Enzymatic and Mutational Analysis of the PruA Pteridine Reductase Required for Pterin-Dependent Control of Biofilm Formation in Agrobacterium tumefaciens

ABSTRACT

Pterins are ubiquitous biomolecules with diverse functions, including roles as cofactors, pigments, and redox mediators. Recently, a novel pterin-dependent signaling pathway that controls biofilm formation was identified in the plant pathogen Agrobacterium tumefaciens. A key player in this pathway is a pteridine reductase, termed PruA, where its enzymatic activity has been shown to control surface attachment and limit biofilm formation. Here, we biochemically characterized PruA to investigate the catalytic properties and the substrate specificity of this pteridine reductase. PruA demonstrated maximal catalytic efficiency with dihydrobiopterin and comparable activities with the stereoisomers dihydromonapterin and dihydroneopterin. Since A. tumefaciens does not synthesize or utilize biopterins, the likely physiological substrate is dihydromonapterin or dihydroneopterin or both. Notably, PruA did not exhibit pteridine reductase activity with dihydrofolate or fully oxidized pterins. Site-directed mutagenesis studies of a conserved tyrosine residue, the key component of a putative catalytic triad, indicated that this tyrosine is not directly involved in PruA catalysis but may be important for substrate or cofactor binding. Additionally, mutagenesis of the arginine residue in the N-terminal TGX₃RXG motif significantly reduced the catalytic efficiency of PruA, supporting its proposed role in pterin binding and catalysis. Finally, we report on the enzymatic characterization of PruA homologs from Pseudomonas aeruginosa and Brucella abortus, thus expanding the roles and potential significance of pteridine reductases in diverse bacteria.

IMPORTANCE Biofilms are complex multicellular communities that are formed by diverse bacteria. In the plant pathogen Agrobacterium tumefaciens, the transition from a free-living motile state to a nonmotile biofilm state is governed by a novel signaling pathway involving small molecules called pterins. The involvement of pterins in biofilm formation is unexpected and prompts many questions about the molecular details of this pathway. This work biochemically characterized the PruA pteridine reductase involved in the signaling pathway to reveal its enzymatic properties and substrate preference, thus providing important insight into pterin biosynthesis and its role in A. tumefaciens biofilm control. Additionally, the enzymatic characteristics of related pteridine reductases from mammalian pathogens were examined to uncover potential roles of these enzymes in other bacteria.

INTRODUCTION

Pterins are molecules containing a 2-amino-4(3H)-oxopteridine core (Fig. 1), which is the same chemical moiety present in folates. Although the functions of folates as key players in one-carbon metabolism are well-defined (1), the functions of diverse pterins in nature are much broader and include roles as cofactors, pigments, and redox mediators (2, 3). The pteridine ring can exist in the fully oxidized form, a dihydro form, and a tetrahydro form (Fig. 1A). During their biosynthesis, pterins are generally in a dihydro form, which is then enzymatically reduced to the biologically active tetrahydro form. Both reduced forms are susceptible to spontaneous oxidation in the presence of molecular oxygen, which generates the fully oxidized pteridine ring. Pteridine reductases produce and maintain the pool of tetrahydropterins and thus are essential for various functions in diverse organisms.

FIG 1.

Structures of pteridine-containing molecules relevant in this study. (A) Different oxidation states of the pteridine ring with the associated enzymes involved in each reduction reaction. (B) Side chains of common pteridine-containing molecules. (C) The reaction catalyzed by PruA to produce H₄MPt, followed by potential methylation to produce 2′-O-methyl-H₄MPt. One or both of these MPt species are likely the signaling molecules that control biofilm formation in A. tumefaciens.

The most well-known and well-studied pteridine reductase is dihydrofolate reductase (DHFR), which uses NADPH to reduce folate and dihydrofolate (H₂F) to generate the ubiquitous cofactor tetrahydrofolate (H₄F) that is used for fundamental one-carbon transfer reactions in the biosynthetic pathways of purines, thymidine, and select amino acids (4). Other established pteridine reductases are unrelated to DHFR in terms of both sequence and structure and instead belong to the short-chain dehydrogenase/reductase (SDR) family (5). These include pteridine reductase 1 (PTR1), which is required for tetrahydrobiopterin (H₄BPt) synthesis in select eukaryotes and which has been most well studied in trypanosomatids (6–8), and FolM, which produces tetrahydromonapterin (H₄MPt), the cofactor of phenylalanine hydroxylase in specific bacteria (9). PTR1 produces H₄BPt from dihydrobiopterin (H₂BPt) or from fully oxidized biopterin (BPt) scavenged from host cells (Fig. 1) (10). In contrast, FolM from Escherichia coli displays activity only with the dihydro form of its pterin substrate (9). Although presumably not their primary physiological functions, both of these SDR pteridine reductases also exhibit DHFR activity in vitro (9–11). Despite the overall low sequence identity among members of the SDR family (∼15% to 30%), a central catalytic YX₃K motif is highly conserved (12, 13), as is an N-terminal glycine motif (TGX₃GXG), involved in cofactor binding and recognition (13–15). Interestingly, the pteridine reductases in the SDR family have an arginine in place of the glycine at position 6 in this motif (TGX₃RXG).

PruA is a recently identified SDR pteridine reductase involved in a pterin-dependent signaling pathway controlling surface attachment and biofilm formation in Agrobacterium tumefaciens (16). The gene encoding PruA (Atu1130) was initially identified through genetic analysis to be important for the negative regulation of biofilm formation (17). Compared to the wild type, a ΔpruA mutant exhibited elevated biofilm formation and increased production of unipolar polysaccharide (UPP), the key adhesin that facilitates polar attachment to surfaces (16) and, hence, elevated biofilm formation. Additionally, analysis of wild-type A. tumefaciens cultures revealed a novel pterin species, identified as 2′-O-methyl-monapterin (2′-O-methyl-MPt), that was absent in the ΔpruA mutant (16). Since the pterins in this study were isolated and chemically converted to their oxidized forms before analysis, the physiologically relevant form of this novel pterin is likely to be the fully reduced 2′-O-methyl-H₄MPt (Fig. 1C).

The mechanism of PruA-dependent regulation of biofilm formation is under active investigation, but it is clear that the enzymatic activity of PruA is essential for the proposed pterin-dependent regulatory pathway. In the current working model (16), PruA is required for the biosynthesis of a reduced pterin species, likely H₄MPt and/or 2′-O-methyl-H₄MPt (Fig. 1C), that presumptively associates with and activates a pterin-binding protein (PruR). The PruR-pterin complex then interacts with a dual-function diguanylate cyclase/phosphodiesterase A enzyme (DcpA), biasing its activity toward a phosphodiesterase-dominant state and resulting in diminished levels of cyclic diguanylate monophosphate (cdGMP) (16). cdGMP is a well-established second messenger involved in regulating the motile-to-sessile switch and subsequent biofilm formation. In general, high levels of cdGMP favor the sessile state by enhancing processes required for surface attachment, while lower levels of cdGMP favor the motile planktonic state (18). Thus, the enzymatic activity of PruA leads to decreased cdGMP and promotes maintenance of the motile lifestyle.

Here, we biochemically characterized PruA to determine its catalytic activity and thereby gain insight into how this activity then influences DcpA and surface attachment. We further compare this newly described pteridine reductase to other related bacterial enzymes involved in diverse physiological processes.

RESULTS

Biochemical characterization of PruA from A. tumefaciens.

The gene encoding PruA was introduced into pET15b and overexpressed as an N-terminal His₆-tagged protein in E. coli. Following nickel affinity purification (Fig. 2A), size exclusion chromatography revealed that PruA is ∼60 kDa, consistent with the protein existing as a homodimer (monomer size, 28 kDa) (see Fig. S1 in the supplemental material). In the presence of NADPH and dihydromonapterin (H₂MPt), PruA exhibited maximal pteridine reductase activity at pH 6.0; however, the enzyme had a pH profile somewhat broader than that of other pteridine reductases and, notably, exhibited ∼20% of maximal activity at pH 8.0 (Fig. 2B). In contrast, FolM is reported to be completely inactive at pH 8.0 (9), and PTR1 is also inactive with most substrates at pH 8.0 (10).

FIG 2.

Purification and pH study of PruA from A. tumefaciens. (A) SDS-PAGE gel stained with Coomassie showing purified His₆-PruA. (B) pH dependence of PruA pteridine reductase activity in the presence of 100 μM H₂MPt and 100 μM NADPH.

To determine the substrate profile of PruA, we performed kinetic studies with various potential pterin and folate substrates. PruA exhibited maximal catalytic efficiency with H₂BPt, while both H₂MPt and dihydroneopterin (H₂NPt) (Fig. 3 and Table 1) also served as competent substrates. It is important to mention that in our original preliminary enzymatic characterization of PruA, pteridine reductase activity was not detected with H₂NPt (16), while here we have confirmed that the kinetic parameters observed for PruA with H₂NPt are similar to those observed with H₂MPt. This is due to both the optimization of PruA purification to produce a stable active enzyme and the modified conditions for activity assays. Interestingly, PruA showed no detectable pteridine reductase activity with H₂F or any fully oxidized pteridine molecules (biopterin, neopterin, or folate).

FIG 3.

Kinetic characterization of PruA from A. tumefaciens (A) and its homolog from P. aeruginosa (PA3437) (B). Reactions were carried out in the presence of 100 μM NADPH with various concentrations of each indicated pterin substrate. GraphPad Prism (version 8.0) software was used for data analysis. H₂MPt, dihydromonapterin; H₂NPt, dihydroneopterin; H₂BPt, dihydrobiopterin; H₂F, dihydrofolate.

TABLE 1.

Summary of kinetic parameters and substrate specificities for PruA from A. tumefaciens compared to those of homologs from B. abortus (BAB1_0721) and P. aeruginosa (PA3437)

Substrate	A. tumefaciens PruA (Atu1130)				Sp act (μmol/min/mg) of B. abortus BAB1_0721a	P. aeruginosa PA3437
	Sp act (μmol/min/mg)	kcat (s−1)	Km (μM)	kcat/Km (M−1 · s−1)		Sp act (μmol/min/mg)	kcat (s−1)	Km (μM)	kcat/Km (M−1 · s−1)
H2MPt	4.77 ± 0.23	2.3	71.3 ± 11.1	3.2 × 104	3.72 ± 0.18	4.30 ± 0.36	1.9	170.0 ± 31.2	1.1 × 104
H2NPt	5.03 ± 0.22	2.4	95.1 ± 11.9	2.5 × 104	3.16 ± 0.30	1.63 ± 0.13	0.7	90.0 ± 20.3	7.8 × 103
H2BPt	9.83 ± 0.24	4.7	34.7 ± 3.8	1.3 × 105	4.02 ± 0.25	1.87 ± 0.11	0.8	75.7 ± 13.5	1.1 × 104
H2F	No activity				No activity	7.36 ± 0.31	3.2	59.3 ± 8.3	5.4 × 104

^aThe complete kinetic parameters for the PruA homolog from B. abortus were not obtained due to protein insolubility and instability upon purification. Specific activity is reported at 300 μM H₂MPt and 100 μM NADPH.

Characterization of putative pteridine reductases in other bacteria.

Besides PruA, the only other bacterial SDR pteridine reductase to be enzymatically investigated to date is the dihydromonapterin reductase (FolM) from E. coli, which generates H₄MPt for an unknown biological function (9). Given the unique role of PruA in controlling biofilm formation in A. tumefaciens, we sought to expand our investigation of putative pteridine reductases in other bacteria to gain insight into the functions of these widespread enzymes, especially with respect to their potential involvement in regulating biofilm formation. Homologs of PruA exist in several related members of the Rhizobiales order and more broadly in the Alphaproteobacteria class. Several of these homologs were previously shown to complement the A. tumefaciens ΔpruA mutant (16), including BAB1_0721 from Brucella abortus (55% identity and 70% similarity to A. tumefaciens PruA [AtPruA]), which is the protein encoded by the B. abortus genome with the best reciprocal sequence match with AtPruA (Fig. S2). To compare the in vitro activities of these enzymes, we overexpressed and purified BAB1_0721 in the same manner that we did for AtPruA. The recombinant enzyme was, unfortunately, only sparingly soluble over a range of expression conditions (low temperature, lower isopropyl-β-d-thiogalactopyranoside [IPTG] concentration, etc.) and also was unstable and tended to precipitate after purification. Thus, we were unable to perform a full kinetic study with the various potential substrates. However, we did confirm that BAB1_0721 is active with H₂BPt, H₂MPt, and H₂NPt with specific activities of 4.02 μmol/min/mg, 3.72 μmol/min/mg, and 3.16 μmol/min/mg, respectively, in the presence of 300 μM dihydropterin substrate (Table 1). Similar to the findings for AtPruA, this pteridine reductase from B. abortus did not display activity with H₂F or with any fully oxidized pteridine substrates.

We next examined the PruA homolog in Pseudomonas aeruginosa PA3437 (30% identity and 47% similarity to AtPruA) (Fig. S2). PA3437 was previously defined as FolM and was implicated as being responsible for H₄MPt production, required for phenylalanine hydroxylase activity in P. aeruginosa (9). Expression of PA3437 in the A. tumefaciens ΔpruA mutant partially corrected its phenotypes for UPP biosynthesis and surface attachment (Fig. S3), indicating that this enzyme has activity toward the same substrate as PruA that is relevant in the biofilm signaling pathway. After overexpression and purification of the His₆-tagged PA3437, the enzyme proved much more amenable to in vitro investigation than the B. abortus PruA homolog, and we performed a full kinetic study with various pterin substrates (Fig. 3B and Table 1). Interestingly, PA3437 exhibited maximal enzymatic activity with H₂F, while H₂MPt, H₂NPt, and H₂BPt also served as suitable substrates. Similar to the findings for PruA and FolM, we did not observe activity with any of the fully oxidized pterins tested.

Vibrio cholerae is a model pathogen for the study of biofilm formation and regulation; thus, we were interested to potentially connect the pterin-dependent signaling pathway identified in A. tumefaciens to V. cholerae. Therefore, we expressed, purified, and tested the potential pteridine reductase activities of two SDR family proteins with the highest sequence similarity to PruA in V. cholerae (VC_A0301 and VC_A0691; both share ∼25% identity and 45% similarity with AtPruA). Although both of these enzymes were expressed well in a soluble form, we did not detect pteridine reductase activity with any of the pterin substrates tested. Consistent with this observation, expression of either of these genes in A. tumefaciens also failed to complement the ΔpruA phenotype (Fig. S3). Primary sequence analysis revealed that both of these homologs have an N-terminal TGX₃GXG motif instead of a TGX₃RXG motif (in which the underlining indicates the altered amino acid), supporting the result that although they may be short-chain dehydrogenases, these two enzymes do not function as pteridine reductases. In fact, V. cholerae does not encode an SDR homolog with the TGX₃RXG motif.

Mutational analysis of the PruA YX₃K motif.

PTR1 is the most well-characterized pteridine reductase belonging to the SDR family. In the PTR1-catalyzed reduction of fully oxidized pterins, such as BPt or folate, to the dihydro form (Fig. 1A), the critical proton donor in the reaction is a tyrosine residue which is part of a highly conserved YX₃K motif (19) (Tyr 194 and Lys 198 in Leishmania major PTR1) (Fig. S2). The importance of the tyrosine and lysine residues in the YX₃K catalytic motif has also been studied extensively in other SDR members that are not pteridine reductases, including 15-hydroxyprostaglandin dehydrogenase (20) and 11β-hydroxysteroid dehydrogenase (21). The lysine residue presumably lowers the pK_a of tyrosine by stabilizing the tyrosinate intermediate (5). Importantly, in the second reduction reaction catalyzed by PTR1 to generate the fully reduced tetrahydropterin product (Fig. 1A), the tyrosine residue is not involved and the proton donor is instead proposed to be a water molecule (22) in a mechanism similar to that for DHFR (23–25). Despite the conservation of the YX₃K motif in PruA (Tyr 163 and Lys 167) (Fig. S2), our in vitro experiments reported here showed that PruA does not display pteridine reductase activity with any fully oxidized pterins. We therefore sought to explore the importance of the YX₃K motif, especially the key tyrosine residue, in the PruA-catalyzed reaction.

The previously generated A. tumefaciens ΔpruA strain (16) was used as a background to investigate the activities of select pruA variants in vivo. The ability of pruA variants to complement the deletion strain was determined by measuring biofilm formation and evaluating polysaccharide (UPP and cellulose) production through colony pigmentation on solid medium using the polysaccharide-reactive dye Congo red. UPP production is required for surface attachment and biofilm formation, and both UPP and cellulose contribute to the Congo red pigmentation of colonies (17). Wild-type A. tumefaciens exhibited basal levels of biofilm formation in laboratory culture and formed pale orange or white colonies when grown on solid growth medium supplemented with Congo red (Fig. 4A). In contrast, the ΔpruA mutant exhibited elevated biofilm formation and abundant polysaccharide production independent of surface contact, demonstrating that PruA is an important factor in controlling the motile-to-sessile transition in A. tumefaciens (Fig. 4A) (16). As we reported previously, the plasmid-borne expression of wild-type PruA fully corrects this deficiency, while a plasmid-encoded PruA Y163A variant expressed in the ΔpruA strain had no effect on these ΔpruA mutant phenotypes, with its phenotype appearing similar to the phenotype of the deletion strain alone (Fig. 4A) (16). This indicates that the PruA Y163A variant is inactive and that the enzymatic activity of PruA is required for its involvement in the pathway. However, a more subtle alteration predicted to abolish the ability of Y163 to act as a proton donor would be mutation of the tyrosine to phenylalanine. Thus, we generated PruA Y163F and expressed the enzyme from a plasmid in the ΔpruA mutant. Interestingly, biofilm formation and Congo red colony pigmentation returned to wild-type levels (Fig. 4A), indicating that PruA Y163F retained catalytic activity. To address this in vitro, we kinetically characterized these two variants in the presence of H₂MPt as a substrate. Consistent with the phenotypic results, we found that PruA Y163A was completely inactive, while PruA Y163F demonstrated a catalytic efficiency comparable to that of the wild type (Table 2).

FIG 4.

Phenotypic analysis of A. tumefaciens strains expressing PruA variants. A select portion of the PruA amino acid sequence highlighting residues of interest (top), quantitation of biofilm formation (middle), and Congo red colony phenotypes (bottom) of wild-type (WT) A. tumefaciens compared to A. tumefaciens ΔpruA and the ΔpruA mutant complemented with the PruA variants are shown. (A) Data for tyrosine variants to determine the importance of the putative catalytic triad (highlighted in blue). (B) Data for arginine variants within the N-terminal TGX₃RXG motif. Solid medium for Congo red staining was supplemented with 400 μM IPTG to express plasmid-borne P_lac fusions. A. tumefaciens biofilm formation was measured on PVC coverslips after 48 h of static growth at 28°C. Adherent biomass was quantified by staining with crystal violet (CV), and acetic acid-solubilized CV was measured by determination of the A₆₀₀. In parallel, the OD₆₀₀ of planktonic culture was determined and the CV absorbance was normalized to the culture growth by calculating the A₆₀₀/OD₆₀₀ ratio. IPTG (400 μM) was added to all cultures.

TABLE 2.

Kinetic parameters of PruA variants compared to those of wild-type PruA with H₂MPt

PruA variant	In vivo activitya	Sp act (μmol/min/mg)	kcat (s−1)	Km (μM)	kcat/Km (M−1 · s−1)
Wild type	✓	4.77 ± 0.23	2.3	71.3 ± 11.1	3.2 × 104
Y163A		No activity
Y163F	✓	8.84 ± 0.78	4.2	212.7 ± 36.6	2.0 × 104
Y161A	✓	2.88±0.13	1.4	50.5 ± 8.5	2.7 × 104
Y161A Y163F		No activity
Y161F Y163F	✓	4.13 ± 1.06	2.0	333.2 ± 122.6	6.0 × 103
R24A	✓	1.71 ± 0.23	0.8	169.4 ± 50.2	4.8 × 103
R24G	✓b	0.28 ± 0.11b	NDc	ND	ND

^aThe in vivo activity was assessed by the abilities of each respective variant to complement the A. tumefaciens ΔpruA strain (Fig. 4).

^bThis variant exhibited biofilm formation levels comparable to those of the wild type but displayed elevated levels of polysaccharides, similar to those for strains containing an inactive PruA (Fig. 4). This intermediate phenotype indicates intermediate levels of cdGMP due to the very low enzymatic activity of PruA R24G (see the text for more details).

^cND, not determined. The activity of PruA R24G was so low that we were unable to obtain accurate kinetic parameters. The specific activity reported for this variant was obtained with 300 μM H₂MPt and 100 μM NADPH.

Given the direct involvement of the corresponding tyrosine residue in PTR1 catalysis (19), it was surprising that PruA Y163A was inactive, whereas the Y163F variant was not catalytically impaired. We noted a tyrosine residue (Y161) located two amino acid residues upstream of Y163 (Fig. S2), and it seemed possible that this proximal residue could instead participate in catalysis and mask the effect of the mutation. Therefore, we generated a Y161A variant and two double variants, Y161A_Y163F and Y161F_Y163F. PruA Y161A retained activity both in vivo and in vitro (Fig. 4A and Table 2), indicating that the tyrosine at the 161 position is not required for catalysis. The double variant Y161A_Y163F lacked activity both in vivo and in vitro; however, the Y161F_Y163F variant retained the ability to negatively regulate biofilm formation (Fig. 4A) and still displayed pteridine reductase activity, although it had significantly decreased catalytic efficiency (Table 2). Therefore, these tyrosine residues are likely not directly involved in catalysis but may be important for substrate binding and orientation and/or the structural integrity of the active site. Taken together, these mutagenesis studies demonstrate that catalysis by PruA and the phenotypes that depend upon this in A. tumefaciens do not involve the tyrosine residue in the YX₃K motif known to be essential for the PTR1-catalyzed reduction of fully oxidized pterins. This is consistent with the model in which the YX₃K catalytic motif is involved only in the reduction of fully oxidized pterins to the dihydro state (19, 26) and is not involved in the reduction of dihydropterins to tetrahydropterins (22). However, it is intriguing that the motif is so well conserved in pteridine reductases, such as PruA and FolM (Fig. S2), that do not catalyze the reduction of fully oxidized pterins.

Mutational analysis of the N-terminal TGX₃RXG motif.

As mentioned earlier, the established pteridine reductases in the SDR family contain an N-terminal TGX₃RXG motif rather than the TGX₃GXG motif more generally conserved among other SDRs (5). To examine the importance of the arginine in this motif and to establish a criterion by which to identify other pteridine reductases that could potentially be involved in regulating biofilm formation pathways, we generated two PruA variants, PruA R24G and PruA R24A, and analyzed their in vivo and in vitro activities. The resulting purified enzymes exhibited significantly impaired pteridine reductase activity, with the R24G variant activity being so low that we were unable to perform a complete kinetic study (Table 2). However, both plasmid-carried mutant alleles were able to partially complement the elevated biofilm formation of the ΔpruA mutant, decreasing it to levels closer to wild type (Fig. 4B). Despite their ability to rescue the elevated surface attachment phenotype, the ΔpruA mutant expressing these alleles retained a portion of the elevated Congo red pigmentation, with PruA R24G being most pronounced in this respect, suggesting incomplete complementation and an impaired but not an abolished in vivo functionality of the plasmid-expressed PruA variants. Taken together, the results indicate that the arginine in the TGX₃RXG motif is important but not essential for pteridine reductase activity.

DISCUSSION

The established roles of pterins in nature include serving as enzyme cofactors, antioxidants, and insect pigments, with the redox properties of the pteridine ring being central in all of these roles. The recently described involvement of pterins as signaling molecules controlling biofilm formation in A. tumefaciens is a previously undefined function for these diverse molecules. One of the key players in this signaling pathway in A. tumefaciens is the pteridine reductase PruA. Given the divergent physiological functions and substrate specificities of established pteridine reductases, we sought here to further characterize PruA to provide insight into its impact on pterin metabolism and the resulting control of biofilm formation.

PruA substrate specificity and physiological implications.

Kinetic characterization of PruA from A. tumefaciens revealed the highest catalytic efficiency with H₂BPt as a substrate (Fig. 3A and Table 1). The product of this reaction, H₄BPt (Fig. 1), is found primarily in eukaryotic organisms, where it is used as a redox cofactor for hydroxylation reactions catalyzed by amino acid hydroxylases and nitric oxide synthases (27). In mammals, H₄BPt is synthesized de novo from GTP, while in parasites, such as trypanosomes, the oxidized cofactor is scavenged from host cells. After H₄BPt-dependent hydroxylation reactions, the resulting pteridine product is H₄B-4a-carbinolamine, which requires the action of two enzymes for conversion back to H₄BPt, pterin 4a-carbinolamine dehydratase and quinonoid dihydropteridine reductase (QDPR) (27). Similar to the vast majority of other bacteria, A. tumefaciens does not appear to possess the genes encoding enzymes that are known to utilize H₄BPt, and it additionally does not have the genes required for H₄BPt biosynthesis and complete regeneration. Thus, it is likely that the high in vitro activity of PruA toward H₂BPt is not physiologically relevant.

PruA exhibited similar catalytic efficiencies with the stereoisomers H₂MPt and H₂NPt (Fig. 1 and 3A; Table 1). H₂NPt is an intermediate in H₄F biosynthesis and can serve as a precursor for H₂MPt synthesis, but neither H₂NPt nor H₄NPt has been reported to have any independent physiological functions. H₄MPt is biosynthesized via a branchpoint in H₄F biosynthesis, and MPt was isolated as the most abundant pterin present in E. coli (28). Genetic studies have shown that H₄MPt can serve as a cofactor for the P. aeruginosa phenylalanine hydroxylase, and E. coli FolM was shown to catalyze the conversion of H₂MPt to H₄MPt in vitro (9). Interestingly, the function of H₄MPt in bacteria that do not have a phenylalanine hydroxylase gene, such as E. coli, remains unclear. Our prior work identified an MPt derivative, 2′-O-methyl-MPt, extracted from A. tumefaciens cells that was absent in the ΔpruA strain (16); thus that result, taken together with the current data, indicates that the primary biochemical function of PruA is likely to reduce H₂MPt to H₄MPt. Since the A. tumefaciens genome does not encode any amino acid hydroxylase homologs and since PruA was identified to be a key component regulating biofilm formation in this organism, we propose that an important physiological role of PruA in A. tumefaciens is to produce H₄MPt as part of a signaling cascade to control biofilm formation.

Comparison of diverse SDR pteridine reductases.

The pteridine reductase BAB1_0721 from B. abortus has the same substrate profile as PruA from A. tumefaciens (Table 1). This is consistent with their high sequence identity as well as previously reported data showing that the gene for this protein can fully complement the A. tumefaciens ΔpruA mutant (16). Additionally, the genomic context of pruA and BAB1_0721 is conserved and syntenous (see Fig. S5 in the supplemental material). Thus, it is reasonable to conclude that this pteridine reductase from B. abortus has a physiological role(s) similar to that of PruA from A. tumefaciens. However, B. abortus does not have orthologs of DcpA or PruR, and, therefore, the pterin-dependent control of biofilm formation may not be relevant in this organism. This would indicate that PruA has an additional yet-to-be-identified physiological function.

The pteridine reductase in P. aeruginosa (PA3437) has a substrate profile (Fig. 3B and Table 1) distinct from that of PruA and other previously characterized pteridine reductases. Although PA3437 was originally defined as FolM and is encoded in a gene cluster with other genes involved in H₄MPt biosynthesis (FolE and FolX) (Fig. S5), its high activity with H₂F implicates this pteridine reductase as a potential backup DHFR. The observation that PA3437 also complements the A. tumefaciens ΔpruA mutant (Fig. S3) suggests that it can fulfill the same role as PruA in the A. tumefaciens biofilm control pathway. The two SDR family members in V. cholerae with the highest sequence similarity to PruA are not pteridine reductases, as indicated by their lack of in vitro activity with any of the potential pterin substrates and their inability to complement the A. tumefaciens ΔpruA mutant (Fig. S3). These putative SDRs possess an N-terminal TGX₃GXG motif, as opposed to the TGX₃RXG motif characteristic of pteridine reductases (Fig. S2).

The major characterized SDR pteridine reductases investigated before this work were PTR1, QDPR, FolM, and PTR2. PTR2 from Trypanosoma cruzi differs from the T. cruzi PTR1 in only nine amino acid residues. Interestingly, PTR2 does not exhibit pteridine reductase activity with oxidized pterins (folate and BPt) but shows roughly equivalent specific activities for H₂F and H₂BPt that are significantly higher than the specific activities of T. cruzi PTR1 for the same substrates (29, 30). Therefore, similarly to the bacterial pteridine reductases studied here as well as FolM (9, 11), PTR2 does not reduce fully oxidized pterins, despite the fact that it possesses the established catalytic triad residues known to be involved in folate and BPt reduction in the PTR1 reaction. Interestingly, the amino acid residues that are different in PTR2 and PTR1 do not correspond to any catalytic or substrate/cofactor binding residues, and the crystal structure of T. cruzi PTR2 shows that these residues are mostly found on the surface and/or in loop regions (31).

Pteridine reductase catalysis and structure.

The highly conserved catalytic motif YX₃K is a defining characteristic of SDR enzymes, and the importance of both the tyrosine and the lysine in this motif has been demonstrated in several SDR members, including PTR1 (19). The tyrosine acts as the proton donor after hydride transfer, and the lysine residue presumably increases the acidity of the tyrosine through electrostatic interactions (5, 14) (Fig. S6A). A third residue completes a catalytic triad in SDR enzymes, which is a serine in most cases (5) but an aspartate in the case of PTR1 (Asp 181 in the L. major PTR1). This aspartate acts as a key hydrogen bond donor to the catalytic tyrosine (22) (Fig. S6A). Although PruA and other characterized pteridine reductases possess the complete catalytic triad of PTR1 (Fig. S2), interestingly, they do not catalyze the reduction of fully oxidized pterins (Table 3). For the reduction of dihydropterins to the tetrahydro form (Fig. 1A), the catalytic triad is not involved, and instead, the proton donor is proposed to be a water molecule (Fig. S6B).

TABLE 3.

Comparison of the relative activities of select characterized pteridine reductases from the short-chain dehydrogenase family

Enzyme (source)	Relative activity
	H2MPt	H2NPt	H2Fol	H2BPt	BPt	Folate
FolM (Escherichia coli)a	+++	−	++	+	−	−
PTR1 (Leishmania major)b	NDd	ND	+	+	+++	++
PTR2 (Trypanosoma cruzi)c	ND	ND	+++	++	−	−
PruA (Agrobacterium tumefaciens)	++	++	−	+++	−	−
PA3437 (Pseudomonas aeruginosa)	++	++	+++	++	−	−

^aFolM data are from Giladi et al. (11) and Pribat et al. (9).

^bPTR1 data are from Nare et al. (10).

^cPTR2 data are from Senkovich et al. (29).

^dND, not determined.

SDR enzymes are comprised of single α/β domain, based on the Rossman fold, with 6 or 7 parallel β-sheets sandwiched between 3 or 4 α-helices on each side (5). The quaternary structures of these enzymes vary, with homodimers and homotetramers being the most common, although cooperativity is generally not observed (5). We showed here that AtPruA exists as a dimer (Fig. S1), different from the archetypal pteridine reductase, PTR1, which is tetrameric in solution as well as in the crystal structure (22, 26, 32). The crystal structure of the PruA homolog from Brucella suis has been determined (33), and a homology model of AtPruA using this Brucella structure aligns well with the PTR1 structure (Fig. S7). Importantly, the positions of the catalytic triad residues in the B. suis PruA and AtPruA structures nearly perfectly overlay with the corresponding PTR1 residues (Fig. S7B and C). The AtPruA homology model places the Tyr 161 outside the active site (Fig. S7B), supporting the results of our mutagenesis studies that indicated that this residue is not involved in catalysis (Table 2). The overall conservation of the position of the catalytic triad residues in the PruA active site (Tyr 163, Lys 167, and Asp 150) further highlights the conundrum associated with the lack of activity observed for most pteridine reductases with fully oxidized pterins. The distinct substrate specificities of pteridine reductases (Table 3) provide an interesting system of structure-function relationships that will require more detailed structural and mechanistic investigation in the future.

All pteridine reductases studied to date, except for QDPR, which acts upon the quinonoid form of its pterin substrate, contain an arginine in position 6 of the N-terminal cofactor binding motif (TGX₃RXG), whereas other SDR family members have a glycine in this position. The importance of this arginine is supported by our mutagenesis studies with PruA (Table 2), which show a marked decrease in catalytic efficiency for both variants tested, especially PruA R24G. Thus, our data are consistent with the arginine at position 6 in the TGX₃RXG motif being a defining characteristic of pteridine reductases. This arginine plays a critical role as a contact residue to both the cofactor and the pterin substrate (Fig. S7C) and is proposed to participate in catalysis through stabilization and orientation of the enol intermediate (22) (Fig. S6B). It is important to note that in the PruA sequence, there is an additional arginine (underlined) at position 5 in the TGX₃RXG motif (TGAARRLG) (Fig. 4 and Fig. S2). In many SDR family members, this residue confers cofactor specificity, where a basic residue (R or K) promotes the binding of NADP(H) and a nonbasic residue results in a preference for NAD(H) (15). However, the PTR1 structure revealed that, although there is a basic residue in this position (Lys 16), it does not interact with the 2′-phosphate of NADPH, and instead, the phosphate binding pocket is achieved through backbone interactions (22). Our in silico analysis revealed that this cofactor binding mode is conserved in both B. suis PruA and A. tumefaciens PruA, where the side chain of the basic residue at position 5 in the cofactor binding motif does not interact with NADP(H).

Role of PruA catalytic activity and surface attachment in A. tumefaciens.

Our prior work revealed that PruA catalytic activity is required to bias the activity of dual-function DcpA (diguanylate cyclase/phosphodiesterase) toward the phosphodiesterase dominant state (16). The loss of pruA leads to the switching of DcpA to a strongly diguanylate cyclase-biased state, thus resulting in increased levels of cdGMP. We posit that it is the pteridine reductase activity of PruA that is important and, thus, that the production of the tetrahydropterin species is the relevant activity in this pathway. Given that A. tumefaciens produces a unique MPt metabolite (2′-O-methyl-MPt) and that the loss of PruA abolishes production of this product (16), the simplest model is that H₄MPt and/or 2′-O-methyl-H₄MPt (Fig. 1C) is active for DcpA control. The fact that the PruR protein, a divergent homolog of the SUOX molybdopterin cofactor (MoCo)-binding protein family, is required for this PruA-dependent control is also entirely consistent with a model in which the H₄MPt product is the species active in maintaining the DcpA protein in its phosphodiesterase-dominant state, keeping cdGMP levels low through PruR, and preventing attachment under inappropriate conditions (16). Modulation of this pterin-dependent control of DcpA could therefore promote or inhibit biofilm formation.

Mutations which compromise the catalytic activity of PruA have impacts on DcpA-regulated activities, such as polysaccharide production and biofilm formation. In most cases, the findings of the catalytic profiling of PruA mutants and PruA homologs from other bacteria were entirely consistent with their control over these phenotypes in vivo. PruA enzymes with diminished activity in vitro were less able to regulate these A. tumefaciens phenotypes, and those proficient in vitro were also proficient in vivo. This relationship was, however, not completely consistent for the mutations in the N-terminal motif (TGX₃RXG), where the R24A and R24G enzymes were highly compromised for in vitro activity but were able to partially complement the A. tumefaciens ΔpruA mutant (Table 2 and Fig. 4). Since the variants in this study were highly expressed from a plasmid, it is likely that the PruA R24A and R24G variants retain enough residual activity in vivo to produce the observed phenotypes. The elevated biofilm phenotype of the ΔpruA null mutant was largely corrected by expression of either the PruA R24A mutant or the PruA R24G mutant, consistent with sufficient enzymatic activity to impart UPP control through DcpA. Their enzymatic deficiency was, however, more observable for the Congo red phenotype, which reflects multiple cdGMP-regulated polysaccharides, including UPP and cellulose, with the R24G allele manifesting a more dramatic defect than the R24A allele. Control pathways for cdGMP in bacteria have been described as complex, multimodal networks with differential regulation of various target functions (34). The divergences in the attachment and Congo red phenotypes that we observed when the partially deficient R24 mutant alleles were expressed are likely due to the differential sensitivity for the production of specific polysaccharides (as well as other regulated functions) to the various intracellular levels of cdGMP. There clearly remains much to be learned about how the levels and metabolic flux of the appropriate pterins differentially impact cdGMP levels and, in turn, how these levels synergize to control output phenotypes in A. tumefaciens. The physiological relevance of this pterin-dependent control is also under investigation.

MATERIALS AND METHODS

Chemicals.

Biopterin, 7,8-l-dihydrobiopterin, folate, and dihydrofolate were obtained from Sigma-Aldrich. 7,8-l-Dihydromonapterin and 7,8-d-dihydroneopterin were prepared from their fully oxidized counterparts via zinc reduction in sodium hydroxide (35). All other pterins used in this study were obtained from Schircks Laboratories (Bauma, Switzerland). The concentrations of all pterin stock solutions were determined by measuring their absorbance at pH 1.0 and using the reported extinction coefficients (36).

Cloning and site-directed mutagenesis.

To construct the PruA purification plasmid, the coding sequence was amplified and then ligated into pET15b, which had been digested with NdeI and XbaI. Plasmids were then transformed into chemically competent E. coli DH5α cells and plated on lysogeny broth (LB) plates containing 100-μg/ml ampicillin. Construction was confirmed through genetic sequencing. For the PruA homologs from B. abortus, E. coli, P. aeruginosa, and V. cholerae, the coding sequence of the respective gene was amplified and ligated into pET15b that had been digested with NdeI and XbaI via isothermal assembly using an NEBuilder HiFi DNA assembly kit (New England Biolabs) following the supplier’s instructions. Site-directed mutagenesis of the active site of pruA was achieved using a QuikChange II site-directed mutagenesis kit (Stratagene) following the manufacturer’s instructions, with the wild-type plasmid being used as the template. Mutations were confirmed via sequencing (ACGT, Wheeling, IL). Site-directed mutagenesis of pruA R24G/A was done using a Q5 high-fidelity site-directed mutagenesis kit (New England Biolabs) following the protocol described in the instructions. Mutated plasmids were transformed into chemically competent E. coli DH5α cells and plated on LB plates containing 100-μg/ml ampicillin. Mutations were confirmed through genetic sequencing (ACGT, Wheeling, IL). To assess the complementation ability of PruA, its active-site mutations, and the homologs from B. abortus and E. coli, the coding sequence was amplified and ligated into pSRKGm that had been digested with NdeI and XbaI. The P. aeruginosa and V. cholerae homologs were constructed using Gibson assembly, as described above. The plasmids were then transformed into chemically competent E. coli DH5α cells and plated on LB plates containing 25 μg/ml gentamicin. Following sequence verification (ACGT, Wheeling, IL), the plasmids were transformed into electrocompetent A. tumefaciens cells and plated onto ATGN solid medium (AT minimal medium [37] supplemented with 0.5% [wt/vol] glucose and 15 mM ammonium sulfate) containing 300-μg/ml gentamicin.

Protein expression and purification.

PruA, its variants, and homologs from P. aeruginosa (PA3437), B. abortus (BAB_0721), and V. cholerae (VC_A0301 and VC_A0691) were all expressed and purified in the same manner, as follows. The pET15b plasmid, harboring the respective pteridine reductase with an N-terminal hexahistidine tag, was transformed into chemically competent E. coli BL21(DE3) cells. One colony from the plated culture was used to inoculate 5 ml LB supplemented with 50-μg/ml ampicillin and grown overnight at 37°C with shaking at 215 rpm. The entire overnight culture was used to inoculate 0.5 liter LB supplemented with 50-μg/ml ampicillin and then incubated at 37°C with shaking at 215 rpm until the optical density at 600 nm (OD₆₀₀) reached 0.7, at which point protein expression was induced with 1 mM IPTG and the cells were incubated for an additional 4 h. Cells were harvested by centrifugation for 20 min at 6,000 rpm and stored at −20°C.

For purification, the cell pellet (∼5 g from 2 liters of culture) was resuspended in 15 ml buffer A (50 mM NaH₂PO₄·H₂O, 300 mM NaCl, 25 mM imidazole, 5 mM 2-mercaptoethanol, pH 8.0). The cells were lysed by sonication on ice, followed by centrifugation at 15,000 rpm for 30 min (4°C). The supernatant was filtered and loaded onto a 5-ml HisTrap FF column (GE Healthcare) connected to an Äkta Start fast-performance liquid chromatograph that had been equilibrated with buffer A. The fast-performance liquid chromatography method consisted of a 5-column-volume buffer A wash, followed by a 10-column-volume linear gradient to 100% buffer B (50 mM NaH₂PO₄·H₂O, 300 mM NaCl, 500 mM imidazole, pH 8.0) at a constant flow rate of 5 ml/min. The fractions containing the protein of interest (typically eluting at ∼50% buffer B) were combined and concentrated with an Amicon centrifugal filter (10,000-molecular-weight cutoff; 15 ml; EMD Millipore) to 2.5 ml. The sample was then applied to a PD-10 desalting column with Sephadex G-25 M resin (GE Healthcare) for buffer exchange into 50 mM NaH₂PO₄, pH 8.0. The protein was supplemented with 10% glycerol before being flash frozen and stored at −80°C in 500-μl aliquots for use in subsequent kinetic assays. Protein concentrations were determined using the Bradford method with bovine serum albumin as a standard (38). We typically obtained ∼15 mg of purified PruA per liter of E. coli culture.

Enzyme assays.

The enzymatic activities of the pteridine reductases were determined spectrophotometrically by measuring NADPH depletion in the presence of various concentrations of pterin substrates. Assays (reaction volume, 200 μl) were performed at room temperature in reaction buffer (100 mM K₂HPO₄, 10 mM 2-mercaptoethanol, pH 6.0), and the assay mixtures contained 5 μg purified protein (except with H₂BPt as a substrate, which contained 2.5 μg wild-type PruA), 0.1 mM NADPH, and a pterin substrate (25 μM to 400 μM). For pH dependence experiments, the following buffers (100 mM) were used: NaHCO₂ (pH 4.0), NH₄CH₃CO₂ (pH 5.0), K₂HPO₄ (pH 6.0 and 7.0), Tris (pH 8.0 and 9.0), and glycine (pH 10.0).

The reactions were initiated with the addition of the respective pterin substrate and proceeded for 30 s before quenching by adding 800 μl 100 mM glycine, pH 10.0 (PruA is inactive at pH 10) (Fig. 2). The values of the absorbance at 340 nm (A₃₄₀) of the resulting 1-ml samples were measured in a 1-cm-path-length quartz cuvette. The assays were run in triplicate (all reactions were from the same protein purification) and included two control assay mixtures lacking enzyme to correct for the nonenzymatic breakdown of NADPH. The specific activities of the enzymes were determined from the control corrected A₃₄₀ values using the extinction coefficients for the coupled oxidation/reduction of NADPH and the pterin substrate determined at pH 10.0: H₂MPt and H₂NPt, 9,500 M⁻¹cm⁻¹; H₂BPt, 9,150 M⁻¹cm⁻¹; and H₂Fol, 11,050 M⁻¹cm⁻¹. The specific activities of each triplicate, associated with the control A₃₄₀ average, were fitted to Michaelis-Menten curves using Prism (version 8.0) software.

Congo red staining assay.

Early-stationary-phase cultures of A. tumefaciens were spotted (volume, 2 μl) onto ATGN plates supplemented with 100 μg/ml Congo red with either no additions or 400 μM IPTG to induce plasmid expression. The plates were incubated for 48 h at 30°C prior to photographing. All of the colony images were taken from a single plate, and the entire image was processed uniformly prior to cropping out single colonies to be shown in Fig. 4.

Biofilm formation quantification.

Polyvinyl chloride (PVC) coverslips were placed upright in a 12-well tissue culture plate (Corning, Inc.) and sterilized with UV light. Mid-exponential-phase A. tumefaciens cultures were subcultured to a starting OD₆₀₀ of 0.05 in ATGN supplemented with 22 μM iron sulfate and 400 μM IPTG to induce plasmid expression and incubated statically at room temperature for 48 h. To quantify cellular attachment, coverslips were washed with water and stained with 0.1% (wt/vol) crystal violet (CV). To measure adhered cells, the CV was solubilized in 1 ml of 33% acetic acid, and the absorbance at 600 nm (A₆₀₀) was measured. Readings were normalized against the OD₆₀₀ of the planktonic cells remaining in the well.

In silico structural analysis.

AtPruA homology models were generated using both the I-TASSER (39) and Phyre2 (40) programs. Model validation was performed using the SWISS-MODEL structure assessment tool (41), which revealed that the Phyre2-generated model had the overall highest quality for further analysis. Structural analysis, alignment, and image generation (see Fig. S7 in the supplemental material) were carried out using PyMOL (version 2.3.2) software.