Sphingomonas sp. KT-1 PahZ2 Structure Reveals a Role for Conformational Dynamics in Peptide Bond Hydrolysis

Abstract

Poly(aspartic acid) (PAA) is a common water-soluble polycarboxylate used in a broad range of applications. PAA biodegradation and environmental assimilation were first identified in river water bacterial strains, Sphingomonas sp. KT-1 and Pedobacter sp. KP-2. Within Sphingomonas sp. KT-1, PahZ1_KT-1 cleaves β-amide linkages to oligo(aspartic acid) and then is degraded by PahZ2_KT-1. Recently, we reported the first structure for PahZ1_KT-1. Here, we report novel structures for PahZ2_KT-1 bound to either Gd³⁺/Sm³⁺ or Zn²⁺ cations in a dimeric state consistent with M28 metallopeptidase family members. PahZ2_KT-1 monomers include a dimerization domain and a catalytic domain with dual Zn²⁺ cations. MD methods predict the putative substrate binding site to span across the dimerization and catalytic domains, where NaCl promotes the transition from an open conformation to a closed conformation that positions the substrate adjacent to catalytic zinc ions. Structural knowledge of PahZ1_KT-1 and PahZ2_KT-1 will allow for protein engineering endeavors to develop novel biodegradation reagents.

Graphical Abstract

INTRODUCTION

Polymers are ubiquitous in our daily lives and are easily recognized in the form of packaging materials and single-use containers. The general public is acknowledging that these versatile synthetic products do not naturally decompose and are readily accumulating in landfills and the environment, as evidenced by the ever expanding Great Pacific Garbage Patch.^1,2 These tangible items have rightfully garnered significant interest about their environmental impacts,^3-6 shortcomings associated with recycling,⁷ and potential benefits through biodegradation of persistent plastics like poly(ethylene terephthalate).^8-12 However, perhaps equally as important is to investigate those polymers which we cannot see and are being directly or indirectly released into the environment: water-soluble polymers (WSPs).^13,14 Just because they are not seen does not warrant labeling all WSPs as being environmentally friendly or biodegradable.¹⁵

WSPs are found in a myriad of industrial applications ranging from washing agents to wastewater treatment to pharmaceuticals.¹³ The broad use of WSPs has increased as market demand has shifted away from water-insoluble polymers that are visibly damaging to ecological systems. This transition is motivated by the assumption that WSPs are environmentally friendly, biodegradable, and generally nontoxic. However, it is well-known that water-soluble polymers such as polyacrylamides and polycarboxylates exhibit slow biodegradation contributing to significant environmental persistence.^16,17 The negative effects of specific biodegradation products are clear in some cases; polyacrylamide breakdown yields the potent neurotoxin acrylamide. Acrylamide monomer concentrations in wastewater derived from oil-and-gas industrial applications have been reported to exceed the threshold for ecotoxic effects at 10–1000 mg/L.^14,18,19 In other cases, increased WSP use may lead to increased environmental emissions leading to unintended and unexpected ecological consequences not related to toxicity effects.¹⁴ Ecological effects may instead be related to WSP behavior as flocculants, detergents, or water/soil conditioners in environments extending beyond those intended.

Polycarboxylates like poly(acrylate) and its derivatives are WSPs that have found utility as superabsorbent materials in diapers and feminine hygiene products in addition to being used as detergents, as antiscaling agents, and as a fertilizer synergist.^20-22 Reports have found these synthetic polymers to be environmentally stable and undergo minimal biodegradation.^23,24 As such, there is concern about their environmental accumulation through leaching into ground or surface water, which may also result in metal contamination due to the ability of polycarboxylates to chelate metal ions.^25,26 While poly(acrylate) is generally considered safe, unless ingested in very high quantities,²⁷ there is a need to shift toward greener alternatives to mitigate environmental accumulation. Poly(aspartic acid) (PAA) is an eco-friendly alternative to polycarboxylates like poly(acrylate) that has been shown to be biodegradable.

PAA is easily synthesized using naturally occurring aspartic acid. Thermal synthesis of PAA (tPAA) is accomplished by heating aspartic acid to form a poly(succinimide) that is ring-opened by the addition of sodium hydroxide to yield a tPAA with α- and β-amide linkages. The process results in formation of an atactic polymer of racemized aspartates, irregular end groups, and preferential branching (~70%) through the β-linkage.²⁸ Biodegradation and assimilation of tPAA were first identified in two river water bacterial strains, Sphingomonas sp. KT-1 and Pedobacter sp. KP-2.^29,30 Subsequent studies of these bacteria identified, isolated, and initially characterized three different proteins capable of tPAA degradation: PahZ1_KT-1, PahZ2_KT-1, and PahZ1_KP-2. Within Sphingomonas sp. KT-1, tPAA degradation is an intracellular process whereby PahZ1 cleaves the β-amide linkages of low-molecular-weight tPAA (<5000 Da) leaving oligo(aspartic acid) to be cleaved at both the α- and β-amide linkages by PahZ2 to yield monomeric aspartic acid.^29,31-34 PahZ1_KP-2 was shown to extracellularly cleave both low-and high-molecular-weight tPAA; however, a corresponding PahZ2 has not been identified in Pedobacter sp. _KP-2.^29,35

Recently, we determined the first structure of a poly(aspartic acid) hydrolase, PahZ1_KT-1. This serine protease was shown to have a dimeric assembly and displays a positively lined trough of residues that directs the polyanionic tPAA substrate toward the catalytic active site allowing for endolytic cleavage.³⁶ Here, we further build upon our mechanistic understanding of tPAA degradation with the crystal structure of PahZ2_KT-1 which was solved to 1.85 Å. PahZ2_KT-1 is also a dimer but is structurally related to M28 metallopeptidases, a family of dinuclear Zn(II)-dependent exopeptidases.³⁷ PahZ2_KT-1 subunits are similar to two other well-characterized M28 family members, carboxypeptidase G2 and DapE, containing both a dimerization and a catalytic domain that harbors two Zn binding sites.^38-40 On the basis of activity assays and molecular dynamics (MD) simulations, we show the OAA substrate binds at the interface between these two domains, through electrostatic interactions, and the presence of NaCl enhances the activity of PahZ2_KT-1. The latter is likely through a conformational change that shifts the protein from an open-complex to a closed-complex allowing for enhanced catalytic activity.

Having the structures of PahZ1_KT-1 and now PahZ2_KT-1 allows for an atomistic understanding of how PAA can be recycled back to monomeric aspartic acid. While the activity of PahZ2_KT-1 does not yield complete degradation of OAA to Asp, knowing its structure provides the framework for future protein engineering endeavors to enhance the catalytic activity of PahZ2_KT-1, which is exceptionally important from a green chemistry perspective of using PAA, or potentially other WSPs, in a complete cradle-to-cradle process.

EXPERIMENTAL SECTION

Materials.

All solutions used to generate data reported here were prepared with double-distilled water produced from a Purelab Ultra Genetic System (Siemens Water Technology). The DNA coding for the mature PahZ2_KT-1 sequence³¹ was cloned into the pET15b vector using the 5′ Ndel and 3′ Xhol restriction sites. DNA synthesis, cloning, and site-directed mutagenesis were performed by GenScript (Piscataway, NJ). PahZ2_KT-1 E155A, E156A, D121A, D122A/D184A, and E156A/H374A mutant constructs were created using the PahZ2_KT-1 plasmid described above.

Preparation of Digested Poly(aspartic acid) (dPAA).

It has been shown that PahZ2_KT-1 does not show enzymatic activity against tPAA due to the presence of irregular end groups.³¹ To produce the substrate for PahZ2_KT-1, thermally synthesized poly(aspartic acid), tPAA,⁴¹ was solubilized to 100 mg/mL in 50 mM HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid), pH 7.0. Subsequently, the tPAA was digested overnight at 37 °C with 0.5 mg/mL PahZ1_KT-1.³⁶ The sample was then incubated at 70 °C to precipitate the PahZ1_KT-1, and the supernatant containing the digested tPAA (dPAA) served as the substrate for PahZ2_KT-1 in all gel permeation chromatography (GPC) assays.

Methods.

Protein Expression and Purification.

Plasmid encoding for PahZ2_KT-1 or PahZ2_KT-1 mutant constructs was transformed into BL21(DE3) Escherichia coli cells (New England Biolabs, Ipswich, MA), and all followed the same expression and purification protocol as described. A single colony was used to inoculate 50 mL of lysogeny broth (LB) media containing 50 μg/mL final carbenicillin. Cells were incubated overnight at 37 °C, with shaking at 225 rpm. After incubation, 10 mL of this growth was used to inoculate 500 mL of LB-media in a 2 L Erlenmeyer flask where cells were grown to an optical density at 600 nm (OD₆₀₀) equal to 1.0 at the same temperature and shaking speed. Expression of PahZ2_KT-1 was induced with a final concentration of 0.1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG, GoldBio, St. Louis, MO), and the flask was incubated at 20 °C for 4 h with shaking at 250 rpm. The cells were harvested by centrifugation at 10 000g for 10 min at room temperature. Cell pellets were stored at −20 °C prior to further use.

Pelleted cells were resuspended in 50 mM tris(2-amino-2-(hydroxymethyl)-1,3-propanediol), 300 mM NaCl, and 20 mM imidazole and lysed by sonication for 45 min on ice. Turbonuclease (50 μL, Accelagen, San Diego, CA) was added and allowed to rock for 45 min at room temperature to increase lysis and remove genomic DNA. The cell lysate was clarified by centrifugation at 10 000g for 20 min at 4 °C and subsequently loaded on a HisPur Ni-nitriloacetic acid (NTA) 5 mL cartridge (Thermo Scientific, Waltham, MA) equilibrated in 50 mM Tris, 300 mM NaCl, and 20 mM imidazole. After loading, PahZ2_KT-1 was eluted from the column with 50 mM Tris, 300 mM NaCl, and 500 mM imidazole using a gradient of 0–100% over 60 min. The eluted peak corresponding to PahZ2_KT-1 was collected and analyzed via SDS-PAGE which showed that the protein was >98% pure after this single purification step. Protein was dialyzed against 50 mM HEPES, pH 7.40, or 20 mM Tris, pH 8.0, and stored at 4 °C.

X-ray Structure Determination and Refinement.

PahZ2_KT-1 crystallization screening was initially performed at the Hauptman–Woodward Institute High-Throughput Crystallization Screening Center,⁴² with crystals observed in a variety of conditions. Single PahZ2_KT-1 crystals were obtained by hanging-drop vapor diffusion at 22 °C by mixing 2 μL of protein (13–15 mg/mL PahZ2_KT-1 in 20 mM Tris pH 8.0) with 2 μL of reservoir solution placed over a 500 μL reservoir solution containing 15% w/v PEG 3350, 0.1 M HEPES pH 6.5, and the silver bullets D4 (Hampton) cocktail (0.005 M gadolinium(III) chloride hexahydrate, 0.005 M samarium(III) chloride hexahydrate, 0.05 M benzamidine hydrochloride, 0.25% w/v salicin, and 0.02 M HEPES pH 6.8). Crystals grew to full size in approximately 2–3 weeks. For preparation for data collection, crystals were quickly transferred through a solution containing 15% w/v PEG 3350, 0.1 M HEPES pH 6.5, silver bullets D4, and 30% v/v glycerol, and then flash-cooled by plunging into liquid nitrogen. X-ray data were collected at a wavelength of 1.11608 Å at beamline 8.3.1 at the Advanced Light Source on a Dectris Pilatus3 S 6 M detector. Data were processed using XDS.⁴³

A strong anomalous signal was detected during data processing, likely due to the Gd³⁺ and/or Sm³⁺ present in the crystallization condition. The structure of PahZ2_KT-1 was determined by single-wavelength anomalous diffraction (SAD) phasing using the program AutoSol⁴⁴ in the Phenix software suite.⁴⁵ AutoSol successfully solved the structure with a figure of merit of 0.412 and an R_work and R_free of 23.5% and 24.3%, respectively. The heavy atom sites and density modified maps from AutoSol were then fed into Autobuild,⁴⁶ which was able to successfully build residues 7–404 of both protein monomers in the asymmetric unit, with an R_work and R_free of 21.0% and 21.8%, respectively. Manual structure building was followed by xyz coordinate, real space, TLS, and individual B-factor refinement using the program phenix.refine. Noncrystallographic restraints were included during refinement.

For the Zn²⁺-bound PahZ2_KT-1 structure, crystals grown in 15% w/v PEG 3350, 0.1 M HEPES pH 6.5, and silver bullets D4 were soaked in a solution containing 15% w/v PEG 3350, 0.1 M HEPES pH 6.5, silver bullets D4, and 25 mM ZnCl₂ for 4 h at room temperature. No back-soaking was performed in an effort to prevent the active-site Zn²⁺ metals from being replaced by the Gd³⁺/Sm³⁺ present in the crystallization solution, which led to 28 total zinc atoms present in the final structure. After soaking, the crystals were quickly transferred through a solution containing 15% w/v PEG 3350, 0.1 M HEPES pH 6.5, silver bullets D4, 25 mM ZnCl₂, and 30% v/v glycerol, then flash-cooled by plunging into liquid nitrogen. X-ray data were collected at a wavelength of 1.28273 Å at beamline 8.3.1 at the Advanced Light Source on a Dectris Pilatus3 S 6 M detector. Data were processed using XDS.⁴³ The structure of PahZ2_KT-1 bound to zinc was determined by SAD phasing to 2.50 Å using the program AutoSol⁴⁴ in the Phenix software suite.⁴⁵ AutoSol successfully solved the structure with a figure of merit of 0.322 and an R_work and R_free of 27.1% and 29.5%, respectively. The heavy atom sites and density modified maps from AutoSol were then fed into Autobuild,⁴⁶ which was able to successfully build residues 7–102 and 122–405 of chain A, and residues 12–404 of chain B, with an R_work and R_free of 20.6% and 22.9%, respectively. Manual structure building was followed by xyz coordinate, real space, TLS, and individual B-factor refinement using the program phenix.refine. Noncrystallographic restraints were included during refinement. A summary of data collection and refinement statistics for both structures is provided in Table 1. X-ray structure figures, as well as electrostatic surface potential maps, were generated using the program PyMol.⁴⁷

Table 1.

Data Collection and Refinement Statistics for the PahZ2_KT-1 Crystal Structures^a

	PahZ2KT-1 Zn2+	PahZ2KT-1 Gd3+/Sm3+
	Data Collection
space group	P212121	P212121
cell dimensions
a, b, c (Å)	48.40, 146.17, 197.24	48.67, 147.87, 197.14
α, β, γ (deg)	90, 90, 90	90, 90, 90
resolution (Å)	117.4–2.50 (2.59–2.50)	98.57–1.85 (1.92–1.85)
total reflections	98 445 (9,103)	244 755 (24 036)
unique reflections	49 265 (4588)	122 397 (12 022)
CC1/2	0.997 (0.591)	0.999 (0.886)
CCb	0.999 (0.862)	1 (0.969)
Rmerge (%)	4.60 (50.8)	3.16 (27.6)
I/σ	16.43 (1.40)	23.23 (2.53)
completeness (%)	99.34 (94.64)	94.50 (82.82)
redundancy	2.0 (2.0)	2.0 (2.0)
	Refinement
no. reflns	49 257 (4587)	115 801 (9982)
Rwork/Rfree (%)	18.21/21.00 (30.68/36.92)	14.51/16.16 (25.33/27.34)
no. atoms
protein	5926	5955
water	202	879
metals	28	3
B-factors
protein	67.34	36.50
water	63.34	48.23
metals	113.17	34.01
	Stereochemical Ideality
bond lengths (Å)	0.002	0.013
bond angles (deg)	0.54	1.21
φ, ψ most favored (%)	97.99	99.37
φ, ψ allowed (%)	2.01	0.63
φ, ψ outliers (%)	0.00	0.00

^aCoordinates have been deposited in the Protein Data Bank as entries 7LJH and 7LJI.

^bValues in parentheses are for highest-resolution shell.

PahZ2_KT-1 Activity Assays in the Presence of Metals.

PahZ2_KT-1 was exhaustively dialyzed against 50 mM HEPES, 1 mM EDTA, and pH 7.0 and concentrated to 1.27 mg/mL as determined from absorbance measurements at 280 nm. All metals were the chloride salt except for magnesium which had sulfate as the counterion. Metal solutions were made in 50 mM HEPES, pH 7.0, at a concentration of 50 mM (except Zn²⁺), added to PahZ2_KT-1 to a 2.5 mM final concentration, and incubated on ice overnight. Since zinc is poorly soluble at physiological pH, PahZ2_KT-1 was dialyzed against 50 mM HEPES, pH 7.0, 2 mM ZnCl₂, at 4 °C. Activity assays were set up as 100 μL reactions containing 80 μg PahZ2_KT-1 and 24.5 μL of dPAA spiked with 3 mg/mL thyroglobulin (MilliporeSigma, Burlington, MA). Samples were incubated at 37 °C overnight, and the reactions were stopped by heating at 80 °C for 10 min to precipitate the PahZ2_KT-1. Samples were stored at −80 °C until they were loaded on the GPC. PahZ2_KT-1 activity was assayed using a Shimadzu GPC with a 3 μm SEC-2000, 300 mm × 7.8 mm (Phenomenex, Torrance, CA) column equilibrated in 50 mM HEPES, pH 7.0. Injections were 10 μL, and the thyroglobulin served as an internal loading control. As PahZ2_KT-1 degrades dPAA, it forms aspartate which can be evaluated with changes in intensity in the peak with a retention time of 11.18 min. Percent activity was calculated using a ratio of this peak height to thyroglobulin peak at 4.95 min. Alternatively and equivalently, the decrease in dPAA oligomer peak height at 7.0 min can be used to assess enzyme activity by again using the ratio of the peak height to thyroglobulin. Figure S1 shows the inverse correlation between product formation at time ~11.2 min corresponding to polymer degradation at time ~7.0 min. Since the presence of Zn²⁺ showed the greatest activity, all other metals were compared relative to Zn²⁺.

PahZ2_KT-1 Salt Activity Assay.

Digested tPAA was dialyzed against 50 mM HEPES pH 7.0 without NaCl, or with 150 mM NaCl or 500 mM NaCl. Dialyzed dPAA samples were lyophilized and resuspended to a final concentration of 100 mg/mL. PahZ2_KT-1 was prepared by an initial dialysis against 50 mM HEPES with 2 mM ZnCl₂ at pH 7.0 and subsequent dialysis against 50 mM HEPES pH 7.0 with or without NaCl. Assay samples were made with 24.5 mg of dPAA and 0.1 mg/mL PahZ2_KT-1 in a final reaction volume of 100 μL and incubated at 37 °C. Aliquots of 20 μL were taken at the following time points: 5 min, 30 min, 1 h, 2 h, 3 h, 6 h, 12 h, and 24 h. Reactions were stopped by incubating at 80 °C for 10 min and stored at −80 °C. Prior to running on the GPC, samples were spiked with 10 μL of 2.2 mg/mL thyroglobulin. A new GPC 3 μm SEC-2000, 300 mm × 7.8 mm column from the same manufacturer but from different lots was used for the salt assays; however, the same corresponding peak as the metal assays was analyzed for the 0 mM salt assays. The activity of PahZ2_KT-1 was determined by analyzing the peak formed from the formation of aspartate with a retention time of 11.02 min, on this new column, relative to the thyroglobin for the 0 mM NaCl assays. However, the presence of NaCl shifted the chromatogram, and the aspartate peaks for 150 and 500 mM NaCl were at 10.75 and 10.80 min, respectively. Figure S2 shows chromatogram deviations based on the presence of salt and includes monomeric Asp along with Asp + dPAA as controls to show that PahZ2_KT-1 cleaves dPAA to Asp.

PahZ2_KT-1 Catalytic Mutant Activity Assays.

The procedure mimicked the 0 mM NaCl activity assays described above with the exception of the only time point being assayed at 24 h. Given the abolishment of activity with the catalytic mutants, assays were assessed using the degradation of the dPAA peak at a time of 7.0 min because there was more noise in the measurements at time ~11 min where the salt peak and product (monomeric aspartic acid) peak overlap.

Molecular Dynamics Simulations. Docking and Ligand Preparation.

Molecular docking of poly(aspartic acid) and other ligands to the PahZ2_KT-1 active site was performed in several steps. First, Protein Data Bank (PDB) models of PahZ2_KT-1 were generated using the build function of the Pymol molecular visualization program³⁰ or downloaded from the Protein Data Bank.⁴⁸ All subsequent preparation prior to docking was completed using UCSF Chimera.⁴⁹ Structure files were converted to the requisite PDBQT file type, and polar hydrogens were added where applicable. Search coordinates were constructed corresponding to the known active site, and docking was carried out without constraints in Autodock Vina.⁵⁰ The docked ligand coordinates were then exported, and the SwissParam small molecule force field generation tool⁵¹ was used to build parameter sets for later use in molecular simulations.

Production Molecular Dynamics.

All models prepared for in silico studies were structurally refined with 200 ns of production molecular dynamics (MD) via the Gromacs molecular simulation package.^52,53 Briefly, all models were solvated in triclinic boxes with simple point charge water and neutralizing ions. Where applicable, 500 mM NaCl was also added. Ligand-bound simulations feature parameter sets generated by the SwissParam small molecule force field generation tool (see Docking and Ligand Preparation).⁵¹ Solvated systems were minimized using a steepest descent energy minimization protocol. Except where otherwise noted, all minimized structures were thermally equilibrated for up to 5000 ps of position restrained NVT and NPT ensembles with a Berendsen thermostat.^53,54 Unrestrained production MD was then carried out for 200 ns with Nose–Hoover^55,56 and Parinello–Rahman^57,58 temperature and pressure coupling, respectively, using the Charmm-36 all-atom force field.^59,60

Free Energy Landscapes.

Free energy landscapes (FEL) were generated using the sham analysis package as included in the Gromacs suite of molecular simulation tools. To prepare for sham input, all models were equilibrated by 200 ns of production MD as described above. Simulations performed with high salt conditions were solvated in simple point charge water and supplemented with 0.5 M NaCl. Following equilibration, root-mean-square deviation (RMSD) and radius of gyration (RoG) calculations were performed for each system. Gromacs-generated average structures were extracted and used as reference files for these calculations. A text file containing RMSD and RoG data per frame was constructed and passed to sham. The resulting FEL data was then visualized by the Gnuplot suite of plotting tools.⁶¹

Principal Component Analysis.

Following 200 ns of unrestrained production MD (see Production Molecular Dynamics section above), all protein coordinates were extracted as DCD format trajectories. These files were then read into Bio3d, a powerful R package for molecular simulation and structure analysis.^62,63 C_α atoms were selected and used for subsequent superposition of the trajectory frames in preparation of principal component calculations. Plotting tools provided by R Studio⁶⁴ and Bio3d^62,63 were used for all component visualization.

Center-of-Mass (COM) Pulling.

All PahZ2_KT-1 models with bound Zn²⁺ atoms were prepared and simulated as described above. A 200 ns frame representing conformational convergence was taken from the resulting trajectory. The catalytic domain featuring the Zn²⁺ binding site was then extracted and used as input for the subsequent COM pulling simulations using Gromacs.^52,53 The catalytic domain was first solvated in a TIP3P water box^65,66 measuring 13 nm along the reaction coordinate (ξ). Neutralizing ions were added, and the system was brought to an energetic minimum by steepest descent energy minimization. Convergence of thermodynamic parameters was accomplished following 5000 ps of NPT ensemble dynamics in the presence of position restraints. Restraints were removed, and each Zn²⁺ atom was then pulled along ξ in a stepwise procedure. Unrestrained NPT dynamics were used for 1000 ps after pulling the first Zn²⁺ atom to allow for relaxation of active-site residues before extracting the remaining Zn²⁺ atom. All pulls used for umbrella sampling were performed at a rate of 0.01 nm/ps and a force constant of 500 kJ mol⁻¹ nm². Additional pulls for each Zn²⁺ were also performed across a wide range of rates of constants between 0.001 and 0.01 nm ps⁻¹ and between 500 and 1000 kJ mol⁻¹ nm⁻², totaling five complete COM pulling simulations for each bound Zn²⁺ atom.

Umbrella Sampling and Potential of Mean Force (PMF) Calculations.

Configurations with varying COM distances between 0.02 and 0.5 nm along ξ were extracted and prepared for umbrella sampling simulations. Additional configurations were used as necessary to ensure sufficient sampling, resulting in 32 and 34 windows for the first and second Zn²⁺ steered molecular dynamics dissociations, respectively. Each chosen configuration along ξ was then brought to thermodynamic convergence with 100 ps of restrained NPT ensemble equilibration. Restraints were removed, and each configuration was simulated for 10 ns with an applied umbrella potential at a force constant of 500 kJ mol⁻¹ nm². The resulting force data was then used as input for subsequent potential of mean force calculations by the WHAM Gromacs package.⁶⁷ Adequate sampling was confirmed by histogram analysis, and error estimation was obtained by bootstrap methods (n = 200).

Dynamic Light Scattering (DLS) to Assess PahZ2_KT-1 Dimerization in Solution.

To determine if differing concentrations of NaCl would disrupt the dimeric structure of PahZ2_KT-1, purified PahZ2_KT-1 was dialyzed against 50 mM HEPES, pH 7.0, and supplemented with NaCl to final concentrations of 0, 150, and 500 mM. All protein solutions were at 0.5 mg/mL final and centrifuged at 17 000g for 5 min at room temperature to remove any particulate. Samples were analyzed on a Zetasizer Nano ZS instrument (Malvern Panalytical, Westborough, MA) with a size detection range of 0.3 nm to 10 μm where 500 μL of sample was added to a disposable cell.

RESULTS AND DISCUSSION

Overview of the PahZ2_KT-1 Crystal Structure.

Screening for crystallization conditions was performed at the Hauptman–Woodward Institute High-Throughput Crystallization Screening Center, which revealed PahZ2_KT-1 crystals in a variety of conditions. Single crystals grown in the presence of Gd³⁺/Sm³⁺ diffracted to 1.85 Å, whereas crystals soaked in 25 mM Zn²⁺ diffracted to 2.50 Å (Table 1). Metal-bound structures were determined independently by single-wavelength anomalous diffraction (SAD) phasing methods. For the PahZ2_KT-1 structure in the presence of Gd³⁺/Sm³⁺, refinement was carried out with Gd³⁺ placed in the structure. However, both metals have similar anomalous scattering coefficients at the wavelength used for data collection, so differentiation between the bound metals is ambiguous. It is likely that a mixture of both Gd³⁺ and Sm³⁺ occupy the metal coordination sites in the structure, and activity assays described below show comparable PahZ2_KT-1 activity in the presence of these metals. Therefore, we conclude that the structure presented accurately represents the structure with either Gd³⁺ or Sm³⁺ bound.

The structures each contain two PahZ2_KT-1 monomers in the asymmetric unit, and Figure 1A shows the believed biological PahZ2_KT-1 dimer. Support for this dimer assembly is drawn from structurally related enzymes, size-exclusion chromatography experiments, and dynamic light scattering experiments discussed below. Individual PahZ2_KT-1 subunits are related by a 2-fold rotational symmetry at the dimer interface and reveal two distinct domains: (1) dimerization domain and (2) catalytic domain. Figure 1B illustrates that the two domains are connected by a hinge region, with the dimerization domain occurring as an insertion spanning V196–A318. The PahZ2_KT-1 catalytic domain is represented by residues M1–G195 and G319–Q405. Structures determined with bound Gd³⁺/Sm³⁺ or Zn²⁺ superimpose well, with root-mean-square deviation equal to 0.652 Å for the dimer (767 residues) and 0.558 Å for the monomer (400 residues) superpositions (Figure 1A,B) and do not reveal significant differences in side-chain positioning. Additional discussion of each PahZ2_KT-1 structural domain is provided below.

Figure 1.

Overview of PahZ2_KT-1 structure. (A) Superposition of the PahZ2_KT-1 dimers bound with Gd³⁺/Sm³⁺ or Zn²⁺ in the active sites. Both monomers of the Gd³⁺/Sm³⁺ dimer are shown in purple, while the two Zn-bound dimers are colored blue and green. The dimers superimpose with an RMSD of 0.625 Å over 767 residues. (B) Superposition of the PahZ2_KT-1 monomers bound with Gd³⁺/Sm³⁺ (purple) or zinc (green) in the active sites. The two structures superimpose with an RMSD of 0.558 Å over 400 residues. Labels are included to highlight the location of the dimerization and catalytic domains as well as the connector hinge region. PahZ2_KT-1 structures are represented in cartoon view. (C–E) Calculated electrostatic surface potentials for PahZ2_KT-1 (C), carboxypeptidase G2 (D), and DapE (E) monomer structures. All structural representations were generated using the Pymol software package and Protein Data Bank accession files 1CG2 and 5VO3 for carboxypeptidase G2 and DapE, respectively.

A structural comparison using the DALI server^68,69 reveals multiple closely related structural homologues. Structural homologues that have been experimentally characterized are observed to commonly be dimeric with Zn²⁺ bound to the catalytic domain. Examples include Pseudomonas sp. RS-16 carboxypeptidase G2⁴⁰ (PDB ID 1CG2, Z-score = 35.6) and Haemophilus influenza N-succinyl-l,l-diaminopimelic acid desuccinylase (DapE)^38,39 (PDB ID 5VO3, Z-score = 31.8), which share 21% and 18% sequence identity, respectively, with PahZ2_KT-1. Structural superposition of PahZ2_KT-1, carboxypeptidase G2, and DapE reveals a shared fold, but electrostatic surface potential maps demonstrate a marked difference in surface charge. Figure 1C-E illustrates these differences by showing calculated electrostatic surface potential maps for the PahZ2_KT-1 (Figure 1C), carboxypeptidase G2 (Figure 1D), and DapE (Figure 1E) monomer structures. Each structure harbors nonpolar content at the core of the dimer interface (white, Figure 1C-E), consistent with the formation of a protein/protein interface. In contrast, the PahZ2_KT-1 catalytic domain surface is enriched in positive charge (blue, Figure 1C), which contrasts the solvent accessible surfaces for carboxypeptidase G2 and DapE. However, the active site is conserved in each enzyme, thereby leading to similar charge distribution in this region. Calculations of the sum of formal charges on DapE, carboxypeptidase G2, and PahZ2_KT-1 predict net charges equal to −18, −14, and +2, respectively. Differences in electrostatic surface potential for PahZ2_KT-1 solvent accessible surfaces relative to other enzymes may reflect an evolved ability to interact with negatively charged oligo(aspartic acid) substrates.

The overall structures of PahZ1_KT-1 and PahZ2_KT-1 are notably different in domain architecture and classification (Figure S3A,B). While both enzymes exist in solution as a homodimer, PahZ1_KT-1 monomers consist of only a single domain, while PahZ2_KT-1 monomers each have distinct dimerization and catalytic domains as discussed here. Visualization of the electrostatic surface potential for each enzyme also highlights how each has evolved the ability to distinguish between poly(aspartic acid) and oligo(aspartic acid). Figure S3C highlights the electrostatic surface for PahZ1_KT-1, where a large trough exists that is lined with basic residues. In our previous report of the PahZ1_KT-1 structure,³⁶ we utilized in silico molecular dynamics methods to examine how the enzyme utilizes this cationic surface to distinguish between α- and β-linkages in its substrates consistent with previous reports of a β-linkage specificity.³⁴ In contrast, no such cationic trough is observed for PahZ2_KT-1 (Figure S3D), which is consistent with previous kinetic data examining the length-dependent kinetic properties for PahZ2_KT-1 catalyzed oligo(aspartic acid) degradation. Hiraishi and co-workers reported that PahZ2_KT-1 exhibits a V_max and K_M that are dependent and independent, respectively, on substrate length such that optimum activity is observed for a pentameric substrate.³³ Thus, PahZ2_KT-1 does not possess a binding surface that would accommodate tPAA similar to that found in PahZ1_KT-1. Last, PahZ1_KT-1 and PahZ2_KT-1 each belong to unique protease families. Figure S3E,F highlights the active-site architectures observed in PahZ1_KT-1 (Figure S3E) and PahZ2_KT-1 (Figure S3F), which are consistent with membership in serine protease and metalloprotease families, respectively.

PahZ2_KT-1 Dimerization Domain.

Examination of the PahZ2_KT-1 dimer interface reveals symmetric interactions that occur between subunits involving two separate contact points for both Gd³⁺/Sm³⁺- and Zn²⁺-bound structures. Figure 2A highlights the two-tiered arrangement wherein dimerization is mediated through interactions that occur between α-helical and β-sheet layers of associated subunits. Intersubunit interactions involve a mixture of nonpolar and polar residues that occur in both tiers. Figure 2B provides insight into how polar and nonpolar residues are distributed about the interface. The symmetric interface is depicted such that one dimerization domain is shown as an electrostatic surface and the other is shown in cartoon representation. From this, it is apparent that nonpolar residues populate the core of the dimer interface with increased positive and negative charge existing at the periphery of the interface. Furthermore, protein–protein contact occurs primarily through a single α-helix and β-strand on each subunit consisting of residues M222–K234 and S248–T254, respectively. The interaction of each secondary structural element is symmetric such that the α-helix or β-strand is positioned antiparallel to the same feature on the adjacent subunit.

Figure 2.

Dimerization domain is a nonpolar interface. (A) The dimer interface of PahZ2_KT-1 is shown in cartoon view to highlight interactions occurring between secondary structures positioned antiparallel to one another between a single α-helix and β-strand on each subunit, represented by residues M222–K234 and S248–T254, respectively. (B) An electrostatic surface reveals the core of the dimer interface to be largely nonpolar.(C) Summary of proteins, interfaces, structures, and assemblies (PISA) analyses for the wild-type PahZ2_KT-1 X-ray structure.

To identify interactions that stabilize the PahZ2_KT-1 dimer, we subjected the interface to analysis using the Protein Data Bank in Europe Proteins, Interfaces, Structures, and Assemblies (PDBePISA)⁷⁰ interactive tool. The summarized PDBePISA output for the higher-resolution interface derived from the Gd³⁺/Sm³⁺-bound structure is presented in Figure 2C with data tabulated for each individual subunit as well as the overall dimer complex. Of the 799 residues represented by both PahZ2_KT-1 chains, 11.8% are found at the dimer interface. Figure 2C further highlights the contribution of subunits A and B to the interface surface as 1687.7 Ų (10.0% of subunit A surface area) and 1682.5 Ų (10.0% of subunit B surface area), respectively. The overall dimer interface area is equal to the mean of individual subunit surface areas, 1685.1 Ų. On the basis of the difference in solvation energies for the isolated chains versus the dimer structure, PDBePISA also provides an estimate of solvation free energy gain associated with complex formation, ΔGⁱ. The solvation free energy gain is estimated as ΔGⁱ = −21.8 kcal mol⁻¹ (−91.2 kJ mol⁻¹), where the strongly negative value here indicates the presence of a hydrophobic interface. By comparison, estimates of overall dimer interface area and solvation free energy gain have been reported for the PahZ1_KT-1 dimer as 626.2 Ų and −7.3 kcal mol⁻¹, respectively, consistent with reports of increased thermal stability for PahZ2_KT-1 versus PahZ1_KT-1.^36,71

Examination of the interface amino acid composition suggests a mixture of nonpolar and polar residues that collectively promote dimerization. All interactions are observed to be symmetric due to the antiparallel orientation of dimerization domains relative to one another. Multiple salt bridges are present at the dimer interface that include 19 total interactions occurring as side-chain/side-chain, side-chain/main-chain, and main-chain/main-chain arrangements. Of these, 4 hydrogen bonding interactions occur between side-chain functional groups R200–N257, N257–R270, N220–Y233, and N268–S212 with distances equal to 2.68, 3.88, 2.7, and 3.20 Å, respectively. Additional hydrogen bonding interactions involving main-chain atoms include K242–G214, R299–Y233, N220–A247, G252–T250, Y213–T244, Y233–R299, S248–T254, and V256–R270. Of these interactions, two hydrogen bonding interactions, N220–Y233 and N257–R200, are notable for short distances equal to 2.6 and 2.4 Å, respectively, that may suggest an increased bond strength. Given the symmetric dimer interface, two N220–Y233 interactions are observed to stabilize the terminal ends of the helix–helix interaction spanning residues M222–K234 for each helix and two N257–R200 interactions that stabilize the β-tier of the dimerization domain. A detailed MD inspection of in silico-generated interface residue mutations that either stabilize or destabilize PahZ2_KT-1 dimer formation is presented in the Supporting Information, which generally confirm the role of nonpolar residues in stabilizing the dimer (Figures S4 and S5).

PahZ2_KT-1 Catalytic Site Analysis.

Overall Catalytic Site Description.

The PahZ2_KT-1 structure reveals an active site consistent with classification as an M28 family member.³⁷ The M28 family of metallopeptidases contains both aminopeptidases and carboxypeptidases that each harbor two zinc ions at the active site. From available crystal structures for M28 family members, each zinc ion is expected to coordinate in a tetrahedral molecular geometry via 3–4 amino acids and additional water molecules. Figure 3A,B highlights the metal-coordinating residues in the PahZ2_KT-1 active site when Gd³⁺/Sm³⁺ (Figure 3A) versus Zn²⁺ (Figure 3B) is bound, where anomalous difference electron density allows for unambiguous metal placement in each case. The overlay of active-site residues with bound Gd³⁺/Sm³⁺ and Zn²⁺ shown in Figure 3C illustrates no significant difference in positioning of nearby side-chain functional groups but instead shows a difference in metal placement. The PahZ2_KT-1:Gd³⁺/Sm³⁺ complex reveals the metal to adopt a coordination number of seven involving interactions with side-chain functional groups from E155, D184, H94, D121, E156, and two water molecules (Figure 3A). Inspection of the zinc-bound PahZ2_KT-1 active site reveals two zinc ions that occupy unique binding sites, which is consistent with other M28 family members (Figure 3B).^38-40 Within the M28 family,³⁷ the two Zn²⁺ binding sites are generally annotated as Zn_I and Zn_II as shown in Figure 3B. At the Zn_I coordination site, H94, D184, and D121 function as zinc ligands, whereas ligands found at the Zn_II site include H374, E156, and D121.

Figure 3.

PahZ2_KT-1 catalytic site description. (A) Active site with Gd³⁺/Sm³⁺ bound. Anomalous difference density, contoured at 3σ, is shown in green surrounding the Gd³⁺. Additionally, 2F_o − F_c electron density, contoured at 1σ, in shown in brown for residue E155 to show the overall quality of the data. (B) Active site with Zn²⁺ bound. Anomalous difference density, contoured at 5σ, is shown in green surrounding the Zn²⁺. 2F_o − F_c electron density, contoured at 1σ, is shown in brown for residue E155 to show the overall quality of the data. (C) Overlay of Gd³⁺/Sm³⁺- and Zn²⁺-bound active-site features.(D) Relative activity comparing maximal PahZ2_KT-1 catalyzed dPAA degradation activity observed in the presence of ethylenediamine tetraacetic acid (EDTA) and Zn²⁺, Ni²⁺, Cu²⁺, Mg²⁺, Ca²⁺, Sm³⁺, and Gd³⁺ metal ions. (E) Relative activity comparing maximal PahZ2_KT-1 catalyzed dPAA degradation activity observed for reactions performed with wild-type PahZ2_KT-1 alongside Zn_I and Zn_II site mutant constructs. All bar graph plots are represented with mean and standard deviation derived from triplicate measurements. All structural representations were generated using the Pymol software package with active-site residues presented in stick view. The Gd³⁺/Sm³⁺ and Zn²⁺ metal ions are represented as teal or gray spheres, respectively. Individual Zn_I and Zn_II binding sites are indicated by arrows.

PahZ2_KT-1 Catalysis is Supported by Multiple Metals.

The tetrahedral arrangement of zinc atoms associated with M28 family active sites suggests that Zn²⁺ may not be the only metal capable of supporting PahZ2_KT-1 catalytic function. In fact, Co²⁺, Zn²⁺, Cd²⁺, Mn²⁺, Ni²⁺, Cu²⁺, and Mg²⁺ metal ions are known to support DapE catalytic function.⁷³ To assess whether PahZ2_KT-1 activity can be supported by alternative metals, we performed activity assays reporting on PahZ2_KT-1 catalyzed poly(aspartic acid) degradation as described in the Materials and Methods sections. Activity assays were performed in the presence of ethylenediamine tetraacetic acid (EDTA), Zn²⁺, Ni²⁺, Cu²⁺, Mg²⁺, Ca²⁺, Sm³⁺, and Gd³⁺. As shown in Figure 3D, incubation of PahZ2_KT-1 in the presence of EDTA significantly inhibits catalytic function such that triplicate measurements yield a mean relative activity equal to 5.61 ± 0.9%. By comparison, Hiraishi and co-workers reported a relative activity equal to 16% in the presence of 1 mM EDTA, which suggested metal-dependent PahZ2_KT-1 activity.^71,73 We report here maximum PahZ2_KT-1 catalytic activity when reactions are performed in the presence of 2 mM ZnCl₂. However, we observe relative activities equal to 93.52 ± 2.18% and 80.99 ± 0.52% when reactions are performed in the presence of Ni²⁺ or Cu²⁺, respectively. Interestingly, Ni²⁺ commonly adopts a tetrahedral coordination geometry, whereas Cu²⁺ is capable of adopting either square planar or octahedral arrangements. In contrast, divalent cations Mg²⁺ and Ca²⁺ adopt octahedral coordination geometries, which drives an apparent decrease in relative activity to 29.38 ± 9.02% and 21.94 ± 7.20%, respectively. Taken together, these data suggest that the PahZ2_KT-1 catalytic output is strongly dependent upon the preferred coordination geometry of a bound metal ion.

Our initial PahZ2_KT-1 structures obtained in the presence of samarium or gadolinium salts revealed electron density consistent with a single bound metal ion. To determine whether single metal binding may support PahZ2_KT-1 function, additional activity assays were performed in the presence of Sm³⁺ or Gd³⁺, which each yield relative activities equal to 64.38 ± 2.96% and 70.82 ± 4.37%, respectively. However, Sm³⁺ and Gd³⁺ do not form tetrahedral complexes; instead, they adopt complexes with variable coordination numbers ranging from 6 to 8. Reference to the Gd³⁺/Sm³⁺-bound PahZ2_KT-1 structure (Figure 3A,C) reveals the metal binds directly to the Zn_I site, but such that the Zn_II Site E156 residue also coordinates. As such, we expect that metal binding to the Zn_II site of Gd³⁺/Sm³-bound PahZ2_KT-1 would be blocked, thereby retaining only activity associated with the Zn_I site.

Bioinformatic Analysis.

Figure 3B clearly shows that D121 forms interactions at both the Zn_I and Zn_II sites, which is further consistent with classification as an M28 metallopeptidase family member. M28 metallopeptidase family members are classified on the basis of the stated zinc coordination arrangement discussed above, which positions zinc ligands based on the His–X_aa–Asp and Glu–Glu motifs (X_aa represents any amino acid). A multiple sequence alignment was performed to compare PahZ2_KT-1 primary sequence features to structurally similar proteins identified from the DALI Protein Structure Comparison Server.^68,69 Inspection of the PahZ2_KT-1 primary sequence reveals the sequences HLD (H94–D96) and HSLD (H374–D377) at the Zn_I and Zn_II sites, respectively. Figure S6A illustrates a high degree of sequence conservation at the Zn_I site such that a conserved His–X_aa–Asp (HXD, where X represents a variable amino acid position) motif is readily identified. A similar degree of conservation is observed at the Zn_II site that yields a His–X_aa–X_aa–Asp (HXXD) motif (Figure S6C).

Figure S6B demonstrates invariable conservation of the Glu–Glu motif across all structural homologues examined here. On the basis of structural alignments with related proteins such as carboxypeptidase G2 and DapE,^38-40 the conserved Glu–Glu motif positions an acidic functional group to likely participate in water activation essential to formation of the catalytic nucleophile. Once substrate is bound, DapE has been suggested to utilize E134 (Figure S7) in the deprotonation of a metal-bound water molecule that will undergo nucleophilic attack on the bound peptide substrate.^39,72 Steady-state kinetic studies with DapE performed as a function of pH reveal the maximum reaction velocity to be independent of pH when pH ≥ 7, but to decrease with decreasing pH when pH < 7 with an apparent pK_a equal to 6.6 ± 0.2.⁷⁴ Given similar observations for DapE,⁷⁴ carbonic anhydrase,^75,76 and prolidase,⁷⁷ this pK_a suggests the pH-sensitive functional group to be a bound water molecule, where metal binding is known to depress the acid dissociation constant for water. Moreover, the positioning of E134 adjacent to the metal-bound water as well as kinetic studies with E134 mutants strongly implicate this glutamate as a general acid/base residue.^38,39,78 The observed structural/sequence conservation observed between PahZ2_KT-1 and DapE strongly implicates PahZ2_KT-1 E155 as a general acid/base in the water activation mechanism presented in Figure S6.

Catalytic Site Mutagenesis Reveals Zn²⁺ Binding Site Asymmetry.

To further examine the catalytic importance of the acidic resides discussed above, recombinant PahZ2_KT-1 mutant constructs were prepared. Mutations were made as follows: E155A, E156A, D121A, D122A/D184A, and E156A/H374A. Double-mutations were designed to target acid/metal chelating residues at either the Zn_I or Zn_II sites to disrupt metal binding at either site. Activity assays were performed as described in the Materials and Methods sections in triplicate by incubating PahZ2_KT-1 with dPAA for 24 h followed by GPC-based polymer degradation analysis. The resulting data were reported as percent activities relative to wild-type activity. Figure 3E presents relative activities for each mutant construct relative to wild-type PahZ2_KT-1. Percent relative activities for E155A, E156A, D121A, D122A/D184A, and E156A/H374A were estimated as 0.6 ± 0.2%, 1.1 ± 0.3%, 6.5 ± 0.5%, 5.9 ± 0.3%, and 1.3 ± 0.6%, respectively. The application of ANOVA testing provided support for the conclusion of statistically unique means (P < 0.0001). Further application of pairwise Student’s t tests were performed to determine whether mean activities are statistically unique for each comparison: E155A and E156A (not significant, P = 0.0929), E156A and E156A/H374A (not significant, P = 0.5483), E156A/H374A (Zn_II site) and D122A/D184A (Zn_I site) (significant, P = 0.0003), D121A and E155A (significant, P < 0.0001), and D121A and E156A (significant, P = 0.0001). While all active-site mutations nearly abolish catalytic activity, we highlight that mutations introduced to the Zn_II site have a more negative, statistically significant, effect on relative activity compared to similar mutations introduced at the Zn_I site.

Differences in mean relative activities for each PahZ2_KT-1 active-site mutant present multiple mechanistic possibilities that center on nucleophilic water activation. First, we note that all mutations examined here yield a significant loss of relative activity, which strongly supports labeling these residues as catalytic. The noted differences in relative activities for Zn_I and Zn_II site mutant constructs further suggests that each site is nonidentical in catalytic contribution. E156A and E156A/H374A mutants render more significant catalytic impairment relative to the D122A/D184A construct. On the basis of the PahZ2_KT-1 structure, we expect that each double-mutation would decrease the metal binding affinity at the Zn_I and Zn_II sites. Thus, the relative activities presented in Figure 3E provide the first clue as to whether one site is more critical to water activation or peptide substrate binding/orientation. It is not possible from these data alone to associate water activation versus peptide substrate binding with a specific site.

Identification of PahZ2_KT-1 E155 as a General Acid/Base Catalytic Residue.

Comparison with literature reports for DapE mutants implicate E155 of PahZ2_KT-1 as the only active-site residue functioning as a general acid/base in the water activation mechanism. Consistent with this, an E155A mutation yields 0.6 ± 0.2% relative activity for data points collected after a 24 h incubation. Given that water is not resolved at the active site in the zinc-bound PahZ2_KT-1 structure reported here, we next sought to evaluate the possibility of E155 functioning as a catalytic acid/base by computational methods where water molecules could be readily visualized. Molecular dynamics (MD) simulations were performed as described in the Materials and Methods sections at 4 and 25 °C with the goal of understanding how individual water molecules interact with bound zinc ions and E155. Multiple temperatures have been examined here to account for highly labile interactions involving zinc. Figure 4A provides a snapshot of the PahZ2_KT-1 metal-binding site for the system equilibrated at 4 °C. MD trajectories reveal two distinct PahZ2_KT-1 active-site metal coordination geometries. As shown in Figure 4A, the Zn_I site harbors a tetrahedral zinc atom, Zn_I, that engages in interactions with E155, D184, and D121 as observed in the X-ray structure. However, these data reveal a rotamer shift for H94 relative to the crystal structure that allows for D122 to complete the zinc coordination sphere. This is consistent with isomerization free energies reported by Dudev and Lim, where the transition from a Zn²⁺ octahedral to tetrahedral geometry occurred with ΔG = −8.4 or −9.6 kcal mol⁻¹ when one imidazole or carboxylate ligand was coordinated, respectively.⁷⁹ Thus, the replacement of H94 with D122 at the Zn_I site likely represents an arrangement with a marginal stability increase.

Figure 4.

Molecular dynamics simulations performed for 200 ns at different temperatures capture a bridging water molecule. (A) MD simulations were performed at 4 °C that reveal a stably coordinated water molecule that interacts with the Zn²⁺_II metal ion and E155. (B, C) Distribution analyses indicate that, with the exception of H94, all active-site residues form stable interactions with nearby Zn²⁺_I (B) or Zn²⁺_II (C) metal ions based on the distance between side-chain and metal. (D) MD simulations performed at 25 °C reveal water coordination as described in panel A. (E, F) Distribution analyses for active-site side-chain/metal ion distances observed at 25 °C for Zn²⁺_I(E) or Zn²⁺_II (F) sites. All structural representations were generated using the Pymol software package with active-site residues and water molecules presented in stick view. The Zn²⁺ metal ions are represented as gray spheres.

The Zn_II site differs from the Zn_I site through the coordination of three ligands from the protein and additional interactions with water. Figure 4A highlights the Zn_II site coordination geometry wherein Zn²⁺ interacts with D121, E156, and H374 alongside a bridging water molecule. We note that this water molecule bridges E155 and the Zn_II ion, which is consistent with the proposed DapE mechanism.³⁹ These data suggest that the Zn_II ion functions to polarize the water molecule, while E155 is positioned to abstract a proton leading to hydroxide formation. Thus, the residual activity reported for E155A in Figure 3E may be the consequence of another residue such as D96 functioning as a general acid/base in the absence of E155. We reason then that the role of the Zn_I metal ion is likely to facilitate both E155 and oligo(aspartic acid) substrate positioning. Support for this model can be found in the observation that all Zn_I and Zn_II site residues are nearly static in position over the course of the low-temperature simulation (Figure 4B,C).

Additional simulations performed at 25 °C demonstrate the presence of a bridging water molecule between the Zn_II metal ion and E155 at a temperature expected to increase water exchange rates. As shown in Figure 4D, a bridging water molecule is observed as described, though it is not positioned correctly for nucleophile generation. Improper positioning of this water molecule is likely the consequence of frame sampling. We expect that finer sampling would reveal the water molecule to adopt a catalytic orientation at some point during the simulation. Analysis of the distance between Zn_I and Zn_II site residues and the bound metal ion at 25 °C reveals stable interactions that do not vary significantly over time (Figure S8). Figure 4E demonstrates mean distances between the side-chain and Zn²⁺ for D121, D122, E155, D184, and H94 equal to 1.97 ± 0.04, 2.03 ± 0.06, 1.96 ± 0.04, 2.04 ± 0.06, and 6.3 ± 0.3 Å, respectively. Mean distances between Zn_II and site 2 residues for D121, H374, and E156 are equal to 1.97 ± 0.04, 2.23 ± 0.07, and 1.98 ± 0.05 Å, respectively (Figure 4F). Inspection of a plot of side-chain–Zn²⁺ distance versus time for E156 reveals side-chain oscillations within the first ~10 ns that may reflect artifactual behavior as the system establishes equilibrium. The MD trajectory shown in Figure S8 is observed to arrange the D156 side-chain at a mean distance equal to 2.06 ± 0.07 Å for the Zn_II metal ion for the first 3.87 ns followed by an increase to 3.7 ± 0.3 for the next 5.64 ns, decrease to 2.1 ± 0.1 Å for 0.49 ns, increase to 3.6 ± 0.3 Å for 0.39 ns, and, finally, decrease to 1.98 ± 0.05 Å for the remaining 190 ns. This increased side-chain flexibility for E156 is reflected in a statistically significant increase in root-mean-square fluctuation (RMSF) for E156 relative to E155 (Figure S9, P = 0.0316). Taken together, these data suggest that all Zn_I and Zn_II site residues appear to be nearly static in position alongside the bound zinc ions. On this basis, the rate of water activation is likely limited by water exchange.

Asymmetric Zn²⁺ Binding at Sites I and II.

The Zn_I and Zn_II sites each contribute unique ligands that function separately and uniquely to coordinate Zn²⁺ ions. From Figure 4A, site I utilizes a tetrahedral coordination geometry exclusively involving carboxylate functional groups, whereas site II contributes the side-chains of D121, E156, and H374 alongside water to function as Zn ligands. On this basis, we next asked if PahZ2_KT-1 binds each Zn²⁺ with unique affinity. To address this question, we utilized a steered molecular dynamics (SMD) technique known as center-of-mass (COM) pulling to stepwise dissociate each Zn²⁺ ion toward the estimation of dissociation free energy values. Figure 5A provides an illustrative look at the method itself. A Zn²⁺-bound PahZ2_KT-1 model for the isolated catalytic domain that has previously been subjected to 200 ns of production molecular dynamics simulation is subjected to an applied force centered on the Zn²⁺ bound at the Zn_II site. Through a continued application of force, the metal ion dissociates and moves further away from the protein with increasing time. The same process can then be repeated for the metal bound at the Zn_I site (Figure 5B). For experiments removing Zn²⁺ from sites I and II, the centers-of-mass between Zn²⁺ and PahZ2_KT-1 are ~1 Å prior to pull force application but reach 6 Å at the end of the 500 ps simulation. In order to use SMD trajectories for thermodynamic calculations, the bias introduced through the application of pull force must be removed. To this end, umbrella sampling techniques were applied to the extraction of frames along the SMD trajectory for re-equilibration. Configuration of umbrella sampling simulations involved sampling 27 and 34 intersubunit COM distances for the Zn_II and Zn_I sites, respectively. Each frame was subjected to a 10 ns simulation to allow for equilibrium establishment and then used for PMF calculations summarized in Figure 5C,D. Each SMD trajectory was judged by weighted histogram analysis (Figure 5E,F) as sufficiently sampled for use in PMF calculations. Each plot of PMF versus distance shown in Figure 5C,D is observed to exhibit a sharp increase in energy with increasing COM distance until a distance of ~3 nm is reached and forms an apparent plateau. Calculation of the difference between the PMF maximum and minimum provides an estimate of the apparent free energy change, ΔG, describing the dissociation of Zn²⁺ from each binding site. Figure 5C,D reveals estimates for the dissociation free energies at Zn_II and Zn_I as ΔG_ZnII = 42 ± 4 kJ mol⁻¹ and ΔG_ZnI = 62 ± 2 kJ mol⁻¹, which can be converted to dissociation equilibrium constants equal to 49 nM and 16 pM, respectively. These data indicate that binding of Zn²⁺ to each PahZ2_KT-1 metal binding site occurs with unique affinity.

Figure 5.

In silico calculation of PahZ2_KT-1 affinity for zinc metal binding. Steered molecular dynamics techniques were employed alongside umbrella sampling methods to simulate stepwise metal dissociation events at the Zn_II and Zn_I sites. For each dissociation event, frames observed along the SMD pulling trajectory are shown for the (A) Zn_II and (B) Z_I sites. The PahZ2_KT-1 catalytic domain is position restrained, while each zinc metal ion is sequentially pulled along a defined path by application of a static force vector. The resulting SMD trajectories were then subjected to umbrella sampling techniques in order to calculate the potential of mean force (PMF). The amplitude of a resulting plot of PMF versus distance between metal and PahZ2_KT-1 centers-of-mass yields an estimate of the dissociation free energy. The zinc dissociation free energies from the Zn_II (C) and Zn_I (D) sites is estimated as 42 ± 4 and 62 ± 2 kJ mol⁻¹, respectively. Adequate sampling was confirmed by weighted histogram analysis (E, F), and error estimation was obtained by bootstrap methods (n = 200).

We can categorize Zn_I and Zn_II sites as high- and intermediate-affinity binding sites. By comparison, the suppressor of fused (SUFU), metal responsive element-binding transcription factor (MTF1), carbonic anhydrase II, and rhodopsin each bind zinc with an affinity equal to 0.5 nM,⁸⁰ 30 pM,⁸¹ 0.8 pM,⁸² and 0.1 μM,⁸³ respectively. High-affinity zinc binding allows for the cell to keep the concentration of free zinc low, which is necessary to avoid accidental binding of other metals such as iron.⁸⁴ In fact, the concentration of free zinc in Escherichia coli has been estimated to be on the order of femtomolar,⁸⁵ which allows for activity modulation based on the affinities discussed here. From this, we can predict the dominant PahZ2_KT-1 ligation state under physiologically relevant conditions. If we assume a free [Zn²⁺] = 500 fM, the distribution of free PahZ2_KT-1 (E), singly bound (EZn) species, and doubly bound (EZn₂) can be calculated as 96.9%, 3.0%, and ~0%, respectively, based on a two-site, independent binding model. Figure S10 demonstrates that a further increase to 16 pM promotes a distribution of 49.9% E, 49.9% EZn, and 0.02% EZn₂. The occurrence of asymmetric PahZ2_KT-1 metal binding sites leading to unique Zn²⁺ binding affinities provides a path to the regulation of enzyme activity. The DapE structural homologue displays 60% relative activity when only a single Zn²⁺ ion is bound, but binding of two zinc ions is required for full enzymatic activity.^39,74 This observation for DapE provides the basis for a model wherein PahZ2_KT-1 activity is directly dependent upon free zinc available from the local environment. That is to say, PahZ2_KT-1 likely has a single high-affinity zinc atom always bound at the Zn_I site that would support activity, though full activity would only result when both Zn_I and Zn_II sites are metal-bound. Thus, the metal binding at the Zn_II site may serve to regulate PahZ2_KT-1 activity with varied [Zn²⁺]. PahZ2_KT-1 may have evolved to have a tunable enzymatic activity that allows for increased function when Sphingomonas sp. KT-1 encounters locally high [Zn²⁺] associated with metal-chelated poly(aspartic acid) in river water.

NaCl Promotes PahZ2_KT-1 Activity.

To date, all studies reporting on the activity of PahZ2_KT-1 catalyzed hydrolysis of oligo(aspartic acid) have utilized buffer conditions lacking supplemented salt.^32,73 This strategy ignores the fact that PahZ2_KT-1 function under physiologic conditions would necessarily include the presence of salt ions that may impact how PahZ2_KT-1 interacts with its anionic substrate. To examine the effect of salt on PahZ2_KT-1 activity, we performed additional PahZ2_KT-1 activity assays wherein enzyme was incubated with poly(aspartic acid) substrate in the presence of [NaCl] = 0–0.5 M. For each condition, time points were quenched at specific time intervals and subjected to analysis by gel permeation chromatography (GPC). The resulting chromatograms can then be used to quantify product formation as a function of time. Figure 6A illustrates the resulting plot of relative product formation as a function of time. In the absence of NaCl, a ~360 min lag phase is observed prior to product formation. By determining the slope of the line describing product formation from 360 to 1500 min, we can place an estimate on the reaction initial velocity. In contrast, conditions including 0.15 and 0.5 M NaCl yield the observation of product formation within 5–10 min of incubation (Figure 6B,C). Figure 6D presents a plot of apparent initial velocity as a function of [NaCl]. The solid line represents a linear least-squares fit to a straight line with slope equal to (377 ± 6) × 10⁻⁶ relative product min⁻¹ mM⁻¹ NaCl. Moreover, the observation of a linear and positive increase in initial velocity with increasing [NaCl] without a significant decrease in apparent time course amplitude (Figure 6B,C) suggests that salt stimulates PahZ2_KT-1 catalytic function without significant impairment to substrate binding leading to product formation.

Figure 6.

Initial velocity analysis reveals PahZ2_KT-1 catalyzed dPAA degradation is stimulated by [NaCl]. (A–C) Plots of relative product formation as a function of incubation time in the presence of 0 (A), 150 (B), and 500 (C) mM NaCl were subjected to linear least-squares analysis to estimate the slope describing product formation as a function of time for the linear phase observed in each plot. (D) A plot of initial velocity as a function of [NaCl] reveals a linear dependence with slope equal to (377 ± 6) × 10⁶ relative product min⁻¹ mM⁻¹ NaCl. All data points represent the mean of triplicate measurements with associated standard deviation presented as vertical error bars. All insets depict linear data utilized for initial velocity analysis alongside indicated correlation coefficient.

Poly(aspartic acid) Binding Site Prediction.

Inspection of the PahZ2_KT-1 crystal structure does not sufficiently address two critical components of mechanistic importance: (1) poly(aspartic acid) binding site identification and (2) the role of NaCl in promoting PahZ2_KT-1 activity. To first predict where substrate binding may occur, we turned to the cocrystal structure of DapE bound to products succinic acid and 2,6-diaminopimelic acid, which reveals protein/ligand interactions that occur in both the catalytic and dimerization domains.³⁸ The interaction of DapE with 2,6-diaminopimelic acid is supported by R258 and S181, which are each contributed from the dimerization domain. By comparison, DapE R258 is conserved in both sequence and structure in PahZ2_KT-1 as R270. Similarly, DapE S181 is conserved in PahZ2_KT-1 sequence and structure as S198. In contrast, similar conserved elements were not located in the PahZ2_KT-1 catalytic domain.

To identify the putative poly(aspartic acid) binding site on PahZ2_KT-1, an α-linked oligo(aspartic acid) pentapeptide substrate, PAA₅, was docked to PahZ2_KT-1 without constraints as described in the Materials and Methods sections. Following PAA₅ docking, the protein:substrate complex was brought to equilibrium over 200 ns in the absence and presence of 0.5 M NaCl in order to assess the impact of solution ionic strength on stabilizing electrostatic interactions between protein and ligand. Inspection of the equilibrated PahZ2_KT-1:PAA₅ complex reveals that substrate binds at the hinge between dimerization and catalytic domains such that contacts are made across both domains independent of incubation conditions (Figure 7A,B). The overall positioning of PAA₅ in the PahZ2_KT-1 binding site is similar to contacts made between the structural homologues DapE and carboxypeptidase G2 with their substrates where intermolecular interactions occur between substrate and both dimerization/catalytic domains.^38,86,87

Figure 7.

In silico identification of ligand binding site. PAA_S ligand docks to a PahZ2_KT-1 surface spanning the dimerization and catalytic domains with NaCl absent and Zn²⁺ present (A) as well as with both NaCl and Zn²⁺ present (B). (C, D) The PAA₅ position is shown for each simulation condition and reveals hydrogen bonding interactions with R162, R165, R200, R270, and R316 that exhibit NaCl-specific behaviors. (E) Analysis of overall hydrogen bonding frequency between PahZ2_KT-1 arginine residues and PAA5 in the absence (solid blue line) or presence (dashed red line) of NaCl. Mean hydrogen bonding frequency under each condition is described in the text.

We note condition-dependent differences that exist for PahZ2_KT-1 with respect to the orientation of PAA₅ in the binding site not previously reported for structural homologues. Figure 7A reveals that incubation of Zn²⁺/PAA₅-bound PahZ2_KT-1 in the absence of salt promotes an open conformation that does not position the catalytic zinc ions adjacent to bound polypeptide substrate. In contrast, the incubation of Zn²⁺/PAA₅-bound PahZ2_KT-1 with 0.5 M NaCl promotes a closed conformation that brings catalytic zinc atoms into proximity of the bound PAA₅ substrate (Figure 7B). Inspection of the local electrostatic environment for each PahZ2_KT-1 conformation indicates that each case involves substrate binding to surfaces enriched with cationic residues (Figure 7C,D). However, in all cases, we note that the C-terminal end of the substrate is positioned adjacent to catalytic zinc atoms, which is consistent with the existing literature reports that conclude PahZ2_KT-1 is an exopeptidase with a preference for cleavage of C-terminal aspartic acid units.^28,31

PAA₅ binding to open versus closed PahZ2_KT-1 involves two unique substrate binding poses. Figure 7C demonstrates that PAA₅ binding to the open conformation involves interactions in the dimerization domain, the hinge region, and the catalytic domain. Hydrogen bonding interactions are observed between PAA₅ and R270 (dimerization domain), R316 (dimerization domain), S356 (hinge region), R162 (catalytic domain), and R165 (catalytic domain). In contrast, incubation conditions that include 0.5 M NaCl and Zn²⁺/PAA₅-bound PahZ2_KT-1 promote a closed conformation that brings catalytic zinc atoms into the proximity of bound substrate. However, the closed PahZ2_KT-1 conformation promotes a unique substrate binding pose that eliminates interactions in the hinge region. Instead, interactions between PAA₅ with enzyme involve dimerization domain residues R270, R200, and R316 as well as catalytic domain residues R165 and R162 (Figure 7D).

Calculation of hydrogen bonding distance distributions from MD trajectories for interactions between PAA₅ and R162, R165, R200, R70, and R316 reveal NaCl-dependent behaviors. Figure 7E reveals three general groupings for these residues. We observe that the distribution of R162:PAA₅ hydrogen bonding interactions is bimodal in the absence of NaCl with a median distance equal to 2.58 Å but transitions to a unimodal distribution with median distance of 4.11 Å when salt is present. From this, R162:PAA₅ interactions apparently weaken in the presence of NaCl. In contrast, R165 and R200 interactions with PAA₅ undergo stabilization in the presence of NaCl. The median R165 interaction distances in the absence or presence of 0.5 M NaCl are 5.10 and 3.79 Å, respectively. However, the R165 interaction distance distribution is bimodal in the presence of 0.5 M NaCl with a major population adopting distances equal to 1.72 Å alongside a noninteracting minor population with a mean distance equal to 7.21 Å. Similarly, R200 interaction distances shorten in the presence of NaCl such that median distances equal 6.60 and 2.93 Å in the absence and presence of NaCl, respectively. Thus, R165 and R200 appear to form stronger interactions with PAA₅ in the presence of NaCl, where evidence for this conclusion is drawn from the apparent shortened hydrogen bonding distances observed in the presence of NaCl relative to conditions without NaCl. The last general grouping includes only R270, which is observed to interact with PAA₅ without impact from NaCl. Observed median interaction distances for R270 in the absence or presence of NaCl are 1.84 and 1.90 Å, respectively.

An exception to this overall grouping strategy is R316, which transitions from a median hydrogen bonding distance equal to 2.68 Å in the absence of NaCl to a highly bimodal distribution with a median distance equal to 4.98 Å in the presence of NaCl. R316 cannot be strictly classified alongside R162 as undergoing hydrogen bond weakening in the presence of NaCl. The R316:PAA₅ hydrogen bonding distance distribution calculated in the presence of NaCl includes two populations, where one increases the frequency of short hydrogen bonds and the other effectively does not interact. Taken together, these data provide insight into how PahZ2_KT-1 conformation switching between open and closed states may directly translate into catalytic activity.

NaCl Influences Conformational Dynamics.

The observation of an apparent NaCl-dependent shift in PahZ2_KT-1 between open and closed conformations provides a structural rationale for the [NaCl] promoted oligo(aspartic acid) hydrolysis reported in Figure 6. Given that a closed conformation brings catalytic zinc atoms into the proximity of bound substrate and that NaCl may promote such a transition, it is reasonable to infer that salt may consequently stimulate catalytic function by shifting the equilibrium between open and closed states. Additional molecular dynamics experiments were performed in order to better understand the impact of NaCl on the PahZ2_KT-1 conformational landscape. MD trajectories were prepared as described in the Materials and Methods sections in the presence of 0 and 0.5 M NaCl for 200 ns each in triplicate with bound PAA₅ ligand. A plot of free energy as a function of radius of gyration and root-mean-square deviation (RMSD) provides a topological perspective that relates protein conformation to free energy. Figure 8A presents such a conformational landscape plot for PahZ2_KT-1 incubated in the absence of supplemental NaCl and illustrates a restricted set of accessible low-energy conformations that are described by radii of gyration between ~3.6 and 3.7 nm and RMSD values equal to 0.2–0.3 Å. In contrast, MD simulations including 0.5 M NaCl yield a broadened accessible conformational space that spans radii of gyration between ~3.7 and 3.9 nm and RMSD values equal to 0.2–0.5 Å (Figure 8B). Therefore, these data allow for the conclusion that NaCl promotes an increase in accessible conformational switching between similar low-energy states.

Figure 8.

Impact of NaCl on PahZ2_KT-1 dimer structure and conformational dynamics. (A, B) Molecular dynamics simulations were performed wherein the PahZ2_KT-1 dimer was incubated in the (A) absence or (B) presence of 0.5 M NaCl, which indicates that NaCl promotes expanded conformational access. Three-dimensional plots of RMSD, radius of gyration, and free energy (x, y, z) provide an estimate of accessible PahZ2_KT-1 conformations for each incubation condition. Calculated free energy values are colored on the indicated scale from low to high from dark purple to yellow, respectively. All simulations were performed as described in Materials and Methods sections. (C) Dynamic light scattering experiments indicate that incubation of PahZ2_KT-1 with varied [NaCl] does not perturb dimer integrity. From Z-average distributions that plot fractional scattering intensity as a function of particle size, the mean particle sizes observed in the presence of 0.00, 0.15, and 0.50 M NaCl are estimated as 8.78 ± 0.74, 8.91 ± 0.92, and 10.0 ± 2.0 nm, respectively. Each sphere represents an individual particle size measurement with horizontal lines highlighting mean and standard deviation values.

Complementary dynamic light scattering (DLS) experiments were performed as a function of [NaCl] to assess dimer integrity over the discussed experimental conditions. DLS experiments were performed in the presence of [NaCl] = 0–0.5 M. Figure 8C presents a plot of signal intensity as a function of particle diameter. In the absence of salt, the mean particle diameter is estimated as 8.8 ± 0.7 nm. By comparison, the PahZ2_KT-1 crystal structure predicts a diameter of ~12 nm, which is consistent with the DLS data shown in Figure 8C. In the presence of 0.15 or 0.5 M NaCl, particle diameters are estimated from DLS experiments as 8.9 ± 0.9 and 10 ± 2 nm, respectively. A comparison of mean particle diameter estimates for each condition by the Student’s t test indicates no significant differences in means across all conditions examined here. We note an increase in peak width (polydispersity) for conditions that include 0 versus 0.5 M NaCl, which can be quantitatively described from the apparent increase in standard deviation for the high-salt condition. However, application of pairwise F-tests indicates that all combinations of variances are statistically identical.

Size-exclusion chromatography experiments were performed to complement DLS data which indicate that PahZ2_KT-1 incubation with NaCl retains the dimeric complex intact. We performed size-exclusion chromatography (SEC) experiments in the presence of 0 or 0.5 M NaCl. Samples were prepared for SEC analysis by incubation of PahZ2_KT-1 (0.75 mg/mL) overnight at 25 C° in the presence or absence of 0.5 M NaCl. Figure 9A presents the resulting chromatograms for each condition where conditions with 0 or 0.5 M NaCl are colored as solid blue or broken teal lines, respectively. Elution volumes corresponding to PahZ2_KT-1 monomers, dimers, and trimers have been indicated as A₁, A₂, and A₃, respectively. Equilibration conditions including 0 or 0.5 M NaCl yield elution profiles that are unique with respect to predicted molecular weight. Figure 9A,B highlights these observations wherein the PahZ2_KT-1 molecular weight is estimated as 86.6 ± 0.8 and 106.6 ± 0.9 kDa in the presence of 0 or 0.5 M NaCl, respectively. The introduction of NaCl causes an apparent shift in PahZ2_KT-1 complex stoichiometry from 1.94 ± 0.02 to 2.4 ± 0.02 when 0 or 0.5 M NaCl is present, respectively. Thus, the PahZ2_KT-1 complex assembles into an apparent dimer in the presence and absence of NaCl. The apparent shift in complex molecular weight may reflect a [NaCl]-dependent conformational change that alters migration through the size-exclusion column. Taken together, these data are consistent with the formation of a stable PahZ2_KT-1 dimer independent of NaCl, though with condition-dependent changes in dimer conformational landscape.

Figure 9.

NaCl promotes altered behavior by size-exclusion chromatography. (A) Samples were prepared for SEC analysis by incubation of PahZ2_KT-1 overnight at 25 C° in the presence or absence of NaCl. Chromatograms resulting from incubation conditions including 0 or 500 mM NaCl are represented as blue solid or broken teal lines, respectively. Elution volumes that correspond to predicted PahZ2_KT-1 monomeric, dimeric, or trimeric species are labeled as A₁, A₂, and A₃, respectively. (B) Double-Y plot of apparent molecular weight and apparent stoichiometry determined in the presence of 0 and 500 mM NaCl. All experiments were performed in triplicate. Error bars represent the corresponding standard deviation.

Principal Component Analyses Reveal NaCl-Dependent Domain Movements.

From our activity data reported in Figure 6, it is apparent that NaCl stimulates PahZ2_KT-1 catalyzed oligo(aspartic acid) degradation. Combined molecular dynamics (Figures 7A,B and 8A,B) and size-exclusion (Figure 9A) data collectively suggest that PahZ2_KT-1 incubation with NaCl also leads to an altered conformational landscape involving open and closed states. On the basis of these observations, we asked the following: How do incubation conditions impact PahZ2_KT-1 conformational motions leading to the stabilization of a closed conformation that positions active-site zinc ions adjacent to bound peptide? To address this question, additional 200 ns MD simulations were performed in varied combinations of NaCl, Zn²⁺, and PAA₅ as (1) −NaCl, −Zn²⁺, −PAA₅; (2) +NaCl, −Zn²⁺, −PAA₅; (3) −NaCl, +Zn²⁺, −PAA₅; (4) −NaCl, +Zn²⁺, +PAA₅; (5) +NaCl, +Zn²⁺, −PAA₅; and (6) +NaCl, +Zn²⁺, +PAA₅. For each experiment including bound Zn²⁺, the cation was placed in the active site using coordinates from the X-ray structure without modification. Each production quality MD trajectory was then subjected to principal component analysis (PCA) to identify protein motions that contribute to overall conformational dynamics. Principal component analysis has found widespread application in the analysis of MD trajectories since it offers a straightforward path to the reduction of multidimensional data sets into simplified data representations with lower dimensionality while still capturing most of the information contained initially.^88-92 Thus, we can use PCA to identify the principal components contributing to overall system variance leading to overall groups of correlated atom motions.

Figure 10 presents the results of PCA for each condition examined by molecular dynamics alongside the major structural motion, principal component 1 (PC1). For conditions lacking NaCl, Zn²⁺, or PAA₅, three principal components drive ~80% of system variance, where PC1 and PC2 each contribute 46.5% and 21.6%, respectively, as indicated in Figure 10A. Inspection of extracted structural models that describe PC1 and PC2 reveals subtle motions related to the catalytic domain. Calculation of the contribution per residue to PC1 and PC2 reveals a minimal contribution from the dimerization domain but a greater contribution from the catalytic domain in both chains A and B (Figure S11). Structural movements captured in PC1 can best be described as a rotational motion parallel to the plane formed by the dimerization domain. In contrast, the structural movements associated with PC2 occur as an open–close motion perpendicular to the plane of the dimerization domain. However, the range of motion for PC2 combined with its contribution to overall system variance suggests that while the closed PahZ2_KT-1 conformation is sampled under these conditions, it is not significantly stabilized.

Figure 10.

Principal component analysis (PCA) indicates that PahZ2_KT-1 conformational dynamics are condition-dependent. Scree plots illustrate the contribution of individual principal components to overall system variance as a percentage for (A) apoprotein and conditions that include (B) +NaCl, −Zn²⁺, −PAA₅; (C) −NaCl, +Zn²⁺, −PAA₅; (D) −NaCl, +Zn²⁺, +PAA₅; (E) +NaCl, +Zn²⁺, −PAA₅; and (F) +NaCl, +Zn²⁺, +PAA₅. Y-Axis labeling indicates the contribution of major principal components to system variance. Internal plot labeling indicates total variance captured with each additional principal component. Each condition includes a structural representation of the motion represented in principal component 1 (PC1). All structural representations are shown as surface views with the direction of motion indicated by an arrow.

In contrast to conditions lacking salt, PCA resulting from MD trajectories collected in the presence of 0.5 M NaCl without Zn²⁺ or PAA₅ suggests condition-dependent conformational dynamics. Figure 10B illustrates similar principal component contributions relative to conditions lacking salt such that PC1 and PC2 each contribute 43.5% and 29.7%, respectively, to overall system variance. However, inspection of extracted motions for PC1 and PC2 reveals significant differences between 0 and 0.5 M NaCl conditions. In the presence of NaCl, PC1 is observed to describe a significant open–close event involving movement of the catalytic domain perpendicular to the plane formed by the dimerization domain. While similar to PC2 in the absence of salt, the range of movement is greater for PC1 when 0.5 M NaCl is present such that PC2 (0 M NaCl) and PC1 (0.5 M NaCl) involve ~6 and 11 Å motions, respectively. Furthermore, PC1 observed with salt present contributes more than twice as much to overall system variance relative to PC2 (0 M NaCl), thereby predicting a significant open–close motion for PahZ2_KT-1 when salt is present. However, it is worth noting that the conformational landscapes presented in Figure 8A,B for conditions with 0 versus 0.5 M NaCl revealed an increased overall RMSD when salt is present, which suggests that this open–close event captured by PC1 in the presence of salt likely does not lead to a stable closed conformation. Instead, PahZ2_KT-1 likely experiences frequent transitions between open and closed conformational states when in the presence of NaCl without Zn²⁺ or PAA₅.

Additional PCA analysis of MD trajectories performed in the presence of different combinations of NaCl, Zn²⁺, and PAA₅, but with not all present, do not suggest the formation of a stable PahZ2_KT-1 closed conformation. Figure 10C,D highlight that conditions including Zn²⁺ without or with bound PAA₅ do not promote catalytic domain motion leading to a closed conformation. Supplemental free energy landscape calculations support this notion through increased RMSD and radii of gyration estimates (Figure S12). For each of these conditions, PC1 is observed to correspond to catalytic domain movements yielding an open or semiopen conformation with a high degree of active-site solvent accessibility. Similarly, simulation conditions including 0.5 M NaCl for Zn²⁺-bound PahZ2_KT-1 reveal little preference for motions supporting a stable closed conformation (Figure 10E). Examination of extracted structural motions for PC1 reveals these motions to represent a rocking motion wherein the catalytic domain rotates perpendicular to the dimerization domain such that the protein moves between open and semiopen states.

In contrast, Figure 10F demonstrates that incubation of Zn²⁺/PAA₅-bound PahZ2_KT-1 with 0.5 M NaCl promotes a closed conformation that is unique compared to that observed under any other conditions examined here. Extracted structural motions reveal a stable closed conformation wherein the catalytic and dimerization domains frequently contact one another via rocking motions that never lead to a fully open conformation. In fact, these motions attributed to PC1 under conditions with 0.5 M NaCl and bound Zn²⁺/PAA₅ contribute 60.3% to total system variance. A supplemental calculation of conformational landscapes indicates that the incubation of Zn²⁺/PAA₅-bound PahZ2_KT-1 with 0.5 M NaCl promotes a restriction in accessible low-energy states with decreased RMSD and radii of gyration relative to other incubation conditions (Figure S12). These data support a model wherein PahZ2_KT-1 binding of Zn²⁺ and PAA₅ substrate only adopts a closed conformation in the presence of NaCl, where the closed conformation likely stabilizes the position of all relevant active-site elements for catalysis.

CONCLUSION

PahZ2_KT-1 adopts a stable dimeric structure supported by nonpolar interactions occurring at the subunit interface. The X-ray structure reveals that each subunit contains two distinct domains, a dimerization and a catalytic domain. The catalytic domain is characterized as an M28 metallopeptidase family member that displays optimal enzymatic activity upon binding of dual zinc metal ions, where zinc binding occurs at Zn_I and Zn_II sites with dissociation equilibrium constants equal to 16 pM and 49 nM, respectively. Oligo(aspartic acid) substrate binding involves electrostatic interactions that occur across both PahZ2_KT-1 domains. The data reported here indicate that optimal catalysis occurs only in the presence of NaCl, Zn²⁺, and peptide substrate due to stabilization of a closed PahZ2_KT-1 conformational state that precisely positions catalytic zinc metal ions adjacent to bound peptide. The linear dependence in PahZ2_KT-1 catalyzed oligo(aspartic acid) degradation as a function of [NaCl] suggests that salt promotes a shift in the equilibrium between the open and closed conformation toward the stabilization of a closed state enzyme of catalytic relevance. Importantly, knowledge of the PahZ2_KT-1 structure alongside the PahZ1_KT-1 structure recently reported provides a structural picture of the enzymatic participants responsible for conversion of poly(aspartic acid) into free aspartic acid. Therefore, the current work provides a complete structural and mechanistic framework to build from in protein engineering studies that bring PahZ1_KT-1 and PahZ2_KT-1 together toward the goal of developing novel biodegradation agents toward poly(aspartic acid) and other water-soluble polymers.

ABBREVIATIONS

MD

molecular dynamics

SMD

steered molecular dynamics

COM

center-of-mass

dPAA

digested poly(aspartic acid)

PAA

poly(aspartic acid)

OAA

oligo(aspartic acid)

SEC

size-exclusion chromatography

PC

principal component