Solution structural model of the complex of the binding regions of human plasminogen with its M-protein receptor from Streptococcus pyogenes

Abstract

VEK50 is a truncated peptide from a Streptococcal pyogenes surface human plasminogen (hPg) binding M-protein (PAM). VEK50 contains the full A-domain of PAM, which is responsible for its low nanomolar binding to hPg. The interaction of VEK50 with kringle 2, the PAM-binding domain in hPg (K2_hPg), has been studied by high-resolution NMR spectroscopy. The data show that each VEK50 monomer in solution contains two tight binding sites for K2_hPg, one each in the a1- (RH1; R¹⁷H¹⁸) and a2- (RH2; R³⁰H³¹) repeats within the A-domain of VEK50. Two mutant forms of VEK50, viz., VEK50[RH1/AA] (VEK50^△RH1) and VEK50[RH2/AA] (VEK50^△RH2), were designed by replacing each RH with AA, thus eliminating one of the K2_hPg binding sites within VEK50, and allowing separate study of each binding site. Using ¹³C- and ¹⁵N-labeled peptides, NMR-derived solution structures of VEK50 in its complex with K2_hPg were solved. We conclude that the A-domain of PAM can accommodate two molecules of K2_hPg docked within a short distance of each other, and the strength of the binding is slightly different for each site. The solution structure of the VEK50/K2_hPg, complex, which is a reductionist model of the PAM/hPg complex, provides insights for the binding mechanism of PAM to a host protein, a process that is critical to S. pyogenes virulence.

Graphical Abstract

1. Introduction

Approximately 250 serotypically distinct strains of Group A Streptococcus pyogenes (GAS) have been identified that present patients with mild to severe symptoms. The different GAS strains are serotyped through the nature of their characteristic surface M- and M-like proteins, with one skin-trophic subgroup (Pattern D) containing a direct human host plasminogen (hPg)/plasmin (hPm) surface M-protein receptor (PAM) (Berge and Sjobring, 1993). Binding of hPg to PAM-type M-proteins greatly facilitates its activation to the protease, hPm, by GAS- secreted streptokinase (SK), a step that is essential for its virulence (Sun et al., 2004). After activation, hPm remains bound to GAS, resulting in exacerbation of GAS pathogenicity through proteolytic disruption of host innate immune barriers to dissemination (Sanderson-Smith et al., 2008), e.g., the extracellular matrix and tight cellular junctions (Lahteenmaki et al., 2000; Sumitomo et al., 2013; Sumitomo et al., 2016). The N-terminal A-domain of various PAM-type M-proteins usually consist of one or two peptide modules, viz., a1- and a2- repeats, having 13–17 residues each (Qiu et al., 2018). Both repeats have been shown to be critical determinants for hPg binding. (Ringdahl et al., 1998; Rios-Steiner et al., 2001; Sanderson-Smith et al., 2007; Schenone et al., 2000; Wang et al., 2010b; Yuan et al., 2017).

Contained within the hPg protein, the lysine binding site (LBS) of its small functionally-independent kringle 2 (K2_hPg) domain has been identified as the major receptor site for PAM binding (Wistedt et al., 1998). The LBS of four of the five kringle domains of hPg generally interact with activation effectors (Castellino and Ploplis, 2003), as well as with cellular receptors (Pancholi and Fischetti, 1998), via C-terminal lysine residues of the receptor (Miles et al., 1991). However, the prototypical Pattern D GAS receptor for hPg, e.g., PAM from isolate AP53 (PAM_AP53), does not contain a C-terminal lysine residue and interacts with hPg via an internal pseudo-lysine arrangement of amino acid side-chains (Rios-Steiner et al., 2001; Wang et al., 2010b; Yuan et al., 2017). In this regard, adjacent Arg and His residues in both the a1- and a2-repeats of the PAM A-domain have been shown to be essential for hPg binding (Rios-Steiner et al., 2001; Sanderson-Smith et al., 2007; Schenone et al., 2000).

Since the binding domains of PAM and hPg were originally defined (Wistedt et al., 1995; Wistedt et al., 1998), a small peptide, VEK30 from PAM_AP53, that contains only its a1-repeat, as well as VKK38 from PAMNS455, that is composed of both the a1- and a2-repeats, along with K2_hPg have been employed as minimalistic structural models to investigate the essential aspects of the very specific and tight binding of PAM proteins to hPg. The side-chains of residues R¹⁷H¹⁸ and E¹⁹E²⁰ of VEK30 serve as positive and negative clusters around one turn of an α-helix. The residues are spaced similarly to the α-COOH and ε-NH₂ of a free lysine and fit into the LBS of K2_hPg. This has been interpreted as an a1-located lysine side-chain isostere (RH1) in PAM_AP53 (Rios-Steiner et al., 2001; Wang et al., 2010b). Similarly, the side-chains of the more extended VKK38, containing residues R³⁰H³¹ and D³²H³³D³⁴, constitute a second a2-positioned lysine isostere (RH2) in PAMNS455 (Yuan et al., 2017). We showed previously that VKK38 can accommodate two K2_hPg peptides on the VKK38 α-helix, likely on opposite faces of this helix (Yuan et al., 2017).

We aimed to elucidate the roles of each of the two hPg/hPm binding sites in PAM in order to assess their equivalency in the structure and function of PAM, especially since other Pattern D GAS strains are highly substituted in this region. Additionally, some Pattern D GAS isolates only contain one hPg/hPm binding site in the PAM A-domain (the a2-repeat), e.g., PAMNS88.2. These PAMs display different hPg acquisition properties than those with two hPg/hPm binding sites (Qiu et al., 2018). Thus, an understanding of the mechanism of binding of hPg to PAM has become an important factor in the study of GAS virulence. In an associated recently submitted paper (Quek et al, submitted for publication), a larger peptide, VEK75, sequentially containing a C-terminal portion of the hypervariable region (HVR)-the full A-domain-and a partial N-terminal portion of the B-domain of PAM_AP53 (Bhattacharya et al., 2014), in complex with K2_hPg, has been crystallized and the structure solved, showing that the two binding sites in this PAM_AP53 peptide can simultaneously bind to K2_hPg. In order to resolve potential generic issues surrounding the influence of crystal packing forces in determining the final structure, and/or rearrangement of the conformations for optimal crystal formation, and to obtain a more direct observation of the interchain binding interactions in solution, we have additionally determined the solution structure of a VEK50/K2_hPg complex, using a functionally similar, but slightly shorter (for ease in assigning NMR chemical shifts), fragment of PAM_AP53 containing its full A-domain. VEK50 is extended in sequence from VEK30 and VKK38 to more rigorously assess possible exosites in binding. The high equivalency of the studies with a variety of peptides from the A domain of PAM clarifies the binding modality of hPg to PAM_AP53.

2. Materials and Methods

2.1. Construction of expression plasmids

The construction of plasmids for peptide and protein expression was accomplished as previously described (Wang et al., 2010b). VEK50, cloned from PAM_AP53, and VEK50 mutant peptides, generated by site-directed mismatch mutagenesis with synthetic oligonucleotides, were expressed in Escherichia coli BL21 (DE3) cells employing the His₆-tagged-GB1 domain fusion expression system (Wang et al., 2010b). The final constructs contained sequentially from the 5’: [ATG initiation codon - purification His₆ tag - GB1 domain for enhanced solubility - 9-residue linker - thrombin cleavage site, LVPR/GS, followed by the VEK peptide]. This cassette was inserted into pET-15b (Novagen) (Wang et al., 2010b). Thus, all peptides cleaved with thrombin possessed an exogenous GS dipeptide at the N-terminus. In addition, a codon for a Tyr was intentionally placed at the C-termini of the VEK peptides for 280 nm absorption properties (Bhattacharya et al., 2014). All plasmid inserts were subjected to full nucleotide sequencing and only those identical to the expected sequences were employed for expression.

2.2. Expression and purification of VEK peptides

The expression of VEK50 peptides was carried out in E. coli BL21 using previously published protocols (Wang et al., 2010b). After induction with 0.8 mM isopropyl-1-thio-β-D-galactopyranoside (IPTG), proteins extracted from the cell pellet were loaded onto a Ni²⁺-Sepharose affinity chromatography column (HisTrap HP; GE Healthcare) at 4° C, washed, and eluted with 60 mM imidazole, pH 8.0. The concentrated eluates were further purified after cleavage with 1000 U of thrombin (ERL). The resulting cleaved fragments were separated using a HisTrap HP affinity column (GE Healthcare). At this step, the flow-through fraction, containing VEK50 peptides, was applied to p-aminobenzamidine agarose (Sigma) to remove thrombin. To prepare ¹⁵N- and ¹⁵N/¹³C-VEK50 samples for NMR experiments, expression and purification were similarly accomplished as described previously (Wang et al., 2010a; Wang et al., 2010b). For uniform labeling, ¹⁵NH₄Cl (99%, Cambridge Isotope Laboratories) and/or ¹³C-glucose (99%, Isotec) were used as the sole nitrogen and carbon sources, respectively.

2.3. Expression and purification of K2_hPg

A triple variant of human K2_hPg (C⁴G/E⁵⁶ D/L⁷²Y) that displays ~5-fold enhanced affinity. without loss of selectivity for PAM, as well as the inability to aggregate through disulfide bonding, compared to WT-K2_hPg, was expressed in Pichia pastoris GS115 cells as described in detail previously (Nilsen et al., 1997; Wang et al., 2010a; Wang et al., 2010b). The tighter binding of this variant allowed separation of K2_hPg from unbound components during purification. For uniform ¹⁵N- and ¹⁵N/¹³C-K2_hPg, the medium employed was (¹⁵ NH₄)₂SO₄ (99%, Cambridge Isotope Laboratories)/¹³C-glucose (99%; Isotec)/¹³C-methanol (99%, Isotec). The K2_hPg contained seven exogenous residues at the N-terminus of K2_hPg (enumerated as −7 to −1) as a necessary result of the construction of the expression plasmid (Nilsen et al., 1997). The final construct is as follows:

$${\text{NH}}_2-{\text{Y}}^{-7}\text{VEFSEE}\left[{\text{C}}^{+1}-\text{K}{2}_{\text{hPg}}-{\text{C}}^{78}\right]{\text{AA}}^{80}$$

The Y⁻⁷VEF is from the polylinker site of the plasmid and SEE represents the linker between K1_hPg and K2_hPg in hPg numbering begins at C⁺ and concludes at C ⁷⁸

The integrity of all labeled and unlabeled proteins and peptides was determined by MALDI-TOF mass spectrometry on an Autoflex III spectrometer (Bruker Daltonics). For all uniformly-labeled ¹⁵N- and ¹³C/¹⁵N-peptides, single mass peaks were obtained of the correct molecular weights, indicating complete incorporation of the heavy isotopes.

2.4. Extinction coefficients of peptides and proteins

Precise concentrations of the peptides and proteins are required for accurate titrations of the VEK-peptides and K2_hPg. For this, we directly determined extinction coefficients of all samples at 280 nm. Since the VEK peptides are naturally devoid of aromatic residues, a C-terminal Tyr residue was added at the COOH-termini of these peptides to provide an observable A_280nm. Next, we determined the A_280nm combined with the refractive indices (RI) on the same samples. The procedure made use of a combination of size exclusion chromatography (SEC), to separate any impurities in the sample, with the column flow-through passed in-line through a 2013variable wavelength detector (VWD), set to monitor UV absorbance at 280 nm, a light scattering detector to ensure that the molar mass of the eluted peak matched that of the peptide of interest, followed by a refractive index detector (RID). The equipment used was an Agilent 1260 infinity II HPLC system with a Wyatt WTC-030S5 (7.8 × 300 mm, 5 μm, 300Å) SEC column coupled to Wyatt Treos II multi-angle (three angles) light scattering (MALS) detector, an Agilent VWD, and a Wyatt Optilab T-Rex RID. The flow path, together with the SEC column, was equilibrated with PBS, pH 7.4. The sample (100 μl) in the same buffer was added to the column via an autosampler and passed through the SEC column. Approximately 95% of all samples applied reached the detectors. Since the refractive index increments of the peptides are constant at 0.185 ml/g, the A_280nm and RI values allow the extinction coefficient of the sample to be readily calculated (Wen et al., 1997).

The experimental extinction coefficients (ml mg⁻¹ cm⁻¹) determined from this approach (compared to those calculated from ExPASY (Gasteiger et al., 2003) are: K2_hPg = 2.38 (2.40); VEK50 = 0.20 (0.24); VEK50^△RH1 = 0.21 (0.25); VEK50^△RH2 = 0.21 (0.25); hPg (for a positive control) = 1.75 (1.72). In all cases, the experimental extinction coefficients were used for concentration determinations.

2.5. Isothermal titration calorimetry (ITC)

ITC measurements were performed at 25° C on a VP-ITC 200 Microcal calorimeter (Malvern) in 50 mM sodium phosphate, pH 7.4. For VEK50/K2_hPg binding, VEK50 peptides (40 μM) was stirred in a cell at ~800 rpm and K2_hPg (900 μM) was injected at a rate of 4 μl/s for WT-VEK50 and 2 μl/s for VEK50 mutants with 120 sec equilibration time. Control experiments were performed by titrating K2_hPg to buffer instead of VEK peptides. The integrated heats of interaction were normalized to the molar ratio of VEK50 peptide to K2_hPg, and the data were fit with Origin-ITC 7.0 software.

2.6. Analytical ultracentrifugation (AUC)

The molecular weights of individual proteins were determined by AUC equilibrium analysis with a Beckman Optima XLI ultracentrifuge. Purified K2_hPg (10 μM) and VEK50 peptides (150 μM), at 0.1 – 0.3 μg/ml, were analyzed in 20 mM Tris-Cl/0.1M NaCl, pH 7.4, at 25° C. A_280nm data from rotor speeds of 32,000 and 40,000 rpm were recorded every hr until the scans were identical, thus indicating that equilibrium was attained. The partial specific volumes of the VEK50 peptides and K2_hPg were calculated from their amino acid sequences using Sednterp (Laue et al., 1992). AUC data were analyzed by the Optima XL-A/XL-I data analysis software (Beckman Coulter), and apparent molecular weights were obtained throughout the concentration gradient generated in the cell (Bhattacharya et al., 2014).

For complexes of K2_hPg and each VEK50, the peptides were mixed in a 3:1 molar ratio of K2_hPg:VEK50 and loaded onto a S75 gel filtration column (GE Healthcare) equilibrated with 20 mM Tris-Cl/0.1 M NaCl, pH 7.4. The column was operated with an AKTA FPLC system at a flow rate of 0.5 ml/min and elution was monitored by A_280nm. The complex was resolved from free K2_hPg and the first peak to elute was concentrated by membrane filtration (3K-MWCO), and analyzed on the AUC as described above.

2.7. NMR

2.7.a. Data collection

Protein concentrations used for NMR samples of the apo-VEK50 peptides, apo-K2_hPg, and the VEK50/K2_hPg complexes ranged from 0.2 –1.0 mM. Purified samples of the VEK50s and K2_hPg were dialyzed against NMR sample buffer (20 mM Bis-Tris-d19/2 μM DSS/5% ²H₂O/95% H₂O, pH 6.7) prior to NMR measurements. In order to solve the bound-form structures of the VEK50 peptides with K2_hPg, ¹⁵N/¹³C-VEK50, ¹⁵N/¹³C-VEK50[RH1/AA] (VEK50^△RH1), and ¹⁵N/¹³C-VEK50[RH2/AA] (VEK50^△RH2) samples were mixed with unlabeled K2_hPg at molar ratios of 1:2, 1:1, and 1:1, respectively.

Data were acquired at 298 K on a Bruker AVANCE 800 spectrometer equipped with a 5-mm triple resonance (TCI, ¹H, ¹³C, ¹⁵N) cryoprobe. Standard triple resonance NMR experiments were used to assign ¹⁵N/¹³C-labeled VEK50 peptide samples. The following spectra were collected: ¹⁵N-HSQC (Kay et al., 1992; Mori et al., 1995), ¹⁵N-TOCSY-HSQC (Marion, 1989 #13328} (80 ms mixing time), HNCO/HN(CA)CO (Clubb et al., 1992; Kay et al., 1990), HNCA (Grzesiek and Bax, 1992), HNCACB/CBCA(CO)NH (Grzesiek et al., 1993a; Wittekind and Mueller, 1993), C(CO)NH (Grzesiek et al., 1993b) for the backbone and aliphatic side chain resonance assignments, as well as ¹⁵N-NOESY-HSQC (Talluri and Wagner, 1996) and ¹³C-NOESY-HSQC (Ikura et al., 1990), to collect intramolecular NOE distance constraints for the structure calculations. NMR data were processed using Bruker TopSpin 3.5 software and analyzed using NMRFAM-Sparky (Lee et al., 2009). ¹H chemical shifts were referenced to internal DSS, while ¹³C and ¹⁵N chemical shifts were referenced indirectly to DSS (Markley et al., 1998).

For docking studies, intermolecular distance constraints were determined by ¹⁵N/¹³C-half-filtered/edited 3D NOESY-¹H-¹⁵N/¹³C-HSQC experiments (120 ms mixing time), performed on samples of ¹⁵N/¹³C-VEK50 peptides complexed to unlabeled K2_hPg , and ¹⁵N/¹³ hPg C- K2_hPg complexed with unlabeled VEK50 peptides. 2D NOESY experiments with ¹⁵N- and ¹³C-filtered in both dimension (120 ms mixing time) were acquired on the same samples under identical conditions as controls.

2.7.b. Structure calculations of VEK50 peptides

The structures of VEK50 peptides in their apo- and bound-forms were first predicted by the CS-ROSETTA program (Shen et al., 2008) based on their residue-specific chemical shifts obtained from the NMR experiments listed above. Next, the structure from the converged results was used as the template, and further refined by Xplor-NIH 2.36 program (Schweiters et al., 2018). Backbone torsion angle (ϕ and ψ) restraints predicted from TALOS-N distance restraints from NOESY spectra, and RDCs were applied in a simulated annealing protocol in the Xplor-NIH program. For each VEK50 peptide, 200 structures were calculated, from which 20 structures with the lowest energy restraint values were further refined with implicit water (Chen et al., 2004). The quality of the structures was analyzed with PROCHECK 3.5.4 (Laskowski et al., 1993) and MolProbity (Chen et al., 2010). Visualization of the structures was performed using the PyMOL program.

2,7,c. Chemical shift perturbations (CSP)

Titrations for binding studies were carried out by recording ¹H/¹⁵N-HSQC experiments on ¹⁵N-VEK50 peptides (0.2 mM) with increasing molar ratios of unlabeled K2_hPg. The amide nitrogen and hydrogen CSPs were mapped for each amino acid according to $\sqrt{{(\mathrm{\Delta }\text{δHN})}^2+{(\text{ΔδN}/6)}^2}$, where Δδ^HN and Δδ^N represent the chemical-shift changes of ¹H and ¹⁵N atoms between the apo- and bound-forms, respectively (Williamson, 2013). CSP values for K2_hPg were collected from reversed-labeled samples of the complexes, i.e., ¹⁵N- K2_hPg titrated with increasing molar ratios of unlabeled VEK50 peptides.

2.7.d. RDC measurements

Residual dipolar coupling (RDC) measurements were performed on ¹⁵N-VEK50, ¹⁵N-VEK50^△RH1, and ¹⁵N-VEK50^△RH2 in their apo-forms and in complexes with unlabeled K2_hPg at molar ratios of 1:2, 1:1, and 1:1, respectively. Additionally, ¹⁵N-K2_hPg, complexed with unlabeled mutant VEK50 peptides, was analyzed at a molar ratio of 1:1. In all cases, the concentration of labeled protein ranged from 0.2 mM to 0.5 mM, and the unlabeled protein was present in sufficient excess to ensure that the labeled protein was entirely in the bound state. ¹D_NH RDCs were measured using two-dimensional IPAP[¹⁵N/¹H]HSQC experiments (Ottiger et al., 1998). RDC values for each residue were obtained by taking the difference in the corresponding J-splittings measured in magnetically oriented (~10 mg/ml) Pf1 phage and isotropic in H₂O. The magnitudes of the axial component of the tensor, D_a^NH (−16.5 Hz) and rhombicity, η (0.45), components of the alignment tensor, were determined based on the NMR structure of VEK30/K2_hPg (PDB code:2KJ4) using the DC program (https://spin.niddk.nih.gov/bax/). For CSP and RDC data analyses, sequence-specific backbone ¹⁵N-¹H assignments for K2_hPg, published previously (Wang et al., 2010a), were used.

2.7.e. Restraint-based docking of the VEK50/K2_hPg complex

Restraint-based docking calculations for the complexes of VEK50 peptides with K2_hPg were performed on the HADDOCK web server (van Zundert et al., 2016) and the Xplor-NIH program. The topologies and starting structural coordinate files for input into HADDOCK were generated from the refined NMR structures. For generation of ambiguous interaction restraints, residues R¹⁷H¹⁸ of VEK50^△RH2 and residues R³⁰H³¹ of VEK50^△RH1 that are directly involved in interaction with K2_hPg were defined as active-site residues, and residues with significantly different CSP values (^△δ > 0.2 ppm) were defined as passive residues. For K2_hPg, residues exhibiting CSP values larger than 0.1 ppm were defined as active residues, i.e. the ones at binding site. The passive residues of K2_hPg were automatically picked by HADDOCK server program based on its 3D structure. RDC orientation constraints and intermolecular NOEs from VEK50 peptides and K2_hPg were used as unambiguous restraints for structure calculation to obtain the complex of VEK50^△RH2/K2_hPg and VEK50 ^△RH1/K2_hPg at molar ratios of 1:1. Segments of VEK50 peptides and K2_hPg that constituted these active and passive residues, plus two sequential residues on either side of the active and passive residues, were kept semiflexible during the docking steps. N- and C-residues of VEK50 peptides (residues 1–8 and 48–50) were kept flexible throughout the docking. The alignment tensor was calculated using the DC program with default parameters for singular value decomposition (SVD) fitting (https://spin.niddk.nih.gov/bax/). Intervector projection angle restraints for VEAN were generated with the python script provided in HADDOCK.

The simulated structure model of the VEK50/K2_hPg complex at a molar ratio of 1:2 was obtained from HADDOCK combined with the CSP data from NMR ¹⁵N-HSQC titration experiments for VEK50/K2_hPg, along with previous results for VEK30/K2_hPg (Wang et al., 2010b). The dihedral angle restraints for the bound-form of K2_hPg and VEK50 predicted from TALOS-N were also included in HADDOCK calculation. For purposes of docking, intermolecular NOEs and RDCs of K2_hPg bound at R¹⁷H¹⁸ of VEK50^△RH2, and K2_hPg bound at R³⁰H³¹ of VEK50^△RH1 were used to define the relative positions of the two K2_hPg molecules in their complexes with VEK50. The structural models were clustered and the best model with the highest HADDOCK score was selected to present the structure of the complex. The statistical data for the structures derived from HADDOCK are listed in Supporting Information Table S1.

2.8. Data deposition

The coordinates of the calculated structural ensembles for the bound-form of VEK50, as well as the structural ensembles for the complex of K2_hPg with VEK50^ΔRH1, and K2_hPg with VEK50^ΔRH2, have been deposited in the Protein Data Bank with accession codes 6OQ9, 6OQJ, and 6OQK respectively. Chemical shift assignments for the VEK50 peptides in complex with K2_hPg and the corresponding experimental restraints used in the structure calculation have been deposited in the BioMagResBank with accession numbers 30603, 30605, and 30606, respectively.

3. Results

3.1. Binding stoichiometry of the VEK50 peptides to K2_hPg

3.1.a. ITC titrations

VEK50 is a truncated peptide of PAM_AP53 that sequentially includes: [the C-terminus of the hypervariable region (HVR) - the entire A-domain, including its component a1a2-repeats and the N-terminus of the flexible B-domain]. This peptide begins at V⁹⁷ of PAM_AP53, which is residue-1 in VEK50 (Table 1). To quantitate binding of VEK50 to K2_hPg, ITC measurements were employed. The heats of binding were measured at 25° C as a function of molar ratio of K2_hPg to VEK50. As shown in Figure 1A, the binding isotherm is best-fit by a molar stoichiometry of 2:1 for K2_hPg:VEK50, with an average K_d value of 9 nM. In order to identify the binding properties of the RH motifs separately, two mutant peptides, VEK50^△RH1 and VEK50^△RH2, were designed by separately replacing RH1 (R¹⁷H¹⁸) or RH2 (R³⁰H³¹) in VEK50 (Table 1) with two Ala residues. This yields peptides, VEK50^△RH1 and VEK50^△RH2, respectively. Using these peptides in the titrations, a molar binding stoichiometry of 1:1 was observed for K2_hPg/VEK50^△RH1, with a K_d of 9 ± 2 nM (Figure 1B) and for K2_hPg/VEK50^△RH2 of 42 ± 4 nM was best-fit to the data (Figure 1C). These results confirmed that the RH motifs are the active-site residues that are directly involved in the binding to K2_hPg. The binding affinities of these two mutants also show that K2_hPg binds more tightly to RH2 than RH1.

Table 1.

Amino acid sequences of truncated peptides from PAM_AP53 and hPg

Peptides	Amino Acid Sequencesa
VEK50	GS(V1EKLTADAELQRLKNER17H18EEAELERLKSER30H31DHDKKEAERKALEDKLAD)Y
VEK50ΔRH1	GS(V1EKLTADAELQRLKNEA17A18EEAELERLKSER30H31DHDKKEAERKALEDKLAD)Y
VEK50ΔRH2	GS(V1EKLTADAELQRLKNER17H18EEAELERLKSEA30A31DHDKKEAERKALEDKLAD)Y
K2hPg	Y-7VEFSEE(C1MHC4SGENYDGKISKTMSGLECQAWDSQSPHAHGYIPSKFPNKNLKKNYCRNPDRE56LRPWCFTTDPNKRWEY72 CDIPRC78)AA

^aResidues outside the parentheses in all cases are exogenous to PAM or K2_hPg.

Fig. 1. Isothermal titration calorimetric (ITC) analysis of the binding of VEK50 peptides to K2_hPg.

Binding isotherms at 25° C for the titration of K2_hPg (900 μM stock) into: (A) 40 μM VEK50 in the cell, (B) 40 μM VEK50^△RH1 in the cell, and (C) 40 μM VEK50^△RH2 in the cell. The heats liberated upon binding were measured as a function of the concentration of K2_hPg. The red lines in Panels A-C represent the best-fits of the experimental data (black symbols) to a two-site model (A) and one-site models (B, C).

3.1.b. AUC analyses

The binding stoichiometry between VEK50 peptides and K2_hPg was further assessed by determining the molecular masses of the complexes of VEK50/K2_hPg using analytical AUC. The complexes of VEK50 peptides with K2_hPg were prepared using three different initial concentrations of K2_hPg (13 μM, 19 μM, and 25 μM) added to VEK50 (5 μM, 7.5 μM, and 10 μM), each representing a ~2.5-fold molar excess of K2_hPg/VEK50 at different total concentrations of the complex. The complexes were separated from the nonbound materials by gel filtration. The theoretical molecular mass of K2_hPg is calculated to be 10,151 Da, that of VEK50 is calculated as 6,159, and, for the two mutant VEK50 peptides, the calculated molecular masses are 6,008 Da, all in agreement with the MALDI data listed in Table 2. From the AUC data of Table 2, it is clear that a 2:1 molar complex of K2_hPg/VEK50 is found, whereas for the two mutant complexes, viz., VEK50^△RH1/K2_hPg and VEK50^△RH2/K2_hPg, the molecular masses of the isolated complexes corresponded to a 1:1 molar stoichiometry (Table 2). In all cases, a single molecular mass species was found throughout the concentration gradient in the centrifugation cell. This single species that exists over a wide concentration range supports the ITC data showing that very tight binding occurs.

Table 2.

Molecular masses of truncated peptides from PAM_AP53 and the complexes with K2_hPg

Peptide	Theoretical (Da)	MALDI-TOF (Da)a	AUC (Da)b
			Apo-peptides	Peptide + K2hPg
VEK50	6,159	6,166	6,300 ± 800	27,300 ± 850
VEK50ΔRH1	6,008	6008	5,600 ± 500	16,150 ± 1,000
VEK50ΔRH2	6,008	6010	5,700 ± 600	16,400 ± 1,000
K2hpg	10,151	10,155	10,260 ± 800	above

^aMolecular masses of the protein linear sequence from MALDI-TOF.

^bWeight-average molecular mass of proteins in solution from AUC experiments at 25° C.

3.1.c. HSQC titrations of VEK50 peptides to K2_hPg

1H/¹⁵N-HSQC experiments were recorded on ¹⁵N-VEK50 peptides with increasing unlabeled K2_hPg at 1:0.5, 1:1, 1:1.5, 1:2.0, and 1:2.5 molar ratios of VEK50:K2_hPg respectively (Figure 2A–D; Supporting information Figure S1). During the titrations the chemical shifts of residues at the N- and C-terminii, viz., V¹-K³ and D⁴⁹-Y⁵⁰, respectively, change only slightly and only one peak for each residue is observed in NMR spectra. This suggests that these residues experience fast exchange (Figure 2A). However, the amide resonances of most residues in VEK50 peptides shift significantly and start to appear at two locations. As one example of the data, when K2_hPg was titrated to 1:0.5 ratio, residue A⁴⁸ in VEK50 (Figure 2B), -VEK50^△RH1 (Figure 2C), and -VEK50^△RH2 (Figure 2D) showed two peaks, corresponding to the amide resonances in the apo- and bound-forms. respectively. After a 1:1 molar ratio was reached, all of the amide resonances in VEK50^△RH1 and VEK50^△RH2 are shifted to the positions of the bound forms, as exemplified by A⁴⁸ (yellow) in Figure 2C, D. These residues remained at the same locations when additional unlabeled K2_hPg was added to the system. Meanwhile, amide resonances of some residues in VEK50 show three peaks indicating the existence of a mixture containing apo-, 1:1, and 1:2 molar complexes of VEK50 to K2_hPg. These residues are located in both the a1- and a2-repeats, suggesting that K2_hPg similarly binds to each of these two motifs. The amide resonances were continually shifted until a 1:2 molar ratio of VEK50/K2_hPg was reached, as exemplified by A⁴⁸ (colored blue) in Figure 2B. These results further confirm that the VEK50 peptide within PAM_AP53 contains two binding sites for K2_hPg and that the binding is mediated by RH motifs in each repeat of the a1a2 domain. For each of the two VEK50 mutants, a 1:1 binding stoichiometry is also confirmed.

Fig. 2. NMR spectroscopic analysis of the interactions between VEK50 peptides and K2_hPg.

The stoichiometry between VEK50 and K2_hPg was confirmed from chemical shift changes of some residues in overlaid ¹⁵N-HSQC spectra of: (A, B) ¹⁵N-VEK50; (C) ¹⁵N-VEK50^△RH1; and (D) ¹⁵N-VEK50^△RH2 by titrating ¹⁵N-labeled VEK50 peptides with unlabeled K2_hPg at molar ratios of (VEK50 peptide:K2_hPg) of 1:0.5 (orange), 1:1 (yellow), 1.1.5 (green), 1:2 (cyan), and 1:2.5 (blue). The sample buffer was 20 mM Bis-Tris-d19, pH 6.8 at 25° C. (D-F) Chemical-shift differences (^△δ ppm) between apo- and bound-forms of VEK50 peptides. The ^△δ ppm values of: (E) WT-VEK50, (F) VEK50^△RH1, and (G) VEK50^△RH2 are plotted against the residue numbers. (H–J) Plots of ¹D_NH RDCs as a function of the residue number for ¹⁵N-VEK50 (H), ¹⁵N-VEK50^△RH1 (I), and ¹⁵N-VEK50^△RH2 (J), each in complex with K2_hPg at molar ratios of 1:2, 1:1, and 1:1. The measurements were performed on samples in filamentous phages Pf1 (10 mg/ml) using two-dimensional IPAP[¹⁵N/¹H]HSQC experiments (Ottiger et al., 1998). Residues with severe peak overlap or poor signal-to-noise ratios were excluded.

On the basis of these titration experiments, the combined CSPs for the three VEK50 peptides are summarized in Figure 2 (E–G). These data show that residues displaying large chemical shift perturbations upon binding to K2_hPg are not limited to R¹⁷H¹⁸ in the a1-repeat and R ³⁰H³¹ in the a2-repeat, but also occur in other residues in that region, demonstrating the existence of probable exosites identified using this longer peptide from PAM_AP53. The results also demonstrate that conformational transitions occur in VEK50 peptides upon binding to K2_hPg. For example, the residues in VEK50^△RH1 showing most significant CSPs are in the RH2 binding region, i.e., residues L²⁶-K³⁶ (Figure 2E). In the mutant VEK50^△RH2 the chemical shift changes were found between residue A⁸ to E²⁰, which consists of the RH1 binding motif (Figure 2F). Since WT-VEK50 contains both the RH1 and RH2 binding motifs, most of the residues in VEK50, except those in the N-, and C-terminal regions, show significant chemical shift changes at most locations upon binding to K2_hPg (Figure 2D).

In addition, ¹H-²H exchange experiments were performed for three VEK50 peptides in the apo-forms and in the complex with K2_hpg by dissolving lyophilized sample in ²H₂O, followed by temporal acquisition of HSQC spectra. For the apo-form peptides and the complex of two mutant VEK50 peptides with K2_hPg, amide resonances disappear during the first 20 min period. Meanwhile, amide resonances of some residues in the WT-VEK50/K2_hPg complex remain, and those from residues L²⁶, K²⁷, and E²⁹ are still detectable after 20 hr (Figure 3), indicating that these residues become less exposed to solvent upon VEK50 binding with two K2_hPg molecules.

Fig. 3. ¹H/²H exchange of VEK50.

Overlays of the ¹H-¹⁵N HSQC spectra of the complex of ¹⁵N-VEK50/unlabeled K2_hPg obtained in BisTris-d19, pH 6.8 (black), at 20 min (red) and 24 hr (blue) after the lyophilized complex powder was re-dissolved in ²H₂O at 25° C. The assignments of the backbone amide signals are indicated by the respective single-letter codes and residue numbers.

3.2. Binding sites in K2hPg for a1- and a2-repeats

3.2.a. Chemical shift differences show that RH1 and RH2 bind to the same LBS of K2_hPg

Chemical shift changes in K2_hPg are also ligand-induced, and primarily found in the LBS (Figure 4). The binding of K2_hPg to either wild-type VEK50 or its mutant peptides, VEK50^△RH1 and VEK50^△RH2, is detected in HSQC spectra by monitoring changes in ¹H/¹⁵N chemical shifts from ¹⁵N-labeled K2_hPg as a function of the concentration of VEK50 peptides (Supporting Information Figure S2). Three segments in K2_hPg, viz., G³⁴-K⁴⁶, D⁵⁴-W⁶⁰, and K⁶⁸-E⁷¹, exhibit the most dramatic CSPs upon binding to VEK50 peptides, suggesting these are binding residues in K2_hPg (Figure 4). Several residues in these regions, such as D⁵⁴ and R⁵⁵, show weak or very broad signals in the apo-form of K2_hPg, but shift to a new position in the bound-form (Supporting Information Figure S2). These results are consistent with previous studies for the K2_hPg/VEK30 complex (Wang et al., 2010b), but, additional residues in K2_hPg are affected when K2_hPg binds to the larger VEK50 peptides, further indicating available exosite interactions. The CSP data also suggest that K2_hPg uses the same domain for binding to either VEK50^△RH1 or VEK50^△RH2 (Figure 4A,B). When three HSQC spectra of K2_hPg bound to VEK50 and two mutant peptides were overlaid (Supporting Information Figure S2C), most of peaks appeared at the same chemical shifts. However, in the HSQC spectrum of K2_hPg bound to VEK50, the residues with larger CSP always show at two locations with different peak intensities. For examples, D⁵⁴, L⁵⁷, and W⁶⁰ of K2_hPg have two fresonance signals corresponding to two K2_hPg molecules bound to RH1 or RH2 of VEK50 respectively (Figure 4C). Notably, the signals from K2_hPg bound to RH2 show a 3–5-fold stronger peak intensities, suggesting that in the presence of excess of VEK50, RH2 is more favorable for K2_hPg binding. This is also consistent with the ITC results for two mutant VEK50 peptides. Thus, the difference of K_d values observed from the mutants truly represents the stronger binding ability of the RH2 site in the WT-VEK50, which is not affected by the steric crowding of two K2_hPg molecules in the complex. Slight differences are also observed in the segment comprising residues G⁶ - N⁸, which is close to the flexible N-terminus, therefore, the differences are likely not directly related to ligand binding.

Fig. 4. Differences in the binding sites of K2_hPg upon its interaction with VEK50^△RH1 and VEK50^△RH2.

Combined chemical shift changes (^△δ ppm) along the sequence of K2_hPg between apo-K2_hPg and K2_hPg bound to: (A) VEK50^△RH1 and (B) VEK50^△RH2. (C) Residues in the VEK50-bound K2_hPg exhibit two resonance signals as shown in the expanded ¹⁵N-HSQC spectra. (D) The refined solution structural backbones of K2_hPg bound to VEK50^△RH1 (red) and VEK50^△RH2 (green) are overlaid and shown in as ribbons.

3.2.b. The binding interface in K2_hPg for RH1 and RH2

Although only one binding domain is detected in K2_hPg by NMR titration, the binding interface formed between K2_hPg and the a1- or a2-repeat in two mutant VEK50 peptides is somewhat different. The differences are observed from the NMR solution structures of K2_hPg in the complex with the two mutant peptides, which are refined by residual dipolar coupling (RDC) data. A total of 63 amide RDCs (¹D_NH) from ¹⁵N-K2_hPg/unlabeled VEK50^△RH1 and 60 amide RDCs (¹D_NH) from ¹⁵N-K2_hPg/unlabeled VEK50^△RH2 were obtained from RDC measurements using 2D IPAP- HSQC (Supporting information Figures S3 A, B). The RDC data obtained from these two complexes were fitted to the solution structure of the bound-form of K2_hPg in complex with VEK30 (PDB code: 2kj4) (Wang et al., 2010b), which contains only the RH1 binding motif. It was found that residues on N- and C-termini exhibit larger differences (>5 Hz) between the calculated and observed RDCs (¹D_NH). The residues in the ligand binding region, such as D⁵⁴-L⁵⁷ and W⁶⁰, which show differences in the sign and intensity of the absolute RDC values in Figures S3 A,B, also exhibit larger differences (>3 Hz) when the experimental RDCs are fitted to the bound-form of K2_hPg (Figures S3 C,D). Therefore, we used these experimental RDCs to refine the bound-form of K2_hPg with VEK30 in order to provide more precise structures for K2_hPg with VEK50 peptides. As depicted in Figure 4D, the binding region (colored red and green) in the two complexes can be overlaid with a RMSD of 1.30 Å, while the rest of K2_hPg, except the residues at the flexible N- and C-termini, which are exogenous to K2_hPg, have a RMSD of 0.6 Å. Additionally, the binding region, D⁵⁴-P⁵⁹, has a more open conformation in the complex with VEK50^△RH1.

3.3. Changes in solution structure of VEK50 peptides upon binding to K2_hPg

3.3.a. Apo-structures of three VEK50 peptides

The solution structures of VEK50 peptides in their apo-forms were calculated based om NMR data (Figure 5). The residues in the N-terminal region, viz., V¹ - D⁷, and in the C-terminal region, viz., D⁴⁹ - Y⁵⁰, have only sequential connectivities, suggesting that these regions exist as coils. In apo-VEK50, three short α-helical segments, viz., A⁸ - K¹⁴, E²² - R³⁰, and K³⁷ - K⁴⁶, are connected through two flexible loops, which contain the RH1 and RH2 motifs (Figure 5A). The overall structures of two mutant peptides are very similar to that of WT-VEK50, although one of the RH groups has been replaced by Ala-Ala residues. The RH2 group of VEK50^△RH1 (Figure 5B) and the RH1 group in VEK50^△RH2 are also in the loop regions in the apo forms (Figure 5C).

Fig. 5. Comparison of the solution structures of apo- and K2_hPg-bound VEK peptides.

(A, D) VEK50; (B, E) VEK50^△RH1; and (C, F) VEK50 ^△RH2. A, B, and C represent the apo-forms of the indicated peptides; D, E, and F represent the peptides bound to K2_hPg. The RH1 and RH2 backbones are shown as red sticks and Ala-Ala backbones are colored green in the ribbons.

3.3.b. Bound structures of three VEK50 peptides

The binding regions, RH1 and RH2, in VEK50 undergo significant structural changes and become more rigid upon its binding to K2_hPg. Thus, as shown in Figure 5D, the helical segment from residue A⁶ - D³² in VEK50 is formed and two RH groups (R¹⁷H¹⁸ and R³⁰H³¹) are positioned on opposite faces of the helix. The mutant peptides also exhibit conformational changes but those changes mainly occur at the binding regions (Figure 5E,F). The alanine replacements in VEK50^△RH1 enhance the rigidity of the region consisting of A⁶ - E¹⁹ in the apo-form. However, the conformation of this region does not significantly change and remains bent in the complex. The conformation of VEK50^△RH2 exhibits high similarity to WT-VEK50 in its bound form, especially the RH1 binding regions.

Residual dipolar coupling (RDC) data provide additional structural and dynamic information in a μsec timescale for the bound-form VEK50 peptides. The plot of ¹H-¹⁵N RDCs as a function of the residue number (Figure 2H–J) confirm that most of the residues in the a1a2- repeat of VEK50 have RDC absolute values above 5 Hz, while the residues at N- and C-terminal are flexibly disordered, presenting significantly smaller RDCs (< 3Hz) (Figure 2H). The plot of the RDCs for VEK50^△RH1 show that the residues presenting large RDCs are not limited to those in the binding region, i.e., the a2-repeats (Figure 2I), whereas the residues in a1-repeat of VEK50^△RH2 exhibit significantly larger RDCs (Figure 2J), consistent with the more rigid structure formed upon binding.

All three peptides in the bound form contain a helical segment (L²³ - E²⁹) between the two RH motifs. The helical feature gained in the bound form of these peptides fades at the end of the a2-repeat. Residues 1–7 at the N-terminus and residues 49–50 at the C-terminus are flexible and show sharp resonance signals in HSQC spectra. But residues 44–48 are always present at two locations in HSQC spectra, indicating that this fragment is in slow exchange between locally different structures in the bound form.

3.4. HADDOCK calculated structural models of the K2_hPg/VEK50 complexes

In this study, we present a structural basis of the binding of VEK50, an active fragment of PAM_AP53, to K2_hPg and provide evidence for the high affinity and specificity of this peptide. From a previous study of the VEK30/K2_hPg complex (Wang et al., 2010b), the hydrophobic groove in K2_hPg formed by Y³⁵, F⁴⁰, W⁶⁰, W⁷⁰, and Y⁷² in combination with the anionic center of D ⁵⁴ and D⁵⁶, comprised the binding site of K2_hPg/RH1. These same residues in K2_hPg also exhibit significant chemical shift perturbations upon its binding to either the a1-repeat in VEK50^△RH2 or a2-repeat in VEK50^△RH1 (Figure 4), suggesting that there is only one potential interaction site for K2_hPg in the complex with VEK50 peptides.

For the VEK50 peptides, the active-site residues, which directly participate in binding, were assigned based on the mutations studied previously and the CSPs observed from NMR titrations in this study (Figure 2). It is noted that both VEK50^△RH1 and VEK50^△RH2 contain the complete a1- and a2-repeats, except that one of the RH groups has been replaced by Ala-Ala, thus inactivating the K2_hPg binding site in each repeat. Thus, only the fragment having the intact RH motif exhibits large CSPs upon the binding to K2_hPg (Figure 2F, G). This further confirms that the RH motifs are critical residues for tight K2_hPg binding. Other residues, i.e., 10–22 in VEK50^△RH1 and 24–34 in VEK50^△RH2, with dramatic chemical shift changes, were also directly related to binding, and were defined as passive residues for docking calculations.

The interface of VEK50 peptides and K2_hPg was also identified by the intermolecular NOEs between ¹³C/¹⁵N-K2_hPg and unlabeled VEK50 peptides, or ¹³C/¹⁵N-labeled VEK50s and unlabeled K2_hPg. For example, methyl resonances from L²⁶ of VEK50^ΔRH1 exhibited NOEs to residues H³³, G³⁴, Y³⁵, and I³⁶ of K2_hPg. Similar NOEs were also observed form L¹³ of VEK50^ΔRH2 with corresponding residues in K2_hPg, suggesting that these residues are in close contact with each other. Using ¹³C/¹⁵N-filtered, ¹³C/¹⁵N-edited NOESY-HSQC experiments, 12 and 14 intermolecular NOEs between K2_hPg/VEK50^△RH1 and K2_hPg/VEK50^△RH2 were observed, respectively, and used as unambiguous distance restraints.

In addition to these distance restraints, RDC data collected from ¹⁵N-labeled K2_hPg (Figure 4C) and ¹⁵N-labeled VEK50 peptides (Figure 2 H–J) in the corresponding complex were included to provide the orientation restraints for docking. When the complex samples were oriented in Pf1 phage medium, no noticeable structural perturbations were observed. A total of 90 and 93 RDCs (¹D_NH) obtained from the complexes of VEK50^ΔRH1/K2_hPg, and that of VEK50^ΔRH2/K2_hPg, respectively, were applied as orientation restraints in docking. The resulting structures calculated in HADDOCK showed no distance violations and only residues in the mobile N-terminal and the C-terminal regions in K2_hPg and VEK50 peptides show RDC outliners (>3 Hz).

Previous work has shown that the LBS of K2_hPg contains an anionic center that interacts with positive side-chains of the pseudo-lysine ligand. In the modeled structures of the pseudo-lysine binding sites, the anionic center of K2_hPg (D⁵⁴, D⁵⁶) is involved in electrostatic interactions with side-chains of residues, K¹⁴, R¹⁷, and H¹⁸ in the a1-repeat (Figure 6A) and K²⁷, R³⁰ and H³¹ in the a2-repeat (Figure 6B). In these structural clusters, hydrogen bonds can also form between the side-chains of these same residues in the a1- or a2-repeat with carbonyl groups of D⁵⁴ and D⁵⁶ in K2_hPg . Meanwhile, in VEK50^ΔRH2, acidic residues D⁹ of the a1-repeat and E²⁰ of the a2- repeat are within close contact distances with basic residues R⁵⁵ and R⁶⁹ of K2_hPg, respectively (Figure 6A). Although VEK50^ΔRH1 only contains one binding motif, viz., RH2, the acidic residue, D⁹, and basic residues, R¹² and K¹⁴ in the a1-repeat are all present in this mutant peptide. Since this region is structured as a bent helical fragment, these residues could act as exosites through electrostatic interactions with K2_hPg as illustrated between R¹² of VEK50^ΔRH1 and D⁵⁴ of K2_hPg (Figure 6B). Additionally, D³⁴ in the a2-repeat of VEK50^ΔRH1 forms an H-bond with the side-chain of R⁶⁹ of K2_hPg. Overall_, more electrostatic interactions were found between VEK50^ΔRH1 and K2_hPg.

Fig. 6. Solution binding models of the K2_hPg/VEK50 peptides derived from Xplor-NIH and HADDOCK.

The lowest-energy conformation was used for the representation of: (A) K2_hPg /VEK50^ΔRH2 and (B) K2_hPg/VEK50^ΔRH1. (C) Superposition of backbone traces of the 20 lowest-energy NMR structures of K2_hPg /VEK50 is shown. VEK50^ΔRH1 and VEK50^ΔRH2 bind to K2_hPg at a molar ratio of 1:1, whereas VEK50 binds to K2_hPg at a molar ratio of 1:2. Residues having H-bond interactions with K2_hPg and having close contact (~3 Å) are labeled and shown as sticks. K2_hPg is colored yellow whereas the a1- and a2-repeats of VEK50 are colored magenta and cyan, respectively.

To dock two K2_hPg molecules onto VEK50, we used the multi-body interface module on the HADDOCK 2.2 web server. As described earlier, a semi-flexible mode of docking was performed, where the side-chains of the active and passive residues are designated as flexible. The bound-form of VEK50 was determined by NMR (Figure 5D), and the bound-form K2_hPg was refined using RDCs. At least, 24 intermolecular NOEs collected from VEK50^△RH1/K2_hPg and VEK50^△RH2/K2_hPg were used to define each binding interface in VEK50 with K2_hPg. The resulting structural model using 20 of the lowest energy structures of VEK50:K2_hPg at a molar ratio of 1:2 is shown in Figure 6C. In order to validate the HADDOCK-calculated structures, the observed RDCs of VEK50 were compared with predicted RDCs based on the model and they correlated well with a low Q-factor of 0.5 (Supporting information Figure S4). In these binding models, no further major conformational changes were observed for the NMR-determined bound structures of either VEK50 or K2_hPg. The bound-form of VEK50 exists as an α-helix between residue D⁷-D³⁴, while two K2_hPg molecules were bound to opposite faces of the helix. Since both of the a1- and a2-repeats of VEK50 became more structured, these two fragments interact with K2_hPg in a very similar manner. For example, the H-bond of R¹⁷ and R³⁰ in VEK50 with the LSB of K2_hPg can be consistently formed in the simulated structure model. The interactions identified from this structural model are consistent with the observation from NMR experiments, in which the residues in the K2_hPg bound to RH2 always show stronger resonance signals. Additionally, L²⁶ and K²⁷ of VEK50 are located in the middle of the helix, are buried in the contact surface of the a2-repeat and K2_hPg. This finding is consistent with their slower ¹ H/²H exchange rate when present in complexes with K2_hPg.

Thus, from these docking calculations, we conclude that two RH motifs have separate clusters to form the required pseudo-lysine ligand that interacts with the LBS of K2_hPg. RH1 (R¹⁷-H¹⁸) in the a1-repeat combines with D⁹, K¹⁴, and E²⁰ with to form the lysine isostere in VEK50^△RH2, while RH2 (R³⁰-H³¹) has another group of residues, including E²⁴, K²⁷, and D³⁴ to form its own lysine isostere in VEK50^△RH1.

4. Discussion

The ubiquitous high copy number surface M-protein of GAS, encoded by the emm gene, is a major virulence factor in the pathogenesis of these bacteria, primarily due to its ability to protect against various human host innate responses to infection. Further, the ability of M-proteins to interact with host cells and host proteins confers unique properties to the bacterial cells that allow their invasion and dissemination. There are more than 250 serotypes of M-proteins, primarily identified by the highly variable 5’-nucleotide sequences of the emm gene, and at least five subfamily patterns of M and M-like proteins (A-E), revealed by nucleotide sequences of the more conserved 3’-nucleotide sequences encoding peptidoglycan spanning domains of these proteins (Bessen, 2016). These serotypes and pattern types are of epidemiological value through allowing characterization of infectious breakouts and their possible consequences.

The mature form of M-protein is covalently anchored to the bacterial cell wall at the C-terminus of the protein via a typical sortase A-catalyzed linkage. This leaves the N-terminal region of the elongated M-protein to protrude through the GAS outer capsule to interact with host proteins. The sequence variety at the N-termini of M-proteins allows different types of interactions with the host to occur, and there are further differences depending on the serotype of the M-protein. For example, complement-mediated opsonization inhibitors, e.g., C4BP and Factor H, interact with the N-terminal hypervariable regions (HVR) of some emm-expressed M-proteins of Patterns A-C and E GAS strains (Buffalo et al., 2016; Gustafsson et al., 2013), and in this manner inhibit opsonization of the bacteria. These same complement inhibitors do not bind to the emm gene product (PAM) of Pattern D strains, but do functionally interact with other M-like proteins of these strains, e.g., Enn and Fba, each present along with PAM in the multiple gene activator (mga) regulon, and thereby perform this same function (Agrahari et al., 2013; Liang et al., 2013).

Our major interest in GAS-host interactions is centered in the important hemostasis/inflammation responses to GAS infections. Some Pattern A-C strains of GAS interact with fibrinogen in their centralized B-domain (Glinton et al., 2017) and in this way provide anti-opsonic activity to the microorganism (Courtney et al., 2006; Whitnack and Beachey, 1982). All pattern D strains of M-proteins, e.g., PAM from strain AP53, are unique in their ability to bind hPg tightly and directly to their N-terminal A-domains, with profound consequences to their virulence. Activation of hPg to hPm on the GAS surface by GAS-secreted SK2b generates a surface bound protease, hPm, that can aid dissemination in the host by digesting extracellular matrix proteins and cellular epithelial and endothelial tight junctions through this new proteolytic front.

Pattern D M-proteins exist as elongated fibrous proteins, which, at least in solution, have been identified as irregular weakly associated helical coiled-coils (Ghosh, 2011; McNamara et al., 2008; Qiu et al., 2019; Stewart et al., 2016). The COOH-terminal C- and D-domains of PAM that are present close to the GAS capsule, are primarily responsible for dimerization, with large irregularities in the PAM NH₂-terminal HVR-, A-, and B-domains. These irregularities, also seen in coiled-coil proteins such as tropomyosin (Nitanai et al., 2007), appear to allow binding of fibrinogen to B-domains of some M-proteins (McNamara et al., 2008), and thus are of functional significance.

A critical interaction in survival and virulence of Pattern D strains of GAS is the binding of hPg to PAM (Sun et al., 2004) and our work has focused on the nature of this interaction as shown in numerous published studies. The question addressed in this manuscript is the relevance to binding of each of the two adjacent hPg binding sites in the A-domain of PAM. We have employed reductionist approaches in these studies by utilizing peptide secondary structural stretches of PAM, along with the PAM binding K2 domain of hPg. Since PAM is composed solely of secondary structure, with no observed tertiary structure, and K2_hPg is an independent domain in hPg (Castellino et al., 1981), this approach should isolate intact binding sites to allow us to examine the properties of each of the regions of interest in these proteins.

In previous studies using X-ray crystallography (PDB code 1I5K) (Rios-Steiner et al., 2001) and NMR (PDB code 2KJ4) (Wang et al., 2010b), very similar structural models of the K2_hPg/VEK30 complex have been proposed. The solved structures suggest that there are two interaction clusters formed in the complex. The first is the anionic center of the LBS of K2_hPg formed by residues D⁵⁴ and E/D⁵⁶, which interact with a cationic cluster of side-chains, viz., K¹⁴, R ¹⁷, and H¹⁸ of VEK30. The second is the cationic locus of the K2_hPg binding pocket, R⁶⁹ which forms a salt-bridge with E²⁰ of VEK30. The role of each residue in the a1-repeat to the binding to K2_hPg has been investigated by a variety of mutants, but less has been done for the a2-repeat. From our current study, the interactions between residues K²⁷, R³⁰, and H³¹ in the a2-repeat act as K¹⁴, R¹⁷, and H¹⁸ in the a1-repeat. The interactions between the LBS of K2_hPg and K¹⁴, R¹⁷, H¹⁸, and E²⁰ play major roles in forming a rigid helical structure in the a1-repeat. In the bound form of VEK50, RH1 is in the middle of a rigid helical region from the N-terminal of a1 (D⁷) to the C-terminal of a2 (D³²), while RH2 is close to the flexible C-terminal of a2-repeat. Although both H¹⁸ in a1 and H³¹ in a2 have close contact with the residues of LBS, the H-bond between these two His residues and the LBS are not observed in all models. Meanwhile, the H-bond of R¹⁷ and R³⁰ with the LBS of K2_hPg can be consistently formed in all of the lowest energy simulated structures.

VEK50 is a truncated peptide from PAM_AP53, which contains both the a1- and a2-repeats. The replacement of either RH1 or RH2 by Ala-Ala slightly changes the binding affinity of mutant VEK50 peptides to K2_hPg. The differences observed from their binding affinities to K2_hPg are related to the secondary structure of these truncated peptides. VEK50 can bind with K2_hPg on the a1- and/or a2-repeats, while its mutants can only bind using either a1 or a2. As discussed above, the a2-repeat has more interactions than the a1-repeat with the LBS in K2_hPg. Thus, compared to VEK50^△RH2, VEK50^△RH1 binds to K2_hPg tighter via its complete a2-repeat. As shown in the apo-forms, two RH binding motifs are located in flexible regions in apo-VEK50 and become rigid in the bound-form to obtain complete interactions. The interactions between the LBS of K2_hPg and K¹⁴, R¹⁷, H¹⁸, and E²⁰ play the major roles in forming a rigid helical structure in the a1-repeat. When the RH1 is inactivated by Ala-Ala substitutions, residues in the a1-repeat remain as a flexible loop. The selective inactivation of RH2 results in a peptide that can only use its a1-repeat to bind to LBS in K2_hPg. Similarly, this a1-repeat alone appears to have fewer interactions with K2_hPg compared to the a2-repeat. Therefore, VEK50^△RH1 has a slightly higher affinity than VEK50^△RH2 for K2_hPg.

Further affirmation that the entire hPg binding ability is contained within the A-domain of PAM-type M-proteins is evidenced by studies with another class of M-protein, M1 from a Pattern A GAS strain. M1 does not directly interact with hPg, but instead tightly binds to fibrinogen via its B-domain. We replaced the entire B-domain of M1 with the A domain of PAM_AP53. This mutated M1 no longer interacted with fibrinogen but interacted with hPg nearly identically to PAM_AP53 (Chandrahas et al., 2015). These results affirm that the a1a2 module in PAM proteins presents the complete epitope with regard to hPg binding, whereas effects from other domains in PAM are of little influence in this regard. In the wider scope, we suggest that the domains of M-proteins have evolved independently and have been incorporated into other M-proteins by recombination to generate new GAS strains with advantages for their survival against the immune responses of the host (Bao et al., 2016a; Bao et al., 2016b). The variability of M-proteins, that are relatively rapidly genetically adapted to survive in various host human niches, lies at the basis of the numerous serotypes of GAS that are singularly directed to humans, who cannot evolve as rapidly to combat these bacteria. In addition, the ability of GAS to undergo rapid genetic adaptation is detrimental to M-protein-directed vaccines to combat these infections.

5. Conclusions

Group A Streptococcus (GAS) is a bacterial pathogen unique in its remarkable ability to exploit a range of cellular and tissue environments to establish colonization and progress to a wide spectrum of disease states in the human host. A major determinant of GAS host tropism and virulence is the M- and M-like protein (PAM), expressed on the surface of all GAS strains. M-protein forms have evolved to serve highly specialized roles in GAS survival, including its specific ability to recruit and activate hPg. In this report, we provide a detailed structural view of the VEK50/K2_hPg complex, a reductionist approach to understanding the nature of the PAM/hPg complex, that offers insights at the molecular scale for the binding mechanism of PAM to hPg. Broadly considered, our findings suggest that the domains of M-proteins have likely been exchanged through horizontal recombination events and independently further evolved in order that emerging GAS strains optimize survival advantages against the host immune response (Bao et al., 2016a; Bao et al., 2016b). The ability of GAS to utilize multiple genetic mechanisms to produce highly variable M-protein forms that successfully allow GAS to adapt to survive in various host human niches, lies at the basis of the numerous serotypes of GAS that are singularly directed to humans, who cannot evolve as rapidly or as specifically to combat these bacteria. Studies aimed at elucidating the mechanistic details of how GAS virulence determinants, such as PAM, engage host targets will provide future insights into improved anti-virulence strategies as well as alternative approaches to improve current vaccine strategies against GAS.

Highlights.

Solution structural determination by high-resolution NMR of the A-domain of PAM reveals occupancy by two molecules of K2_hPg on opposite faces of the A-domain helix.

K2_hPg binds to PAM through repeating sequences (a1a2-repeats) in the A domain of PAM and uses the same lysine binding site for its interaction with each of the two a-repeats.

K2_hPg binds more tightly to the a2-region than the a1-region due to different sets of exosite interactions.

Abbreviations used:

GAS

Group A Streptococcus pyogenes

hPm

human plasmin

hPg

human plasminogen

K2_hPg

kringle 2 domain of human plasminogen

LBS

lysine binding site

PAM

plasminogen-binding group A streptococcal M-like protein

SPR

surface plasmon resonance

CSP

chemical shift perturbation