Identification and Structural Characterization of a Novel Myeloperoxidase Inhibitor from Staphylococcus delphini

Abstract

Staphylococcus aureus and related species are highly adapted to their hosts and have evolved numerous strategies to evade the immune system. S. aureus shows resistance to killing following uptake into the phagosome, which suggests that the bacterium evades intracellular killing mechanisms used by neutrophils. We recently discovered an S. aureus protein (SPIN for Staphylococcal Peroxidase INhibitor) that binds to and inhibits myeloperoxidase (MPO), a major player in the oxidative defense of neutrophils. To allow for comparative studies between multiple SPIN sequences, we identified a panel of homologs from species closely related to S. aureus. Characterization of these proteins revealed that SPIN molecules from S. agnetis, S. delphini, S. schleiferi, and S. intermedius all bind human MPO with nanomolar affinities, and that those from S. delphini, S. schleiferi, and S. intermedius inhibit human MPO in a dose-dependent manner. A 2.4 Å resolution co-crystal structure of SPIN-delphini bound to recombinant human MPO allowed us to identify conserved structural features of SPIN proteins, and to propose sequence-dependent physical explanations for why SPIN-aureus binds human MPO with higher affinity than SPIN-delphini. Together, these studies expand our understanding of MPO binding and inhibition by a recently identified component of the staphylococcal innate immune evasion arsenal.

Graphical abstract

INTRODUCTION

Neutrophils are the most abundant white blood cells in human circulation and play a critical role in the acute phase of inflammation. Neutrophils serve as the first line of defense against invading bacteria [1, 2], and eliminate bacterial pathogens following phagocytosis. Efficient phagocytic killing is a complex process dependent upon the various antimicrobial proteins and peptides contained within neutrophils’ subcellular granules (Reviewed in [1–5]). As the phagosomal compartment matures, an enzymatic system that leads to generation of diverse reactive oxidant species assembles within its membrane. Activity of the multipartite NADPH oxidase converts O₂ into O₂^−•, which dismutates either directly or enzymatically into H₂O₂. Although H₂O₂ on its own is mildly cytotoxic, it also serves as a substrate for the abundant heme-dependent granule enzyme, myeloperoxidase (MPO). MPO generates highly toxic and reactive oxidant species, most notably hypochlorous acid (HOCl). Neutrophil granules likewise contain high levels of antibacterial peptides and proteases, such as neutrophil elastase (NE), cathepsin G (CG), and proteinase-3 (PR3). While these proteases on their own are sufficient to kill certain bacteria [6–8], there is evidence to suggest that they function synergistically with MPO-derived oxidants to enhance killing of bacteria trapped within the phagosome [9]. Thus, the abundant and overlapping anti-bacterial systems acting within the neutrophils’ phagosomal compartment present a formidable innate defense against infection.

Staphylococcus aureus is a Gram-positive pathogenic bacterium that causes a broad range of infections in humans and animals [10]. Along with closely related staphylococcal species, S. aureus has an increasingly negative impact worldwide due to a rapidly expanding incidence of antibiotic resistance as well as a generally enhanced capacity for virulence. Even though neutrophils play a central role in the overall response to S. aureus infection [11, 12], the bacterium has evolved a broad repertoire of strategies to resist both opsonisation and phagocytosis [13–16]. Furthermore, S. aureus shows resistance to killing following uptake into the phagosome, which suggests that the bacterium can actively evade specific intracellular killing mechanisms used by neutrophils [13]. In this regard, we identified a family of proteins secreted by S. aureus that bind non-covalently to and block function of the neutrophil granule proteases NE, CG, and PR3 [17]. More recently, we discovered a previously uncharacterized protein, SPIN (for Staphylococcal Peroxidase INhibitor), that binds tightly to MPO and inhibits its enzymatic activity [18].

Since SPIN shares no sequence relationships to other known proteins, we have relied on structure-based approaches to better understand the molecular basis for its function. A co-crystal structure of S. aureus SPIN bound to a recombinant form of human MPO (rhMPO) indicated that this inhibitor acts as a molecular stopper to prevent exchange of substrates/products with the MPO active site [18]. This structure also revealed that the ~8.3 kDa SPIN protein consists of two functionally distinct regions. Whereas the C-terminal 60 amino acids adopt a compact three α-helical bundle fold, the N-terminal 13 residues comprise a unique β-hairpin motif [18]. Subsequent studies showed that the α-helical bundle domain is responsible for driving interaction of SPIN with MPO, while the β-hairpin region makes only minor contributions to SPIN/MPO binding [19]. This latter feature is consistent with solution NMR spectroscopy studies on SPIN, which demonstrated that the N-terminal β-hairpin is actually disordered in the absence of MPO [19, 20]. Nevertheless, the N-terminal residues of SPIN are required for inhibiting MPO activity [19], since the β-hairpin they comprise inserts into the MPO active site cavity and blocks substrate/product exchange [18].

Although molecular level analysis of S. aureus SPIN (hereafter SPIN-aureus) has provided fundamental insights into the structure and function of this novel MPO inhibitor, our current understanding is restricted to sequence/structure features specific to the SPIN-aureus protein. To circumvent this limitation, we sought to identify SPIN-aureus homologs in other staphylococcal species and characterize their interactions with and effects on human MPO. We report here the identification of eight SPIN homologs, three of which bind to and inhibit human MPO in a dose-dependent manner. We also present a 2.4 Å co-crystal structure of SPIN from Staphylococcus delphini (SPIN-delphini) bound to rhMPO. This work not only allows the first comparative analyses of SPIN proteins, it also broadens our appreciation of a unique class of MPO inhibitor that functions in staphylococcal evasion of neutrophil-mediated immunity.

EXPERIMENTAL

Proteins

Two different forms of human myeloperoxidase were used during the course of this work. Native human myeloperoxidase (MPO) that had been chromatographically isolated from purulent sputum was obtained from Elastin Products Corp. (Owensville, MO; catalog # MY862) and was used for biochemical and functional analyses. A recombinant form of human MPO (rhMPO) bearing a C-terminal 10-His tag was purchased from R&D Systems (Minneapolis, MN; catalog # 3174-MP-250) and was used for structural studies. The lyophilized proteins were reconstituted and handled as suggested by the supplier.

SPIN from S. aureus strain Newman was expressed and purified as described previously [18]. The sequences of the predicted mature form of various SPIN homologs were codon optimized for E. coli expression using the Sequence Manipulation Suite Reverse Translate tool (http://www.bioinformatics.org/sms2/rev_trans.html) [21]. The corresponding DNA fragments were synthesized as gBlocks Gene Fragments (Integrated DNA Technologies; Coralville, IA, USA) with BamH1 and Not1 sites appended at the 5’ and 3’ ends, respectively. Each of the coding fragments were subcloned into the BamHI and Not1 sites of a modified form of the prokaryotic expression vector pT7HMT [22]. The integrity of each insert was verified by DNA sequencing prior to transformation into E. coli BL21(DE3) cells for protein expression as previously described [18, 22]. The vector pT7HMT encodes an N-terminal 6-His affinity tag that is used for Ni²⁺-affinity chromatography purification of the protein, but which can be removed by digestion with Tobacco Etch Virus (TEV) protease [22]. Following TEV cleavage, the recombinant form of each SPIN protein contains an artificial “GSTGS” amino acid sequence at its N-terminus. Analogous residues arising from the subcloning procedure are not visible in the co-crystal structure of SPIN-aureus bound to human MPO [18], and thus are not believed to influence the physical nature of the SPIN/MPO complex. All SPIN proteins were analyzed by MALDI-TOF mass spectrometry for both purity and integrity prior to their usage in subsequent experiments.

Surface Plasmon Resonance

Direct binding studies of SPIN proteins to native human MPO were performed on a Biacore T-200 instrument (GE Healthcare). All experiments were carried at 25 °C using a running buffer of 20 mM HEPES (pH 7.4), 140 mM NaCl, and 0.005% (v/v) Tween-20 and a flowrate of 30 µl/min. Experimental surfaces were created on CMD 200M sensor chips (XanTec Bioanalytics GmbH; Dusselorf, Germany) by coupling native human MPO on three separate flow cells via random amine chemistry according to manufacturer’s suggestions. First, a solution of N-hydroxysuccinimide (NHS; 0.1 M in ddH₂O) and N-ethyl-N'-(dimethylaminopropyl) carbodiimide hydrochloride (EDC; 0.4 M in ddH₂O) was mixed immediately prior to injection over the naïve surface to convert the carboxymethyl groups into reactive N-hydroxysuccinimidyl esters. Subsequently, a solution of 25 µg/mL MPO in 5 mM sodium acetate buffer (pH 5.5) was injected over the surface for 7 min to allow covalent coupling of protein-borne amines to the surface-derived esters via amide bond formation. Finally, any uncoupled NHS-esters were deactivated by injecting 1 M ethanolamine (pH 8.5), prior to re-equilibrating the surface in running buffer. MPO was immobilized at different levels on each flow cell to allow for replicate measurements while ensuring that surface density did not substantially affect the interaction parameters; two flow cells had a relatively high level of MPO at 7155 and 7311 resonance units (RU), respectively, while a third flow cell was derivatized with 2829 RU of MPO. A reference surface was also prepared by activation followed by immediate inactivation with ethanolamine. A concentration series for each SPIN protein was injected over the flow cells for 3 min, followed by a 4 min dissociation phase. Regeneration to baseline was achieved by two consecutive 0.5 min injections of 0.1 M glycine (pH 10.0). Data processing was performed using Biacore T-200 Evaluation Software v3.0 (GE Healthcare). Each reference subtracted injection series was analyzed using a 1:1 binding model (Langmuir) and fitting R_max locally. The fitted binding curves and were imported and graphed in Igor Pro (Wavemetrics) and the resonance units (RU) were normalized to the maximal response for each experimental series.

MPO Activity Assays

MPO activity was measured spectrophotometrically as previously described using H₂O₂ as a substrate and o-dianisidine as a redox indicator [18]. Each reaction contained 0.2 U/mL MPO isolated from human sputum, 0.5 mM H₂O₂, and 1.9 µM of o-dianisidine dihydrochloride (Sigma-Aldrich) dissolved in 45 mM phosphate buffer (pH 6.0). The absorbance at 450 nm was measured every 45 s for 1 h at 37 °C using a FLUOst ar Omega (BMG LABTECH) microplate reader. The slope before saturation was taken as a measure of MPO activity. For enzyme inhibition studies, the MPO protein was incubated for 1 h at 37 °C with a dilution series (2000 nM to 1 nM) of each SPIN protein prior to substrate addition.

Crystallization, Structure Determination and Refinement

A sample of the SPIN-delphini/rhMPO complex was prepared by mixing the purified monomers in an equimolar ratio. The sample was then concentrated to 5 mg/mL total protein and exchanged by ultrafiltration into a buffer of 5 mM tris (pH 7.4), 50 mM NaCl. Initial crystallization trials were carried out by vapor diffusion of hanging drops at 20 °C using Hampton Research Crystal Screen I and II and various customized buffers. Crystals suitable for X-ray diffraction studies were grown over the course of 7–10 days from drops that contained 1 µL of complex mixed with 0.5 µL distilled water and 0.5 µL of a precipitant solution consisting of 0.1 M sodium acetate (pH 4.6), 25% (v/v) PEG 3350. The crystals were briefly soaked in a cryoprotectant solution of 0.1 M sodium acetate (pH 4.6), 25% (w/v) PEG 3350 and 10% (v/v) PEG 4000 prior to flash cooling in liquid N₂.

X-ray diffraction data were collected at beamline 22-ID of the Advanced Photon Source of Argonne National Laboratory. A total of 360 images were collected with an oscillation angle of 1°, an exposure time of 1 s, radiation of λ=1.000 Å, and a sample to detector distance of 300 mm. The reflections were indexed, integrated, and scaled using the HKL-2000 package [23]. The structure was solved by molecular replacement using and the refined polypeptide coordinates of SPIN-aureus/rhMPO (PDB entry 5UZU) as a search model [18] and PHASER [24] as implemented in the PHENIX software suite [25, 26]. The final model was constructed by an iterative combination of automated and manual rebuilding, followed by crystallographic refinement using PHENIX.REFINE [25, 26]. 96.08% of the modeled polypeptide residues lie in favored regions of the Ramachandran plot, with only 0.31% in regions classified as outliers. In addition to the two polypeptides, the final model contains 105 ordered solvent molecules, a Ca²⁺ ion coordinated entirely by groups derived from rhMPO, a single Cl⁻ ion, and a covalently bound heme prosthetic group within the MPO active site. A more detailed description of the cell constants, diffraction data quality, and properties of the final model can be found in Table 3. All structural analyses, including calculation of buried surface areas and identification of potential hydrogen bonds and salt bridges, were performed using EBI-PISA (http://www.ebi.ac.uk/msd-srv/prot_int/cgi-bin/piserver). Representations of protein structures were generated by PyMol (http://www.pymol.org/).

Table 3.

X-ray Diffraction Data Collection and Refinement

	SPIN-delphini/rhMPO
Data Collection
Space group	P 21 21 21
Cell dimensions
a, b, c (Å)	84.63, 90.67, 125.66
Resolution (Å)	45.34-2.40 (2.49-2.40)*
Rpim	0.045 (0.382)
I / σI	15.5 (1.7)
Completeness (%)	99.7 (97.2)
Redundancy	13.5 (7.7)
Refinement
Resolution (Å)	45.34-2.40
Number of Reflections	36,188
Rwork / Rfree	18.5/21.8
No. atoms
Protein	5143
Ligand/ion	87
Solvent	105
Mean B-factors (Å2)
Protein	56.8
Ligand/ion	65.3
Solvent	50.9
R.M.S. deviations
Bond lengths (Å)	0.015
Bond angles (°)	1.48
PDB accession code	6BMT

^*Values in parentheses are for the highest-resolution shell.

RESULTS

Identification of SPIN Homologs in a Subset of Staphylococcal Species

The spn gene is nearly ubiquitous among both human and animal-derived clonal lineages of S. aureus [18]. Furthermore, aside from S. aureus strains Mu3 and Mu50 which encode a substantially truncated and therefore inactive protein, there is little variability in the SPIN coding sequence between S. aureus strains [18]. Although these previous studies indicated that SPIN is highly conserved among S. aureus strains, we wondered whether genes encoding SPIN homologs might be found in related staphylococcal species. To test this possibility, we used the BLASTP algorithm to query the NCBI non-redundant protein database for any uncharacterized molecules with sequence homology to SPIN-aureus from strain Newman. Indeed, we identified eight such proteins from closely related staphylococcal species that are documented pathogens of humans, livestock, and/or domestic companion animals. Since SPIN is targeted for secretion from the bacterial cell via an N-terminal signal peptide, we used the SignalP server (http://www.cbs.dtu.dk/services/SignalP/) to deduce the predicted matured sequences of the SPIN homologs [27]. We then compared these sequences to one another and with SPIN-aureus using a combination of CLUSTAL OMEGA [28] and MultAlin [29] (Table 1).

Table 1.

Amino Acid Identities and Similarities Among the SPIN Sequences from Various Staphylococcal Species.

SPIN	sciuri	aureus	Chromogenes	agnetis	hyicus	schleiferi	delphini	intermedius	pseudintermedius
sciuri	100	52	54	64	59	56	62	63	63	similarity
aureus	38	100	70	73	64	77	80	79	73
chromogenes	35	48	100	85	84	74	77	73	73
agnetis	41	52	75	100	90	74	77	73	76
hyicus	38	46	69	85	100	67	70	69	71
schleiferi	29	56	61	59	51	100	91	81	83
delphini	33	53	58	54	48	69	100	86	90
Intermedius	39	57	53	50	47	59	74	100	91
pseudintermedius	38	57	55	55	49	69	75	81	100
	identity

We found agreement in our phylogenetic analysis of these novel SPIN-like sequences and previous analyses that identified relationships between these staphylococcal species [30] (Fig. 1A). For example, S. hyicus and S. agnetis belong to the same clade of staphylococci [31]. Comparison of SPIN-hyicus and SPIN-agnetis revealed less than 5 point accepted mutations (PAM) and ~85% sequence identity between these two proteins. A similar relationship was observed between S. schleiferi and established members of the Staphylococcus intermedius group (SIG) (i.e. S. intermedius, S. pseudintermedius, and S. delphini) [32]. SPIN sequences within this cluster share no less than ~59% identity with one another (i.e. SPIN-intermedius v. SPIN-schleiferi), and also display the highest overall sequence identities with SPIN-aureus (~53–57%). By contrast, the distant phylogenetic relationship between S. sciuri and other staphylococci [33, 34] was also reflected in the relatively low sequence identity of SPIN-sciuri with the other SPIN sequences we analyzed here (~29–39% identity).

Fig. 1.

Analysis of Sequence Conservation Across Putative SPIN Homologs. A group of hypothetical sequences homologous to SPIN-aureus was identified by BLAST searching in the NCBI non-redundant protein database. The predicted mature form of each protein following cleavage of the signal peptide was determined via the SignalP server [27]. (A) Phylogenetic tree based upon the matured amino acid sequences of SPIN-aureus and eight other species of staphylococci. The tree is drawn to scale with branch lengths measured in the number of substitutions per site. (B) The structure of rhMPO-bound SPIN-aureus is shown as a ribbon diagram (PDB code 5UZU [18]). Residues invariant in all SPIN homologs and found at the SPIN/rhMPO interface are colored orange, while invariant residues that do not play a direct role in rhMPO contact are colored pink. Numbering reflects the sequence of matured SPIN-aureus. The N-terminal β-hairpin is shown at the top right of the image for purposes of orientation. (C) The sequences of putative SPIN homologs were aligned to that of SPIN-aureus using Clustal Omega [28], and displayed using EsPript [42]. Invariant residues are shown in reverse blue typeface, while residues conserved in a majority of sequences and conservative substitutions are shown in purple typeface. The secondary structure of rhMPO-bound SPIN-aureus is displayed above the alignment. Residues invariant across all SPIN homologs and found at the SPIN/rhMPO interface are designated with an orange circle.

All published structure/function analyses thus far have been carried out using either natively-produced or recombinantly-expressed SPIN-aureus [18]. In this regard, the overall sequence identity of the matured homologs to SPIN-aureus varies between 38–57% (Fig 1A and Table 1). Surprisingly, examination of a multiple sequence alignment revealed that only 10 of 73 positions (~14%) are invariant across the nine sequences analyzed (Fig. 1B–C). Of these, only six residues (G³⁹, L⁴², D⁴⁴, D⁴⁵, H⁵¹, and Y⁷⁵ per SPIN-aureus numbering) were found at the SPIN-aureus/rhMPO interface, as judged by the co-crystal structure [18]. Interestingly, four of these residues (G³⁹, L⁴², D⁴⁴, and D⁴⁵) lie within the N-terminal region of the SPIN protein. Although this region of SPIN-aureus appears to make only minor contributions to SPIN binding of MPO, it is essential for inhibiting MPO enzymatic activity [19]. Thus, the relatively low level of absolute identity within the known MPO-binding site of SPIN suggested that valuable insights might be gained from further structure/function studies on these SPIN homologs.

A Subset of SPIN-aureus Homologs Bind to and Inhibit the Enzymatic Activity of Human MPO

Since genomic DNA was not readily available for all of these staphylococcal species, we designed synthetic coding sequences for each of the SPIN homologs, subcloned these fragments into a plasmid vector that directs expression of N-terminally His-tagged fusion proteins [22], and used these plasmids to produce highly purified forms of each putative SPIN. The integrity of each protein was confirmed by MALDI-TOF mass spectrometry. Unfortunately, the SPIN-hyicus protein appeared prone to degradation, and was thereby excluded from further analysis. We then examined the ability of each remaining SPIN protein to bind immobilized native human MPO by a surface plasmon resonance (SPR) approach. Injections of an increasing concentration of each recombinant SPIN gave clear evidence of binding for several of the homologs (Fig. 2). SPIN-chromogenes and SPIN-sciuri appeared to bind human MPO much more weakly than the others and were not analyzed further (Data Not Shown), while SPIN-pseudintermedius gave no evidence of binding whatsoever (Fig. 2). The reference subtracted sensorgrams for all other series were analyzed using a kinetic model of 1:1 binding to derive association (k_a), dissociation (k_d), and affinity constants (K_D) for each interaction (Table 2).

Fig. 2.

Characterization of SPIN Homolog Binding to Native Human MPO by Surface Plasmon Resonance. A two-fold dilution series of recombinant SPIN-aureus and various homologs was injected over three separate flow-cells of native human MPO that had been randomly immobilized at different surface densities. The reference-subtracted sensorgrams for each injection series (black traces) were fit to a 1:1 binding model (red traces) and normalized to their respective maximal responses. Representative sensorgram series are shown for native MPO binding to (A) SPIN-aureus, (B) SPIN-delphini, (C) SPIN-intermedius, (D) SPIN-schleiferi, and (E) SPIN-agnetis. (F) Representative traces for injections of SPIN-pseudintermedius over native MPO surfaces. SPIN-pseudintermedius does not appear to bind significantly to native human MPO, since the observed response values were very low for this homolog. As a consequence, the data were not normalized. Comparatively weak binding to human MPO was observed for SPIN-chromogenes and SPIN-sciuri, so these are not presented here in the interest of space.

Table 2.

Binding Analysis of SPIN Homologs to Native Human MPO as Determined by SPR^* and IC₅₀ Values as Determined from MPO Activity Assay^**.

SPIN	ka (M−1s−1) × 104	kd (s−1) × 10−3	KD (nM)	IC50 (nM)
aureus	20.10 ± 2.22	3.16 ± 0.04	15.9 ± 2.1	4.6 (2.7 – 7.8)
agnetis	5.64 ± 0.14	2.36 ± 0.10	41.8 ± 1.1	-
schleiferi	1.12 ± 0.14	2.52 ± 0.17	230 ± 42.1	261.2 (153.5 – 444.2)
delphini	1.08 ± 0.08	3.34 ± 0.17	310 ± 42.5	29.7 (23.0 – 51.4)
intermedius	0.35 ± 0.08	3.34 ± 0.12	984 ± 274	221.7 (159.1 – 308.9)

^*Values represent the mean plus or minus the standard deviation obtained from replicate injections across three independent flow cells derivatized with native human MPO.

^**Values represent the best fit value obtained from three replicate experiments with the 95% confidence interval provided in parenthesis.

We previously reported that SPIN-aureus binds immobilized native human MPO with an apparent K_D of 10±0.6 nM [18]. Replicate measurements performed during this present study yielded a similar value of 15.9±2.1 nM (Table 2), which indicated a high level of precision for the SPR interaction assay across several independent experiments. We found considerable variability in the interactions between human MPO and the SPIN homologs, however (Fig 2 and Table 2). Although no SPIN homolog bound MPO as well as SPIN-aureus, SPIN-agnetis displayed the second highest affinity for MPO, as judged by its K_D of 41.8±1.1 nM. The affinity of the remaining homologs ranged from 230±42.1 nM for SPIN-schleiferi to 984±274 nM for SPIN-intermedius. These values represent approximately 14.5 and 61.9-fold decreases in affinity for human MPO, respectively, when compared to SPIN-aureus. We also noted that all of the SPIN homologs that bind human MPO had similar dissociation rate constants (k_d), which ranged from 2.36×10⁻³ to 3.34×10⁻³ s⁻¹ as a group. Conversely, the association rate constants (k_a) for binding to MPO varied widely, from a high of 20.1×10⁴ M⁻¹s⁻¹ for SPIN-aureus to a low of 0.354×10⁴ M⁻¹s⁻¹ for SPIN-intermedius. Thus, nearly all of the differences in affinity for human MPO displayed by these SPIN homologs are attributable to variations in their association rate constants.

Our co-crystal structure of SPIN-aureus/rhMPO indicates that SPIN acts as a molecular plug to block solute exchange to and from the MPO active site. [18]. However, subsequent studies on a series of site-directed and deletion mutants of SPIN-aureus revealed that binding to MPO is necessary, but not sufficient for inhibiting its activity [19]. We therefore sought additional information on whether the SPIN homologs that bound MPO in the SPR assay could also block MPO function in an enzymatic activity assay. We used a colorimetric activity assay to examine the ability of each SPIN to inhibit MPO function in vitro (Fig 3). Each SPIN homolog was added to the assay in triplicate across a dilution series that spanned three orders of magnitude. Consistent with our previous work, we found that SPIN-aureus significantly inhibited MPO activity under these conditions (IC₅₀ ~8 nM) [18, 19]. Conversely, we also found that all SPIN variants that failed to bind human MPO (i.e. SPIN-sciuri, SPIN-chromogenes, and SPIN-pseudintermedius) also failed to inhibit its activity.

Fig. 3.

A Subset of SPIN Homologs Inhibit the Enzymatic Activity of Native Human MPO. The enzymatic activity of native human MPO was investigated across a dilution series of various recombinant SPIN homologs. Initial reaction velocities were determined in triplicate at each concentration point prior to curve fitting and determination of IC₅₀ values. Whereas SPIN-delphini, SPIN-schleiferi and SPIN-intermedius all inhibit MPO in a dose-dependent manner, SPIN-agnetis fails to do so even though it binds with low-nanomolar affinity to native MPO. SPIN-aureus is included as a control for inhibition [18]. Bars express the mean plus or minus the standard deviation (n=3). A legend is inset.

Out of the homologs that remained, we found that three retained a clear inhibitory capacity against human MPO. Whereas SPIN-intermedius and SPIN-schleiferi inhibited MPO with IC₅₀ values of ~200 nM, SPIN-delphini displayed an approximately 6.7-fold more potent IC₅₀ value of ~30 nM. These differences in IC₅₀ values were somewhat unexpected, given that these three SPIN homologs are part of the same phylogenetic clade and share over 60% identity with one another (Fig. 1A and Table 1). The most surprising results were obtained for SPIN-agnetis. This protein failed to significantly inhibit MPO at concentrations up to 2 µM, even though it binds human MPO with a K_D of 41.8 nM as judged by SPR (Fig. 2 and Table 2). In this regard, the functional properties of SPIN-agnetis most closely resemble those of a site-directed mutant in SPIN-aureus where the conserved sidechains of H⁴³, D⁴⁴, and D⁴⁵ were simultaneously changed to alanine [19]. This mutant also failed to inhibit MPO despite the fact that it binds MPO with a K_D near 30 nM.

The Crystal Structure of SPIN-delphini Bound to Recombinant Human MPO at 2.4 Å Resolution

We required additional structural information on rhMPO in complex with a tightly binding SPIN homolog in order to better understand the similarities and differences among these newly identified SPIN family members. We succeeded in growing single crystals of SPIN-delphini bound to rhMPO that diffracted synchrotron X-rays to 2.4 Å limiting resolution, and solved and refined this structure to R_work/R_free values of 18.5 and 21.8%, respectively (Fig. 4A and Table 3^#). As expected, the structure of SPIN-delphini/rhMPO compares favorably to the structure of SPIN-aureus/rhMPO (Fig. 4B), as the 539 Cα positions from the two coordinate sets that align with one another superimpose with an RMSD of 0.289 Å. When we compared only the structures of rhMPO-bound SPIN-aureus and SPIN-delphini to one another, the 61 Cα positions align with an RMSD of 0.812 Å (Fig. 4C). This high level of structural identity was preserved throughout the entirety of the models, save for some minor deviations at the proteins’ N and C-termini.

Fig. 4.

The Structural Basis for Inhibition of MPO by SPIN-delphini. (A) A 2.4 Å co-crystal structure of rhMPO (purple) bound to SPIN-delphini (orange). Proteins are represented as ribbon diagrams, while the covalently bound heme prosthetic group is drawn in ball-and-stick convention with carbon atoms in yellow, nitrogen in blue, oxygen in red, and the iron ion as a red sphere. Ordered Ca²⁺ and Cl ions are drawn as light blue and green spheres, respectively. (B) Superposition of the SPIN-aureus/rhMPO co-crystal structure (PDB code 5UZU [18]) onto the SPIN-delphini/rhMPO structure as shown in panel A. SPIN-aureus is drawn as a cyan ribbon. (C) Superposition of rhMPO-bound forms of SPIN-aureus (cyan) and SPIN-delphini (orange). The proteins are drawn in wire convention with the N-terminal β-hairpin at the right-hand side of the image. (D) F_o-F_c electron density map (green mesh contoured at 2.5σ) within the rhMPO active site after initial structure solution by molecular replacement. The proteins chains are colored identically to the panels above for clarity, while the locations of the Ca²⁺ and Cl ions are included for reference. (E) Properties of the heme prosthetic group in the final model of SPIN-delphini/rhMPO. The heme is drawn as in panel A, while noteworthy rhMPO sidechains (labels insert) are drawn as stick models (carbon atoms light purple). Note the inclusion of three covalent bonds as described in the text. 2F_o-F_c electron density (blue mesh contoured at 1.3σ) is shown as an indicator of model to map correlation. (F) Visualization of the SPIN binding site on MPO in the context of the native human MPO dimer. Two copies of the SPIN-delphini/rhMPO structure were superimposed on native human MPO (PDB code 1CXP [35]). MPO is drawn in purple as a molecular surface, while SPIN-delphini is drawn as an orange, space-filling mesh. The heme is drawn as in panel A, with the exception that the iron ion is colored green to enhance contrast.

For reasons that remain unclear, our crystal structures of rhMPO bound to either full-length or an N-terminally deleted SPIN-aureus are characterized by low occupancy of the heme prosthetic group [18, 19]. Absence of the heme was not required for SPIN binding though, as full-length and N-terminally deleted SPIN-aureus bind equally well to both rhMPO and native MPO that had been isolated from purulent human sputum [18, 19]. Interestingly, examination of the F_o-F_c electron density map calculated after initial placement of the polypeptide models in the SPIN-delphini/rhMPO asymmetric unit revealed an obvious planar density at contour levels up to 3σ within the rhMPO active site (Fig. 4D). Following model building and refinement, our final SPIN-delphini/rhMPO coordinates have full occupancy of the heme prosthetic group with a median B-factor of 57.6 Ų. This value is consistent with the structure as a whole, whose all atom mean B-factor is 56.8 Ų (Table 3).

The final 2F_o-F_c electron density maps were consistent with the existence of four covalent bonds between SPIN-delphini bound rhMPO and the heme group (Fig. 4E). However, we included only three of these four covalent bonds in our final model: H⁵⁰² to heme Fe at 2.51 Å, M⁴⁰⁹ to heme CBB at 1.95 Å, and D²⁶⁰ to heme CMD at 1.58 Å. The final covalent bond between E⁴⁰⁸ and heme CMB was visible as contiguous density within the 2F_o-F_c map at levels up to 1.4 σ, but at 2.16 Å fell outside an acceptable bond length for this linkage (c.f. 1.55 Å for the corresponding bond in PDB entry 1CXP [35]). We believe this discrepancy likely resulted from the lower positional certainty of coordinates at 2.4 Å resolution when compared to higher resolution structures, such as the 1.8 Å resolution structure of halide-bound native human MPO referenced above [35]. These minor differences aside, the structure of SPIN-delphini/rhMPO presented here demonstrates that binding of SPIN family proteins is not mutually exclusive with high occupancy of the heme prosthetic group in rhMPO.

Native MPO exists as a ~150 kDa dimer comprised of two heavy and two light chains, respectively [19]. Moreover, crystal structures of native MPO from both human [35] and canine [36] sources revealed the presence of dimers within the asymmetric unit. The crystals of SPIN-aureus bound to rhMPO contain only a single copy of each polypeptide in the asymmetric unit [18], however, and examination of the protein-protein interfaces created by crystallographic symmetry operators showed no dimeric features equivalent to those found in native MPO. Similarly, the crystals of SPIN-delphini/rhMPO likewise contain only a single copy of the complex within the asymmetric unit (Fig. 4A), and there was no dimer equivalent to native MPO visible via crystallographic symmetry operations. Despite the fact that both SPIN-aureus and SPIN-delphini crystallized with rhMPO as binary complexes, we found that each of these structures is still compatible with the dimeric assembly found in native MPO (Fig. 4F). Superposition of both sets of coordinates onto the structure of dimeric native MPO revealed no obvious steric clashes that would prevent binding of either SPIN to such an arrangement. In fact, this is consistent with analytical gel filtration chromatography studies that show clear evidence for binding of SPIN-aureus to native human MPO (Fig. S1). Thus, the differences in oligomerization state observed in our structural studies most likely resulted from the rhMPO we used in this work, rather than reflecting bona fide consequences of SPIN binding to MPO.

Sequence-Dependent Structural Differences Between SPIN-delphini and SPIN-aureus Offer an Explanation for Their Differences in Affinity for Human MPO

Whereas SPIN-delphini and SPIN-aureus form complexes with rhMPO that closely resemble one another (Fig. 4B), the affinities of these proteins for MPO differ by nearly 20-fold at 310 nM and 15.9 nM, respectively (Table 1). Since SPIN-delphini shares 53% identity with SPIN-aureus, we considered whether the differences in affinities observed by SPR were due to alterations of key sidechains found at the SPIN/rhMPO interfaces. We analyzed the SPIN-delphini/rhMPO complex using the PISA server [37] and used the output to construct both an interface map (Fig. 5A) and a list of likely intermolecular interactions (Table 4). We then compared these data from SPIN-delphini/rhMPO to analogous results obtained for SPIN-aureus/rhMPO [18] (Table 4).

Fig. 5.

Molecular Analysis of the SPIN-delphini/rhMPO Interface and Comparsion with SPIN-aureus/rhMPO. (A) Representation of the SPIN-delphini/rhMPO structure highlighting the residues comprising the SPIN-delphini binding site on rhMPO. SPIN-delphini is shown as an orange ribbon, while rhMPO is drawn as a molecular surface with non-interacting residues in purple and interfacing residues in grey. The heme prosthetic group is drawn as in Fig. 4 for reference. (B) The structure of SPIN-delphini drawn as an orange ribbon. Sidechains found at the rhMPO interface are drawn in stick convention with the carbon atoms colored grey. Selected sidechains are individually labeled. Note that this representation of SPIN-delphini is rotated approximately 180° in the plane of the page relativ e to panel A. (C) Polar interactions between SPIN-aureus sidechains (cyan) and N³⁵² of rhMPO (blue). Likely interactions (cyan dashes) and their corresponding distances are labeled. (D) Polar interactions between SPIN-delphini sidechains (orange) and N³⁵² of rhMPO (blue). Likely interactions (orange dashes) and their corresponding distances are labeled. Grey dashes represent interactions not formed due to excessive distance between donor and acceptor groups. (E) Sequence comparison between SPIN-aureus (numbering above) and SPIN-delphini (numbering below) with identical residues shown reverse blue typeface. The secondary structure of rhMPO-bound SPIN-aureus is displayed above the alignment. Identical residues found at the rhMPO interface of each complex that form non-polar interactions are indicated by a cyan circle, while those that form polar interactions are indicated by a magenta circle. Residues invariant across all SPIN homologs and found at the SPIN/rhMPO interface are designated with an orange circle. Note the participation of invariant residues D⁴⁵, H⁵¹, and Y⁷⁵ (per SPIN-aureus numbering) in interactions with N³⁵² of rhMPO, and how this varies in SPIN-delphini by comparing panels C and D above.

Table 4.

Comparative Interface Analysis of SPIN-delphini/rhMPO and SPIN-aureus/rhMPO.

SPIN-delphini	Bond type	SPIN-a ureus	Bond type	MPO residue
		33 LYS	HS		GLU346
29 VAL		34 VAL
30 THR		35 TYR	H		GLU346
31 SER	H	36 SER	H	GLU268
32 GLN	H	37 GLN		THR265
33 ASN	H	38 ASN	H	GLU268
34 GLY		39 GLY
35 ILE		40 LEU
36 ILE		41 VAL
37 LEU		42 LEU
38 HIS	S	43 HIS	HS	ASP380
39 ASP		44 ASP	HS*		ARG272*
40 ASP		45 ASP	H		ASN352
42 ARG	H	47 ASN	H	ASN381
		47 ASN	H		HIS383
43 MET	H	48 PHE	H	ASN381
44 LEU		49 LEU
45 ASP	S	50 GLU	HS	ARG368
46 HIS	H	51 HIS	H	ASN352
47 GLU
49 GLN		54 SER
50 TYR	H	55 TYR	H	MET356
		55 TYR	H		ARG354
53 VAL		58 VAL
62 GLN
63 THR		68 THR
66 ARG		71 ASN
67 LEU		72 LEU
70 TYR	H	75 TYR	H	ASN352
		78 ASP
74 GLN		79 LYS	H		ASN352
		93 GLN
89 ASP	H	94 ASP	H	ASN381
		95 GLY
91 LEU		96 PHE
92 ASP

We determined that the total number of inhibitor-derived residues at the SPIN-delphini/rhMPO interface (30) is nearly identical to that of SPIN-aureus/rhMPO (31) (Fig. 5A–B). Consistent with the overall identity between these two proteins, however, only 17 interfacing residues are identical between SPIN-delphini and SPIN-aureus. Given that even minor changes at the interface might result in altered surface compatibility between rhMPO and the two SPINs, we analyzed the surface complementarity of both structures using the program s.c. [38]. Both SPIN-delphini/rhMPO and SPIN-aureus/rhMPO have similar overall s.c. coefficients of 0.715 and 0.704, respectively; these values are slightly lower than those of other high-affinity protein/protein interactions relevant to staphylococcal innate immune evasion, such as that of the neutrophil serine protease inhibitor, EapH1, bound to neutrophil elastase (s.c. = 0.77) [17]. Nevertheless, since previous work had shown that the SPIN-aureus/MPO interaction primarily arises via the α-helical bundle region of the inhibitor [19], we recalculated the s.c. coefficients of these structures in the absence of the N-terminal β-hairpin motifs. This yielded an s.c. value of 0.715 for SPIN-delphini/rhMPO, but of 0.774 for SPIN-aureus/rhMPO. We considered the rather large difference between these two s.c. values as evidence that the α-helical bundle region SPIN-aureus is far better suited to packing tightly against rhMPO than is SPIN-delphini. Consistent with this prediction, SPIN-delphini buries only 1365 Ų of its surface area when bound to rhMPO. This value is only ~85% of that previously reported for SPIN-aureus [18].

We also found that the number of polar interactions (i.e. hydrogen bonds and salt bridges) formed with rhMPO is considerably smaller for SPIN-delphini than for SPIN-aureus. Of the 31 residues found at the interface between SPIN-aureus and rhMPO, 14 appeared to participate in one or more polar contacts with groups from rhMPO. By contrast, of the 30 residues found at the SPIN-delphini/rhMPO interface, only 11 were judged to form polar contacts with rhMPO. Whereas seven of these residues are identical between the two SPINs and deemed to form equivalent interactions (i.e. S³¹, N³³, H³⁸, H⁴⁶, Y⁵⁰, Y⁷⁰, and D⁸⁹ per SPIN-delphini numbering), the other conserved position (i.e. D⁴⁰) fails to form a hydrogen bond in the SPIN-delphini/rhMPO structure even though it does so in the SPIN-aureus/rhMPO structure. Furthermore, three other positions in SPIN-delphini (i.e. Q²⁸, T³⁰, and Q⁷⁴) represent non-conservative changes when compared to SPIN-aureus and thereby disrupt polar contacts.

Since only the Q⁷⁴ from SPIN-delphini resides in the α-helical bundle, we surmised that this change from K⁷⁹ in SPIN-aureus is likely the most significant alteration in terms of MPO binding. In that regard, we noted that the residues D⁴⁵, H⁵¹, and K⁷⁹ of SPIN-aureus (corresponding to SPIN-delphini D⁴⁰, H⁴⁶, and Q⁷⁴) are all involved in hydrogen bonding to groups from position N³⁵² of rhMPO (Fig. 5C–D). In the case of SPIN-aureus D⁴⁵ versus SPIN-delphini D⁴⁰, the observed distance between acceptor and donor is 3.7 Å in SPIN-aureus/rhMPO compared to 4.2 Å in SPIN-delphini/rhMPO. A similar situation is found for SPIN-aureus K⁷⁹ versus SPIN-delphini Q⁷⁴, where the observed distance to the N³⁵² sidechain is 3.2 Å compared to 5.6 Å. Although both SPIN-aureus and SPIN-delphini utilize a conserved histidine sidechain (i.e. H⁵¹ and H⁴⁶, respectively) to form hydrogen bonds of ~3.0 Å distance to the backbone oxygen from N³⁵² in rhMPO, disruption of the other interactions with the N³⁵² sidechain as described above may be a contributing factor to the reduced affinity of SPIN-delphini for MPO. In summary, while SPIN-delphini maintains an ability to bind human MPO and inhibit its activity (Figs. 2–3), the quantitative differences between SPIN-delphini and SPIN-aureus vis-à-vis affinity for and inhibition of MPO appear to be predicated on sequence/structure changes that affect both the overall shape and positioning of functional groups at the MPO binding site of these closely related proteins (Fig. 5E).

DISCUSSION

Though neutrophils serve many essential roles in the innate immune response, the significance of their contributions in preventing bacterial infection is underscored by the manifold strategies S. aureus deploys to evade opsonophagocytosis and efficient intracellular killing by these leukocytes [13]. In this regard, the recently identified inhibitor of MPO, SPIN, is one component of a multipartite S. aureus evasion program that acts within the maturing phagosome. Although investigations of SPIN-aureus have provided vital information on this novel protein [18, 19], the studies we present here are an important step toward a broader understanding of the structure/function relationships of the SPIN protein family. Herein, we have identified eight different SPIN homologs (Fig. 1), shown that four homologs bind human MPO (Fig. 2), and further demonstrated that three of these block enzymatic activity of human MPO (Fig. 3). We have also determined a 2.4 Å co-crystal structure of SPIN-delphini bound to rhMPO (Fig. 4), thus allowing the first comparative structural analyses between SPIN proteins in complex with rhMPO (Fig 5).

Our work has confirmed that functional SPIN proteins are distributed amongst several bacterial species closely related to S. aureus. However, this has also opened up avenues for further investigation. For example, whereas S. aureus is a known pathogen of both humans and livestock (e.g. dairy cattle), several staphylococcal species that encode a SPIN-like sequence (e.g. the Staphylococcus intemedius group [32]) are primarily veterinary pathogens. We have previously shown that SPIN-aureus exhibits a pronounced binding selectivity and inhibitory capacity toward MPO from known hosts of this organism [18]; this feature is perhaps best illustrated by tight binding to and inhibition of human MPO by SPIN-aureus, but an apparent lack of interaction with or inhibition of murine MPO [18]. Consequently, it seems reasonable that several of the SPIN homologs that failed to bind or inhibit human MPO (Fig. 2–3) might instead display a similar selectivity for MPO proteins derived from their preferred hosts. While additional studies will be necessary to determine whether or not this is the case, an experimental approach similar to those we employed here would be well-suited to addressing this issue.

Separately, our comparative analysis of the SPIN-aureus/rhMPO and SPIN-delphini/rhMPO complexes has allowed for a more thorough understanding of SPIN/MPO interactions at the sequence/structure level. First, we found no substantially greater level of conservation of the residues at the SPIN/MPO interface (i.e. 17/30 or 17/31 = ~56%) when compared to the two proteins overall (53%) (Figs. 4–5 and Table 1). This result was somewhat surprising, given that SPIN-delphini still maintains a relatively robust interaction with human MPO (K_D = 310 nM) (Fig. 2 and Table 2). Second, and along these lines, we noted that the numerous changes in SPIN-delphini when compared to SPIN-aureus manifest themselves as a notably lower overall surface complementary for rhMPO and buried surface area within the SPIN-delphini/rhMPO complex. This is particularly true within the α-helical bundle region of the SPIN protein, which drives interaction with MPO [19]. Though it is difficult to infer or test experimentally which of these subtle changes was most detrimental to a highly optimized and cooperative system like a protein/protein interaction, we believe that the cumulative effects of these alterations must have impacted the observed affinity of the complexes (Fig. 2 and Table 2). Finally, we found that the SPIN-delphini/rhMPO interface contains a significantly smaller number of polar interactions than does SPIN-aureus/rhMPO (Fig. 5 and Table 4). While some of the differences are due to loss of specific sidechains (e.g. SPIN-delphini Q⁷⁴ vs SPIN-aureus K⁷⁹), others are due to unfavorable distances in the SPIN-delphini/rhMPO structure as compared to SPIN-aureus/rhMPO (e.g. SPIN-delphini D⁴⁰ vs SPIN-aureus D⁴⁵). It seems then that changes in both shape complementarity and presentation of key polar groups at the MPO interface are contributing factors to the distinct MPO-binding and inhibitory properties of SPIN-aureus and SPIN-delphini.

SPIN is not the only staphylococcal innate immune evasion protein that shows strong selectivity for a target molecule from a biologically-relevant host of S. aureus [18]. Species selectivity has also been described for the S. aureus SCIN family of proteins [39], as SCINs block activity of the alternative complement pathway in humans but not mice [39, 40]. The bi-component pore forming leukocidins of S. aureus and related staphylococci likewise show strong preference for interaction with cellular receptors from specific species, thereby explaining their differential effects on target leukocytes from various hosts [41]. Unfortunately, our current understanding of this selectivity is limited primarily to qualitative descriptions of binding and function. While these considerations are clearly useful, we believe that there must also be specific structural and bio/physical-chemical features which underlie such marked host selectivity. This is likely to be true for not only S. aureus innate immune evasion proteins, but for any molecules involved in pathogen/host interactions in a much broader sense. Considering that multiple SPIN sequences are now available and that quantitative biochemical, functional, and comparative structural information on these SPINs bound to MPO from various host species can in theory be obtained, we suggest that the SPIN/MPO interaction can be developed into a valuable model system for understanding the structure/function principles that underlie host specific evolution of virulence proteins.

HIGHLIGHTS.

• S. aureus secretes a protein, SPIN, that potently inhibits human Myeloperoxidase

• A number of closely related staphylococci also encode SPIN homologs

• SPIN homologs vary in their binding to and inhibition of human Myeloperoxidase

• SPIN from S. delphini has been co-crystallized with recombinant human MPO

• Comparison of SPIN co-crystal structures explains differences in binding behavior