Expansion of the known poly(aspartic acid) hydrolases through the identification of four new PahZ1 homologs

Abstract

Polyaspartic acid (PAA) is a biodegradable polymer with various industrial applications. To date there are only three known PAA hydrolases (from the gene PahZ) capable of degrading PAA. These enzymes are expressed in two different bacteria, Sphingomonas sp. KT-1 (PahZ1_KT-1 and PahZ2_KT-1) and Pedobacter sp. KP-2 (PahZ1_KP-2). PahZ1_KT-1 and PahZ2_KT-1 form a two-component system degrading tPAA to oligoaspartic acid (OAA) and subsequently into aspartic acid. This study aims to expand the diversity of PAA hydrolases and inform efforts to improve PAA degradation. To further understand the known PahZ1 homologs, the X-ray crystal structure of PahZ1_KP-2 was determined to examine its structural homology with PahZ1_KT-1. Crystallographic analysis revealed PahZ1_KP-2 is monomeric, contrasting with the dimeric PahZ1_KT-1, yet both share a conserved serine protease catalytic triad. With the aim of expanding the PahZ1 family, four putative homologs were identified using bioinformatics and AI-based structural modeling, all of which retained the α/β hydrolase domain. Importantly, all homologs exhibited measurable PAA-degrading activity and each was classified as either monomer or dimer to further expand the diversity of PahZ1 enzymes and provide a broader toolbox for sustainable polymer degradation.

Introduction

Polymeric materials occupy an indispensable place in global society with roles in a range of applications spanning from manufacturing of plastics (Geyer et al. 2017) to use as water-retention agents (Chang et al. 2021; Abdallah 2019; Li et al. 2014; Xu et al. 2015; Yang et al. 2014; Guilherme et al. 2015) in agriculture (Namazi 2017). The central reason for this trend is rooted in their physical durability (Vroman & Tighzert 2009), which allows for use under a range of conditions without concern for significant polymer degradation. However, these durability properties also limit the extent to which polymers, primarily those petroleum-derived, are biodegraded in the environment (Vroman & Tighzert 2009). This has led to multiple strategies to promote downstream biodegradation of polymer materials. Of relevance to the current work, one strategy is the development of polymers that harbor hydrolyzable backbones including polyesters, polyamides, polyurethanes, polyureas, poly(amide-enamine)s, and polyanhydrides (Vroman & Tighzert 2009; Chandra & Rustgi 1998; Nair & Laurencin 2007).

Hydrolyzable polymers commonly find use as replacement materials for traditional polymers that are poorly, or not at all, biodegradable. For example, polycarboxylates such as poly(acrylate) are water soluble polymers (WSPs) (Huppertsberg et al. 2020) that have been used as superabsorbent materials in diapers, feminine hygiene products as well as detergents, antiscaling agents, and as a fertilizer synergist (Thombre n.d.; Paik et al. 1996; Mei-mei et al. 2019). However, multiple studies have concluded that polycarboxylates such as poly(acrylate) are environmentally stable and undergo minimal biodegradation (Bai et al. 2015; Wilske et al. 2014). Resistance to biodegradation may lead to environmental accumulation and possible pollution of ground or surface water supplies (Amass et al. 1998; Freeman et al. 1996). While poly(acrylate) is considered to only be toxic when consumed in large quantities (Dorman et al. 2018), polycarboxylates are capable of metal ion chelation (Amass et al. 1998; Freeman et al. 1996), thereby presenting the possibility of water contamination with metals as a consequence of environmental accumulation. As an alternative to poly(acrylate), poly(aspartic acid) (PAA) has found use as a biodegradable replacement material due to similarities in superabsorbent properties and the presence of a hydrolyzable amide backbone (Alford et al. 1994).

PAA can be readily synthesized from naturally occurring aspartic acid monomers using a thermal synthetic route (Thombre n.d.). When heated, aspartic acid will form a poly(succinimide) that can be ring-opened by treatment with sodium hydroxide to form thermally-synthesized poly(aspartic acid) (tPAA) (Fig. S1). Thermal PAA synthesis results in an atactic polymer with 30% α- and 70% β-amide linkages as well as irregular end groups (Hiraishi & Maeda 2011). The preferential formation of β-amide linkages consequently promotes significant branching in the polymer structure; however, such branching via β-linkages may limit amide backbone accessibility to common proteases and, therefore, limit overall biodegradation potential. Indeed, known species capable of tPAA biodegradation is limited to two river water bacterial strains, where Sphingomonas sp. KT-1 and Pedobacter sp. KP-2 have been reported to both biodegrade and assimilate tPAA (Tabata et al. 2000; Tabata et al. 1999).

PAA represents a promising candidate for implementing true circular economy principles in polymer science. Unlike traditional polymers that follow a linear ‘take-make-dispose’ model, PAA could enable closed-loop systems where the polymer is enzymatically degraded back to its constituent aspartic acid monomers, which can then be repolymerized. This cradle-to-cradle approach would not only minimize waste but also reduce dependency on new raw materials. However, realizing this circular vision requires deeper understanding of the natural enzymatic machinery that can efficiently break down PAA under various conditions. By identifying and characterizing diverse PAA hydrolases from different organisms, we can develop a robust enzymatic toolkit optimized for different degradation environments and PAA variants.

Studies of Pedobacter sp. KP-2 and Sphingomonas sp. KT-1 identified, isolated, and characterized three enzymes capable of catalyzing tPAA hydrolysis: PahZ1_KP-2, PahZ1_KT-1, and PahZ2_KT-1 (Hiraishi et al. 2003a; Hiraishi et al. 2009; Hiraishi et al. 2003b; Tabata et al. 2001). Hiraishi and coworkers concluded that PahZ1_KP-2 is functional as a periplasmic enzyme that cleaves tPAA into oligo(aspartic acid) (OAA) (Hiraishi & Maeda 2011; Hiraishi et al. 2009). Similarly, tPAA degradation in Sphingomonas sp. KT-1 involves two periplasmic enzymes, PahZ1_KT-1 and PahZ2_KT-1, which catalyze sequential reactions converting intact tPAA into (OAA) that is further digested into free aspartic acid. A PahZ2 homolog has not been identified in Pedobacter sp. KP-2. For both PahZ1 homologs, tPAA degradation involves cleavage of β-amide linkages to yield oligo(aspartic acid) (OAA). Regardless of which PahZ1 enzyme acts on tPAA, the OAA can be cleaved at α- or remaining β-amide linkages by PahZ2 to produce monomeric aspartic acid (Tabata et al. 2000; Hiraishi et al. 2003a; Hiraishi et al. 2003b; Tabata et al. 2001; Hiraishi et al. 2004).

Structural details of the PAA hydrolase proteins has been limited to our previously reported structures of PahZ1 and PahZ2 enzymes from Sphingomonas sp. KT-1 (Brambley et al. 2021; Brambley et al. 2020). The X-ray structure for PahZ1_KT-1 was observed to adopt a dimeric assembly both in the crystal structure and in solution, whereby the subunits form a dimer via antiparallel positioning of an α-helix formed by residues 244–265 located at the subunit interface (Brambley et al. 2020). A later report revealed that a single E251K mutation was sufficient to disrupt PahZ1_KT-1 dimerization to yield a monomer species (Lamantia et al. 2022). From the dimer crystal structure, each PahZ1_KT-1 monomer adopts an α/β hydrolase fold with a solvent-accessible surface harboring significant positive charge that is predicted to be responsible for PAA substrate binding through molecular dynamics (MD) methods (Brambley et al. 2020). Based on in silico predictions, enzyme activity analysis confirmed its classification as a serine protease through the incorporation of D225A, S157A, and H280A single-point mutations. In addition, R246 was identified as contributing significantly to catalytic function; R246 mutations to alanine, glutamic acid, or lysine resulted in ablation of PAA degradation activity. MD simulations predicted R246 to drive substrate positioning, thereby allowing for discrimination between α- and β-amide linkages. Taken together, this work represented a reconciliation of biochemical and structural knowledge to pave the way for future PahZ1_KT-1 engineering with the goal of developing reagents useful for industrial tPAA degradation as well as potential tPAA synthesis in cradle-to-cradle processes.

The catalytic efficiency of PAA hydrolases can be contextualized relative to other well-characterized polymer-degrading enzymes. PET hydrolases have been extensively studied, with reported turnover numbers, k_cat, ranging from 0.02 to 0.7 s⁻¹ for enzymes acting on intact PET substrates (Castro et al. 2023). The engineered variant LCC^ICCG, currently among the most efficient PET hydrolases, achieves productivities of approximately 5.8 g L⁻¹ h⁻¹ under industrial-relevant conditions (Arnal et al. 2023). For cellulases, particularly the well-characterized cellobiohydrolase Cel7A from Trichoderma reesei, turnover numbers of 2–10 s⁻¹ have been reported for the inner catalytic cycle; however, the maximal specific rate at quasi-steady state is typically an order of magnitude lower (0.1–1 s⁻¹) due to rate-limiting association and dissociation steps on the insoluble substrate (Cruys-Bagger et al. 2012; Cruys-Bagger et al. 2013; Kari et al. 2014).

A critical determinant of apparent catalytic efficiency for all polymer hydrolases is substrate accessibility; crystalline or highly ordered polymer regions present fewer cleavable bonds to the enzyme active site compared to amorphous regions (Cruys-Bagger et al. 2013). For PAA hydrolases, the atactic nature of thermally synthesized PAA, with its mixture of α- and β-linkages and significant branching, likely influences both substrate binding and catalytic turnover (Hiraishi & Maeda 2011). The preferential cleavage of β-linkages by PahZ1 enzymes, combined with the requirement for substrate positioning via electrostatic interactions with the polyanionic backbone, suggests that catalytic rates may be limited by productive substrate binding rather than the chemical step of hydrolysis. Understanding these accessibility constraints will be essential for engineering PAA hydrolases with improved industrial performance.

Polyaspartic acid is an emerging polymer with growing industrial relevance, but its efficient biodegradation remains poorly understood. Here, we report the structure for PahZ1 from Pedobacter sp. KP-2 (PahZ1_KP-2) determined by X-ray crystallography for the purpose of expanding our structural knowledge of PahZ1 hydrolase oligomeric and active site structures. Additionally, we identified four novel PahZ1 hydrolases that share structural and functional homology, despite lacking a high degree of shared primary sequence identity. These previously uncharacterized enzymes are confirmed to be PahZ1 hydrolases based on a combination of GPC-based activity assays, de novo structure prediction, monomer molecular weight comparisons, and conservation of a unique ability amongst polymer hydrolases to bind deoxynucleic acids (DNA). This work hints at a broader network of yet-undiscovered homologs with similar catalytic capabilities. Through unlocking the diversity and functionality of these enzymes, we approach a critical step toward describing the natural pathways for PAA breakdown wherein this knowledge offers an opportunity to engineer innovative enzymes with tailored properties, paving the way for transformative advances in sustainable polymer processing and environmental biotechnology.

Experimental

Materials

All solutions used to generate the data reported here were prepared with double-distilled water produced from a Purelab Ultra Genetic System (Siemens Water Technology).

Methods

PahZ1_KP-2 construct design, expression, and purification

The mature wild-type PahZ1_KP-2 nucleotide sequence coding for amino acid residues 42–306 was cloned into the pET15b vector using 5′ NdeI and 3′ BamH1 restriction sites as previously described (Hiraishi et al. 2009; Brambley et al. 2020). DNA synthesis and cloning was performed by GenScript (Piscataway, NJ). Cloning into this vector adds the following upstream residues, to include a 6x His tag to aid in purification: MGSSHHHHHHSSGLVPRGSHMDEG. The start of the mature protein is underlined with the aspartate corresponding to residue 42 of the Universal Protein Resource (Uniprot) accession code: B6VQA9. Site-directed mutagenesis to create mutant constructs (F125H, S126A, D194A, R215A, R215E, R215K) was performed by GenScript using the aforementioned PahZ1_KP-2 in pET 15b as a template. Mutant residue numbering throughout this manuscript corresponds to their position in the deposited crystal structure. PahZ1_KP-2 was transformed, expressed, and purified using the same conditions as PahZ1_KT-1 as described previously (Brambley et al. 2020). All candidate PahZ1 homolog expression plasmids were also created by GenScript (Piscataway, NJ) as described for PahZ1_KP-2 above.

X-ray structure determination and refinement

PahZ1_KP-2 crystallization screening was initially performed at the Hauptman-Woodward Institute High-Throughput Crystallization Screening Center (Luft et al. 2003) with PahZ1_KP-2 in 20 mM Tris, pH 7.4 at 13 mg/mL. Crystals were observed in a variety of conditions. Single crystals were obtained by hanging-drop vapor diffusion at 22°C by mixing a 1 μL solution containing 165 μM protein and 198 μM 23mer ssDNA of sequence 5′-GGTAGGTTACGAATCAAGTCAGT with 1 μL reservoir solution placed over a 500 μL reservoir solution containing 22% PEG3350, 0.2 M sodium fluoride, and 0.1 M Tris pH 7.5. Although DNA was included in crystallization, no electron density for the DNA was observed in the structure. The purpose of adding DNA was based on our previous report which suggested DNA binding to an allosteric site, representing a novel binding activity not observed in other known polymer hydrolases (Couch et al. 2024). To prepare for freezing, a solution containing 35% PEG3350, 0.2 M sodium fluoride, and 0.1 M Tris pH 7.5 was prepared, and the crystals were quickly transferred into this solution and flash frozen by plunging into liquid nitrogen. X-ray data were collected at a wavelength of 1.11585 Å at beamline 8.3.1 at the Advanced Light Source on a Dectris Pilatus3 S 6 M detector. The data was processed using XDS (Kabsch 2010). During data processing it was noted that possible twinning may be present. To identify potential twinning, we analyzed the data in lower symmetry space groups but could find no compelling evidence that twinning was present and a lower symmetry space group was correct. The structure was thus refined against the highest symmetry space group possible (P6₄22), and the low R factors and low phase error show that our structure agrees well with the data.

The PahZ1_KP-2 structure was determined by molecular replacement using the program Phaser (McCoy et al. 2007) in the Phenix software suite (Adams et al. 2010), with an AlphaFold (Jumper et al., 2021) model of PahZ1_KP-2 (average pLDDT confidence of 96.8 (a high confidence)) serving as the starting model. Phaser correctly located the position of one PahZ1_KP-2 molecule in the asymmetric unit. Manual structure building was followed by xyz coordinate, real space, individual B-factor, and TLS refinement using phenix.refine. The final refined model contains all PahZ1_KP-2 residues except the C-terminal residue, forty-four waters, and two residues of the N-terminal His tag, with an R_work/R_free of 22.4%/ 24.9% and good geometry. A summary of data collection and refinement statistics are provided in Table 1. Structure factors and coordinates have been deposited in the PDB as entry 9O7U.

Table 1.

Data collection and refinement statistics for the PahZ1_KP-2 crystal structure

	PahZ1KP-2
Data collection
Space group	P6422
Cell dimensions
a, b, c (Å)	165.78, 165.78, 51.20
α, β, γ (°)	90, 90, 120
Resolution (Å)	82.96–2.00 (2.07–2.00)a
Wavelength (Å)	1.11585
Total Reflections	56 846 (5444)
Unique Reflections	28 423 (2722)
CC 1/2	1.0 (0.708)
R merge (%)	1.4 (30.90)
Rmeas (%)	2.0 (43.70)
I / σ	21.5 (1.40)
Wilson B-factor	47.85
Completeness (%)	99.7 (97.4)
Redundancy	2.0 (2.0)
Refinement
No. reflections	28 409 (1930)
R work / Rfree (%)	22.4/24.9 (33.2/29.7)
No. atoms
Protein	2152
Water	44
Ligands	0
Average B-factors
Protein	78.90
Water	63.18
RMS Bond lengths (Å)	0.09
RMS Bond angles (°)	1.39
ϕ,ψ most favored (%)	98.49
ϕ,ψ allowed (%)	1.51
ϕ,ψ outliers (%)	0.00

Coordinates have been deposited in the Protein Data Bank as entry 9O7U.

^aValues in parentheses are for highest-resolution shell.

Deep learning-based structure prediction

Protein sequences of interest were identified with the Basic Local Alignment Search Tool (BLAST) courtesy of the Uniprot (UniProt 2021). Sequences of homologues expressing greater than ~40% identity relative to PahZ1_KT-1 were downloaded and saved as FASTA formatted files. Each FASTA sequence was subsequently uploaded to the Robetta web server for deep learning-based structural determination via RoseTTAFold (Baek et al. 2021; Yang et al. 2020; Song et al. 2013; Raman et al. 2009). The resulting models were ranked according to mean pLDDT scores on a 0–1 scale, with those demonstrating the highest confidence reported here. For comparison, all structures predicted using RoseTTAFold were superimposed on the corresponding structures obtained using AlphaFold, where each methodology yields predicted structures that are consistent with each other. All rendered images and electrostatic surface maps were generated in Pymol (Schrödinger 2015). Pymol was also used for pairwise alignment of homologues via the align function to the PahZ1_KT-1 crystal structure and subsequent Root Mean Square Deviation (RMSD) determination.

Docking and molecular dynamics simulations

Initial models for in silico studies were generated using AlphaFold (Jumper et al. 2021; Abramson et al. 2024). Structures for the mutations were generated using PyMol (v.3.0.0). Molecular docking was done blindly using Autodock VINA (v1.2.7) with the grid box covering the entirety of the receptors, and with an exhaustiveness setting of 100. The best docking poses for each system was selected as the starting structure of molecular dynamics simulations. The Solution Builder module of CHARMM-GUI was used to prepare simulation boxes solvated in a periodic 500 mM NaCl TIP3P water box with 10 Å padding. The protein was modelled with the charmm36 force field and ligand, a 5-mer β-PAA, with the CHARMM General Force Field. All simulations were performed using GROMACS (v. 2023.3 accelerated with CUDA v. 12.1.1). The system was minimized by steepest descent followed by a 250 ps equilibration step at the constant volume and temperature ensemble (NVT) and a series of equilibration steps at the constant pressure and temperature ensemble (NPT) gradually releasing the positional restraints on the protein and the ligand. Trajectories for analysis of each system were generated from 200 ns of unrestrained production simulation at NPT. All simulations were performed at 303.15 K and 1 bar using the V-rescale thermostat and C-rescale barostat. Built-in gmx tools were used to analyze the trajectories.

PahZ1 homolog construct design, expression, purification, and size exclusion chromatography

Identified PahZ1 homologs were identified from Sphingopyxis sp. Root1497 (Uniprot primary accession: A0A0Q7YJ49), Sphingomonadales bacterium (Uniprot primary accession: A0A4Q5S1H9), Brevifollis gellanilyticus (Uniprot primary accession: A0A512MCT2), and Erythrobacter insulae (Uniprot primary accession: A0A547PD95). All predicted, mature proteins were cloned in the pET15b vector using 5’NdeI and 3’XhoI sites. Mature residues for each construct corresponded to the following: Sphingopyxis sp. Root1497 residues 21–313, Sphingomonadales bacterium residues 21–301, Brevifollis gellanilyticus residues 23–644, Erythrobacter insulae residues 21–308. Gene synthesis, cloning, and sequencing was performed by GenScript. All constructs were transformed into BL21(DE3) Competent Cells (New England BioLabs, Ipswich, MA) and expression and purification followed the same protocol used for PahZ1_KT-1 (Brambley et al. 2020; Lamantia et al. 2022; Couch et al. 2024).

To assess the oligomeric state of each construct, protein samples were dialyzed against 50 mM sodium phosphate, pH 7.4 and concentrated to ~3.0 mg/mL. Each sample was prepared for analysis by loading 1.0 mL onto a HiPrep™ 16/60 Sephacryl™ S-200 HR column equilibrated in 50 mM sodium phosphate, pH 7.4. The column was run at a flow rate of 0.5 mL/min using a GE AKTA Purifier FPLC and protein elution was monitored at λ₂₈₀.

Gel Permeation Chromatography (GPC) Activity Assays

The tPAA substrate was synthesized according to Andrzejak et al (Andrzejak et al. 2018), and 100 mg was solubilized in 1 mL of 50 mM HEPES, pH 7.0. Then, 3 mg of thyroglobulin (MilliporeSigma, Burlington, MA) was added to the substrate to serve as an internal standard. Each sample was brought to a final sample volume of 100 μL, with 0.01 mg/mL of protein and 24.5 μL of the substrate. Samples were incubated at 37°C for 10 min, 15 min, 20 min, 30 min, 45 min, 1 hr, 2 hr, 3 hr, and 4 hr. Following incubation, each sample was heated to 90°C for 10 mins to precipitate the enzyme and stop activity, centrifuged to clarify the solution and the supernatant was then stored at −80°C.

To analyze degradation of the tPAA substrate, 10 μL of each sample was loaded onto a Shimadzu GPC with a Yarra 3 μM SEC-2000, 300 × 7.8 mm column (Phenomenex, Torrance, CA) equilibrated in 50 mM HEPES, pH 7.0. Samples were detected by a refractive index detector, and thyroglobulin was used to account for any differences in load amounts or chromatographic shifts. The common oligomeric aspartic acid (OAA) product peak appeared at a retention time of 10.5 min, and a ratio of the product peak height compared to the internal standard height, at a retention time of 5.0 min, was used to calculate the total amount of tPAA degradation at each timepoint.

The same assay protocol was followed with constructs S126A, D194A, R215A, R215E, R215K, and F125H with incubation for a single timepoint of 24 hrs. The formation of product for each mutant was compared to the wild-type at a retention time of 10.5 min, in which the wild-type was set as 100% product formation.

Results and discussion

Characterization of PahZ1_KP-2 as a structural homolog to PahZ1_KT-1

Discovering new homologs expands our understanding of enzyme diversity and functionality by providing insights into the natural mechanisms of PAA degradation and thereby offers potential for engineering novel biocatalysts with enhanced capabilities based on nature. However, the identification of homolog enzymes is challenging given there are only two known PahZ1 hydrolases, PahZ1_KT-1 and PahZ1_KP-2, and they only share 35% primary sequence identity. These enzymes are known to be functional homologs based on biochemical studies (Hiraishi & Maeda 2011; Hiraishi et al. 2009; Hiraishi et al. 2003b; Hiraishi et al. 2004; Hiraishi 2016), but it remains unclear whether the two enzymes are structural homologs as well. These PahZ1 hydrolases are known to differ in oligomeric state; the PahZ1_KT-1 crystal structure, small-angle X-ray scattering, and size-exclusion chromatography experiments indicate the presence of a dimer in solution (Brambley et al. 2020), whereas prior reports for PahZ1_KP-2 suggest a monomer (Hiraishi et al. 2009). As from these data, little information has been reported that clearly defines a common structural scaffold for PahZ1 hydrolases that would position these enzymes as structural, and functional, homologs belonging to a common PAA hydrolase family. Because knowledge of the structural relationship between these two enzymes will significantly aid in both the identification of new PahZ1 homologs, and in the engineering of commercially useful PahZ1 constructs, we set out to determine the structure of PahZ1_KP-2 to elucidate whether it is appropriate to classify PahZ1 enzymes based on a common structural scaffold.

Overview of PahZ1_KP-2 crystal structure

As first steps to determine the structure for PahZ1_KP-2, initial screening experiments, conducted at the Hauptman-Woodward Institute High Throughput Crystallization Screening Center (Luft et al. 2003) revealed single PahZ1_KP-2 crystals in a variety of conditions. Diffraction quality crystals were obtained in 22% PEG3350, 0.2 M sodium fluoride, and 0.1 M Tris pH 7.5 and cryoprotected in the crystallization condition with the PEG3350 increased to 35%. Diffraction was observed to 2.0 Å with structure determination by molecular replacement (Table 1). The structure reveals a monomer with end-to-end distance equal to 6 Å with no crystal packing interactions present that suggest higher oligomeric forms (Figs. 1A and S2). This observation is consistent with previous PahZ1_KP-2 literature (Hiraishi et al. 2009) and measurements made by Dynamic Light Scattering that conclude a mean particle diameter for PahZ1_KP-2 equal to 7.3 ± 0.5 nm; the mean particle diameter for a previously reported PahZ1_KT-1 engineered monomer (Lamantia et al. 2022) equals 6.8 ± 0.6 nm.

Figure 1.

Overview of PahZ1_KP-2 structure. A) PahZ1_KP-2 monomer presented in orientations differing by 90°. Protein is shown in cartoon representation. The serine protease catalytic triad location is highlighted in the black box. B) Superposition of PahZ1 monomer structures from Pedobacter sp. KP-2 (PahZ1_KP-2, green) and Sphingomonas sp. KT-1 (PahZ1_KT-1, blue). The monomers superimpose with an RMSD of 0.993 over 267 residues. C) Active site superimposition from PahZ1_KP-2 and PahZ1_KT-1 colored green and blue, respectively. D-E) Electrostatic surfaces calculated for PahZ1_KP-2 (D) and PahZ1_KT-1 (E) reveal a channel of positive charge spanning the front face of the protein. All structural representations were generated using the Pymol software package with active site residues presented in stick view. PahZ1_KT-1 PDB accession number: 6VE6.

Figure 1A highlights that the PahZ1_KP-2 structure adopts a canonical α/β hydrolase fold, including a putative catalytic triad, like the PahZ1 homolog from Sphingomonas sp. KT-1 (Brambley et al. 2020). Indeed, Fig. 1B illustrates strong structural homology shared between PahZ1_KT-1 and PahZ1_KP-2 with an RMSD = 0.978 observed upon superimposition of the two monomer structures. A deeper inspection of the structural overlay reveals strong similarity in side-chain positioning for catalytic triad residues, where each PahZ1_KT-1 residue has been previously demonstrated as critical for PAA degradation by activity assays involving point mutants (Brambley et al. 2020). The overlay in catalytic triad residues is presented in Fig. 1C, where active site residues from PahZ1_KT-1 and PahZ1_KP-2 are colored blue and green, respectively. As expected, PahZ1_KP-2 harbors a serine protease catalytic triad formed by D194, H249, and S126 (Fig. 1C). We note the conservation of a nearby arginine, R215, analogous to PahZ1_KT-1 (Fig. S3). In Sphingomonas sp. KT-1, R246 was experimentally observed to be important for tPAA degradation; R246A, R246E, and R246K mutations each contributed to impaired catalytic output similar to catalytic serine or aspartic acid (S157A and D225A PahZ1_KT-1 mutant constructs) mutations. Supporting Molecular Dynamics simulations predicted PahZ1_KT-1 R246 to be necessary for positioning of substrate amide bonds adjacent to the catalytic serine (Brambley et al. 2020).

Comparison of Figs. 1D and E reveal additional structural similarities between PahZ1_KP-2 and PahZ1_KT-1. Figures 1D and E present electrostatic surfaces for each protein with positively- and negatively-charged surfaces colored blue and red, respectively. Each protein harbors a trough of positive charge with the catalytic triad oriented perpendicularly. Previous computational docking experiments have predicted this to be the PAA binding site for PahZ1_KT-1 (Brambley et al. 2020).

Experimental validation of PahZ1_KP-2 catalytic triad by GPC

The Pedobacter sp. KP-2 PahZ1 crystal structure predicts that the likely PahZ1_KP-2 serine protease active site forms through contributions from D194, H249, and S126 (Fig. 1C) similar to that observed in the active site of PahZ1_KT-1 (Brambley et al. 2020), which is supported by sequence conservation in known and putative PahZ1 hydrolases. To validate these predictions, PahZ1_KP-2 active site constructs were prepared that each harbored specific point mutations testable using our previously reported GPC-based activity assay (Brambley et al. 2021; Brambley et al. 2020; Lamantia et al. 2022). Briefly, each PahZ1_KP-2 mutant was incubated with tPAA, followed by subsequent heat inactivation to precipitate the enzyme. Similar to Sphingomonas sp. KT-1 PahZ1 (Lamantia et al. 2022), incubation of PahZ1_KP-2 at temperatures greater than ~55 C° promotes irreversible formation of insoluble aggregate species (Fig. S4). Thus, heat treatment of all samples enables reaction quenching by removal of enzyme by precipitation and subsequent centrifugation. Following enzyme precipitation, all samples were injected onto a GPC column for analysis as described in the Methods section and reported previously (Brambley et al. 2020; Lamantia et al. 2022).

To demonstrate wild-type PahZ1_KP-2 behavior in this assay, a time course was first constructed as shown in Figs. 2A and B. Figure 2A presents chromatograms collected as a function of increasing time. As shown by the red solid line in Fig. 2A, injection of tPAA alone yields the observation of a dominant peak with retention time equal to ~9.3 minutes. After incubation of PahZ1_KP-2 with tPAA for ~120 minutes, a shift occurs to yield a major peak retention time equal to ~9.8 minutes. Over the same time period, oligo(aspartic acid) (OAA) formation is observed through formation of a peak with retention time equal to ~10.5 minutes. The observed increase in peak intensity at later retention times reflects the redistribution of polymer mass from larger polydisperse species to smaller, more uniform oligomeric products as enzymatic cleavage progresses. Using OAA formation as a signal, calculation of normalized peak area allows for the construction of a plot of percent degradation versus time (Fig. 2B), which indicates that the appearance of reaction completion occurs after ~125 minutes under these conditions. By comparison, the same reaction involving PahZ1_KT-1 exhibits a plateau in a plot of relative OAA accumulation versus reaction time after ~125 minutes also (Brambley et al. 2020). We note that absolute substrate conversion percentages are not reported here, as the GPC assay monitors relative OAA product formation normalized to wild-type enzyme levels to enable comparative analysis of catalytic efficiency across constructs.

Figure 2.

Experimental identification of PahZ1_KP-2 active site. A) Gel Permeation Chromatography (GPC) chromatograms showing PahZ1_KP-2 activity as a function of increasing PAA hydrolysis reaction time. With increasing reaction time, chromatograms exhibit a right-shift to yield peak maxima at later retention times from t = 0 minutes (solid red line) to 240 minutes (solid black line). Chromatograms are represented by solid red, orange, green, blue, violet, pink, light blue, and black lines corresponding to t = 0, 10, 20, 30, 60, 120, 180, and 240 minutes, respectively. All measurements were performed in triplicate (n = 3). OAA peak position indicated by dashed arrow. B) Average percent PAA degradation as a function of reaction time based on increasing OAA formation. C) End-point comparison of PahZ1_KP-2 mutant catalyzed tPAA degradation after 24 hour incubation by GPC. Chromatograms are represented as solid black, red, orange, pink, purple, blue, green, and mint lines for experiments performed with wild-type enzyme, tPAA alone, S126A, D194A, F125H, R215A, R215E, and R215K, respectively. D) Maximal PahZ1_KP-2 catalyzed tPAA degradation activity observed for reactions performed with mutant constructs relative to wild-type enzyme. All bar graph plots are represented with mean and standard deviation derived from triplicate measurements. The results of ANOVA testing are represented as P < 0.0001 (****).

PahZ1_KP-2 active site mutants were assessed for tPAA degradation activity using the described GPC assay. Each mutant was incubated with tPAA for 24 hours to capture any residual activity resulting from diminished catalytic output associated with each specific mutation. Qualitative inspection of the resulting endpoint chromatograms presented in Fig. 2C reveals appreciable OAA formation relative to wild-type enzyme only for R215A and F125H point mutants. Calculation of PAA degradation activity relative to wild-type enzyme confirms this qualitative observation by revealing near-complete ablation of relative activity for S126A (5 ± 2%), which is in agreement with a previous report (Hiraishi et al. 2009). Additional active site mutations revealed similar activity ablation: R215E (2.9 ± 0.5%), R215K (2.3 ± 0.8%), and D194A (2.8 ± 0.9%) (Fig. 2D). In contrast, R215A and F125H point mutants retain partial function with relative activities equal to 29.9 ± 0.9 and 61.6 ± 0.5%, respectively. Application of ordinary one-way and Brown-Forsyth ANOVA testing reveals statistically significant differences amongst mean values in the dataset as indicated by P < 0.0001 (****) calculated by each method. A Tukey’s multiple comparisons test reveals that all point mutants exhibit diminished catalytic activity relative to wild-type enzyme (P < 0.0001, ****). However, activity observed from R215A is statistically unique with larger mean relative activity compared to either R215E or R215K constructs (P < 0.0001, ****), though the difference between mean relative activities for R215E versus R215K is nonsignificant (P = 0.9805, ns). Similarly, the mean relative activity observed for F125H is statistically unique relative to wild-type enzyme as well as all other point mutants including R215A.

These data strongly support the catalytic triad identification including D194 and S126 but also extend the active site to include R215 and F125. Investigation into the latter was due to the presence of a non-conserved Phe in the lipase box (Fig. 3A) which is necessary for catalytic activity of PahZ1_KP-2. The F125H mutation reduces catalytic activity by 40%, indicating that the phenylalanine phenyl group contributes significantly to optimal catalytic efficiency, likely through hydrophobic substrate interactions or maintenance of active site geometry. The partial activity retention with histidine’s imidazole suggests this position requires occupancy but tolerates chemically distinct side chains with reduced efficiency.

Figure 3.

Identification of PahZ1 homologs based on primary sequence and structural homology with characterized PahZ1 enzymes. A) Multiple sequence alignment including PahZ1 enzymes from Pedobacter sp. KP-2 and Sphingomonas sp. KT-1 alongside novel candidate enzymes reveals significant conservation between sequences. Emphasis is placed on the lipase box associated with active site residues. All residue positions are colored based on amino acid classification. Asterisks indicate invariant sequence conservation. Residue numbers are based on full-length protein sequences identified from the Uniprot database. B) Predicted protein structures from protein sequences identified from Panel A. All known and candidate PahZ1 proteins are represented in cartoon view and as an electrostatic surface in the same orientation as depicted in Fig. 1B. Sequence identity and similarity estimates are provided. Boxes and labels have been included to indicate the position of the catalytic triad and catalytic serine, respectively. All structural representations were generated using the Pymol software package with species, gene name, and shared sequence identity with PahZ1_KT-1 indicated.

Identification of novel PahZ1 hydrolase proteins

AI assisted modeling of putative PahZ1 homologues reveals modularity of PahZ1-like domains

To better understand the extent to which the PahZ1 domain architecture occurs in nature, we built upon the observed structural homology shared between PahZ1_KP-2 and PahZ1_KT-1 by application of bioinformatic techniques to identify potential PahZ1-like domains in uncharacterized proteins. To begin, we utilized the Basic Local Alignment Search Tool (BLAST) (Altschul et al. 1990) with the PahZ1_KT-1 primary sequence to identify a series of putative homologues sharing greater than ~40% sequence identity. Our goal was to find putative homologs with a range of sequence similarities varying between PahZ1_KT-1 and PahZ1_KP-2, given they are structural homologs. While this limits the potential scope, it allowed for a manageable number of candidates to characterize given our initial goal of expanding the known PahZ1 family members beyond two. Resultant sequences demonstrating known signature motifs including the lipase box characterized by the sequence GHSAG (Boel et al. 1988), and PahZ1_KT-1 itself as a controlled reference, were subsequently used for theoretical structure prediction using Robetta, a RoseTTAFold web server, alongside AlphaFold, as described in the Methods section. Figure 3A presents a sequence alignment of the lipase box region for PahZ1 homologs from Pedobacter sp. KP-2 and Sphingomonas sp. KT-1 alongside four novel, BLAST-identified candidate homologs. The lipase box motif is observed to be invariably conserved in all proteins with the exception of residue PahZ1_KP-2 F165 (numbered as F125 in the crystal structure), which is conserved in all other proteins as a histidine. As expected, the catalytic serine, S176 and S166 in PahZ1_KT-1 and PahZ1_KP-2, respectively, (numbered as S157 and S126 in the recombinant constructs used for crystallization) is invariable in conservation for all PahZ1 homologs. All residue numbering in Fig. 3A is representative of the intact protein sequence obtained from the Uniprot (UniProt 2021) database.

Figure 3B reveals Robetta-predicted structures for each PahZ1 candidate identified from the discussed BLAST search. All predicted structures retain the PahZ1 α/β domain architecture despite varying degrees of primary sequence conservation. For example, the predicted PahZ1_KP-2 structure not only agrees well with the experimental crystal structure reported here but also exhibits structural homology with the Sphingomonas sp. KT-1 structure as expected despite the two proteins sharing only 35% primary sequence identity. Similar observations are made for all candidate PahZ1 domains presented in Fig. 3B. As complementary methods that also align with the crystal structure, AlphaFold models were prepared for each putative PahZ1 homolog alongside experimentally characterized enzymes from Sphingomonas sp. KT-1 and Pedobacter sp. KP-2. Superimposition of structures predicted by either AlphaFold or RoseTTAFold overlay well within each grouping (Figs. 4A-F), with predicted structures for PahZ1_KP-2 (Fig. 4E) and PahZ1_KT-1 (Fig. 4F) demonstrating strong agreement when superimposed on their respective X-ray crystal structures (Fig. S5).

Figure 4.

Case-by-case comparison of AlphaFold versus RoseTTAFold methodologies (A-F). For each case, putative PahZ1 homolog structures calculated by AlphaFold or RoseTTAFold are superimposed on one another and labeled as AF or RF, respectively, to indicate calculation method. Structural overviews are provided alongside an active site highlight to clearly demonstrate the strong similarities between structures predicted by different methods. Predicted structures are provided for PahZ1 homologs from Pedobacter sp. KP-2 and Sphingmonas sp. KT-1 as controls, which each agree with experimentally determined X-ray crystal structures. All structural representations were prepared using the Pymol software package.

To provide a more robust assessment of structural similarity independent of sequence length differences, TM-align analysis was performed for all PahZ1 homologs using the PahZ1_KT-1 crystal structure (PDB: 6VE6) as the reference (Table 2) (Brambley et al. 2020; Zhang & Skolnick 2005). TM-align provides length-independent TM-scores that range from 0 to 1, where scores above 0.5 indicate proteins sharing the same overall fold. All five PahZ1 homologs exhibited TM-scores greater than 0.90 when aligned to PahZ1_KT-1, with values ranging from 0.904 (PahZ1_Bfg) to 0.970 (PahZ1_Spyx), strongly supporting classification of these proteins within the same structural fold family. Notably, the sequence identity values derived from TM-align structural alignments (Table 2) correspond closely to those obtained from BLAST sequence alignments (Fig. 3B). For example, PahZ1_KP-2 exhibits 39.2% sequence identity by TM-align compared to the 41% identity reported from BLAST analysis, while PahZ1_Spyx shows 92.7% identity by TM-align, consistent with its status as the most closely related homolog identified in our BLAST search. The concordance between sequence-based (BLAST) and structure-based (TM-align) identity calculations provides additional confidence that these homologs share both sequence and structural homology with the characterized PahZ1 enzymes. The RMSD values for the structural alignments ranged from 0.14 Å (PahZ1_Spyx) to 1.12 Å (PahZ1_KP-2), with all values falling well below the 2 Å threshold typically used to indicate high structural similarity. The number of aligned residues (248–263 out of 268 reference residues) demonstrates that the structural conservation extends across essentially the entire PahZ1 domain for all homologs examined. In each case, overall PahZ1 domain architecture is predicted with conserved positioning of all relevant active site residues including the PahZ1-specific arginine implicated in PAA substrate specificity (Brambley et al. 2020).

Table 2.

TM-align structural comparison of PahZ1 homologs against PahZ1_KT-1 reference structure

PahZ1 Homolog	Chain Length (residues)	Aligned Length*   (residues)	RMSD (Å)	Sequence Identity (TM-align)	TM-scorea
PahZ1KP-2	265	260	1.12	39.2%	0.941
PahZ1Ebac	265	263	0.97	45.2%	0.959
PahZ1Bfg	249	248	1.03	48.8%	0.904
PahZ1Spd	264	261	0.49	59.8%	0.968
PahZ1Spyx	260	260	0.14	92.7%	0.970

^a TM-align structural alignments were performed using the PahZ1_KT-1 crystal structure (PDB: 6VE6, 268 residues) as the reference chain. TM-scores were normalized to the reference chain length. TM-scores > 0.5 indicate proteins sharing the same fold. Sequence identity values represent the fraction of identical residues in the structurally aligned regions.

The largest structural deviation from PahZ1_KT-1 reported here is associated with the candidate PahZ1 homolog from Brevifollis gellanilyticus (Gene Name: BEG01NITE_38030) shown in Figs. 3B, 4C, and S6. The protein sequence associated with the gene was obtained from the Uniprot database using accession number A0A512MCT2 and reveals a predicted multidomain complex (Fig. S6). We note that Fig. 3B and 4C highlight only the PahZ1-like domain, which shares strong structural homology with PahZ1_KT-1 (Table 2). In the Brevifollis gellanilyticus protein, a PahZ1-like domain is observed to be fused to a sequence sharing considerable homology with a capsular poly--glutamic acid (PGA) synthetase (capA) from Bacillus subtilis (Ashiuchi et al. 1999; Presecan et al. 1997; Makino et al. 1989). CapA is thought to contribute to the synthesis of Bacillus sp. capsules via transport of PGA. A search of the Protein Data Bank for CapA yielded no structures comparable to our predicted result. However, a deposition in the Alphafold structure prediction database identified one such model with high confidence (pLDDT 90). The result of a pairwise alignment of the Brevifollis gellanilyticus PahZ1-like domain with the crystal structure for PahZ1_KT-1 and the Brevifollis gellanilyticus CapA-like domain with the Alphafold-predicted CapA (AfCapA) is shown in Fig. S6. The multidomain Brevifollis gellanilyticus protein, predicted by Robetta, is shown in blue, while the PahZ1_KT-1 structure and the AfCapA Alphafold model are shown in violet and green, respectively.

Molecular dynamics simulations reveal conserved roles for basic residues in substrate positioning across PahZ1 homologs

To assess whether the conserved basic residues identified by sequence alignment contribute to substrate binding across the PahZ1 family, all-atom molecular dynamics (MD) simulations were performed for each PahZ1 homolog with a β-linked aspartic acid pentamer (β-PAA₅) docked to the cationic substrate-binding surface. Each homolog, including those from Sphingopyxis sp. Root 1497, Brevifollis gellanilyticus, Pedobacter sp. KP-2, Sphingomonadales bacterium, Erythrobacter insulae, and Sphingomonas sp. KT-1, was subjected to 200 ns of production MD following equilibration using the CHARMM36 all-atom force field. Wild-type simulations were compared against constructs harboring alanine substitutions at conserved basic positions flanking the active site: R264A/H249A/H248A in Sphingopyxis sp. Root 1497, R230A/H215A/H214A in Brevifollis gellanilyticus, R255A/R239A in Pedobacter sp. KP-2, R249A/H234A/H235A in Sphingomonadales bacterium, R260A/H245A/H244A in Erythrobacter insulae, and R246A/H230A/H231A in Sphingomonas sp. KT-1. Notably, Pedobacter sp. KP-2 lacks the conserved histidine pair found in other homologs, instead possessing two arginine residues (R255 and R239).

Analyses of hydrogen bonding frequency and protein-ligand atomic contacts reveal variable effects of mutagenesis across the PahZ1 family (Fig. 5). The most pronounced differences between wild-type and mutant enzymes were observed for PahZ1_Spyx (Fig. 5A and B). The wild-type enzyme displays a broad hydrogen bonding distribution with substantial populations at 2 and 4 bonds, while the R264A/H249A/H248A mutant shifts dramatically toward zero, with approximately 55% of frames lacking any protein-ligand hydrogen bonds. The contact analysis corroborates this finding, showing a clear leftward shift toward lower contact numbers in the mutant construct. Similar, though more modest, effects were observed for PahZ1_Bfg (Fig. 5C and D) and PahZ1_Spd (Fig. 5G and H), where wild-type enzymes maintain measurable populations at 2 to 4 hydrogen bonds that are diminished upon mutagenesis.

Figure 5.

Molecular dynamics simulations reveal conserved roles for basic residues in substrate positioning across PahZ1 homologs. Frequency distributions of protein-ligand hydrogen bonds (left column) and protein-ligand atomic contacts (right column) comparing wild-type (purple) and triple alanine mutant (gray) PahZ1 constructs from six bacterial species. A β-linked aspartic acid pentamer (β-PAA₅) was docked to the cationic substrate-binding surface of each homolog and subjected to 200 ns of production molecular dynamics using the CHARMM36 all-atom force field. Panels A-B, Sphingopyxis sp. Root1497 (PahZ1_Spyx); panels C-D, Brevifollis gellanilyticus (PahZ1_Bfg); panels E-F, Pedobacter sp. KP-2 (PahZ1_KP-2); panels G-H, Sphingomonadales bacterium (PahZ1_Spd); panels I-J, Erythrobacter insulae (PahZ1_Ebac); panels K-L, Sphingomonas sp. KT-1 (PahZ1_KT-1). Mutant constructs harbor alanine substitutions at conserved basic residues flanking the active site as indicated in each panel legend. Wild-type enzymes generally exhibit higher hydrogen bonding occupancies and broader contact distributions compared to mutant constructs, consistent with a conserved role for these basic residues in electrostatic anchoring of the polyanionic tPAA substrate.

In contrast, the remaining homologs exhibited minimal hydrogen bonding in both wild-type and mutant conditions. For PahZ1_KP-2 (Fig. 5E and F), PahZ1_Ebac (Fig. 5I and J), and PahZ1_KT-1 (Fig. 5K and L), both wild-type and mutant enzymes primarily occupy a zero-bond state, with differences limited to the frequency of transient interactions. The contact profiles for these homologs show substantial overlap between wild-type and mutant distributions. Notably, PahZ1_KP-2 maintains relatively low contact numbers for both constructs, which may reflect its unique active site architecture lacking the conserved histidine pair. For PahZ1_KT-1, the wild-type shows a higher frequency of low-contact frames than the mutant, suggesting that R246 may play a specific role in substrate orientation rather than bulk contact stabilization, consistent with prior work highlighting its role in substrate specificity (Brambley et al. 2020).

These results indicate that while conserved basic residues contribute to substrate positioning in some homologs (particularly PahZ1_Spyx and PahZ1_Bfg), the magnitude of this contribution varies considerably across the family. The observation that several homologs maintain low hydrogen bonding frequencies even in the wild-type state suggests that productive substrate binding may involve dynamic, transient interactions rather than stable hydrogen bonding networks. Alternatively, the 200 ns simulation timescale may not fully capture all relevant binding modes.

Taken together, the structural and functional studies with PahZ1_KP-2, combined with MD simulations across all characterized homologs, reveal both conserved features and informative differences within the PahZ1 enzyme family. All homologs share tertiary architecture (TM-scores >0.90) and active site organization, including the serine protease catalytic triad and substrate-positioning arginine. However, the MD simulations indicate that the specific contributions of conserved basic residues to substrate binding vary across homologs. This functional convergence despite mechanistic variation suggests that the cationic binding surface represents a permissive framework for substrate recognition rather than a single optimized solution. Based on these findings, we propose that PahZ1 homologs may be classified according to oligomeric state: Class 1 enzymes (dimers, e.g. PahZ1_KT-1, PahZ1_Spyx) and Class 2 enzymes (monomers, e.g. PahZ1_KP-2, PahZ1_Spd, PahZ1_Ebac). Enzymes such as PahZ1_Bfg that dimerize through alternative mechanisms (e.g. disulfide formation) may constitute a third class. This classification framework provides a systematic nomenclature for organizing the expanding PahZ1 family based on quaternary structure and confirmed PAA hydrolase activity.

Recombinant expression and Oligomeric architecture of putative PahZ1 hydrolase enzymes

To evaluate candidate proteins as bona fide PahZ1 hydrolases, each protein was expressed using recombinant methods in bacteria and included an N-terminal His₆-tag to enable affinity purification, as detailed in the Methods section. Briefly, each protein was overexpressed in E. coli using isopropyl β-D-1-thiogalactopyranoside (IPTG) induction, followed by, after cell lysis and centrifugation, a single-step Ni²⁺-nitrilotriacetic acid (Ni-NTA) affinity chromatography purification as described previously for PahZ1_KT-1 (Brambley et al. 2020; Lamantia et al. 2022). Fig. 6A illustrates that this general method is sufficient to obtain protein samples that are >90% pure as judged by SDS-PAGE methods. Based on the predicted structural models presented in Fig. 3B and Fig. 4, Fig. 6A qualitatively illustrates a single gel band at the expected molecular weight equal to approximately 35 kDa for each construct, with the exception of the sequence from Brevifollis gellanilyticus, which is expected to have a monomer molecular weight close to 70 kDa based on the presence of putative PahZ1- and CapA-like domains. We note the presence of a cysteine residue within the Brevifollis gellanilyticus protein; Lanes 7 and 9 in Fig. 6A each include this protein but loaded in the absence (Lane 7) or presence (Lane 9) of reducing agent. Inspection of Lane 7 reveals two distinct bands, where the higher molecular weight band migrates at a position consistent with a possible dimer complex. That this complex is only observed in the absence of reducing agent suggests that it is the consequence of disulfide formation.

Figure 6.

Biochemical and functional characterization of PahZ1 homologs. (A) SDS-PAGE analysis of purified recombinant PahZ1 proteins showing >90% purity. Lane 1: PageRuler Unstained Protein Ladder (Thermo Fisher Scientific); Lane 2: PahZ1_KP-2; Lane 3: PahZ1_Spd; Lane 4: PahZ1_Ebac; Lane 5: PahZ1_KT-1; Lane 6: PahZ1_Spyx; Lane 7: PahZ1_Bfg (non-reduced); Lane 8: Empty; Lane 9: PahZ1_Bfg (reduced). Molecular weight markers (MW) are indicated in kDa. Image captured using the stain free setting on a ChemiDoc MP instrument. (B) Summary of experimental parameters for each PahZ1 homolog including calculated (MW_calc) molecular weights, solvent accessible surface areas (SASA) from AI-predicted models, and theoretical isoelectric points (pI). Averages exclude PahZ1_Bfg (*). (C) Size exclusion chromatography analysis revealing the oligomeric states of PahZ1 homologs, with corresponding molecular weights indicated in parentheses. (D) Enzymatic activity of PahZ1 homologs measured as percent tPAA degradation at 1, 4, and 24 hours. Error bars represent standard deviation from three independent experiments.

Fig. 6B provides a summary of the calculated molecular weight, solvent accessible surface area (based on predicted models), and isoelectric point for each construct, as calculated from the His₆-tagged PahZ1 sequence. We note an average pI equal to 7.7 ± 1.5, which suggests the average PahZ1-like domain would harbor a net positive charge at physiological pH. This is consistent with the PahZ1_KT-1 structure previously reported (Brambley et al. 2020), the PahZ1_KP-2 structure reported here (Fig. 1), and the known ability of both proteins to bind anionic ligands that include tPAA (Hiraishi & Maeda 2011; Hiraishi et al. 2009; Tabata et al. 2001; Hiraishi et al. 2004; Brambley et al. 2020; Lamantia et al. 2022; Hiraishi 2016) and deoxyribonucleic acid (DNA) (Couch et al. 2024). Specifically, the estimated pI for PahZ1_KT-1 and PahZ1_KP-2 is 8.02 and 9.25, respectively. Candidate PahZ1 proteins from Sphingopyxis sp. Root 1497 (denoted in Fig. 6B as PahZ1_Spyx) and Sphingomonadales bacterium (denoted in Fig. 6B as PahZ1_Spd) have similarly high pI values equal to 9.19 and 9.72, respectively. However, Fig. 6B also highlights two notable outliers, the putative PahZ1 homologs from the Brevifollis gellanilyticus (denoted in Fig. 6B as PahZ1_Bfg) and Erythrobacter insulae (denoted in Fig. 6C as PahZ1_Ebac), which have estimated pI values equal to 6.75 and 6.26, respectively. We note that the final yields for PahZ1_Ebac and PahZ1_Bfg are generally lower than the other candidate PahZ1 homologs (Fig. 6C). This is likely the result of a protein isolation protocol that including buffer and pH conditions optimized for the known PahZ1 enzymes from Sphingomonas sp. KT-1 and Pedobacter sp. KP-2. Future work will address alternative protocols that may enable increased yields of these proteins for biochemical studies.

Given PahZ1_KT-1 and PahZ1_KP-2 have been observed to be dimeric and monomeric, respectively, size exclusion chromatography methods were utilized to determine the oligomeric state of each candidate PahZ1 homolog. Figure 6C presents the results of these experiments for each homolog, where oligomer molecular weights have been estimated based on calibration standards of known molecular weight. PahZ1_KT-1 was run alongside unknown samples as a positive control, yielding an estimated molecular weight consistent with a dimeric complex equal to 65.4 kDa. Division of the experimentally determined complex molecular weight by the calculated monomer molecular weight, , yields a ratio of 2.0 for PahZ1_KT-1, which suggests dimers as a dominant species. Similarly, PahZ1_Spyx and PahZ1_Bfg (non-reducing conditions) each appear to adopt dimer assemblies as well, where = 2.0 for each. In contrast, PahZ1_Spd and PahZ1_Ebac elution profiles are most consistent with a monomer state, where = 1.0 for both proteins. Thus, candidate PahZ1 proteins exist as apparent monomers or dimers.

Dynamic Light Scattering (DLS) experiments were performed to provide additional support regarding oligomeric assembly states. Experiments were conducted with known proteins PahZ1_KT-1 and PahZ1_KP-2 alongside unknown proteins that were isolated at compatible yields (PahZ1_Spd and PahZ1_Spyx). DLS data presented in Fig. S7 allow for estimation of the mean particle diameters in solution for PahZ1_KT-1 and PahZ1_KP-2 as 8.4 ± 0.3 and 7.3 ± 0.5, respectively. Based on complementary data from the prior literature, size exclusion chromatography data from Fig. 6C, and the X-ray structures for PahZ1_KP-2 and PahZ1_KT-1, the former and latter are assumed to exist as either monomers or dimers, respectively (Hiraishi et al. 2009; Brambley et al. 2020; Lamantia et al. 2022; Hiraishi 2016). The application of statistical testing to the DLS data presented in Fig. S7 provides additional clues to reinforce the classifications of putative PahZ1 homologs as dimers or monomers. Based on DLS data, the mean particle sizes of candidate PAA hydrolases from Sphingopyxsis sp. Root 1497 and Sphingomonadales bacterium are 9 ± 1 and 6.9 ± 0.4 nm, respectively, and are revealed by Tukey’s multiple comparisons test to be statistically unique with P < 0.0001. Additional comparisons reveal that the candidate protein from Sphingopyxsis sp. Root 1497 is not statistically unique relative to wild-type PahZ1_KT-1, whereas the protein taken from Sphingomonadales bacterium is statistically smaller than wild-type PahZ1_KT-1 (P < 0.005). The opposite result is observed when comparing PahZ1_KP-2 with candidate PAA hydrolases from Sphingopyxsis sp. Root 1497 (P < 0.0001) and Sphingomonadales bacterium (nonsignificant difference, P > 0.1234). Taken tother, the combination of size-exclusion chromatography (Fig. 6C) and DLS (Fig. S7) data allow for preliminary PahZ1 classifications such that Class 1 (PahZ1_KT-1-like, dimer) proteins may include PahZ1_KT-1 and PahZ1_Spyx, and Class 2 (PahZ1_KP-2-like, monomer) proteins may include PahZ1_KP-2, PahZ1_Spd, and PahZ1_Ebac. Classification of PahZ1_Bfg requires the recognition that some PahZ1 proteins may uniquely form dimers based on disulfide formation rather than the characterized residues identified at the PahZ1_KT-1 dimer interface. As such, we propose that PahZ1_Bfg should be classified into a third group, Class 3, where members are capable of dimer formation based on non-standard interaction mechanisms.

Additional support for PahZ1 oligomeric state can be drawn from the conservation of a primary sequence motif identified in the α-helix (L247-H262) that drives dimer formation for PahZ1_KT-1 (Brambley et al. 2020; Lamantia et al. 2022). The PahZ1_KT-1 dimer interface is formed by the interaction of a single α-helix from each subunit, which are positioned antiparallel to one another. The center of the helix is largely hydrophobic but is flanked on either end by E251 and H262. Previous computational calculations predicted that a nonpolar residue, A255, located at the helix center would contribute significantly to dimer stabilization given that its mutation to either glutamate or tryptophan was sufficient to destabilize the dimer as predicted from PDBePISA analysis yielding ΔΔG = 6.3 and 5.7 kcal mol⁻¹ A255E and A255W mutations, respectively (Brambley et al. 2020). The antiparallel helix arrangement positions E251 and H262 to interact on either end of the helix. Indeed, a single E251K mutation is sufficient to disrupt the PahZ1_KT-1 dimer into a monomer state (Lamantia et al. 2022). Inspection of the primary sequence and AI-predicted structures of PahZ1_KT-1 and PahZ1_KP-2 alongside each candidate PahZ1 homolog reveals a sequence motif surrounding PahZ1_KT-1 E251 that may be used predict oligomeric state. Comparison of PahZ1_KT-1 and PahZ1_Spyx sequences reveals the same motif, GEFF, where E is at the 251^st position in the PahZ1_KT-1 crystal structure (Fig. 7; PahZ1_KT-1, G250-F253; PahZ1_Spyx: G268-F271). Similarly, a histidine is found at the opposite end of the helix in these PahZ1 homologs (H262 in PahZ1_KT-1; H280 in PahZ1_Spyx). In contrast, the glutamate is not conserved in the other proteins: GLFYF (PahZ1_Spd: G253-F257), GQTFF (PahZ1_Bfg: G245-F249), GQSFF (PahZ1_Ebac: G275-F279), and GQTFF (PahZ1_KP-2: G270-F274). In addition, no histidine is conserved at the opposite end of the helix as observed in PahZ1_KT-1 and PahZ1_Spyx.

Figure 7.

Sequence alignment of the dimerization interface in PahZ1 homologs. Multiple sequence alignment of the PahZ1 dimer interface region from six bacterial species. Residues are colored according to chemical properties. Arrows mark positions 251 and 262, critical for dimer formation in PahZ1KT-1. Asterisks indicate conserved residues. Dimeric enzymes (top) contain the GEFF motif with conserved E251 and H262 residues that form hydrogen bonds at the interface, while monomeric enzymes (bottom) display alternative sequences at these positions.

It should be noted that Fig. 6C indicates that PahZ1_Bfg is dimeric in solution under non-reducing conditions, though primary sequence analysis demonstrates no GEFF motif. The presence of the CapA domain harboring a cysteine residue in PahZ1_Bfg is likely the reason for this since dimer formation appears to be dependent on whether reducing agent is present or not. Additional evidence for a dimerization interface that differs from PahZ1 dimer hydrolases was observed using the AlphaFold multimer prediction server (Abramson et al. 2024), which did not place the expected helical regions adjacent to one another for interactions (Fig. S8). Moreover, control structures were generated for all other homologs; the α-helix formed in PahZ1_KT-1 residues L247-H262 were correctly placed, based on the X-ray crystal structure for PahZ1_KT-1 (Brambley et al. 2020), adjacent to one another in an antiparallel orientation for both PahZ1_KT-1 and PahZ1_Spyx, but not in any other homolog examined. As such, PahZ1_Bfg likely dimerizes uniquely relative to that predicted from the PahZ1_KT-1 crystal structure. Regardless of oligomeric architecture, a core requirement that must be addressed is whether these proteins are bona fide PahZ1 hydrolases that catalyze the biodegradation of tPAA.

Observation of PAA hydrolase activity

A required characteristic of any PahZ1 enzyme is the ability to catalyze biodegradation of tPAA. As such, GPC experiments were performed with protein as described previously (Brambley et al. 2020). Each candidate PahZ1 hydrolase was incubated with tPAA for 1, 4, and 24 hours, followed by heat quenching to stop the reaction, and GPC analysis. Figure 6D provides the results as determined by percent tPAA degradation relative to wild-type PahZ1_KT-1 levels. All four newly identified proteins, PahZ1_Ebac, PahZ1_Bfg, PahZ1_Spyx, and PahZ1_Spd, were observed to exhibit apparent PAA hydrolase activity, which confirms their classification as PahZ1 family members. Interestingly, inspection of Fig. 6D reveals that PahZ1_Ebac and PahZ1_Bfg each display PAA hydrolase activity as 75 ± 2 and 79 ± 4% of wild-type PahZ1_KT-1 activity after 24 hours. This decreased PAA hydrolase activity is likely the consequence of a surface charge for PahZ1_Ebac and PahZ1_Bfg that lacks the significant cationic charge predicted for the other PahZ1 enzymes. Figure 6B provided calculated pI values for PahZ1_Ebac and PahZ1_Bfg as 6.26 and 6.75, respectively. Therefore, under assay conditions with pH = 7.4 these two enzymes are predicted to have a slight negative overall charge, which may not support efficient binding of the polyanionic tPAA substrate.

Additional GPC experiments were performed to quantitatively compare the catalytic rates for each PahZ1 hydrolase. As discussed above, each PahZ1 homolog was separately incubated with tPAA for a defined time interval, heat inactivated, and subjected to GPC analysis as previously described (Brambley et al. 2020). Fig. 8 presents the resulting time courses as plots of relative tPAA degradation (as a percentage) versus time, with dimers and monomers separately grouped. All PahZ1 enzymes examined exhibit apparent catalytic rates of tPAA degradation that are approximately the same with time courses reaching a plateau within ~100 minutes. A notable outlier is PahZ1Ebac, which exhibits an apparent k_obs that is nearly 4-fold faster than PahZ1_KT-1. However, PahZ1_Ebac and PahZ1_Bfg each fail to catalyze 100% tPAA degradation relative to PahZ1_KT-1, exhibiting final amplitudes equal to ~25 and 75%, respectively. The combination of an apparently faster rate constant with substantially reduced total degradation for PahZ1_Ebac is consistent with enzyme deactivation occurring during the assay rather than intrinsically superior catalysis. When enzyme activity is lost over the assay time course, the observable reaction window is compressed, which can artificially inflate the fitted k_obs value. Given that PahZ1_Ebac has the lowest calculated pI (6.26) among the characterized homologs, it carries a net negative charge at the assay pH of 7.4, which may destabilize the enzyme and/or reduce favorable electrostatic interactions with the polyanionic tPAA substrate. Similar reasoning may explain the reduced total degradation observed for PahZ1_Bfg (pI = 6.75). As such, PahZ1_Ebac and PahZ1_Bfg present potential optimization opportunities for protein engineering to assess pH profiles, as these two may function effectively at lower pH, based on their lower pI (Fig. 6B).

Figure 8.

Time courses of polyaspartic acid (PAA) degradation by PahZ1 homologs. PAA hydrolysis was monitored by measuring oligoaspartic acid (OAA) product formation via GPC following heat inactivation at specified time points. The resultant time courses were normalized to wild-type PahZ1_KT-1 activity recorded at 250 minutes. The left panels (A-C) show results for dimeric PahZ1 variants (PahZ1_KT-1, PahZ1_Bfg, and PahZ1_Spyx), while the right panels (D-F) display data for monomeric forms (PahZ1_KP-2, PahZ1_Ebac, and PahZ1_Spd). Each time course was fit to a single exponential function using nonlinear least squares analysis to determine the observed rate constant (k_obs). Error bars represent standard deviation from multiple independent experiments. The solid lines show the best fit to the data, with the derived k_obs values and their uncertainties shown in each panel. The apparent k_obs for PahZ1_Ebac should be interpreted with caution, as the low amplitude (~25% relative to PahZ1KT-1) suggests possible enzyme deactivation during the assay.

While the current study establishes that all four newly identified proteins are bona fide PahZ1 hydrolases capable of tPAA degradation, detailed mutational analysis of their active site residues was beyond the scope of this work. Future studies will examine whether corresponding R215A mutations in these homologs retain partial activity, as observed for PahZ1_KP-2, or exhibit complete ablation similar to PahZ1_KT-1 R246A. Such experiments will clarify whether functional redundancy among substrate-binding arginines is a conserved feature across the PahZ1 family.

DNA binding activity is common amongst PAA hydrolases

PAA hydrolases from Sphingomonas sp. KT-1 and Pedobacter sp. KP-2 were recently reported to bind to both single- and double-strand DNA with nanomolar affinities (Couch et al. 2024). Indeed, PahZ1_KT-1 was shown to bind to ssDNA and dsDNA with apparent dissociation equilibrium constants, K_D,app = 81 ± 14 and 19 ± 1 nM, respectively. We note that DNA binding has not been observed, to date, in other polymer hydrolases. To determine whether DNA binding activity is a feature shared by PAA hydrolases, equilibrium experiments were performed (Couch et al. 2024), wherein a single-stranded DNA was titrated with increasing amounts of protein and allowed to incubate for 60 minutes to establish the binding equilibrium. The A₃₀ ssDNA used included 5′-FAM as a fluorescent label, which was shown to undergo emissions quenching on protein binding (Couch et al. 2024). Fig. 9A illustrates that 100 nM Flu-5′-(A₃₀)-3′ ssDNA was incubated with [PahZ1] ranging from 0–500 nM, for each PahZ1 homolog. FAM was excited at 494 nm and emissions read at 520 nm. As noted previously (Couch et al. 2024), raw FAM emissions were observed to decrease with increasing protein concentration. Quenched emissions may be due to, in part, a statistically significant decrease in fluorescence lifetime for FAM that is observed on binding of protein. Figure S9 highlights this transition, where mean lifetimes equal to 3.83 ± 0.02 and 3.36 ± 0.13 ns are estimated for conditions that include only FAM-A₃₀ or FAM-A₃₀ and PahZ1_KT-1, respectively. These estimates are in agreement with similar time-correlated single photon counting (TCSPC) measurements that estimate the lifetime of fluorescein in the range of 3.775–4.11 ns depending on pH in water (Hammer et al. 2005). However, additional work is needed to understand the mechanism leading to quenched emissions.

Figure 9.

Fluorescence-based DNA binding analysis of select PahZ1 homologs. PahZ1 variants (0–500 nM) were titrated against 100 nM fluorescently-labeled single-stranded DNA (Flu-5′-(A₃₀)-3′). (A-C) Binding isotherms for PahZ1_KT-1, PahZ1_Spd, and PahZ1_Spyx, respectively, where normalized fluorescence change ((F_free-F_i)/(ΔF)) is plotted against protein concentration. Data points represent experimental measurements with error bars showing standard deviation resulting from a minimum of three independent replicates. Solid lines represent fits to appropriate binding models, yielding apparent dissociation constants (K_D,app) as indicated in each panel. PahZ1_Spyx (Panel C) exhibits cooperative binding behavior, while PahZ1_KT-1 and PahZ1_Spd (Panels A and B) show non-cooperative binding characteristics.

Fluorescence intensities were used to construct curves of fractional binding versus PahZ1 protein concentration (Figs. 9B and C). PahZ1 proteins from Sphingomonas sp. KT-1 (Fig. 9A, included as a positive control), Sphingomonadales bacterium (Fig. 9B), and Sphingopyxsis sp. Root 1497 (Fig. 9C) were each observed to bind DNA with nanomolar affinity as determined from nonlinear least squares analysis (NLLS) of the resultant binding curves. From NLLS analysis with either non-cooperative (for PahZ1_KT-1 and PahZ1_Spd) or cooperative (PahZ1_Spyx) binding models, the Flu-5′-(A₃₀)-3′ ssDNA was estimated to bind to PahZ1_KT-1, PahZ1_KP-2, PahZ1_Spd, and PahZ1_Spyx with K_D,app = 105 ± 9, 144 ± 4 (Couch et al. 2024), 104.5 ± 0.1, and 122 ± 5 nM, respectively. Reference to Fig. 6B also reinforces that these PahZ1 homologs each have isoelectric points greater than or equal to ~8, which predicts that each will be positively charged under binding conditions with pH = 7.5. Binding experiments were also performed with PahZ1_Ebac and PahZ1_Bfg under standard assay conditions; however, the resulting data revealed limiting binding interactions (results not shown). Future work will address whether the limited DNA binding to PahZ1_Ebac and PahZ1_Bfg is the consequence of these proteins’ lower pI relative to other characterized PahZ1 homologs.

The observation that PahZ1 enzymes bind single- and double-stranded DNA with nanomolar affinity raises questions regarding the physiological and practical relevance of this interaction. From a mechanistic perspective, the shared ability of PahZ1 homologs to bind DNA provides additional evidence for conservation of the positively charged substrate-binding surface that mediates PAA recognition. Whether DNA binding has physiological significance in the native bacterial hosts remains unclear. In the periplasmic environment where PahZ1 enzymes are localized (Hiraishi & Maeda 2011; Hiraishi 2016), DNA concentrations are typically low, suggesting that competition between DNA and PAA for enzyme binding sites would be minimal under normal growth conditions. However, during cell lysis or in environmental contexts where extracellular DNA accumulates, competitive inhibition of PAA degradation by DNA could potentially occur. From a practical standpoint, the DNA-binding property provides a useful biochemical signature for identifying and characterizing PahZ1 homologs, as this activity appears unique among characterized polymer hydrolases.

Conclusion

In this study, we have expanded the diversity of PAA hydrolases, offering new insights into their structural and functional characteristics that could be leveraged to improve tPAA degradation in commercial applications. By determining the X-ray crystal structure of PahZ1_KP-2, we found that it is monomeric, in contrast to the dimeric PahZ1_KT-1 from Sphingomonas sp. KT-1, though both enzymes share significant structural homology and a conserved serine protease catalytic triad. Furthermore, bioinformatics and AI-based structural modeling led to the identification of four novel PahZ1 homologs, all of which retained the characteristic α/β hydrolase domain architecture and exhibited measurable PAA-degrading activity. These homologs exist as either dimers or monomers, as determined by size exclusion chromatography and Dynamic Light Scattering, with oligomeric state predictable from conservation of glutamate and histidine residues at the dimerization interface. Collectively, this work addresses the core question of whether PAA hydrolase activity arises from a common PahZ1 structural scaffold: our findings support that a conserved α/β hydrolase architecture with a characteristic serine protease catalytic triad represents the defining feature of PahZ1 enzymes, while oligomeric state and surface charge properties vary across homologs.

Functional relevance of oligomeric diversity

The observation that PahZ1 homologs naturally exist as either monomers or dimers raises the question of whether oligomeric state influences catalytic function. Previous work demonstrated that an engineered PahZ1_KT-1 monomer (E251K) did not exhibit dramatically different catalytic behavior compared to the wild-type dimer, suggesting that dimerization is not required for efficient tPAA degradation (Lamantia et al. 2022). Previous work has demonstrated that the dimerization interface can be rationally manipulated; replacement of the PahZ1_KT-1 dimerization helix sequence with the corresponding sequence from monomeric PahZ1_KP-2 was sufficient to convert the dimer into a functional monomer with catalytic properties comparable to the E251K point mutant (Lamantia et al. 2022). This establishes proof-of-concept for engineering oligomeric state through interface modification, though systematic application of this approach across all newly identified homologs is beyond the scope of the current study.

The current study reinforces this conclusion across a broader set of homologs; kinetic analysis reveals that both monomeric and dimeric variants exhibit similar observed rate constants, with most enzymes reaching plateau conversion within approximately 100 minutes (Fig. 8). Nevertheless, oligomeric state may have functional relevance beyond intrinsic catalytic efficiency. The dimer interface in PahZ1_KT-1 buries approximately 1200 Å² of solvent-accessible surface area per monomer (Brambley et al. 2020), which may contribute to protein stability under challenging conditions (elevated temperature, pH extremes) not systematically explored here. Indeed, characterization of protein unfolding revealed that an the activation energy of unfolding for the dimer is 46 kJ mol⁻¹ higher than for PahZ1_KT-1 monomers (Lamantia et al. 2022). Additionally, dimeric enzymes present two substrate-binding surfaces that could enhance binding to long polymer chains through avidity effects at substrate-limiting concentrations. From an engineering perspective, the identification of both functional monomeric and dimeric scaffolds expands the toolkit for future development. Monomeric variants may be advantageous for protein fusion strategies, while dimeric variants could offer enhanced stability for industrial applications.

Considerations for industrial application

Translation of PahZ1 enzymes from laboratory characterization to industrial application requires consideration of several key performance parameters. Industrial polymer degradation processes typically demand enzymes with high catalytic activity, operational stability at elevated temperatures, tolerance to the ionic strength conditions often present in industrial waste streams, and high expression levels (Zhu et al. 2022; Tournier et al. 2020). Our kinetic analyses reveal that PahZ1 enzymes achieve approximately 50% substrate conversion in 60–100 minutes under standard assay conditions. While these rates demonstrate catalytic competence, industrial applications typically demand faster conversion and lower enzyme loadings. For comparison, engineered PET hydrolases such as TurboPETase can achieve nearly complete polymer depolymerization in 8 hours at industrially relevant substrate concentrations, and FAST-PETase degrades postconsumer PET within 24–168 hours depending on crystallinity and conditions (Cui et al. 2024; Lu et al. 2022). Achieving 10–100 fold rate enhancement through directed evolution or rational design would significantly improve industrial viability, and the structural diversity revealed here provides multiple starting scaffolds for such engineering efforts.

Industrial PAA applications impose diverse stability requirements. Agricultural soil amendments must function across varied pH and temperature ranges; wastewater treatment exposes enzymes to high ionic strength and surfactants; polymer recycling systems could benefit from elevated temperatures (50–70°C) to accelerate kinetics through Arrhenius effects while extending enzyme operational lifetime and facilitating enzyme recovery and recycling (Wei et al. 2019). The characterized PahZ1 enzymes show thermal aggregation above ~55°C (Fig. S4), indicating that thermostability improvements represent an important future direction.

Recent advances in computational protein design offer promising avenues for PahZ1 engineering. Tools such as ProteinMPNN, a deep neural network for protein sequence design, enable rapid in silico optimization of protein sequences for enhanced stability, expression, and function (Sumida et al. 2024). This approach has demonstrated the ability to generate enzyme variants with improved thermostability (elevated melting temperatures of 6–10°C) and enhanced catalytic efficiency while requiring far fewer experimental variants to be tested compared to traditional directed evolution (Sumida et al. 2024; Dauparas et al. 2022). Application of such methods to the PahZ1 scaffold, guided by the structural and functional insights reported here, could accelerate development of industrially relevant PAA hydrolases. Additionally, the identification of both monomeric and dimeric PahZ1 variants in this study provides distinct starting points for engineering efforts, as these oligomeric forms may respond differently to stabilizing mutations or display different kinetic behaviors under process conditions. The natural diversity in isoelectric points observed among homologs (pI 6.26–9.72) suggests variation in pH optima that could be systematically characterized and leveraged for specific applications. Such advances would support a true cradle-to-cradle approach where aspartic acid monomers cycle between polymer synthesis and enzymatic degradation, establishing closed-loop systems for sustainable polymer management.

However, it will be important to continue to expand the PahZ1 family through the implementation of structural bioinformatics, particularly the Foldseek algorithm, that enable rapid searching of protein structure databases including the Protein Data Bank and the AlphaFold Protein Structure Database. Unlike sequence-based methods, structure-based searches can identify distant homologs that share conserved folds despite limited primary sequence identity; a relevant consideration given that PahZ1_KT-1 and PahZ1_KP-2 share only 35% sequence identity while maintaining strong structural homology. By continuing to expand and engineer the PahZ1 family, we move closer to realizing circular management of poly(aspartic acid)-based materials with minimal environmental impact.

Edited by: Dr. Eva Petersen, (Associate Editor, Protein Engineering, Design & Selection)

Author contributions

Justin D Marsee (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Methodology [equal], Writing - original draft [equal], Writing - review & editing [equal]), Chad A Brambley (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal]), Ho Thi (Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal]), Waylan W Callaway (Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal]), Amanda L Jansch (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal]), Kaitlyn Taylor (Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal]), Shelby R Watson-Sanders (Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal]), Britney Williams (Formal analysis [equal], Investigation [equal], Methodology [equal]), Austin Wolvington (Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal]), Catherine T Nguyen (Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal]), Laiba Khan (Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal]), Claros Cabrera (Investigation [equal], Methodology [equal]), Jamie R Wallen (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal], Resources [equal], Validation [equal], Visualization [equal], Writing - review & editing [equal]), Mitch H Weiland (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Funding acquisition [equal], Investigation [equal], Methodology [equal], Project administration [equal], Resources [equal], Software [equal], Supervision [equal], Validation [equal], Visualization [equal], Writing - original draft [equal], Writing - review & editing [equal]), Justin Mark Miller (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Funding acquisition [equal], Investigation [equal], Methodology [equal], Project administration [equal], Resources [equal], Software [equal], Supervision [equal], Validation [equal], Visualization [equal], Writing - original draft [equal], Writing - review & editing [equal]).

Conflict of interest

None declared.

Funding

None declared.

Data availability

All data supporting the findings of this study, including crystallographic coordinates, structure factors, protein sequences, kinetic measurements, and molecular dynamics simulation trajectories, are available from the corresponding authors upon reasonable request. The crystal structure of PahZ1_KP-2 has been deposited in the Protein Data Bank under accession code 9O7U.

Accession number

The coordinates for the Pedobacter sp. KP-2 PahZ1 structure has been deposited in the Protein Data Bank with PDB ID: 9O7U.