Discovery of cyanophycin dipeptide hydrolase enzymes suggests widespread utility of the natural biopolymer cyanophycin

Significance

Cyanophycin is best known for its roles in cyanobacteria, which have developed dedicated pathways around the biopolymer. We have shown that a novel, common, dedicated cyanophycin dipeptide hydrolase family is found in bacteria which possess other cyanophycin metabolizing genes, as well as in those that lack them. The specificity of cyanophycin dipeptide hydrolases is derived from conserved interactions of the active site with both residues of its β-Asp-Arg dipeptide substrate. This specificity is maintained in cyanophycin dipeptide hydrolase enzymes from both cyanophycin producers and scavengers with medical relevance. The prevalence of cyanophycin metabolism genes and the discovery of this dedicated enzyme family highlight the utility of this interesting natural biopolymer.

Abstract

Cyanophycin is a bacterial polymer mainly used for nitrogen storage. It is composed of a peptide backbone of L-aspartate residues with L-arginines attached to their side chains through isopeptide bonds. Cyanophycin is degraded in two steps: Cyanophycinase cleaves the polymer into β-Asp-Arg dipeptides, which are hydrolyzed into free Asp and Arg by enzymes possessing isoaspartyl dipeptide hydrolase activity. Two unrelated enzymes with this activity, isoaspartyl dipeptidase (IadA) and isoaspartyl aminopeptidase (IaaA) have been shown to degrade β-Asp-Arg dipeptides, but bacteria which encode cyanophycin-metabolizing genes can lack iaaA and iadA genes. In this study, we investigate a previously uncharacterized enzyme whose gene can cluster with cyanophycin-metabolizing genes. This enzyme, which we name cyanophycin dipeptide hydrolase (CphZ), is specific for dipeptides derived from cyanophycin degradation. Accordingly, a co-complex structure of CphZ and β-Asp-Arg shows that CphZ, unlike IadA or IaaA, recognizes all portions of its β-Asp-Arg substrate. Bioinformatic analyses showed that CphZ is found in very many proteobacteria and is homologous to an uncharacterized protein encoded in the “arginine/ornithine transport” (aot) operon of many pseudomonas species, including Pseudomonas aeruginosa. In vitro assays show that AotO is indeed a CphZ, and in cellulo growth experiments show that this enzyme and the aot operon allow P. aeruginosa to take up and use β-Asp-Arg as a sole carbon and nitrogen source. Together the results establish the novel, highly specific enzyme subfamily of CphZs, suggesting that cyanophycin is potentially used by a much wider range of bacteria than previously appreciated.

Cyanophycin is a biopolymer produced by a wide range of bacteria found in environments such as marine habitats, sediment, and gut microbiota (1–9). It was first discovered around 140 y ago as large, dark granules in cyanobacterial cells, visible under a simple light microscope (10). Cyanophycin chains are composed of a backbone of L-aspartate residues with L-arginines attached to each aspartate side chain through isopeptide bonds [i.e., (β-Asp-Arg)_n] (11) (Fig. 1). Although most often described as a nitrogen storage polymer (12), cyanophycin can also be useful as a store of carbon and energy (13–15). Cyanophycin as a dynamic nitrogen reservoir is beneficial for cells in a variety of conditions (16–18). For example, nitrogen-fixing cyanobacteria can use it to separate (aerobic) photosynthesis from the strictly anaerobic process of nitrogen fixation. By producing cyanophycin, cells can stockpile excess fixed nitrogen during anaerobic periods of a day/night cycle (19) or in anaerobic cell types (3, 20). This fixed nitrogen can then be mobilized and used in aerobic conditions, because cyanophycin catabolism is insensitive to O₂.

Fig. 1.

The two steps of cyanophycin biodegradation.

Cyanophycin is commonly synthesized by cyanophycin synthetase 1 (CphA1) (21). CphA1 has two synthetic active sites which alternately add Asp and Arg to the nascent polymer in adenosine triphosphate (ATP)-dependent manners (1, 21). Many CphA1s also harbor a hydrolytic active site that generates cyanophycin primers to facilitate synthesis (22). Some lysine can be incorporated into cyanophycin in place of Arg (23), the amount of which depends on the CphA1 enzyme, the native or heterologous host and the growth conditions (24), but it is typically much lower than Arg. An enzyme related to CphA1 (CphA2), found only in cyanobacteria possessing CphA1, can repolymerize β-Asp-Arg dipeptides into cyanophycin using its single active site (25, 26).

Bacteria must degrade cyanophycin polymer into its constituent amino acids to access the stored carbon, nitrogen, and energy (3, 19). This degradation happens in two steps (Fig. 1). First, the polymer is hydrolyzed into β-Asp-Arg dipeptides by cyanophycinase (27, 28). Then, hydrolysis of the isopeptide bond splits β-Asp-Arg into Asp and Arg, which can enter primary metabolism (29). Unrelated enzymes capable of catalyzing this second reaction can be grouped under the label “isoaspartyl dipeptidee hydrolase” (Fig. 1).

Isoaspartyl dipeptide hydrolases are common because they can also participate in a fairly widespread protein damage pathway: Proteins become spontaneously damaged by intramolecular rearrangement of Asp or Asn residues, in which the peptide backbone from the Asp or Asn main chain is transferred to the side chain. This lesion can be repaired by L-isoaspartyl O-methyltransferases (30), or the damaged protein can be degraded by proteases that hydrolyze the protein backbone. However, proteases cannot digest isoaspartyl dipeptides (31, 32), so isoaspartyl dipeptide hydrolases (33) must lyse them to prevent accumulation to toxic levels (34). Any amino acid can become attached to the Asp side chain during protein damage, and accordingly isoaspartyl dipeptide hydrolases accept a wide range of isoaspartyl dipeptides as substrates (29, 35). Cyanophycin dipeptide degradation is thought to be performed by isoaspartyl dipeptide hydrolases moonlighting from their role in the damaged protein pathway.

Two unrelated types of bacterial isoaspartyl dipeptide hydrolase enzymes are known: “isoaspartyl dipeptidase” (IadA) (33, 36) and “isoaspartyl aminopeptidase” (IaaA) (29, 37–39). Both can degrade β-aspartyl dipeptides derived from damaged protein, as well as β-Asp-Arg and β-Asp-Lys derived from cyanophycin (40). However, over a quarter of cyanophycin-producing bacteria do not possess genes that encode IadA or IaaA (2, 40). We recently observed that other putative hydrolases can cluster with cyanophycin metabolizing genes in some bacteria that lack IadA and IaaA (40).

In this study, we perform structural, biochemical, and bioinformatic studies to interrogate the activity and role of a previously uncharacterized enzyme from Acinetobacter baylyi (AbCphZ). We show that, unlike the unrelated, characterized isoaspartyl dipeptide hydrolases IaaA and IadA, AbCphZ is specific for β-Asp-Arg/Lys. Furthermore, we characterize Pseudomonas aeruginosa AotO, a previously uncharacterized enzyme commonly found in non-cyanophycin-producing proteobacteria, to likewise be a cyanophycin dipeptide hydrolase (CphZ). We find that genes in the arginine/ornithine transport (aot) operon, including aotO, allow P. aeruginosa to use β-Asp-Arg as a nitrogen and carbon source. Together with previous results showing existence of cyanophycin in disparate environments (8, 9), the results imply that cyanophycin is a common material and that bacteria may specifically scavenge its β-Asp-Arg degradation product.

Results

AbCphZ Is a β-Asp-Arg/Lys Dipeptidase.

We recently performed a bioinformatic analyses of co-occurrence and clustering of cyanophycin-metabolizing genes (40) and noted that 27% of genomes which encode cyanophycin synthetase lack both iadA and iaaA. Some cyanophycin synthetase-cyanophycinase clusters instead include a gene annotated as “M14 family metallopeptidase” or “succinylglutamate desuccinylase/aspartoacylase family protein.” Notably, they include A. baylyi DSM587, one of the five bacteria noted by Füser and Steinbüchel (2) to encode CphA1 and cyanophycinase but not IadA nor IaaA. We hypothesized that A. baylyi DSM587 M14 family metallopeptidase (WP_004925890.1), which shares ~22% identity with Escherichia coli succinylglutamate desuccinylase (AstE) (41), may perform hydrolysis of β-Asp-Arg/Lys derived from cyanophycin, and named it A. baylyi cyanophycin dipeptide hydrolase CphZ (AbCphZ).

We heterologously expressed AbCphZ in E. coli for activity and structural studies. AbCphZ could be purified to homogeneity and migrates as a dimer in size exclusion chromatography (SI Appendix, Fig. S1A). We tested the enzyme’s ability to hydrolyze various β-aspartyl dipeptides. AbCphZ displays a clear preference toward β-Asp-Arg/Lys dipeptides, hydrolyzing these robustly, but not other β-aspartyl dipeptides (Fig. 2A). AbCphZ also displays specificity toward the β-linkage, as α-Asp-Arg was not efficiently hydrolyzed (Fig. 2A). It is also specific towards the Asp residue of the substrate dipeptide, as it has very low activity on N(2)-acetyl-Arg and no detectable activity on N(2)-succinyl-Arg, an intermediate in the arginine succinyltransferase (AST) arginine catabolism pathway, which is very similar in structure to β-Asp-Arg (Fig. 2B) (41). Likewise, it has nearly no activity on the most similar glutamate containing isodipeptide, γ-Glu-Arg. The overall substrate specificity is remarkable, as we are unaware of any other enzyme that has selectivity for β-Asp-Arg/Lys dipeptides. Furthermore, cyanophycin degradation is the only abundant source of β-Asp-Arg/Lys dipeptides we know.

Fig. 2.

Structural and biochemical characterization of AbCphZ. (A) Asp release assay of AbCphZ with various Asp-dipeptides. AbCphZ displays high specificity towards β-Asp-Arg/Lys dipeptides. Error bars represent the SD of the mean of n = 4 replicates. (B) Arg release assay of AbCphZ with Arg-containing substrates. (i) AbCphZ preferentially cleaves β-Asp-Arg, over the similar compounds drawn in panel (ii). Error bars represent the SD of the mean of n = 3 replicates. (C) Overview of the crystal structure of AbCphZ, which is a dimer. (D) Asp release assay of AbCphZ mutants with β-Asp-Arg. The highly conserved N112, R113, D222, E223, and E251 are essential for activity, while mutation of S216, which is not as highly conserved (Fig. 3F), decreases activity to approximately half. Error bars represent the SD of the mean of n = 3 replicates. (E) Polder electron density map showing unbiased density for the substrate β-Asp-Arg bound to AbCphZ E251A. The map is displayed at contour at 4.5 σ and carved to within 4 Å of β-Asp-Arg. (F) Detailed view of the active site of AbCphZ E251A with bound β-Asp-Arg. Note that the E251A mutant retains zinc, but is unable to position an activated water for nucleophilic attack, as also observed in other zinc proteases (42).

To determine the structural basis for this substrate specificity, we determined the crystal structure of AbCphZ at 2.7 Å resolution (Fig. 2C and SI Appendix, Table S1). AbCphZ is a homodimer of protomers which contains a large domain (a 9-strand β sheet surrounded by α helices) and a small domain (made of β strands), the same fold possessed by E. coli AstE (PDB code 1YW6, RMSD 3.8 Å across 109 C_α pairs, SI Appendix, Fig. S1B) and bovine pancreatic carboxypeptidase A (43, 44) (PDB code 1HEE, RMSD 4.1 Å across 159 C_α pairs, SI Appendix, Fig. S1C). As with AstE and carboxypeptidase A, the active site of AbCphZ contains a single metal ion liganded by a conserved H-H-E triad (AbCphZ H50, E53, and H179, Fig. 2D and SI Appendix, Fig. S1 C and D). The density map indicates the presence of another metal ion near Y62 of one protomer and H315 and S330 of another protomer within the CphZ dimer. Inductively coupled plasma mass spectrometry (ICP-MS) showed our CphZ samples contain both zinc and manganese (SI Appendix, Table S2). We verified that the ion in the active site is Zn²⁺ using single-wavelength anomalous diffraction (SI Appendix, Fig. S1E) and modeled Mn²⁺ in the second metal site (SI Appendix, Fig. S1F). The conserved E251 is positioned close to the Zn²⁺ ion and the putative substrate binding site (SI Appendix, Fig. S1D) and is essential for activity (Fig. 2D), like analogous glutamates in other zinc metallopeptidases (42).

To understand how CphZ specifically binds and cleaves cyanophycin-derived dipeptides, we determined the co-complex structure of the catalytically inactive mutant CphZ_E251A bound to β-Asp-Arg at 2.2 Å resolution (Fig. 2 E and F and SI Appendix, Table S1). The maps show no substantial rearrangement of the protein from the unliganded conformation, but clear density for β-Asp-Arg in the active site (Fig. 2E). The carbonyl of the dipeptide’s scissile bond is positioned 2.0 Å away from the Zn²⁺ ion (Fig. 2F), in good position for catalysis (45). The Zn²⁺ will serve as a Lewis acid and draw electrons from the bond and facilitate its cleavage, presumably by a water molecule activated by E251 (42). The structure also explains how CphZ specifically recognizes and binds the Arg/Lys portion of the substrate: Conserved residues D222 and E223 form a negatively charged pocket suitable for the binding of a positively charged guanidino or amino group (Fig. 2F). In addition, the conserved R102, N112 and R113 bind the Arg carboxyl of the dipeptide in a similar manner to that observed in carboxypeptidase A (SI Appendix, Fig. S1C), orienting the scissile bond for catalysis. Mutation of any of N112, R113, D222 and E223 abolishes enzyme activity, while mutation of S216, which is not as highly conserved, decreases activity to approximately half (Fig. 2D). The Asp residue of the dipeptide interacts with a pocket that shows lower sequence conservation (SI Appendix, Fig. S1G), but D181 and K366 are well conserved and within hydrogen bonding distance from the Asp (Fig. 2F). Isothermal titration calorimetry (ITC) binding assays with β-Asp-Arg, Asn, and Arg shows CphZ_E251A has good affinity for the substrate β-Asp-Arg (K_D = 7.6 ± 0.1 µM), and much lower affinity for Arg (K_D = 470.8 ± 16.8 µM) and Asn (no binding detected) (SI Appendix, Fig. S1H).

P. aeruginosa AotO Is also a CphZ.

The Conserved Domain Database (46) identifies residues 18 to 280 of AbCphZ as a “M14_PaAOTO_like” (accession cd06250) domain. This domain is also present in the uncharacterized protein AotO (47) of P. aeruginosa [WP_128550578.1; PA0891 (48); here PaAotO]. PaAotO and AbCphZ share 38% sequence identity, with particularly high similarity around the active site (SI Appendix, Fig. S2A). Pseudomonas species do not possess genes for cyanophycin synthesis, but the homology between CphZ and AotO led us to ask whether these enzymes possess similar activity.

AotO is encoded in the aot operon, which also encodes a multicomponent ABC transporter (AotJQMP) and an Arg-dependent transcription activator (ArgR) (47). AotJQMP is important in P. aeruginosa for Arg uptake and its use as a carbon source (47, 49). P. aeruginosa aotO does not affect Arg uptake, and its role is unknown (47). To determine whether CphZ and AotO have the same activity, we expressed and purified PaAotO and tested its activity towards β-Asp-Arg, N(2)-succinyl arginine and N(2)-acetyl arginine. Like AbCphZ, AotO is specific for β-Asp-Arg and hydrolyzes it at a rate comparable to AbCphZ and the nonspecific isoaspartyl dipeptide hydrolases RhIaaA and LmIadA (40) (Fig. 3 A and B).

Fig. 3.

CphZ phylogeny and role in P. aeruginosa. (A) Arg release assay with Arg-containing substrates. PaAotO preferentially cleaves β-Asp-Arg. Error bars represent the SD of the mean of n = 3 replicates. (B) Arg release assay of AbCphZ, PaAotO, and two IadA enzymes described in a previous study (40) with the substrate β-Asp-Arg. Error bars represent the SD of the mean of n = 3 replicates. (C) Growth of P. aeruginosa wild-type PAO1 in minimal media supplemented with 20 mM D-glucose and 10 mM NH₄Cl (●), 20 mM D-glucose and 10 mM β-Asp-Arg (■), 10 mM β-Asp-Arg (▲), or 20 mM D-glucose (▼). (D) Viable cell count (log CFU/mL) of PAO1 WT grown in minimal media supplemented with 20 mM D-glucose and 10 mM NH₄Cl (●), 20 mM D-glucose and 10 mM β-Asp-Arg (■), 10 mM β-Asp-Arg (▲), or 20 mM D-glucose (▼). Cell counts in media containing β-Asp-Arg increased from 1.9 × 10⁷ (SD = 0.37) CFU/mL at t = 0 h to 5.7 × 10⁸ (SD = 0.21) CFU/mL at t = 48 h. Data points are from a representative experiment; mean values (n = 3) ± SD. The experiment was repeated twice. (E) Growth of P. aeruginosa PAO1 in 200 µL minimal media in 96-well microplate supplemented with 20 mM D-glucose and 100 mM β-Asp-Arg (●), 20 mM D-glucose and 10 mM β-Asp-Arg (■), 20 mM D-glucose and 1 mM β-Asp-Arg (▲) or 20 mM D-glucose and 0.1 mM β-Asp-Arg (▼). Growth with 20 mM D-glucose and 10 mM NH₄Cl (♦) is also shown. Growth is saturated at 10 mM β-Asp-Arg, likely due to limitation of available glucose. Data points represent mean values (n = 6) ± SEM from two independent experiments. (F) Growth of P. aeruginosa PAO1 WT, aotM::Tn, and aotO:Tn mutants in 200 µL minimal media in 96-well microplate supplemented with 10 mM NH₄Cl (▲), 20 mM D-glucose (■), or 10 mM β-Asp-Arg (●). Data points represent mean values (n = 6) ± SEM from two independent experiments. (G) Phylogenetic tree of CphZ/AotO distribution of CphZ proteins encoded by completely sequenced bacterial genomes (50). AbCphZ and PaAotO are colored in red. The blue and black colored entries are deltaproteobacterial and mycobacterial respectively, presumably resulting from horizontal gene transfer. (H) A Weblogo (51) of the AbCphZ region 45 to 227 shows high conservation of active site and substrate binding residues in the CphZs. E251 (not shown) is also very highly conserved, while K366 can be K, H, or R.

AotO Allows P. aeruginosa to Grow on β-Asp-Arg as Sole Carbon and Nitrogen Source.

To demonstrate that P. aeruginosa can catabolize β-Asp-Arg in cellulo, we cultured the wild type PAO1 strain (52) in minimal media supplemented with β-Asp-Arg. β-Asp-Arg alone is sufficient to support bacterial growth, although with a slower doubling time (36.6 h) than with D-glucose and NH₄Cl (7.0 h) (Fig. 3C). Bacterial replication in media containing β-Asp-Arg was confirmed by a marked increase in viable cell counts from 1.9 × 10⁷ CFU/mL to 5.7 × 10⁸ CFU/mL in 48 h of flask growth (Fig. 3D). Thus, β-Asp-Arg can be catabolized by P. aeruginosa and sustains bacterial growth as the sole carbon and nitrogen source in flasks, as well as in microplates (SI Appendix, Fig. S2 B and C). The related dipeptide β-Asp-Ala does not support growth alone or with D-glucose, consistent with the substrate specificity of CphZs (SI Appendix, Fig. S2D).

β-Asp-Arg supports bacterial growth well in combination with D-glucose, with a clear β-Asp-Arg concentration dependence (Fig. 3E). Doubling times of 17.4 h, 7.3 h, and 5.2 h were observed with 0.1 mM, 1 mM, and 10 mM β-Asp-Arg, respectively. Notably, growth in 20 mM D-glucose and 10 mM β-Asp-Arg is comparable to that in D-glucose and 10 mM NH₄Cl, indicating that β-Asp-Arg serves as a very efficient nitrogen source (Fig. 3 C and E).

To confirm that the CphZ AotO is responsible for β-Asp-Arg catabolism in P. aeruginosa, we assayed growth of the aotO mutant from a sequence-verified transposon mutant library (53). The aotO mutant was unable to grow on β-Asp-Arg as sole carbon and nitrogen source (Fig. 3F). aotM and aotP mutants, which lack functional AotJQMP transporter, were likewise unable to grow on β-Asp-Arg (Fig. 3F and SI Appendix, Fig. S2E). aot transposon mutants grew identically to their wildtype parental strain in media containing NH₄Cl and D-glucose (SI Appendix, Fig. S2 F–L), confirming that AotO and the AotJQMP transporter are specifically required for β-Asp-Arg catabolism.

The Genetic Context of cphZ.

CphZ is a common enzyme: There are nearly 10,000 entries in the Max Planck Institute Bioinformatics Toolkit (54) nonredundant (nr) database with >70% coverage and >30% identity to AbCphZ, and over 2,000 in the nr 90% (nr90) database. To better understand the role of CphZ, we performed an analysis of the distribution and genomic localization of cphZ relative to other cyanophycin metabolism genes. We searched the RefSeq database (50) for complete bacterial genomes that include cphZ and constructed a phylogenetic tree of these using phyloT (55) according to the genome taxonomy database classification (56) (Fig. 3G). This shows CphZ to be principally a proteobacterial enzyme, which is particularly common in the family Pseudomonadaceae and in the order Burkholderiales. A WebLogo (51) shows high conservation of active site and substrate binding residues (Fig. 3H) in these CphZ sequences. While many Burkholderiales encode CphA1 (40), a BLAST (57) search shows that cyanophycin producers are extremely rare in Pseudomonadaceae, with only one sequenced strain possessing a cphA1 gene.

We then constructed a database of all the complete bacterial genomes in the RefSeq database and used Cblaster (58) to analyze the co-occurrence and clustering of cphA1, cyanophycinase (cphB/E/I) and cyanophycin dipeptide hydrolase (cphZ/aotO). Of the 27,349 genomes in that database, 3,095 (~11%) genomes have at least one of cphA1, cyanophycinase or cphZ (Table 1). cphA1 is present in 1,614 (6%), a cyanophycinase gene is present in 840 (~3%) and cphZ is present in 1,364 (~5%). However, only 41 genomes encode all three of cyanophycin synthetase, cyanophycinase and CphZ, and all three genes cluster to within five kilobases in only 12 of these 41 genomes. Co-occurrence and clustering analyses of cyanophycinase and CphZ show that 50 genomes encode cyanophycinase and CphZ (but not CphA1) and that clustering is more common, with 32 of 50 genomes (Table 1) possessing a two-gene cyanophycin-degradation cluster.

Table 1.

Co-occurrence and clustering of cphA1, cyanophycinase genes, and cphZ

	Total	cphA1	cyano- phycinase	cphZ	cphA1 + cphZ + cyanophycinase	cphA1 + cyanophycinase	cphZ + cyanophycinase	cphA1 + cphZ
In genome	3,095	1,614	840	1,364	41	659	50	56
Clustered	563	543	562	33	12	542	32	13

Discussion

The experimental and bioinformatic results demonstrate that CphZ is a novel enzyme subfamily that is common and highly specific for β-Asp-Arg/Lys dipeptides. These CphZs are the first known enzymes dedicated to degradation of cyanophycin dipeptides, unlike IaaAs and IadAs, which degrade cyanophycin metabolites as a secondary function. PaAotO can equally be named PaCphZ since it also a founding member of the CphZ subfamily. Two functionally uncharacterized proteins with structures deposited in the PDB may also be CphZs: Shewanella amazonensis “succinylglutamate desuccinylase aspartoacylase” (accession code 3FMC) shares 27% identity with AbCphZ, and the structures have a rmsd of 1.8 Å across 220 C_α pairs. Shewanella frigidimarina “putative succinylglutamate desuccinylase/aspartoacylase” (3LWU) shares 29% identity and has a 1.7 Å rmsd. These two gammaproteobacterial species also encode cyanophycinases. Both CphZ homologs share lower identity with AbCphZ or PaAotO than the cutoff we use in our bioinformatic analysis, but their identities to AbCphZ are not very different from that of PaAotO (38%) which we confirmed is a CphZ. Inspection of the structures 3FMC and 3LWU reveals conservation of the residues that form the Arg and Asp binding pockets, suggesting these enzymes are likely CphZs and that there may be even more CphZs in nature than our analyses indicate.

CphZs are common in proteobacteria, which represent over half the bacterial sequences in genomic databases. Like IaaA and IadA, CphZ is mostly found in strains that cannot produce cyanophycin [Table 1; (40)]. A study identifying core genes among Pseudomonas groups found aotO to be conserved among common species, and that the whole aot operon also shows good conservation (59). Cyanophycin-degrading Proteobacteria have been isolated from a variety of environments and microbiomes (60–64), cyanophycin itself has been found in disparate environments (8, 9), and many bacteria encode a secreted version of cyanophycinase, CphE [Table 1, (60)]. The observation that cphZ is more commonly found in genomes that lack a recognizable cyanophycinase gene (1,314 genomes with cphZ but no cyanophycinase gene; 50 with cphZ and a cyanophycinase gene) suggests these bacteria (including Pseudomonas) live in communities with other species that degrade polymer into dipeptide (63). There is no known cyanophycin export process, so presumably cyanophycin in the environment comes from cell lysis of producer bacteria. Extracellular cyanophycinase degrades this environmental cyanophycin to β-Asp-Arg, which is imported and used by CphZ (or IaaA / IadA) encoding species.

The frequency of cyanophycin metabolizing genes (Table 1) indicates that it is clearly advantageous and common for bacteria to have mechanisms capable of scavenging cyanophycin and its breakdown product β-Asp-Arg. Indeed, the analogous bioinformatic analyses of glycogen metabolizing genes shows cyanophycin metabolizing genes are present at a ~1:3 ratio to those of glycogen, which is considered a ubiquitous biopolymer (SI Appendix, Table S3).

Herein, we characterized growth on β-Asp-Arg of one species that possesses a CphZ but not a recognizable cyanophycinase: β-Asp-Arg can be used by the human pathogen P. aeruginosa as the sole carbon source and as an excellent sole nitrogen source, nearly equivalent to NH₄Cl. Growth on β-Asp-Arg appeared biphasic (Fig. 3E), suggesting a possible induction of processes required for rapid metabolism of β-Asp-Arg, perhaps aot operon upregulation by the ArgR regulator (47). Transposon mutation of aotO or of the L-Arg transporter subunits aotM and aotP abolishes the ability to use β-Asp-Arg as a nitrogen and/or carbon source. These data confirm the in vivo CphZ function of aotO and also imply that the transporter AotJQMP can import β-Asp-Arg, not only Arg. Multiple mechanisms for L-Arg uptake have been identified, including by P. aeruginosa ArcD, an arginine/ornithine antiporter (65). ArcD seems unable to import β-Asp-Arg, because the aotM mutant cannot grow on β-Asp-Arg, but can grow on Arg. Given the difference in size and nature between Arg and β-Asp-Arg, it is remarkable that AotJQMP can import both compounds. Furthermore, with the apparent advantage strains can derive from use of environmental cyanophycin dipeptide, inhibition of its uptake or hydrolysis by pathogenic strains could be a promising strategy for limiting their proliferation within complex microbiota.

Conclusion.

Although cyanophycin is best known for its roles in cyanobacteria, the discovery of a common, dedicated CphZ family in both cyanophycin producers and scavengers with medical relevance builds on the recent reports of wide environmental availability to highlight the utility of this interesting natural biopolymer.

Materials and Methods

Cloning, Protein Expression, and Purification.

The cloning and protein expression and purification performed in this study were like that described previously for other cyanophycin metabolizing enzymes (1, 22, 26, 28). The gene encoding AbCphZ (WP_004925890.1) was amplified from A. baylyi DSM587 genomic DNA (Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ), Germany). The codon-optimized synthesis of the gene encoding PaAotO (WP_128550578.1) was commissioned from BioBasic, Canada. The genes encoding AbCphZ and PaAotO were subcloned into vector pBacP, a pJ411-derived plasmid that encodes a C-terminal, tobacco etch virus (TEV) protease cleavable, 8× histidine affinity tag. All cloning and mutagenesis steps were accomplished in cellulo, transforming PCR fragments containing overlapping ends into DH5-α E. coli cells. Protein expression was carried out in E. coli BL21(DE3) grown in TB media supplemented with 150 µg/mL kanamycin. Cultures were grown at 37 °C until an OD₆₀₀ of ~1 was reached, at which time the growth temperature was lowered to 18 °C. Protein expression was induced by the addition of 0.2 mM isopropyl β-d-1-thiogalactopyranoside, and culture were grown for ~20 h prior to harvesting by centrifugation. All following steps were carried out at 4 °C. The cell pellets were resuspended in buffer A (250 mM NaCl, 50 mM Tris pH 8.0, 10 mM imidazole, 2 mM β-mercaptoethanol) supplemented with crystals of lysozyme and DNase I and lysed on ice by sonication. The lysate was clarified by centrifugation at 40,000 g for 30 min and then applied to a HisTrap HP column (Cytiva, USA). Following loading, the protein was washed with at least 20 column volumes of buffer B (buffer A with 30 mM imidazole) and eluted with buffer C (buffer A with 250 mM imidazole). For structural studies, the protein was dialyzed overnight against buffer D (250 mM NaCl, 20 mM Tris pH 8.0, 5 mM β-mercaptoethanol) in the presence of TEV protease for removal of the octahistidine tag. Following tag cleavage, the protein was again applied to a HisTrap column and the flow through was collected. For all protein preparations, sample was concentrated using an Amicon^® Ultra Centrifugal filter and applied to a Superdex200 16/60 column (Cytiva, USA) equilibrated in buffer E (100 mM NaCl, 20 mM Tris pH 8.0, 1 mM dithiothreitol). Fractions containing the highest protein purity were pooled, and following concentration were supplemented with glycerol to a final volume of 15% and flash-frozen in liquid nitrogen for storage.

Protein Crystallization, Data Collection, Structure Solution, and Refinement.

AbCphZ crystals were grown using the sitting drop method using a reservoir volume of 500 µL and a drop containing 2 µL of protein sample in buffer E and 2 µL of crystallization buffer. The crystallization buffer for WT AbCphZ (5 mg/mL) contained 0.1 M bis-tris propane pH 7.5, 24% PEG3350, 0.2 M NaBr, 10 mM spermine, and 10 mM LiCl. Crystals were grown at 4 °C, dehydrated by allowing the drop to equilibrate through vapor diffusion against 0.1 M bis-tris propane pH 7.5, 30% PEG3350, 0.2 M NaBr, and 10 mM spermine overnight, and then cryo-protected by dipping them in 0.1 M bis-tris propane pH 7.5, 20% PEG3350, 20% ethylene glycol, 0.2 M NaBr, and 10 mM spermine. Data were collected at the Advanced Photon Source (APS) beamline 24-ID-C. The structure was solved by molecular replacement using PDB codes 3FMC and 3LWU as search models. The crystals of E251A AbCphZ were grown as above and soaked in a cryo-protection solution containing 10 mM β-Asp-Arg for 30 min prior to freezing. Data were collected at the Advanced Light Source (ALS) beamline 5.0.1. All datasets were both indexed and integrated in DIALS (66) and merged in Aimless (67) implemented in CCP4i2 suite (68), and in parallel indexed, integrated, and merged in HKL2000 (69). The structures were refined in REFMAC (70), Rosetta (71), PHENIX (72), and Coot (73). Figures were prepared in PyMOL (Schrödinger, Inc.). For anomalous difference map calculation, data from E251A AbCphZ crystals were collected at the Canadian Light Source (CLS) Canadian Macromolecular Crystallography Facility Bending Magnet beamline (CMCF-BM) at a wavelength of 1.283 Å, processed in DIALS (66), Aimless (67), retaining the anomalous information. The anomalous difference map was calculated using these data in the program CCP4i2 (68).

Enzyme Activity Assays.

An Asp release assay (36) was used to detect Asp release from β-Asp-X dipeptides: Each 100 µL reaction contained 100 mM HEPES pH 8.2, 20 mM KCl, 5 mM α-ketoglutarate, 500 nM purified enzyme, 0.3 U malate dehydrogenase, 1 mM NADH, 2.4 U aspartate aminotransferase, and 1 mM dipeptide substrate. For other Arg-containing substrates, Arg release was monitored by using a free Arg detection kit (K-LARGE, Neogen, USA): The 135 µL reactions contained 15 µL buffer solution, 10 µL NADPH solution, 1 µL GIDH suspension, 2.5 µL urease solution, 1 µL arginase suspension, 1 mM substrate, and 500 nM enzyme. For both assays, reaction progression was monitored by following 340 nm transmittance in 96-well plates. Data were collected using a SpectraMax Paradigm (Molecular Devices, USA) and analyzed using GraphPad Prism (GraphPad, USA). β-Asp-Arg dipeptides were purified as previously described (26). β-Asp-Ala and α-Asp-Arg were purchased from Bachem (Switzerland), N²-acetyl arginine, β-Asp-Lys, and β-Asp-Leu from Toronto Research Chemicals (Canada), β-Asp-Asp from AchemBlock (USA), and N²-succinyl arginine from BLD Pharmatech (USA).

ITC.

ITC experiments were performed using a MicroCal iTC200 (GE Healthcare, USA) at 25 °C. For β-Asp-Arg, the cell contained 200 µM purified E251A AbCphZ in buffer F (100 mM NaCl, 20 mM Tris pH 8.0, 5 mM β-mercaptoethanol) and the syringe contained 1 mM β-Asp-Arg in buffer F. For L-Arg and L-Asn, the cell contained 400 µM purified E251A AbCphZ and the syringe contained 10 mM substrate, both in buffer F. A total of nineteen 2 µL injections were interspaced by 180 s each. Data were analyzed using Microcal Origin 7.0 (OriginLab, USA) with a binding model stoichiometry of 1:1. Measurements were performed in triplicates.

ICP-MS.

AbCphZ was buffer-exchanged into 100 mM (NH₄)₂CO₃ by performing gel filtration chromatography with a Superdex S200 10/300 column equilibrated with that buffer. The protein was concentrated to 100 μM and analyzed by ICP-MS at the Center for Applied Isotope Studies, University of Georgia. A sample of the buffer eluted from the column was used as a control.

Bioinformatic Analyses.

For bioinformatic analyses, protein sequences were considered homologous if above a threshold of >70% coverage and >30% identity. This threshold almost certainly discards bone fide homologs, but is important to exclude false positives: A set of cyanobacterial sequences featuring 21 to 28% identity and 83 to 87% coverage to AbCphZ is returned in BLAST searches if lower thresholds are used (e.g., E value < 0.05). These cyanobacterial sequences contain residues for binding the Zn²⁺ and the Arg moiety but not the Asp portion of β-Asp-Arg, and thus cannot be assumed to be isofunctional with CphZ. The >70% coverage and >30% identity threshold appears effective in limiting false positives, as measured by the conservation of active site residues: i) Zn²⁺ binding residues: H50 99.7%; E53 99.7%; H179 100%; ii) Nucleophilic water activation: E251 99.96%; iii) Asp moiety binding: D181 99.4%; K366 97.7% as K, H or R; iv) Arg moiety binding: R102 99.96%; N112: 99.98%; R113 100%; S216 93.2%; D222 99.8%; E223 99.96%.

For bioinformatic analyses of gene co-occurrence, all complete bacterial genomes in the NCBI RefSeq database were used to create a local database (May 2022). Cblaster (58) was used in local mode to search this database using several templates for CphA1 (accession codes WP_028947105.1, WP_004925893.1 and WP_015942562.1), cyanophycinase (WP_011058003.1, WP_004925892.1, and Q8KQN8.1) and PaAotO and AbCphZ. The resulting binary table was analyzed using Excel.

Bacterial Strains.

The P. aeruginosa strains used in this study are listed in SI Appendix, Table S4. The aotO, aotP and aotM transposon mutants were obtained from the PAO1 two-allele transposon mutant library and confirmed by PCR (52, 53) and sequencing.

P. aeruginosa Growth Conditions.

P. aeruginosa was cultured in minimal media (22 mM KH₂PO₄, 42 mM Na₂HPO₄, 8.6 mM NaCl, 2.0 mM MgSO₄, 5.0 μM FeSO₄, pH 7.0) supplemented with carbon or nitrogen source as indicated in each experiment, as per Lundgren et al. (74). Subcultures were inoculated from agar plates (LB agar for WT, LB agar with 50 µg/mL tetracycline for Tc-R mutants) and grown in 5 mL minimal media supplemented with 20 mM D-glucose and 10 mM NH₄Cl, in 16 mm × 150 mm sterile glass tubes for 6 h (230 rpm, 37 °C). Following subculture, the bacteria were diluted to an OD₆₀₀ of 0.05 in 5 mL of fresh minimal media supplemented with 20 mM D-glucose and 10 mM NH₄Cl and grown for 24 h (230 rpm, 37 °C). Cells were harvested by centrifugation (7,800 × g for 5 min), washed twice in 5 mL of minimal media with no supplementation to remove all residual nutrients, and then resuspended in minimal media at an OD₆₀₀ of ~0.05. The final culture media were then supplemented with one of the following: 20 mM D-glucose and 10 mM NH₄Cl, 10 mM L-aspartate, 10 mM β-Asp-Arg, 10 mM β-Asp-Ala, or 10 mM L-arginine as indicated. For growth experiments, bacteria were incubated in either 15 mL (in 150-mL flasks, 230 rpm, 37 °C) or 200 μL (in 96-well plates, continuous shaking 282 CPM (3 mm), 37 °C) using a BioTek Cytation 5 as indicated. Bacterial growth was monitored by OD₆₀₀ (using a GENESYS UV scanning spectrophotometer or BioTek Cytation 5 plate reader) and by viable bacterial counts (using standard serial dilution and plating on LB agar).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The protein structures solved in this study have been deposited to the Protein Data Bank, under the accession codes 8EIN (AbCphZ) and 8EIP (AbCphZ_E251A + β-Asp-Arg).