the protein was eluted with buffer containing 250 mM imidazole (pH 8.0). The proteins were further purified on a gel filtration chromatographic column (HiLoad^TM 26/610 Superdex^TM 200 pg; GE Healthcare, USA) in lysis buffer. For PhiA, the proteins were further purified using anion-exchange chromatography (HiTrap Q; GE Healthcare) before application to the gel filtration chromatography column. The proteins were concentrated using Vivaspin (Millipore, USA) to 15 mg/ml for SagA and to 12 mg/ml for PhiA.

Crystallization, data collection, and processing

The crystallization conditions were optimized using the hanging-drop diffusion method at 14°C. SagA was crystallized in solution containing 0.2 M ammonium acetate, 0.1 M sodium acetate trihydrate (pH 4.1), 21% (w/v) polyethylene glycol 4,000, and 2 mM Tris (2-chloroethyl) phosphate (TCEP). SeMet-labeled SagA was a crystallized solution containing 0.1 M sodium citrate (pH 5.5), and 27% (v/v) Jeffamine ED-2001. PhiA protein was crystallized in solution containing 0.1 M BIS-Tris (pH 6.0), and 18% (w/v) PEG 10,000. For data collection under cryogenic conditions, SagAΔTM crystals were transferred to 2 μl of the viscous oil Paraton-N and incubated for 1 s. Then, the crystals were flash-cooled in liquid nitrogen at –196°C. The datasets were collected using an Eiger 9M detector (Dectris, Switzerland) at a wavelength of 1.00003 Å in beamline 5C and a Pilatus 3-6M detector (Dectris) at a wavelength of 0.97942 Å in beamline 11C of the Pohang Accelerator Laboratory, Republic of Korea. The diffraction datasets were processed, merged, and scaled using the program HKL-2000 (Otwinowski and Minor, 1997). Table 1 shows the data collection statistics.

Structural determination

For SagA, the initial model was built on phase information from the SAD dataset for SeMet-labeled crystals at 2.7 Å resolution (Terwilliger et al., 2009). The structure was refined against the 2.0 Å resolution native dataset. The structure of PhiA was determined using the molecular replacement method using Phenix Phaser-MR software (McCoy et al., 2007). All of the structures were built using COOT and refined using Phenix.refine software (Adams et al., 2010; Emsley and Cowtan, 2004). The detailed refinement statistics are shown in Table 1.

Muramidase activity assay

All assays were performed using a UV-visible spectrophotometer, Multiscan Go (Thermo Fisher Scientific, USA). The activity of SagA and hen egg-white lysozyme (HEWL) (L6876; Sigma-Aldrich, USA) was measured using freeze-dried Micrococcus lysodeikticus cells (ATCC4698; Sigma-Aldrich) and Bacillus subtilis cells (ATCC6633) suspended in 20 mM potassium phosphate buffer (pH 7.0). Kinetics were determined by optical density at 600 nm at 25°C, and reaction rates were calculated based on the change in absorbance per minute (ΔOD600/min). SagA and lysozyme were treated at a final concentration of 100 nM.

To examine the product inhibition of SagA, we generated a reaction product of M. lysodeikticus cells hydrolyzed by HEWL. M. lysodeikticus cells were suspended in 20 mM of potassium phosphate buffer (pH 7.0) at a concentration of 5 mg/ml and treated with HEWL at a final concentration of 1 μM for 1 h at 37°C. After complete digestion by HEWL, the samples were heat-treated at 100°C for 1 h to inactivate the remaining proteins. The sample was then centrifuged at 10,000g, and the supernatant was used as the reaction product of muramidase. In the experiment examining the product-inhibition mode, 20 μl of the product was added to a 200 μl reaction of M. lysodeikticus cells by SagA or HEWL.

To observe the effect of DAP on the activity of SagA, we added 2,6-diaminopimelic acid (Sigma-Aldrich) to the reaction mixture. SagA (500 nM) or HEWL (500 nM) was added to M. lysodeikticus cells or B. subtilis cells in the presence or absence of 5 mM DAP.

Quantification of bacterial survival

Quantification of the viability of bacteria in SagA- or HEWL-treated samples was measured by the colony counting method. Freeze-dried M. lysodeikticus cells were suspended in 20 mM potassium phosphate buffer (pH 7.0) at a concentration of 0.1 mg/ml. Then, we treated SagA and HEWL at a final concentration of 50 nM for 1 h. After treatment, decimal dilution was performed in the same buffer and seeded onto LB agar plates. Colonies were counted after 48 h of incubation at 37°C.

Statistical analysis

All data are presented as mean ± SD. The values were obtained from at least three repetitive experiments. The experiment results were determined by applying the unpaired t-test using GraphPad Prism version 8.0.1 (GraphPad Software Inc., USA). Significance of analysis is indicated in the figures (ns: not significant, *P < 0.05, ****P < 0.0001).

N-terminal amino acid sequencing

The purified PhiA protein was subjected to SDS-PAGE and electrically transferred onto a polyvinylidene fluoride (PVDF) membrane (Westran Clear Signal; GE Healthcare). After staining with Coomassie Brilliant Blue-R-250 staining solution, the detected band was analyzed by the Edman degradation protein sequencing service (EMASS, Korea).

Size exclusion chromatography coupled with multiangle light scattering (SEC-MALS)

Each sample was subjected to size exclusion chromatography (SEC) on a Superdex 200 increase 10/300 GL column (GE Healthcare). The molecular sizes and oligomerization states of PhiA were measured by MALS (DAWN HELIOS II; Wyatt Technology, USA).

Accession numbers

The atomic coordinates and structure factors (codes 7DNP and 7DPY) have been deposited in the Protein Data Bank (PDB; http://wwpdb.org/).

Structural determination and overall structure of SagA

The recombinant SagA protein, deleted in the C-terminal transmembrane region (hereafter called SagA), was cytosolically produced in the E. coli expression system (Fig. 1A). The monomeric protein in the solution was crystallized using the hanging drop method, as previously described (Fig. 1B) (Hyun et al., 2020). The data collection of flash-frozen crystals was collected and produced a dataset that featured 99.8% completeness at 2.0 Å resolution (Table 1). The crystal structure of SagA was determined by the single anomalous dispersion method using anomalous signals from selenomethionine in the crystals.

The overall structure of SagA is similar to that of StTtsA, consisting of two domains connected by a linker. The crystal structure revealed that SagA has two α-helical domains: an N-terminal domain (residues 1-73; cyan in Fig. 2A) and a C-terminal domain (residues 82-174; pink in Fig. 2A), hinged by a short helix and loops (residues 74-81; orange in Fig. 2A). These structural arrangements are similar to that of T4 lysozyme, consisting of the two domains that are connected by a long helix (Matthews and Remington, 1974). The N-terminal domains of SagA and StTtsA have a high sequence similarity to the N-terminal hydrolase domain of T4 lysozyme, including the catalytic residues (Glu17, Asp26, and Thr31 in the SagA numbering). The N-terminal hydrolase domains of SagA and StTtsA consist of four α-helices (α1-α4) and a long flexible loop (α1-α2 loop; residues 17-37) connecting α1 and α2. The long flexible loop in the N-terminal domain contains conserved catalytic residues. The C-terminal domain of SagA forms a compact α-helical bundle with five α-helices (α5-α9) that show structural and sequence similarity to the PGN binding domain of StTtsA.

As previously observed in the StTtsA and T4 lysozyme structures (Geiger et al., 2020; Kuroki et al., 1993), the putative substrate-binding site of SagA is located in a space between the N-terminal and the C-terminal domains. The substrate-binding site is lined with the α1-α2 loop of the N-terminal domain and the beginning of α9 in the C-terminal domain (Fig. 2A). The closest distances between the α1-α2 loop (residue 26) and the beginning of α9 (residue 151) are 3.6 Å in the SagA structure. The corresponding region is separated by 7-16 Å in the StTtsA structures (Fig. 2B). Thus, the clamping region of SagA is closer than that of the StTtsA structure (Geiger et al., 2020). These findings suggest high structural plasticity between the N-terminal flap domain and the C-terminal domain, presumably resulting in the spatial adjustment of the substrate-binding site.

The substrate-binding site of SagA

We found an extra electron density map at the putative substrate-binding site of SagA, which enabled us to build a part of the PGN repeating unit (GlcNAc-MurNAc-L-Ala-D-Glu) (Fig. 3A). The PGN repeating unit in SagA seemed to be from a digested product of E. coli cell wall components during purification. The GlcNAc moiety at the nonreducing end of the PGN backbone is deep inside the substrate-binding site between the N- and C-terminal domains. The acetyl group at C2 and the hydroxyl group at C3 of GlcNAc make hydrogen bonds with the side chains of Thr39 (in α2) and Tyr69 (in α4) in the N-terminal flap domain (Fig. 3B). The acetyl group also makes hydrophobic interactions with Ile35 in the α1-α2 loop and Tyr65 in α4 in the N-terminal flap domain. Thus, the acetyl group may be necessary for binding to the SagA substrate-binding site.

The α1-α2 loop and α1 in the N-terminal domain surround the MurNAc moiety in the PGN backbone. The α5, α8, and α9 in the C-terminal domain also interact with the MurNAc moiety (Fig. 3B). The acetyl group of MurNAc forms hydrogen bonds in a space between the N-terminal and C-terminal domains with the side chain of His92 in α5 and Trp155 in α9, together with hydrophobic interactions with Phe151 in α9. His92 corresponds to the Asn/Gln residue of the Pho-Asn/Gln motif (Pho indicates hydrophobic residues), which is an essential motif for hydrogen-bond interactions with PGN in the T4 lysozyme family (Pei and Grishin, 2005). Thr39, Tyr69, and Trp155 residues in the binding of the acetyl groups in PGN are conserved in StTtsA and the family of N-acetylmuramidases (Fig. 1A) (Stojkovic and Rothman-Denes, 2007). This finding suggests that the substrate recognition mechanisms are shared with StTtsA and other N-acetylmuramidase family proteins.

We next examined the stem peptide moiety attached to the C3 atom of MurNAc in the crystal structure of SagA. L-Ala and D-Glu in the stem peptide did not form specific polar interactions with the SagA residues. Only van der Waals interactions were found near Leu142 and Leu145 of α8 in the C-terminal domain (Figs. 3A and 3B). Our findings indicate that the acetyl moieties at C2 of GlcNAc and at C2 of MurNAc may contribute to recognizing PGN by SagA rather than stem peptides.

According to the proposed mechanism for T4 lysozyme (Kuroki et al., 1999), a water molecule hydrogen-bonded to Asp20 and Thr26 acted as the nucleophile, attacking the C1 atom of MurNAc. Glu11 of T4 lysozyme, on the opposite side of Asp20 and Thr26, was presumed to act as a proton donor to oxygen O1 of MurNAc (Fig. 3C). The crystal structure of SagA revealed the corresponding and conserved Glu, Asp, and Thr residues (Glu17, Asp26, and Thr31 in the SagA numbering) in the active site, suggesting that T4 lysozyme and SagA share the same reaction mechanism. However, the Glu17 residue of SagA is too far to attack the oxygen atom of MurNAc, and the Asp26 and Thr31 residues of SagA are not within distance to make hydrogen bonds with the water molecule. Thus, our structure suggests that local conformational change of the α1-α2 loop in crystal structure of SagA to close the Asp26 residue toward the Thr31 residue may be necessary for catalysis.

PGN-induced clamping motion

We compared the SagA structure in complex with a part of the PGN repeating unit to the StTtsA structure that contains only a DAP molecule. The structural superpositions between individual N-terminal domains or the C-terminal domains of SagA and StTtsA are well aligned (the root‐mean‐square-deviation [RMSD] between 44 and 47 matched Cα atoms are 0.9 Å and 2.4 Å, respectively). However, a substantial hinge motion was found between the N-terminal and C-terminal domains (Fig. 3D). Approximately 11 Å motion (Cα of Gly28 of SagA and Gly25 of StTtsA) and 20° rotation of the N-terminal domain were measured when the structures were superposed using the C-terminal domains as the reference. The hinge motion is comparable to the “clamping” of the substrate-binding region. Structural comparison to StTtsA suggested that the clamping is coupled with PGN binding to the substrate-binding site of SagA, which is represented by the induced-fit model of the enzymes (Fig. 2B).

We found another conformational change induced by PGN binding in the turn region between α8 and α9 of the C-terminal domain (blue arrow in Fig. 3D). The phenyl ring of Phe151 of SagA is inward to the clamping area by hydrophobic interaction with the acetyl group in MurNAc of PGN. In contrast, the corresponding Phe169 of StTtsA is outward from the clamping area. PGN binding causes conformational changes in the C-terminal domain and clamping motion between the two domains.

Muramidase activities of SagA

We compared the muramidase activities of SagA and hen egg-white lysozyme (HEWL) using B. subtilis cells as DAP-type PGNs. The activities, evaluated by the reduced rate of cellular turbidity, showed a threefold higher muramidase activity of SagA than that of HEWL (Fig. 4A). In a similar experiment with M. lysodeikticus cells for Lys-type PGN, SagA also exhibited a higher activity than HEWL at the initial stage (within ~5 min). However, the activity of SagA sharply decreased after 5 min, resulting in only partial digestion of the cells, unlike HEWL (Fig. 4B). As expected, the SagA treatment could not kill the bacteria with only partial digestion of the bacterial cells, which was different from the 3-log reduction by the HEWL treatment (Fig. 4B).

To investigate the mechanistic reason for the limited activity of SagA on Lys-type PGN, we first suspected depletion of the substrates or the active SagA during the enzyme reaction. The supplementation of the substrate cells or the enzyme SagA in the middle of the reaction with SagA did not affect the reaction rate (Supplementary Fig. S1). However, in the case of HEWL, the addition of the substrate cells or the HEWL enzyme increased the reaction rates. When we added the reaction product of M. lysodeikticus cells by HEWL to the reaction mixture of the cells, the reaction rate of SagA was severely decreased, unlike that of HEWL (Fig. 4C). These observations indicate that the reaction product of muramidase strongly inhibits the activity of SagA. We next tested whether SagA could degrade O-acetylated PGNs, such as from Staphylococcus aureus and Bacillus cereus. The O-acetylated PGNs normally show the resistance to lysozymes. The results showed that the SagA protein failed to degrade S. aureus and B. cereus, as observed by lysozymes (Table 2).

The DAP-binding site

The StTtsA structure contained a DAP molecule in a small pocket in the C-terminal domain, far from the substrate-binding site (Geiger et al., 2020). SagA also has the corresponding DAP binding site surrounded by α7 and its flanking loop. The atomic coordinates of the backbone atoms (Val113, Gly115, and Lys117) and the side chains (Asp111 and Thr118) for the DAP binding site fit well between SagA and StTtsA (Fig. 5A).

To investigate the role of DAP binding in the catalysis of SagA, we compared the muramidase activities of SagA in the presence and absence of DAP. M. lysodeikticus cells were used for Lys-type PGN as substrates of SagA, and B. subtilis cells were used for DAP-type PGN. The addition of DAP reduced the SagA activity on Lys-type PGN, while the SagA activity on DAP-type PGN was not changed at an even higher concentration of DAP (Figs. 5B and 5C). In contrast to SagA, HEWL was not inhibited by the addition of DAP. Our findings suggest that the DAP binding of SagA is involved in Lys-type PGN degradation catalysis. Blocking the DAP binding site inhibited the hydrolysis of Lys-type PGN by SagA. Regarding DAP-type PGN, DAP binding is also crucial for recognizing DAP-type PGN, even though the addition of DAP did not affect the hydrolysis activity of SagA. We expected that the free DAP molecule could not complete the DAP moiety in the PGN bound in SagA.

The DAP binding site is approximately 29 Å apart from the substrate-binding site. In molecular modeling, we extended the bound PGN fragment to the DAP binding site by adding two PGN-repeating units (Fig. 5D). Thus, molecular docking indicates that the DAP binding site is interconnected in potentiation of PGN binding. The DAP binding of SagA might provide the collection or grabbing of the substrate PGN chain, which could be a molecular basis for higher activity than HEWL for DAP-type PGN.

Crystal structures and function of PhiA

Previous cell-based assays presented PhiA as a direct SagA inhibitor (Del Giudice et al., 2019). Moreover, the PhiA gene contains an MliC/PliC domain exhibiting a strong inhibitory effect on mammalian lysozymes (Del Giudice et al., 2019). Thus, the proposed function of PhiA is very plausible even though biochemical evidence is lacking. To confirm the role of PhiA, we noted the updated PhiA open reading frame starting at Met74 of the original entity (WP_002965354). The updated open reading frame was predicted to have a possible signal sequence for secretion to the periplasmic space at residue 60 or 62. To confirm the prediction, we expressed the new PhiA open reading frame with the entire N-terminal sequence in the E. coli system. The expressed PhiA protein was localized in the periplasmic space, and N-terminal amino acid sequencing revealed that residues 1-60 or -62 were cleaved off in the mature forms (Fig. 6A). These findings demonstrated that PhiA has a cleavable signal sequence for secretion to the periplasmic space, as predicted.

We determined the crystal structure of PhiA at 2.0 Å resolution to structurally analyze PhiA. The protomer of PhiA has a flattened β-barrel structure that consists of eight antiparallel β-strands, which is the typical structural fold of MliC and PliC (Fig. 6B). The β-barrel of PhiA is stabilized by the disulfide bonds Cys76 and Cys154. PhiA formed a stable tetramer both in solution and crystals, different from monomeric or dimeric PliC or MliC (Figs. 6A and 6D) (Leysen et al., 2011; Um et al., 2013; Yum et al., 2009). The SGxxY motif is conserved among MliC and PliC for binding to the active site of lysozymes, as observed in the complex structures (Um et al., 2013; Yum et al., 2009). However, the PhiA structure did not contain the SGxxY motif on the primary structure or its equivalent motif at any protruding PhiA loop. The exposed hydroxy group of Ser in the SGxxY motif was critical for the interaction with the catalytic residues of the lysozymes in MliC and PliC. However, any corresponding structural feature was not found in PhiA in the structural superposition of PhiA on the PliC-human lysozyme complex structure (Fig. 6C).

We tested the inhibitory function of PhiA with the purified PhiA and SagA proteins. We failed to observe any direct binding or inhibitory function of PhiA with SagA and HEWL. The absence of the corresponding SGxxY motif might account for the lack of PhiA inhibitory function. Thus, our findings suggest that the inhibitory function of PhiA might act indirectly on SagA.

We determined the crystal structure of the muramidase SagA from B. abortus to be a clamp-like structure between the N-terminal hydrolase domain and the C-terminal PGN-binding domain. The mobile N-terminal hydrolase domain positions to the sessile bonds of PGN with the catalytic residues in the flexible loop, while the C-terminal PGN binding domain firmly holds the PGN unit mostly via van der Waals interactions. This molecular arrangement is reminiscent of the hammer-and-anvil tactic. The overall structural arrangement was similar to that of the T4 lysozyme. The bound PGN repeating unit provided structural details on the recognition of PGN. Comparison to the PGN-free StTtsA structure proposed the substrate-dependent hinge motion of the flap-like N-terminal domain. The structural comparison of SagA to the T4 lysozyme further suggested the closing of the loop containing the Glu, Asp, and Thr residues for catalysis toward the target atoms of the substrate. Thus, our findings indicate that the catalytic mechanism is shared with the T4 lysozyme.

Similar to mammalian lysozymes, SagA has muramidase activity on unmodified PGNs. Notably, SagA showed higher activity on DAP-type PGN than HEWL. This observation is expected because SagA works on the DAP-type PGN in the Brucella cell wall. However, SagA failed to digest Lys-type PGN completely by the product-inhibition mode. Moreover, the exogenous soluble DAP significantly inhibited the activity of SagA on Lys-type PGN unlike HEWL. Thus, our results suggest that SagA is specialized on the complete digestion of DAP-type PGN.

We also determined the crystal structure of PhiA from B. abortus, previously suggested as a direct inhibitor of SagA. Although the PhiA structure showed a similar fold to the MliC/PliC family proteins inhibiting lysozymes, the binding motif for inhibiting lysozymes was not present in PhiA. Since we could not observe any interaction between PhiA and SagA or HEWL, we speculate that it would modulate the activity of SagA in a different mode. It is also puzzling why the SagA/PhiA system is different from the lysozyme/MliC system. We presume that this difference might result from the purpose of the inhibitory proteins. While MliC proteins are the global inhibitor of all the lysozyme molecules, PhiA should act on SagA under the spatial and temporal limitation.

The crystal structure of SagA did not cover the transmembrane region (residues 223-245) or the linker region (residues 175-222) between the C-terminal domain and the transmembrane region (Fig. 1A). We hypothesized that the full-length SagA protein is immobilized to the inner membrane by the transmembrane region in the periplasmic space. The autolytic activity of SagA would be confined within the length of this linker region. If the transmembrane region of SagA interacts with type IV secretion system (T4SS), the localized activity of SagA in the T4SS would be explained. Thus, the confined activity of SagA would be crucial in preventing the further and unnecessary breakdown of the PGN layer.

Lysozymes have been employed to defend against bacteria in blood, tears, and many other body fluids (Hankiewicz and Swierczek, 1974). Due to its antimicrobial effect, lysozymes have been used as food or feed additives to control bacteria. However, many pathogenic gram-negative bacteria have natural inhibitory proteins to confer lysozyme resistance to the bacteria (Abergel et al., 2007; Callewaert et al., 2008; Monchois et al., 2001). Its limited antimicrobial efficacy against gram-negative bacteria restricts its application in the food industry. However, SagA showed higher muramidase activity than HEWL against DAP-type PGN (Fig. 4A). Furthermore, various lysozyme inhibitors from gram-negative bacteria will not inhibit the activity of SagA due to the entirely different primary and 3D structures between SagA and mammalian lysozymes. Thus, SagA has potential as a feed additive to control contaminated bacteria.

In conclusion, this study revealed high-resolution structures of SagA and PhiA. We analyzed the biochemical characteristics of SagA and suggested its substrate-binding mode. These findings have implications for the role of PhiA and the molecular basis for the SagA-PhiA system of B. abortus. The development of an inhibitor of SagA based on structural studies may further contribute to controlling brucellosis by attenuating T4SS, the major virulence factor of Brucella.

This research was supported by Agriculture, Food, and Rural Affairs Convergence Technologies Program for Educating Creative Global Leader (710012-03-1-HD120). This research was also supported by the Bio & Medical Technology Development Program of the National Research Foundation (NRF) funded by the Korean government (MSIT) (NRF-2017M3A9F6029755). We made use of beamlines 5C and 11C at the Pohang Accelerator Laboratory (Pohang, Republic of Korea). This work was also supported by the BK21 Plus Program of the Department of Agricultural Biotechnology, Seoul National University, Seoul, Korea.

Y.H. and N.C.H. designed the research. Y.H., Y.B., C.L., and N.K. conceived and performed experiments. J.A., S.R., and N.C.H. provided expertise and feedback. Y.H. and N.C.H. wrote the manuscript, and N.C.H. secured funding.

The authors have no potential conflicts of interest to disclose.

Data collection and refinement statistics

Muramidase activity of SagA on O-acetylated PGNs