Antimicrobial Peptide Recognition Motif of the Substrate Binding Protein SapA from Nontypeable Haemophilus influenzae

Abstract

Nontypeable Haemophilus influenzae (NTHi) is an opportunistic pathogen associated with respiratory diseases, including otitis media and exacerbations of chronic obstructive pulmonary disease. NTHi exhibits resistance to killing by host antimicrobial peptides (AMPs) mediated by SapA, the substrate binding protein of the sensitivity to antimicrobial peptides (Sap) transporter. However, the specific mechanisms by which SapA selectively binds various AMPs such as defensins and cathelicidin are unknown. In this study, we report mutational analyses of both defensin AMPs and the SapA binding pocket to define the specificity of AMP recognition. Bactericidal assays revealed that NTHi lacking SapA are more susceptible to human beta defensins and LL-37, while remaining highly resistant to a human alpha defensin. In contrast to homologues, our research underscores the distinct specificity of NTHi SapA, which selectively recognizes and binds to peptides containing the charged-hydrophobic motif PKE and RRY. These findings provide valuable insight into the divergence of SapA among bacterial species and NTHi SapA’s ability to selectively interact with specific AMPs to mediate resistance.

Introduction

NTHi is a Gram-negative opportunistic pathogen commonly associated with middle ear infections (otitis media) and respiratory diseases such as chronic obstructive pulmonary disease, acute sinusitis, and pneumonia.¹⁻⁶ The production of AMPs in response to bacterial infection serves to limit infection due to AMP bactericidal activity.^2,7 However, several pathogens resist AMPs through various mechanisms including the Sap transporter.⁸⁻¹⁴ The Sap transporter is part of the Peptide, Opine, and Nickel Uptake Transporter (PepT) family of ATP-binding cassette (ABC) transporters which use ATP hydrolysis to import peptides and heme.¹⁵⁻²¹ Structurally, the Sap transporter consists of two transmembrane domains (TMDs), SapBC, that form the translocation pathway, two highly conserved nucleotide binding domains (NBDs), SapDF, and a substrate-binding protein (SBP), SapA. SapA binds, then delivers the AMPs to the cognate Sap transporter for subsequent degradation in the cytoplasm.¹⁴ This mechanism circumvents accumulation of AMPs in the cell membrane, preventing lysis and subsequent cell death.²²

AMPs are conserved across species and selectively target pathogen cell membranes to disrupt membrane integrity.²³⁻²⁵ The two main classes of AMPs in humans, defensins and cathelicidins, are positively charged amphipathic peptides consisting of 20 to 50 amino acids.^26,27 Defensins are cysteine-rich AMPs with three intramolecular disulfide bridges, forming a conserved γ-core motif composed of a β-hairpin in an antiparallel β-sheet.^26,28 Additionally, each defensin includes an unstructured loop between Cys-2 to Cys-4 which contains charged and hydrophobic residues important for membrane interaction (Figure 1).²⁹⁻³¹ The alpha helical peptide LL-37, the only member of the cathelicidin family found in humans, is produced by a variety of cell types, including the epithelial cells of the airway.²⁷ Defensins, including α-defensin 1 (hNP-1), β-defensin 2 (hBD-2), and β-defensin 3 (hBD-3), are prevalent in the epithelial cells of mucosal surfaces including the nasopharynx and are considered potent AMPs in response to NTHi infections.^7,32−38

Figure 1.

Topology of common AMPs at human mucosal surfaces. Cartoon representations of each AMP (left) with loops (cyan), alpha helices (red), beta sheets (purple), and disulfide bonds (yellow) depicted. The space filling representation (right) is colored by hydrophobicity, with hydrophobic residues (red) and hydrophilic residues (white). The sequences are displayed with cysteine residues highlighted in yellow with the net charge on the right. The AMPs shown are as follows: hBD-3 (A), hBD-2 (B), hNP-1 (C), and LL-37 (D).

In NTHi clinical isolates, exposure to AMPs increases expression of SapA.¹⁴ Moreover, when SapA is absent, NTHi becomes more susceptible to the killing effects of AMPs, highlighting the importance of SapA in virulence.^14,18,22 While the structure of SapA has been determined by X-ray crystallography here and elsewhere, unlike other related PepT SBPs, there are no structures of peptide-bound SapA or comprehensive biophysical characterization of SapA in relation to peptide binding.^{17,21,39−53} Understanding the molecular details of peptide recognition is critical due to the essential role of SapA as a virulence factor in NTHi, involved in not only AMP resistance but also heme uptake, biofilm formation, and host cell invasion.^{14,18−20,54,55}

In this study, the objective was to expand our understanding of how SapA confers resistance to human defensins and LL-37 and to examine the specificity of these interactions. To accomplish this objective, we conducted a comprehensive peptide sequence analysis, including point mutations and truncations of defensins, to identify the residues within the peptide necessary for binding to NTHi SapA. Bioinformatic analysis and mutagenesis of NTHi SapA revealed the key residues in the SapA binding cavity that are critical for peptide binding. Finally, we investigated the conservation of peptide binding residues in NTHi SapA compared with known homologues, identifying domains of low conservation within the binding pocket that highlight diversity in SapA AMP substrates across different species.

Materials and Methods

Peptides

Recombinant hNP-1, hBD-2, hBD-3 (PeproTech), and LL-37 (Phoenix Pharmaceuticals and Anaspec) and synthetic hNP-1 and hBD-2 (Peptide International) were purchased commercially. Custom peptides (>95% pure) were purchased from GenScript (Table S1). All peptides were solubilized according to manufacturer recommendations and stored at 1 mM or according to manufacturer specifications at −80 °C for long-term storage. To evaluate sample quality, dynamic light scattering measurements were conducted with LL-37 at 1 mg/mL, revealing that the LL-37 was aggregated at high concentrations (data not shown).

Cloning, Protein Expression, and Purification

SapA (UniProt Accession ID: Q4QL73) was amplified from the genomic DNA of the clinical strain of NTHi 86–028NP and cloned into the T7 expression vector pMCSG7 by ligation-independent cloning to create a TEV cleavable N-terminal His₆-tagged SapA (residues 32–546) construct.⁵⁶ The plasmid was verified by DNA sequencing (ACGT Inc.). SapA point mutants were generated through site-directed mutagenesis following the manufacturer’s protocol (New England Biolabs) (Table S2).

Recombinant SapA was expressed in Escherichia coli strain BL21(DE3) and purified through nickel-nitrilotriacetic acid affinity chromatography as previously described.¹⁷ All buffers were prepared from the stock chemicals. The cells were lysed in Buffer A (25 mM HEPES at pH 8, 500 mM NaCl, and 15 mM imidazole at pH 8). To improve sample purity, the concentration of imidazole was increased in a wash step with Buffer B (25 mM HEPES pH 8, 500 mM NaCl, 60 mM imidazole pH 8) before elution into Buffer C (25 mM HEPES pH 7.5, 500 mM NaCl, 250 mM imidazole pH 8). The protein was dialyzed overnight in buffer D (25 mM HEPES at pH 7.5, 300 mM NaCl) supplemented with 5 mM BME and 200 μg/mL TEV protease. The cleaved protein was loaded onto a Ni-NTA affinity chromatography column and washed with 8 CV of buffer E (25 mM HEPES pH 7.5, 300 mM NaCl, 60 mM imidazole pH 8) followed by 8 CV of buffer F (25 mM HEPES pH 7.5, 300 mM NaCl, 250 mM imidazole pH 8). The flowthrough and buffer E wash were concentrated with a 50 kDa Amicon Ultra concentrator (15 mL) to a final volume of 6 mL by centrifugation for 10 min intervals at 4000 rpm (3724g), clarified by centrifugation, and injected on a HiLoad 16/600 Superdex 200 size-exclusion chromatography column (GE Healthcare) equilibrated with buffer D. Protein concentration was determined using a nanodrop spectrophotometer (NanoDrop-1000, Thermo Fisher Scientific). SapA mutant proteins were purified as above with the addition of glycerol to the buffers; Buffer A was supplemented with 20% glycerol, buffers B and C with 10% glycerol, and buffer D with 5% glycerol.

Expression constructs for E. coli NikA (UniProt Accession ID: P33590), NTHi OppA (UniProt Accession ID: Q4QLH0), and NTHi HbpA (UniProt Accession ID: Q4QM48) were prepared as previously described and expressed and purified following the previously published protocol.¹⁷

Protein Crystallization

Sitting-drop vapor diffusion and sparse-matrix methods were used to identify potential crystallization conditions for SapA. An initial hit was found in 0.1 M HEPES pH 7.5 and 1.3 M ammonium sulfate at 22 °C. An additive screen (Hampton Research, Cat. No. HR2–428) was used to optimize the initial SapA hit. Using the 2 M sodium thiocyanate additive improved SapA crystal formation, and diffraction-quality bipyramidal crystals were grown overnight at 22 °C. Crystals typically measured about 50 μm × 50 μm × 100 μm. Drops were set up with 1 μL of SapA (15 mg/mL, 25 mM Tris pH 9.5, 150 mM NaCl), 1 μL of crystallization condition (0.1 M Tris pH 9.4, 1.5 M ammonium phosphate), and 0.2 μL of additive condition (2 M sodium thiocyanate) and equilibrated against 500 μL of the crystallization condition in the reservoir. Crystals were briefly soaked in cryoprotectant (0.1 M Tris pH 9.4, 1.8 M ammonium phosphate, and 15% v/v glycerol) and flash-cooled in liquid nitrogen. The crystals were subject to radiation damage, and data was obtained using two crystals from the same drop to collect a complete data set in the outer shell.

Structure Determination of SapA

Crystal diffraction data for SapA was collected at the Advanced Photon Source (Argonne, IL) on LS-CAT beamline 21-ID-G with an MX300 CCD detector (Rayonix). The data were integrated using XDS⁵⁷ and scaled with AIMLESS (CCP4 suite).⁵⁸ The molecular replacement pipeline Balbes was used to build the initial model of SapA.⁵⁹ Model building was improved with ARP/warp from the online CCP4 platform.⁶⁰ The final models were edited in Coot⁶¹ and refined with REFMAC.⁶² The cartoon representations of the solved protein structures were created in PyMOL v2.0 (Schrödinger, LLC). Data collection and refinement statistics of the SapA structure are summarized in Table 2.

Table 2. Data Collection and Refinement Statistics^a.

Parameter, Units		SapA (PDB ID 8TV8)
Statistics
Space group		P41212
Data Collection
Unit Cell Dimensions, Å	a	119.75
	b	119.75
	c	129.59
Unit Cell Angles, °	α	90.0
	β	90.0
	γ	90.0
Resolution	Å	43.97–2.25 (2.32–2.25)
Wavelength	Å	0.9786
Completeness	%	99.9
Rmerge		0.089 (0.718)
Average I/σI		16.1 (4.19)
Redundancy		19.2
Total reflections		869659
Unique reflections		45307
Refinement
Rwork/Rfree		0.2050/0.2330
Number of atoms	All atoms	4266
	Protein	4152
	Water	114
Average B-factor, Å2	All atoms	57.0
	Protein	60.3
	Solvent	48.8
RMSD	Bond lengths, Å	0.009
	Bond angles, °	1.691
Ramachandran statistics	Favored, %	96.6
	Allowed, %	3.0
	Outliers, %	0.4

^aData for the highest-resolution shell are given in parentheses.

Surface Plasmon Resonance (SPR)

Single-cycle kinetic experiments were used to determine the defensin AMP binding affinity of SapA. Defensin AMPs were coupled via a standard amine-coupling method in flow channels 2, 3, and 4 on a CM5 sensor chip (GE Healthcare); 300–1000 RU (response units) of each AMP was immobilized. Flow channel 1 was designated as the control channel. The surface of the chip was equilibrated in running buffer (25 mM HEPES (pH 7.5, 150 mM NaCl, 0.1% Tween 20) at a flow rate of 30 μL/min for at least 3 h. SapA was dialyzed in 25 mM HEPES pH 7.5 and 500 mM NaCl and diluted into 25 mM HEPES pH 7.5, 150 mM NaCl, and 0.1% Tween 20 before serial dilution. Analyte injections of SapA were prepared by 2-fold serial dilutions. The analyte samples were consecutively injected by increasing concentrations over all four channels, followed by a final dissociation step. Experiments were run at a flow rate of 30 μL/min at 25 °C. The sensor surface was regenerated after each experiment with four 30 s injections of running buffer, supplemented with 0.1% SDS, at a flow rate of 50 μL/min.

AMP loop peptides were amine-coupled with 300–400 RU of each peptide immobilized using the above method. The samples were treated with the same conditions as previously mentioned with one exception: two injections were used to regenerate the sensor surface instead of four. After collecting data under oxidizing conditions, the chip was equilibrated with reducing buffer (25 mM HEPES pH 7.5, 150 mM NaCl, 0.1% Tween 20, and 2 mM TCEP) for at least 4 h. The NTHi SapA analyte injections were repeated under reducing conditions.

For SapA mutant SPR experiments, hBD-3 was amine-coupled with 500–600 RU immobilized using the method described above. After immobilization, the chip was equilibrated using the same procedure as full-length defensins. 100 nM SapA WT and SapA mutants were injected across all four channels for a 3 min association followed by a 12 min dissociation step. Experiments were run at a flow rate of 30 μL/min at 25 °C. The chip was regenerated after each sample with three 20 s injections of 10 mM glycine pH 1.75 at a flow rate of 40 μL/min.

All experiments were performed with the Biacore T200 instrument (GE Healthcare) according to the manufacturer’s instructions. The data was obtained in triplicate. Kinetic rate constants and equilibrium dissociation constants were determined by fitting the data globally to the 1:1 two-state reaction model using the Biacore T200 evaluation software v3.0 (GE Healthcare).

In Silico Ligand Docking Studies

Ligand docking was conducted for the custom peptide fragments PKEEQ, RRYKQ, and RRYGT. Ligand pdb files were generated using the Avogadro chemical editor by inputting the amino sequence into the peptide builder feature.⁶³ Peptide–protein docking was modeled using the HPEPDOCK 2.0 web server.⁶⁴⁻⁶⁸ The crystal structure of SapA was utilized as a receptor file for molecular docking. Blind docking was guided by cavity prediction using the default setting on the KVFinder web server⁶⁹⁻⁷¹ for the substrate binding pocket specifying receptor binding residues Y291, Y166, and F549. Each peptide fragment had 1000 peptide conformations to represent peptide flexibility, with an output of 100 peptide binding modes for each docked peptide. The protein–ligand interactions for the highest scoring solutions for each peptide fragment were identified using the protein–ligand interaction profiler (PLIP) and displayed using PyMOL.⁷²

Protein Conservation Analysis

Homologues of SapA, HbpA, and OppA were collected using BLAST queries set to the nonredundant protein sequences database using the blastp algorithm for 1000 target sequences scored using BLOSUM62 and restricted to sequences with 35% identity.^73,74 A multiple sequence alignment (MSA) was generated in MEGA11 for the SapA data set trimmed for redundant sequences using MUSCLE alignment.^75,76 The MSA was analyzed using the Consurf web server with color-coded structure visualized in PyMOL.⁷⁷⁻⁸⁰ An MSA was generated in MEGA11 of 70 sequences with at least 60% identity and 85% query coverage including each homologue from Pasteurellaceae species using the default settings for MUSCLE alignment. A phylogenetic tree was generated using the Neighbor-Joining method with evolutionary distances computed using the p-distance method in MEGA11.⁸¹ An MSA using Clustal Omega was utilized to compare the sequences for SapA from NTHi, Haemophilus ducreyi, Actinobacillus pleuropneumoniae, and Salmonella enterica serovar Typhimurium, displayed in Jalview with the percentage identity color scheme.^82,83 A logo plot was generated using WebLogo 3 for SapA homologues and PepT SBPs.⁸⁴

Circular Dichroism (CD) Spectroscopy

Protein samples were exchanged into CD Buffer (12.5 mM Tris pH 7.5 and 150 mM NaCl) using PD MiniTrap G-25 Sample Preparation Columns (Cytiva), filtered with 0.1 μm centrifugal filters, and clarified by centrifugation. Samples were diluted to 0.5 mg/mL with the CD buffer. CD far-UV spectra were collected using a 1 mm quartz cuvette from 260 to 197 nm in the step scan mode, with 2 nm bandwidth, 2 s response time, and 1.0 nm step speed at 10 °C. Each spectrum is the accumulation of three scans. CD analysis was performed by using a J-815 CD spectrometer (JASCO).

Spectral Shift Assays

SapA was labeled using RED-NHS second-generation dye (NanoTemper) at a 1:3 protein to dye ratio in a labeling buffer of 300 mM NaCl, 25 mM HEPES pH 7.5 but otherwise following the provided product protocol. Protein concentration was quantified by Nanodrop and diluted to a working concentration of 40 nM with assay buffer (150 nM NaCl, 25 mM HEPES pH 7.5, 10 mM MgCl₂, and 0.1% Pluronic F-127). In a 384-well plate, 1.5-fold serial dilutions with a final volume of 10 μL per well were prepared in triplicate for each peptide fragment of interest. For weak binding peptides, the highest ligand concentration for the serial dilution was 100 or 500 μM, while for peptides with higher affinity, it was 100 nM; from the stock solutions, peptides were diluted with assay buffer. Then, 10 μL of 40 nM labeled SapA was added to the bottom of each well. Following a 1 min spin at 1500 rpm (524g), the reactions were incubated for 15 min at room temperature to reach equilibrium. Samples were loaded into capillaries, and the 650/670 nm fluorescence ratio was measured and reported using the Monolith X (NanoTemper). Equilibrium dissociation constants were determined by fitting the data using the K_d fit model on the Monolith X (Nanotemper).

Bactericidal Assays

Construction of the sapA deletion mutant strain (ΔsapA) was carried out using a recombineering approach developed for NTHi.⁸⁵ The NTHi clinical isolate strain 86–028NP was utilized to create the ΔsapA mutant.^86,87 The parental NTHI strain 86–028NP::rpsLA128G was constructed as previously described.⁸⁸ Bactericidal assay protocol was performed as previously described.¹⁸ In brief, bacteria were grown overnight on chocolate agar plates. NTHi were inoculated into prewarmed supplemented brain heart infusion (sBHI) medium, grown for 3 h at 37 °C in 5% CO₂ and then diluted with medium to 10⁸ CFU/mL. The bacteria were diluted further in 10 mM sodium phosphate pH 7.4 to a concentration of 10⁶ CFU/mL. AMPs were diluted to 10 μg/mL from stock concentrations. In a 96-well plate, defensin and cathelicidin AMPs were 2-fold serial diluted with 10 mM sodium phosphate pH 7.4 containing 1% sBHI medium. The bacterial solution was added to each well to a final concentration of 1 × 10⁵ CFU/mL, and the plate was incubated for 1 h at 37 °C in 5% CO₂. Bacteria survival was determined by serial dilution and growth on chocolate agar, and data were expressed as percent survival relative to bacteria not exposed to AMP.

Results

SapA Protects Against Antimicrobial Peptide Killing

Defensins and cathelicidins are important factors of the immune response against NTHi infection.^7,32,33,89 To understand how AMP production impacts NTHi viability in a SapA-dependent manner, the parent strain and the ΔsapA mutant were exposed to serial dilutions of hNP-1, hBD-2, hBD-3, and LL-37 to assess AMP bactericidal activity. As excess concentrations of AMP result in cell death, a range of concentrations based on the physiological range found in humans was chosen for the serial dilution.⁹⁰⁻⁹² The survival of the parent strain was most impacted by LL-37, hBD-3, and, to a lesser extent, by hBD-2, and remained unchanged in response to hNP-1 (Figure 2). The ΔsapA mutant exhibited a concentration-dependent decrease in survival in the presence of hBD-3 and LL-37 (Figure 2). Of note, no discernible difference in sensitivity was observed at 10 μg/mL hBD-3 and LL-37, as neither the parent strain nor the ΔsapA mutant survived at this high concentration. In contrast, the loss of SapA has no significant impact on the survival of NTHi in the presence of hBD-2 across all concentrations tested (Figure 2). At the concentrations tested, the ΔsapA mutant also lacked sensitivity to hNP-1, suggesting that NTHi resistance to hNP-1 is independent of NTHi SapA (Figure S1). The reported survival above 100% for both NTHi and the ΔsapA mutant indicates that the bacteria grow in the presence of sublethal concentrations of hNP-1 (Figure S1). These bactericidal assays reveal that SapA from NTHi is critical in mediating resistance to hBD-3 and LL-37 but not to hBD-2 or hNP-1.

Figure 2.

Antimicrobial activity of common AMPs is mediated by NTHi SapA. AMP bactericidal assays compared the percent survival of the parental strain and ΔsapA mutant strain in the presence of defensin AMPs and LL-37. NTHi survival was assessed after bacteria were incubated for 1 h with increasing concentrations (μg/mL) of (A) LL-37, (B) hBD-3, (C) hBD-2, or (D) hNP-1. The error bars represent data collected from three replicate experiments. Statistical significance was determined by two-tailed t test, ***, p = 0.0009 *, p < 0.02, N. S., not significant.

SapA Binds Defensin AMPs with High Affinity

To determine if NTHi resistance to AMPs is modulated by the affinity of SapA to these peptides, the interactions between NTHi SapA and immobilized AMPs (hBD-3, hBD-2, hNP-1, and LL-37) were investigated using SPR. The SPR data fit a two-state model, representative of either binding-induced conformational changes or heterogeneity in the protein conformation in solution.⁹³ SapA binds full-length defensins hBD-3, hBD-2, and hNP-1 with K_D values in the range of 4 to 17.5 nM (Figure 3, Table 1), revealing similar affinities. As a comparison, PepT SBPs OppA and HbpA, also from NTHi, exhibited no measurable binding specificity for hBD-3 (Figure S2). While the affinities are similar for each defensin, the kinetics of the interaction differ. The association rate for binding (k_a1) and disassociation rate for the SapA-defensin complex (k_d1) are similar for hBD-3, hBD-2, and hNP-1 (Table 1). However, the forward (k_a2) and reverse (k_d2) rate constants for the binding-induced conformational change vary. The increase in k_a2 for hBD-2 and the decrease in k_d2 for hNP-1 suggest a more persistent ligand-bound complex with hBD-2 and hNP-1 in comparison to hBD-3.

Figure 3.

NTHi SapA binds AMPs with high affinity. Kinetic analysis was performed with five analyte injections at increasing concentrations of NTHi SapA, noted on the sensogram, over an immobilized substrate. For the association phase, each analyte sample was injected in series with a final dissociation phase. Representative sensogram data are shown in red and fitted model drawn in black. NTHi SapA displays high affinity for the defensins (A) hBD-3, (B) hBD-2, and (C) hNP-1. SPR sensograms of AMP loop peptides (D) hBD-3_loop, (E) hBD-2_loop, and (F) hNP-1_loop demonstrate the ability of NTHi SapA to bind the loop region without a decrease in affinity.

Table 1. SapA Kinetic and Equilibrium Dissociation Constants for Defensin AMPs.

	ka1 (M–1 s–1)	kd1 (s–1)	ka2 (s–1)	kd2 (s–1)	KD (nM)
Full-length defensin AMP
hBD-3	7.1 (5.4) × 105	1.9 (1.2) × 10–2	5.9 (1.4) × 10–3	2.2 (0.3) × 10–3	10.5 (5.3)
hBD-2	6.3 (9.6) × 105	4.7 (7.1) × 10–2	1.3 (0.8) × 10–2	2.6 (0.1) × 10–3	17.5 (10.1)
hNP-1	3.6 (1.5) × 105	1.2 (0.3) × 10–2	6.1 (1.9) × 10–3	6.9 (1.6) × 10–4	4.0 (1.1)
Oxidized: disulfide bridge
hBD-3 loop	1.0 (0.6) × 106	3.3 (0.9) × 10–2	4.0 (1.6) × 10–3	3.0 (2.1) × 10–3	17.2 (12.0)
hBD-2 loop	7.5 (1.1) × 105	1.9 (0.1) × 10–2	2.4 (0.2) × 10–3	1.4 (0.2) × 10–3	9.5 (2.0)
hNP-1 loop	6.0 (1.0) × 105	2.3 (0.1) × 10–2	2.6 (0.3) × 10–3	1.9 (0.2) × 10–3	17.1 (4.4)
Reduced: two cysteine residues
hBD-3 loop	2.8 (1.4) × 105	2.8 (2.0) × 10–2	1.2 (0.4) × 10–2	1.2 (0.4) × 10–4	0.9 (0.3)
hBD-2 loop	7.6 (1.1) × 105	3.4 (0.1) × 10–2	9.5 (0.2) × 10–3	2.4 (0.2) × 10–4	0.9 (0.2)
hNP-1 loop	8.8 (2.1) × 105	5.0 (0.3) × 10–2	6.4 (0.8) × 10–3	4.5 (0.5) × 10–4	3.8 (0.1)

While the bactericidal assay results show that SapA is involved in resistance to LL-37, assessment of affinity for LL-37 with the peptide both in solution and immobilized was inconclusive due to concentration-dependent aggregation of peptides in solution (Figure S3). As a result, during dilution of the peptide to low concentrations, accurate determination of the peptide concentration was limited. LL-37 immobilized on an SPR sensor chip displayed clear association and dissociation with 3.3 μM NTHi SapA, but none of the other concentrations that were tested (Figure S3A). Immobilization of LL-37 enabled the detection of both association and dissociation; however, we encountered difficulty fitting the model to the data. Similarly, a binding assay measuring interaction of a fragment of LL-37 (residues 1–10) with fluorescently labeled NTHi SapA did not produce a K_D (Figure S3B). These results indicate that LL-37 aggregated at high concentrations in our hands when it was not anchored to a surface.

SapA Binds Independent of AMP Topology

To investigate how the conserved 3D structure of the defensins contributes to SapA recognition, the AMPs were simplified to encompass just the C2–C4 loop region. This was achieved by creating peptides hBD-3_loop, hBD-2_loop, and hNP-1_loop with a disulfide bond connecting the termini of each 13-residue fragment and a C-terminal linker for peptide accessibility after SPR immobilization. The results of the SPR experiments confirmed that these modified loop regions retained their ability to interact with SapA (Figure 3D, E, and F). Remarkably, the binding affinities of SapA to the defensin loops are similar to those of the full length defensin AMPs, with K_D values between 9.5 and 17.2 nM (Table 1). In the presence of reducing agents, SapA exhibited a stronger affinity for the AMP fragments, increasing approximately 5- to 20-fold (Figure S4). This increased affinity in the presence of a reducing agent could be attributed to accessibility of the peptide binding motif within these immobilized linearized loop peptides. Collectively, these experiments demonstrate that SapA maintains high-affinity recognition toward defensins, irrespective of the peptide loop structure.

Identification of the Region and Motif for AMP Binding

Given that SapA maintained high-affinity binding for the truncated loop regions of the defensins, recognition of a common element in the loop could account for the NTHi SapA peptide specificity. The positively charged and hydrophobic residues present in the loop region of the defensins are distinguishing features of AMPs. This amphiphilicity of AMPs is important for their ability to interact with membranes; cationic arginine and lysine are essential for electrostatic interactions with the negatively charged bacterial membrane, while hydrophobic residues including tyrosine, phenylalanine, and tryptophan are involved in AMP oligomerization and membrane insertion.

To ascertain the specific regions of each defensin that contribute to binding, we synthesized a series of peptide fragments for the N-terminal, loop, and C-terminal regions of hNP-1, hBD-2, and hBD-3. Fluorescently labeled SapA was incubated with AMP fragments, and the 670/650 nm fluorescence ratio was measured to determine the affinity to peptides. Data were analyzed by the K_d fit model describing a molecular interaction with a 1:1 stoichiometry. The loop region fragments hBD-3_PKEEQIGK, hBD-2_RRYKQ, and hNP-1_RRYGT had comparable affinity to the linearized loops without cysteines (Figure 4B, C, and D), revealing a minimal region necessary and sufficient for high-affinity peptide binding. All other fragments demonstrated low mM affinity independent of peptide characteristics, including length, charge, and hydrophobicity (Figure S5).

Figure 4.

Interaction of defensin AMP fragments with NTHi SapA. (A) Individual sequences for hBD-3 (green), hBD-2 (blue), and hNP-1 (brown) with AMP shown as a cartoon, and sequences of fragments tested shown (mutations in orange). Regions that bind are highlighted while nonbinding regions are in black. Fluorescently labeled SapA was incubated with a serial dilution of defensin AMP fragments, and the 670/650 nm fluorescence ratio was measured to determine affinity to peptides. The results from one experiment (technical triplicate) were normalized and graphed using GraphPad Prism 9 for hBD-3 (B), hBD-2 (C), and hNP-1 (D) peptides.

To establish whether the charged and hydrophobic residues present in hBD-3_PKEEQ, hBD-2_RRYKQ, and hNP-1_RRYGT were essential for binding by NTHi SapA, the affinities of peptide point mutants were measured. While the hBD-2 and hNP-1 fragments had three residues in common (RRY), the hBD-3 fragment did not (PKE). Substitutions for arginine, lysine, and glutamate were tested to determine the importance of these charged residues on NTHi SapA peptide binding (Figure 4). Each substitution only impacted either charge or hydrophobicity. Substitution of charged residues reduced overall charge of the peptide. Substitutions of tyrosine or proline to phenylalanine or tryptophan increased hydrophobicity, while substitutions to serine decreased hydrophobicity. Upon substitution with serine, a position-dependent reduction in affinity was observed (Table S3). Saturation of binding was not observed even at 500 μM with lysine or glutamate substitution (Figure S6). Double serine substitutions (hBD-2_SSYKQ, hNP-1_SSYGT, and hBD-3_PSSEQ) completely abolished affinity to SapA (Figure S6). Additionally, substitutions of the tyrosine in hBD-2_RRYKQ and hNP-1_RRYGT and the proline in hBD-3_PKEEQ were tested to investigate the role of hydrophobicity in peptide binding by SapA (Figure S6). Substitution of these residues with serine, phenylalanine, or tryptophan drastically reduced the affinity for these peptides. A summary of the results for all substitutions is presented in Table S4.

The truncations and the mutagenesis of the defensin peptides reveal two different three-residue binding motifs, RRY and PKE, associated with the high-affinity peptide binding of SapA. Sequence variation in these motifs was not tolerated and resulted in reduced affinity. However, the degree of impact on affinity was also dependent on position. The identity of the amino acid was also important as lysine was not tolerated in place of arginine, nor could proline or tyrosine be substituted with other hydrophobic residues. These results suggest that the PKE motif forms different interactions with SapA compared with the RRY motif.

Crystal Structure of Unbound NTHi SapA

To study the interactions between NTHi SapA and AMPs at the molecular level, we made efforts to crystallize SapA in the presence of AMPs. However, despite exploring various approaches for cocrystallization, we were unable to obtain crystal structures with clear density for ligands. The crystal structure of apo SapA (residues 32–560) was solved to a resolution of 2.3 Å by molecular replacement (Table 2, Figure 5A). We compared our structure to the four in the RCSB: SapA in closed and open states with dsRNA (PDB ID 7OG0), SapA closed with dsRNA (PDB ID 7OFZ), and SapA with heme (PDB ID 7OFW).²¹ In our structure, the first 10 residues (residues 32–42) are not visible in the electron density in the refined model, and residues 139–162, which could not be modeled in previous structures, have well-defined density (Figure S7).²¹ In the previous structures, SapA was expressed at low concentrations in E. coli and purified with dsRNA bound. The significance of this interaction remains uncertain, as dsRNA is not known to have a defined function in the bacterial periplasm. The previous SapA heme crystal was produced through crystal soaking.²¹ Both dsRNA and heme are located on the exterior surface of the protein between two SapA molecules in the crystal lattice. Alignment of the closed structures with dsRNA or heme produced a range of RMSDs from 2.0 to 2.4295 Å, while our structure crystallized in a conformation similar to the open structure of NTHi SapA (PDB ID 7OG0, chain B; RMSD 0.7357 Å) (Figure 5B). Minute changes are apparent between the different closed structures in complex with ligands; the structure of NTHi SapA in complex with dsRNA (PDB ID 7OFZ) has an RMSD of 2.0439 Å, while SapA with heme (PDB ID 7OFW) has an RMSD of 1.9994 Å. Based on these structural comparisons, we crystallized SapA in an open unbound conformation. Our structure provides additional insight into the binding cavity of SapA.

Figure 5.

Structure of apo NTHi SapA in the open state. (A) Cartoon representation of crystal structure of unbound NTHi SapA with I_A, I_B, and II domains highlighted in wheat, magenta, and cyan, respectively. Helices are numbered in domain I_B. Inset shows newly resolved area. (B) Alignment of NTHi SapA (wheat) with 7OFZ closed (purple; top right) and 7OG0 open (green; bottom right). Box 1 shows movement of H2 and loop between H2 and H3 in I_B and box 2 shows movement of loop between B1 and H1 in I_A with arrows showing distance differences.

The structure of SapA consists of two globular domains connected by a hinge region, forming a substrate binding pocket. In cluster C SBPs like SapA, there is an additional sub domain, I_B (residues 69–229).⁵⁵ In comparison to previous structures, our structure resolves two regions, providing additional information about domains I_B and I_A, which leads to an expansion of the binding cavity. The region from residues 139 to 162 in domain I_B can now be modeled as a loop (residues 131–152) followed by an α helix (residues 153–163, helix 3) (Figure 5). The electron density contoured at 1.0 sigma shows a strong density for the helix backbone, although the side chain density is weaker (Figure S7). Comparison with published structures reveals high flexibility in this region. Rotation of helix 2 leads to movement of the loop between helix 2 and 3 toward domain I_A with a displacement of 15.7 Å at P138 (Figure 5B). This region of resolved density creates additional space at the interdomain interface. Differences in I_A lie in the first loop (residues 41 to 60) exhibiting weak electron density, which correlates with high B factors (60–130 Ų) in the open structure of NTHi SapA with dsRNA (PDB ID 7OG0, chain B) denoting flexibility in this region. In our structure of SapA, resolved portions of the loop (residues 41–43) show a 4 Å movement of S43 compared with the open structure of NTHi SapA with dsRNA (PDB ID 7OG0, chain B) away from the interior of the binding cavity (Figure 5B). Additionally, the loop in domain I_B (131–152) partially occupies the space where the loop in domain I_A is positioned in the open state of NTHI SapA with dsRNA (PDB ID 7OG0, chain B). The high B factors in both loops (residues 41–43 and 131–152) highlight the flexibility of both regions (Figure S7). Based on these observations, the loop in domain I_A has moved toward the exterior of the binding cavity while the loop in domain I_B moved toward domain I_A. Other areas with a difference in the RMSD are associated with crystal contacts. When comparing the backbone, our structure aligns with the open structure of NTHi SapA (PDB ID 7OG0, chain B).

Molecular Docking Predicts AMP Binding Modes

We turned to molecular docking to predict potential interactions between the AMPs and SapA. Since no peptide-bound SapA structures exist, we used the predicted cavity of the SapA structure (Figure S8) to define the binding pocket for molecular docking. When the size of the cavity across structures of NTHi SapA is compared with that of dsRNA (PDB ID 7OG0, 7OFZ), the binding pocket volume varies between 90 and 700 ų. Variability of the binding pocket volume results from the flexibility of loops and side chain orientation (Figure S8 and Table S5). It is important to note that while previously published SapA structures have both heme and dsRNA present (PDB ID 7OFW), these molecules on the exterior surface of the protein are distant from the peptide binding pocket situated between domain I and domain II (Figure S9). In the structure from this paper (PDB ID 8TV8), the cavity volume increases to 1400 ų due to the flexible loop in domain I_A.⁴⁴⁻⁵⁷ The peptide docking program HPepDock, optimized for protein–peptide interactions, was used to predict the SapA–AMP interactions utilizing the calculated binding pocket.⁶⁷ The minimal five residue peptides PKEEQ, RRYKQ, and RRYGT were docked in the open structure of SapA centered around the predicted binding pocket, with top scores of −141, −211, and −200, respectively; more negative scores with HPepDock indicate a higher ranked solution. As a control, the peptide AAAAA was also docked and had a score of −106.

As expected, based on solved structures of other PepT binding proteins bound to peptides, the top scoring poses lie along the interface between domain I and domain II (Figure 7A–C).^{17,40,42,44−48,51−53,94} While the tyrosine in both RRYGT and RRYKQ is predicted to engage in a pi–cation interaction with R533, the first glutamate residue in PKEEQ is predicted to form a salt bridge with R533. Additionally, each peptide had arginine and lysine stabilized by a salt bridge to E404 (Figure 6, Table S6). For each peptide, hydrogen bonds were predicted to interact with residues in the flexible loops of the binding pocket. Many of the interactions between peptides are similar, as domain II residue L459 and domain I_B residues F164 and Y166 were predicted to interact with all three peptides; however, only RRYKQ and RRYGT had an interaction with Y291 of domain I_A. Based on the results of the molecular docking, it seems that while the peptides are positioned along the same cleft in the binding pocket, hBD-3_PKEEQ has different interactions than hBD-2_RRYKQ and hNP-1_RRYGT.

Figure 7.

SPR characterization of key residues in the NTHi SapA binding pocket. Shown as cartoon with residues targeted for mutagenesis shown as spheres (A); residues are labeled in zoomed images of the top half of the binding pocket (B) and the bottom half of the binding pocket (C). (D) Analysis was performed with an analyte injection of 100 nM SapA WT and SapA mutants over immobilized hBD-3 across all 4 channels for a 3 min association followed by a 12 min dissociation step. The maximal binding for each channel in response units (RU) was averaged for the analytes and displayed in a bar graph. Residues from domain I_A, I_B, and II are shown in wheat, magenta, and cyan respectively. Statistical significance was determined by one-way Anova against the control group (SapA WT); compared to WT, all SapA mutants had p-values < 0.0004.

Figure 6.

Molecular docking of AMP fragments. Top-scoring HPepDock solutions shown in sticks over cartoon representation of NTHi SapA with I_A, I_B, and II domains highlighted in wheat, magenta, and cyan, respectively, with ligand shown as sticks with white surfaces, E404 shown as spheres, but R533 is hidden from view behind peptide: (A) hBD-3_PKEEQ, (B) hBD-2_RRYKQ, and (C) hNP-1_RRYGT. The predicted salt bridge with E404 (shown as stick) is displayed for (D) hBD-3_PKEEQ, (E) hBD-2_RRYKQ, and (F) hNP-1_RRYGT. Interactions with R553 (shown as sticks) are displayed for (G) hBD-3_PKEEQ, (H) hBD-2_RRYKQ, and (I) hNP-1_RRYGT.

SapA Mutants Exhibit Decreased Binding to hBD-3

In the absence of a SapA crystal structure bound to AMPs, we employed mutagenesis to verify the peptide binding sites. Prior to molecular docking, when we analyzed the solvent accessible residues comprising the binding pocket, the majority of these residues belonged to Domain I_A, with only a few residues in Domain II and I_B. Subsequently, we introduced G55A, T56A, S57A, S290A, Y291A, R533A, I534A, and F549A mutations in domain I_A; Y166A and Q205A mutations in domain I_B; and a L459A mutation in domain II. Proper folding was verified by circular dichroism (Figure S10). Additional mutagenesis attempts in the predicted binding cavity (domain II mutants E404A, Q406A, R437A, and W458A, domain I_A mutant S551A) encountered issues with expression or refolding. Interactions with peptide were tested with hBD-3, as this AMP not only exhibited SapA mediated resistance but also demonstrated high-affinity binding using SPR, unlike hBD-2, or hNP-1. While Q205A showed increased binding and T56A exhibited a moderate decrease, the remaining mutants displayed a 4 to 10-fold reduction in binding to hBD-3 (Figure 7D). The positioning of Q205 behind the loop (residue 41–60) in domain I_A (Figure 7B) suggests that the smaller size of the alanine may allow space for the peptide. Notably, the aromatic residues in the binding pocket (Y166, Y291, and F549) are in the regions that exhibited significant flexibility between the open and closed states. Through these mutagenesis studies, we confirm that peptides bind in the predicted binding pocket of NTHi SapA with contributions from both Domain I and II.

PepT Substrate Binding Pockets Diverge Across Species

Previous analyses of PepT binding protein cavities revealed binding pockets with a range of sizes.^44,52 Using KVFinder to measure predicted cavity volume, the cavity size of PepT SBPs differs between homologues. For instance, OppA from Lactococcus lactis, which accommodates peptides as long as 35 residues, has a binding pocket volume of approximately 2235 ų (PDB ID 3DRG).⁵² NTHi OppA, which binds peptides between three and 11 residues, measures approximately 1151 ų (PDB ID: 6DTH).¹⁷ The difference in the binding cavity size is a combination of the presence of bulkier side chains within the binding cavity and conformational changes. Rotation of domain I with respect to domain II closes the binding cavity in PepT SBPs, but the degree of closure varies. The angle of rotation is approximately 19° in L. lactis OppA and 35° in NTHi OppA which results in less volume between the domains for NTHi OppA compared with L. lactis OppA.^17,52,53E. coli DppA (PDB ID 1DPP) has a smaller 140 ų binding pocket and a large 55° angle of rotation. The degree of closure also depends on the bound peptide; peptides bind in a range of open, semiopen, and closed states of the SBP. Each of these conformational states is observed in the NTHi SapA structures. With all the PepT SBPs, the peptide substrates remain positioned along the cleft between domain I and II as predicted for NTHi SapA (Figure 8A–C).

Figure 8.

Conservation of ligand binding residues across PepT SBPs. PepT SBP shown as cartoon with peptide shown as spheres: (A) L. lactis OppA (PDB ID 3RYA), (B) E. coli DppA (PDB ID 1DPP), and (C) NTHi SapA molecular docking with PKEEQ. (D) Consurf analysis image generated in PyMOL of SapA homologues. (E) Logo plot showing divergence of NTHi SapA ligand binding residues among SapA homologues. (F) Logo plot of PepT SBPs in Pasteurellaceae species. (G) Phylogenetic tree of PepT SBPs in Haemophilus species with SapA homologues shown in light blue, HbpA homologues shown in orange, OppA shown in green, and OppA2 shown in dark blue.

A sequence comparison of SapA homologues reveals both commonalities and divergence in the peptide binding residues of PepT SBPs. An MSA of homologues with greater than 35% sequence identity to NTHi SapA revealed peptide binding residues from NTHi SapA (G55, Y166, R533, F549) that are highly conserved across SapA homologues (Figure 8D, E). To compare with the broader family of PepT SBPs, we conducted a MSA using sequences from the Pasteurellaceae family with at least 60% identity and 85% query coverage to NTHi SapA, OppA, and HbpA. While the conserved NTHi SapA peptide binding residues (G55, Y166, and F549) show more variation when compared to PepT SBPs, glycine, tyrosine, and phenylalanine remain the most frequent residue at this position (Figure 8F). These conserved residues in domain I likely contribute to hydrophobic patches at the interface between the domains in PepT SBPs. On the other hand, the variation in R533 suggests that this residue may be important for interactions in SapA homologues but not PepT SBPs as a whole. A phylogenetic tree from the PepT MSA shows that SapA is most closely related to HbpA while OppA and OppA2, a secondary oligopeptide binding protein present in some Haemophilus species, are more distantly related (Figure 8G). When comparing the structure of Glaesserella parasuis HbpA bound to glutathione, it becomes evident that the residues involved in side chain interactions in HbpA are not conserved in SapA homologues or across PepT SBPs.⁴⁷ Conversely, residues involved in backbone interactions in NTHi OppA and G. parasuis HbpA show higher conservation across homologues reflective of the broad specificity of this family.^17,47

Discussion

With the ability to confer resistance to AMPs, the Sap transport system plays an essential role in NTHi colonization and pathogenesis in the host environment. In a prior study, transmission electron microscopy images provided evidence that the Sap transporter facilitates the transport of hBD-3 and LL-37 in NTHi. Deletion of the Sap TMDs results in an accumulation of hBD-3 and LL-37 on the inner membrane of NTHi. Conversely, when a functional Sap transporter is present, AMPs are transported across the inner membrane of NTHi and into the bacterial cytoplasm, where they undergo degradation.²⁵ Our bactericidal assays demonstrate the necessity of SapA for resistance to various AMPs, with SapA required for robust resistance to hBD-3 and LL-37 in NTHi. Notably, this mechanism is not exclusive to NTHi as SapA is also involved in AMP resistance in other organisms. However, SapA homologues seem to exhibit different specificities. For instance, in H. ducreyi, SapA mediates resistance to LL-37 but not hBD-2 or hBD-3.¹³

For the first time, this study reveals that SapA binds human defensins with a high affinity. Although SapA has a high affinity for all defensins tested, the bactericidal assay results suggest that resistance to AMPs is not directly linked to affinity. Upon examining the relationship between types of AMPs and resistance, we observe that the survival of the NTHi ΔsapA strain, lacking a functional SapA, is most impacted by LL-37 and hBD-3. In contrast, the impact on NTHi survival is less pronounced with hBD-2. Despite the ability of SapA to bind hNP-1 with the same affinity as the other defensins, we observe minimal to no impact on survival in both ΔsapA and the parent strain.

While SPR and spectral shift assays suggested a low affinity for LL-37, efforts to characterize the interaction between SapA and LL-37 were hampered by the propensity of LL-37 to form aggregates in solution. While hBD-3, hBD-2, and hNP-1 exist as monomers and dimers in solution, LL-37 forms higher order oligomers at the concentrations used for affinity experiments, which only allowed us to collect data across a limited range.⁹⁵⁻⁹⁷ Based on our findings, the dissociation constants alone are not sufficient to explain the sensitivity to AMPs. The studies with full-length AMPs suggest that affinity alone may not be the primary determinant of AMP resistance via the Sap transporter.

The kinetics of the interaction between the ligand and SBP are linked to the rate of transport in ABC importers. When a ligand associates with the SBP, it triggers a significant conformational change in which globular domains enclose the substrate, an action often referred to as the “Venus flytrap” mechanism.⁹⁸ Substrate-bound SBP forms a complex with the TMDs and triggers a series of events that lead to ATP binding within the NBDs coupled with TMD movement, facilitating the release of substrate by the SBP.⁹⁹ Compounds that bind to the SBP but do not trigger SBP closure are not transported. In the case of the Opp transporter, slower substrate dissociation from OppA is directly linked to reduction in transport.^100,101 Furthermore, when substrate dissociation from the SBP is slow or irreversible, it results in reduction or complete inhibition of transport in Type I ABC importers.¹⁰² In the case of the full-length defensins, the SPR data indicate a two-state model fit, suggesting a conformational change occurs following the initial 1:1 binding of the protein to immobilized ligand.⁹³ These data imply that SapA undergoes a shift to an alternate conformation after binding, and both dissociation rates influence final substrate dissociation from SapA. The kinetic rates from the SPR data suggest that substrate dissociation is slower with hBD-2 and hNP-1 and faster with hBD-3. If the transport mechanism is conserved between OppA and the PepT homologue SapA, we can infer that slow substrate dissociation from SapA, like with hBD-2 and hNP-1, leads to reduced transport via the Sap transporter.

In general, we observed a gradient of the peptide specificity among members of the PepT SBP family. OppA homologues tend to favor peptides with certain characteristics but with no particular motif that indicates a binding preference.^42,52,53 Alternatively, as with H. influenzae HbpA and Mycobacterium tuberculosis DppA, deviations from a specific sequence result in a loss of affinity.^40,47 Based on our peptide studies, NTHi SapA recognizes a precise motif, PKE or RRY. Despite the relatively long length of these defensins, our mutagenesis studies reveal that the selectivity is primarily governed by specific amino acids in the loop region of the AMPs. When comparing defensin sequences, these motifs are present in human defensins found in the microenvironment of NTHi (Figure S11).^103−106

The PKE and RRY motifs have different binding orientations based on the molecular docking predictions. While both motifs are anchored by a salt bridge with E404, the interactions involving other side chains vary between the motifs. Additionally, both motifs share common features, including a hydrophobic residue and two charged residues. However, when comparing the position of the hydrophobic residues in the docking solutions for RRY to those of PKE, we observe that the tyrosine residue of the RRY motif forms a pi–cation interaction with R533 in the binding pocket while the proline of the PKE motif does not interact with R533 (Figure 6). This behavior is not unique to NTHi SapA; other PepT SBPs exhibit this behavior where a specific side chain interaction anchors the peptide, but the orientation of the rest of the peptide can vary. As an example, the hydrophobic pocket in OppA from L. lactis preferentially fits isoleucine while the remaining portion of the peptide can be accommodated in different parts of the binding cavity, as neither the N- and C-termini are anchored by a salt bridge.⁵²

When examining how a full-length peptide would bind, it is important to consider the flexibility of the binding cavity as well as the state of bound AMPs. To gain insight into the interaction between folded defensins and SapA, we conducted modeling by superimposing the structures of defensins with the highest scoring molecular docking poses. The process revealed clashes between the SapA and defensin structures, confirming our initial prediction that further conformational changes would be necessary to accommodate them in the binding pocket (Figure 9). In our structural analysis, we noted significant flexibility in both the loop in domain I_A (spanning from 41 to 60) and the region within domain I_B (139 to 162). Movement of loops in the binding cavity and switching of side chain rotamer conformations are two strategies we observed in another PepT SBP, NTHi OppA, to accommodate bulky peptides.¹⁷ Moreover, OppA has been found to accommodate heme within the binding pocket alongside small peptides, although longer peptides hinder heme binding.¹⁷ The flexibility of the NTHi SapA binding pocket offers a potential explanation for its ability to accommodate a larger portion of the peptide. While our studies indicate that NTHi SapA binds peptides independent of structure, as both oxidized and reduced defensin loops exhibit high-affinity binding, we assume defensins are folded, given the oxidizing environment within the bacterial periplasm. However, Blair et al. propose an alternative explanation for how SapA might bind unfolded defensins, suggesting that peptide transport is mediated by chaperone-assisted unfolding, with transport systems implicated in AMP resistance.⁸

Figure 9.

Superimposition of full-length defensins with docked poses of defensin fragments. Full-length defensins were superimposed on the top-scoring HpepDock solutions docked in the active site. Docking solutions (spheres) aligned with full-length defensins (cartoon) are shown with cartoon representations of NTHi SapA visualized in PyMOL: (A) hBD-3_PKEEQ with hBD-3, (B) hBD-2_RRYKQ with hBD-2, and (C) hNP-1_RRYGT with hNP-1. While NTHi SapA binds human cathelicidin and defensins, SapA homologues from other bacterial species interact with different AMPs. In bactericidal assays with nonpolar sapA knockout strains, SapA contributes to cathelicidin resistance in H. ducreyi (LL-37) and A. pleuropneumoniae (PR-39) and protamine resistance in S. enterica.^9,11,13,107 PR-39 is a linear proline-rich peptide (net charge +10), while protamine is an arginine-rich AMP with a predicted hairpin loop structure (net charge +21).^11,108 Notably, A. pleuropneumoniae is not resistant to beta defensins, and the Sap transporter in H. ducreyi and S. enterica is not involved in resistance to defensins.^13,107,109

The capability to recognize cathelicidins in NTHi, H. ducreyi, and A. pleuropneumoniae might stem from shared characteristics in the binding pocket. Based on sequence alignment, the residues involved in NTHi SapA ligand binding in domain I_A along the loop from 41 to 60 (G55, T56) are conserved in H. ducreyi and A. pleuropneumoniae, except for a moderate S57N mutation (Figure S12). In S. enterica, these residues are more hydrophobic (T56L, S57I), but the presence of N59D introduces an increased negative charge, which may facilitate interactions with the arginine side chains in protamine. However, the other SapA residues in domain I_A implicated in defensin binding (S290, Y291) show divergence in the other homologues (Figure 8D). In each, S290 is replaced with an alanine, while Y291 is replaced by a glutamine, leucine, and tryptophan in H. ducreyi, A. pleuropneumoniae, and S. enterica, respectively (Figure S12). In NTHi SapA, Y291 is involved in hydrophobic interactions but can also participate in hydrogen bonding. Consequently, these mutations decrease the possible interactions at that position. Since these residues are important for binding hBD-3 in NTHi SapA, the variability in these residues may explain why the other SapA homologues cannot bind defensins.

The data presented in this article offer insights into a potential mechanism of SapA selectivity for defensin peptides. The high affinity of NTHi SapA for defensins enables SapA to sequester AMPs in the periplasm, with a preference for defensins containing the distinct motifs, PKE and RRY. SapA is upregulated in NTHi both in a chinchilla model of otitis media and in response to treatment with AMPs, increasing copies in the periplasmic space.^14,18 For SapA to access AMPs, first, the AMP molecules must accumulate on the membrane up to a critical concentration before permeabilizing the membrane for access to the periplasm. How SapA can remove AMPs from the membrane has yet to be elucidated. With the ends of the peptide protruding from the binding pocket, the peptide–SapA complex in a closed conformation is hypothesized to interact with the TMDs of the transporter to move the peptide into the translocation pathway and ultimately inside the bacterial cytoplasm upon ATP hydrolysis.

Glossary

ABBREVIATIONS

NTHi

nontypeable Haemophilus influenzae

AMP

antimicrobial peptide

ABC

ATPase binding cassette

Sap

Sensitivity to antimicrobial peptides

PepT

Peptide, Opine, and Nickel Uptake Transporter

TMD

transmembrane domain

NBD

nucleotide binding domain

SBP

substrate-binding protein

hNP-1

α-defensin 1

hBD-2

β-defensin 2

hBD-3

β-defensin 3

SPR

surface plasmon resonance

MSA

multiple sequence alignment

CD

circular dichroism

sBHI

supplemented brain heart infusion

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.biochem.3c00562.

• Additional materials and additional biochemical, structural, and bioinformatics experimental data (PDF)

Accession Codes

NTHi SapA (Q4QL73), E. coli NikA (P33590), NTHi OppA (Q4QLH0), NTHi HbpA (Q4QM48)

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The project described was supported by R01GM140584, R01GM140584–04S1, R01AI139519, The National Institutes of Health Training Grant (T32GM008449) through Northwestern University’s Biotechnology Training Program (E.R.B.), and National Institutes of Health NIGMS Molecular Biophysics Training Program Grant T32GM008382 (K.G.R.).

The authors declare no competing financial interest.