A dedicated diribonucleotidase resolves a key bottleneck for the terminal step of RNA degradation.

Degradation of RNA polymers, an ubiquitous process in all cells, is catalyzed by specific subsets of endo- and exoribonucleases that together recycle RNA fragments into nucleotide monophosphate. In γ-proteobacteria, 3-'5' exoribonucleases comprise up to eight distinct enzymes. Among them, Oligoribonuclease (Orn) is unique as its activity is required for clearing short RNA fragments, which is important for cellular fitness. However, the molecular basis of Orn's unique cellular function remained unclear. Here, we show that Orn exhibits exquisite substrate preference for diribonucleotides. Crystal structures of substrate-bound Orn reveal an active site optimized for diribonucleotides. While other cellular RNases process oligoribonucleotides down to diribonucleotide entities, Orn is the one and only diribonucleotidase that completes the terminal step of RNA degradation. Together, our studies indicate RNA degradation as a step-wise process with a dedicated enzyme for the clearance of a specific intermediate pool, diribonucleotides, that affects cellular physiology and viability.

Introduction

Degradation of RNA is initiated by endonuclease-catalyzed cleavages; the resulting oligoribonucleotide fragments are hydrolyzed to completion by a mixture of exoribonucleases for the maintenance of cellular nucleotide pools (Hui et al., 2014). Unlike the conserved machineries for the synthesis of macromolecules, distinct sets of RNases are used by different organisms to degrade these oligoribonucleotides. Escherichia coli genomes encode eight 3’−5’ exoribonucleases, namely polynucleotide phosphorylase, RNase II, D, BN, T, PH, R and oligoribonuclease (Orn) (Hui et al., 2014; Bandyra and Luisi, 2013). A subset of these enzymes recognizes structural features of the RNA substrate, such as in tRNA, while others act on unstructured polymers (Li and Deutscher, 1996).

As an exoribonuclease, Orn is unique for two reasons. First, orn is required for viability in many γ-proteobacteria, including E. coli (Ghosh and Deutscher, 1999) and other organisms (Palace et al., 2014; Kamp et al., 2013), unlike all other known 3’−5’ exoribonucleases. This indicates that not all exoribonucleases have redundant functions despite acting on the 3’ end of RNA substrates and in several cases, including RNase R and RNase II, sharing an activity toward oligoribonucleotides (Zuo and Deutscher, 2002; Cheng and Deutscher, 2002; Cheng and Deutscher, 2005; Frazão et al., 2006). Hence, Orn appears to catalyze a particularly important step in RNA turnover. Second, Orn is a key enzyme in bacterial cyclic-di-GMP (c-di-GMP) signaling. The nucleotide second messenger c-di-GMP is produced in bacteria in response to environmental cues and controls a wide range of cellular pathways, including cell adhesion, biofilm formation, and virulence (Krasteva and Sondermann, 2017; Römling et al., 2013; Hengge, 2009; Jenal et al., 2017; Hall and Lee, 2018). Since the discovery of c-di-GMP over 30 years ago, it has been known that the signal is degraded by a two-step process with a linear pGpG diribonucleotide intermediate (Ross et al., 1987). While the enzyme for linearizing c-di-GMP to pGpG was discovered early on (Tal et al., 1998; Galperin et al., 1999), the identity of the enzyme that degrades pGpG remained elusive. Two recent studies showed that Orn is the primary enzyme that degrades pGpG in Pseudomonas aeruginosa (Cohen et al., 2015; Orr et al., 2015). In an orn deletion strain, the accrual of linear pGpG has a profound effect on cells. Specifically, the increase in pGpG inhibits upstream phosphodiesterases that degrade c-di-GMP, thereby triggering phenotypes associated with high cellular c-di-GMP levels. However, the molecular basis of Orn’s unique cellular functions in γ-proteobacteria that distinguishes it from all other exoribonucleases remains unexplained.

Since its discovery over 50 years ago (Stevens and Niyogi, 1967), Orn has been presumed to degrade oligoribonucleotides (Niyogi and Datta, 1975; Datta and Niyogi, 1975; Yu and Deutscher, 1995). This notion largely derived from assays utilizing two types of substrates. In one series of experiments, 3H polyuridine (poly(U)) was incubated with Orn, other enzymes, or lysates and analyzed by paper chromatography, which offers limited resolution overall (Niyogi and Datta, 1975; Datta and Niyogi, 1975; Yu and Deutscher, 1995). In a second set of experiments, Orn was incubated with oligoribonucleotides that had been tagged at their 5’ terminus by a large fluorophore. The products of this reaction were resolved by denaturing polyacrylamide gel electrophoresis, thereby allowing for detection of products with one or more nucleotides removed from the 3’ terminus (Cohen et al., 2015; Mechold et al., 2006; Mechold et al., 2007; Fang et al., 2009; Liu et al., 2012). In both instances, it was concluded that Orn could processively degrade ‘short’ oligoribonucleotides. Yet, it was not clear how Orn might selectively target ‘short’ RNAs, rather than simply binding to the penultimate sequence at the 3’ termini of single-stranded RNA of any length.

Here, we sought to rigorously examine Orn’s substrate preferences, revealing its unique properties. To that end, we incubated Orn with 5’ 32P end-labeled RNAs of varying lengths and analyzed the products of that reaction over time. Our data show that Orn exhibits a surprisingly narrow substrate preference for diribonucleotides. This finding is in stark contrast to the previous studies that described a broader substrate length range and that originally gave oligoribonuclease its name. We sought to understand this remarkable substrate selectivity of Orn by determining the crystal structures of Orn in complex with pGpG and other linear diribonucleotides. These data reveal a structural basis for the diribonucleotide preference and identify key residues for recognizing diribonucleotides. Furthermore, we find that Orn is the only diribonuclease in P. aeruginosa. Additionally, we show that other diribonucleotides, in addition to pGpG, can affect bacterial physiology. From this we propose a general model of RNA degradation, wherein a combination of exoribonucleases process oligoribonucleotides down to diribonucleotides and Orn completes RNA recycling by cleaving diribonucleotides to nucleoside monophosphates. In this way, Orn occupies an important but discrete step in the overall RNA degradation pathway.

Results and discussion

Orn functions as a diribonuclease in vitro

To understand the length preference of Orn, recombinant affinity-tagged Vibrio cholerae Orn (Orn) was purified and tested biochemically. We used an established ligand-binding assay (Roelofs et al., 2011) to determine the relative substrate affinities of Orn to 5'-radiolabeled oligoribonucleotides that ranged from two to seven nucleotides in length. Unlike fluorophore-labeled RNA used in several earlier studies (Cohen et al., 2015; Mechold et al., 2006; Mechold et al., 2007; Fang et al., 2009; Liu et al., 2012), radiolabeling with 32P ensures that the substrate structure is unperturbed compared to native ligands. The assays were also performed in buffer lacking divalent cations to prevent complication due to enzyme activity. When spotted on nitrocellulose membranes, protein and protein-substrate complexes are sequestered at the point of application to the membrane, whereas unbound substrate diffuses due to radial capillary action (Roelofs et al., 2011). By this assay, quantification of the diffusion zone reveals that Orn exhibits the highest affinity for diribonucleotide (K pGpG = 90 -/+9 nM), as compared to oligoribonucleotides of greater lengths (Table 1) (Orr et al., 2015). The observed affinity of Orn with the purification tag removed (K pGpG = 80 -/+9 nM) was similar indicating the tags did not alter the interactions (Figure 3—figure supplement 1). Increase in the length to three or four residues reduces the affinity 7- or 10-fold, respectively (Table 1). Substrates with five or more bases show a greater than 28-fold reduction in affinity compared to diribonucleotides. These results suggest that Orn has a strong preference for diribonucleotides over longer oligonucleotides.

Table 1.

Quantitative measurement of length-dependent oligoribonucleotide affinities.

Substratea	Dissociation constant Kd (nM)b
GG	90 -/+9
AGG	630 -/+80
AAGG	890 -/+90
AAAGG	2,560 -/+170
AAAAGG	3,830 -/+370
AAAAAGG	3,750 -/+530

a: RNAs were labeled with 32P at their 5’ end. b: Affinities were measured using an established binding assay for radiolabeled RNAs (Roelofs et al., 2011; Patel et al., 2014).

Figure 3—figure supplement 1.

Contribution of catalytic residues and His10-MBP tag to binding interaction between Orn and pGpG.

The dissociation constants (K) of indicated purified (A) His10-MBP-Orn variants and (B) untagged Orn to pGpG were measured by DRaCALA. Fraction bound from data was plotted and determined by nonlinear regression. All data shown represent average and SD of three independent experiments.

To understand whether the affinity preference reflects nuclease activity with natural substrates that are unmodified at the 5’ end, we incubated Orn with 5'-32P-radiolabeled oligoribonucleotides of varying lengths in the presence of divalent cations that support catalysis. The products of these reactions were resolved by urea-denaturing 20% polyacrylamide gel electrophoresis (PAGE). Under these electrophoresis conditions, the mononucleotides and oligonucleotides that were tested (between 2 and 7 nucleotides in length) can be resolved. The diribonucleotide substrate in this experiment was pGpG (GG), whereas the longer oligoribonucleotides included an increasing number of adenine nucleotides at the 5’ end. This arrangement ensured that the same GG sequence was maintained at the 3’ end while also avoiding stable G quadruplex formation that may be observed with RNAs containing stretches of poly-G (Kwok and Merrick, 2017). At substrate concentrations that far exceed enzyme concentration (200:1), the diribonucleotide substrate was already fully processed to nucleoside monophosphates by 30 min (Figure 1A) (Orr et al., 2015). In contrast, longer substrates (i.e. from 3-mers to 7-mers) were not processed at their 3' end, even at 30 min (Figure 1A and B). These results indicate that the length preference for exoribonuclease activity of Orn on diribonucleotides is even greater than mere differences in binding affinities. This strong substrate preference stands in stark contrast to previous studies arguing Orn acts as a general exoribonuclease that cleaves oligoribonucleotides with two to seven residues in length (Cohen et al., 2015; Niyogi and Datta, 1975; Datta and Niyogi, 1975; Mechold et al., 2006; Mechold et al., 2007).

Figure 1.

Orn has a stark preference for diribonucleotide cleavage in vitro.

RNA nucleotides two to seven residues in length (1 μM, containing the corresponding 32P-labeled RNA tracer) were each subjected to cleavage over time with 5 nM (A, B) or 1000 nM Orn (C, D). Aliquots of each reaction were stopped at indicated times (min), and assessed by denaturing 20% PAGE (A, C). Quantification of the intensities of bands corresponding to the amount of uncleaved initial oligonucleotide over time are plotted as the average and SD of three independent experiments (B, D).

Degradation of 32P-AAAAAGG by purified indicated His10-MBP-Orn variants with 5 nM (A) or 1 µM (B). Samples were stopped at indicated time (min) and analyzed by denaturing 20% PAGE. Representative gel images of two independent assays are shown.

To determine if Orn can indeed cleave substrates longer than a diribonucleotide, the enzyme was incubated with RNA substrates at a 1:1 molar ratio. Under these conditions, the diribonucleotide substrate was completely processed to nucleoside monophosphates by the earliest time point, 20 s (Figure 1C and D). Orn also facilitated the degradation of the longer RNA substrates, but only after significantly longer incubation times. For example, it required 10 min and 30 min to fully degrade 3-mer and 4-mer RNAs, respectively (Figure 1C and D). Cleavage was reduced further for longer RNAs; only 40–53% of the 5-mer, 6-mer and 7-mer RNAs were processed to nucleoside monophosphates at 30 min (Figure 1C and D), correlating with the weak affinities determined for these substrates (Table 1). Nonetheless, cleavage of the 7-mer required enzymatic activity since catalytically inactive Orn variants were unable to degrade the 7-mer (Figure 1—figure supplement 1). Of note, we observed a non-uniform distribution of degradation products for the longer RNA substrates. Specifically, the diribonucleotide intermediate was never observed as a reaction intermediate for the longer RNAs. This reaction pattern indicates that RNAs of more than two residues could accumulate, but that diribonucleotide RNAs were always rapidly processed to nucleoside monophosphates. Together, these results indicate that Orn exhibits a strong substrate preference for diribonucleotides over longer oligoribonucleotides – far greater than reported previously (Cohen et al., 2015; Mechold et al., 2006). While assays here were performed at near-physiological ionic strength, prior studies often utilized buffer lacking salt (NaCl/KCl), which may at least in part contribute to the apparent differences in catalytic activity (Niyogi and Datta, 1975; Ghosh and Deutscher, 1999; Cohen et al., 2015; Mechold et al., 2006).

Figure 1—figure supplement 1.

Catalytic residues of Orn are required for degradation for 32P-AAAAAGG in vitro.

Degradation of 32P-AAAAAGG by purified indicated His10-MBP-Orn variants with 5 nM (A) or 1 µM (B). Samples were stopped at indicated time (min) and analyzed by denaturing 20% PAGE. Representative gel images of two independent assays are shown.

Structure of Orn complexes with pGpG and other linear diribonucleotides

To elucidate the molecular basis for these unique properties of Orn, we set out to gain a deeper understanding of the enzyme’s substrate specificity by determining Orn/substrate co-crystal structures. We initially determined the crystal structure of P. aeruginosa Orn (Orn) with diribonucleotide substrate; however, crystal packing contacts prevented substrate binding (data not shown). Instead, we were able to crystallize two representative homologs, Orn (Kamp et al., 2013) and the human REXO2 (also known as small fragment nuclease or Sfn [Bruni et al., 2013]), bound to the diribonucleotide pGpG (Figure 2A and B; Figure 2—source data 1). Both proteins purified free of divalent cation, which was sufficient to prevent catalysis in crystals. Superimposing the two structures indicates their identical fold (rmsd of 0.75 Å for the protomer) (Figure 2—figure supplement 1A), which is preserved in the substrate-free Orn structure determined previously (rmsd of 0.54 Å for the protomer) (Figure 2—figure supplement 1A) (Chin et al., 2006). The pGpG-bound structures reveal a narrow active site that is lined by the conserved acidic residues of the signature DEDD motif (D12, E14, and D112 of Orn; D15, E17, and D115 of REXO2) and the general base H158 or H162 in Orn or REXO2, respectively (Figure 2A and B and Figure 2—figure supplement 1B and Figure 2—source data 2). In Orn, the bases of the diribonucleotide buttress against aromatic residues W61 and Y129, the latter being contributed from the second half-side of the dimeric enzyme. The corresponding residues, W64 and Y132, are conserved in REXO2. Residue L18 in Orn (L21 in REXO2) wedges in-between the two bases. Most notably, residues S108, R130, S135 and the hydroxyl group on Y129 (S111, R133, S138, and Y132 in REXO2) form hydrogen bonds with the 5' phosphate of pGpG, capping the substrate. This phosphate cap creates a major constriction of the active site, which is not observed in structurally related 3'−5' exoribonucleases such as RNase T or ExoI (Figure 2C and Figure 2—figure supplement 1 and Figure 2—source data 2) (Hsiao et al., 2011; Korada et al., 2013; Hsiao et al., 2012). RNase T and ExoI accommodate longer RNA substrates, facilitated by an expansive active site. The structural analysis correlates closely with sequence conservation of the phosphate cap motif, which is strict in Orn homologs but divergent in RNase T proteins (Figure 2D and Figure 2—source data 2). Our structural analysis also suggests that modifications at the bases may be tolerated consistent with the observation that all native bases can be accommodated at the active site (Figure 2—figure supplement 2). Modifications at the sugar or termini beyond a terminal 5’-phosphate will likely fit sub-optimally if at all.

Figure 2.

Structures reveal Orn’s conserved substrate preference for diribonucleotides.

Crystal structures of pGpG-bound V. cholerae Orn (A) and human REXO2 (B) are shown in comparison to E. coli RNase T bound to substrate (PDB 3nh1; Hsiao et al., 2011) (C), another DnaQ-fold 3’−5’ exoribonuclease with a DEDD(h) active site motif. The top panels show ribbon representations of the dimeric enzymes. The insets are surface representations of the enzymes’ active sites shown in similar orientations. The bottom panel describes the active site residues involved in RNA binding and catalysis. Residue numbering for REXO2 refers to its cytosolic isoform lacking the mitochondria-targeting pre-sequence. The sequence logos in (D) were constructed based on multi-sequence alignments of Orn and RNase T orthologs. Overall sequence identity ranges from 43 to 70% for Orn and 46 to 69% for RNase T. Sequence identifiers are provided in Figure 2—source data 2. Sequence logos were plotted using WebLogo (Crooks et al., 2004). Conserved residues of the active site’s DEDD motif (‘act.’; red), for ribonucleotide base binding (‘nuc.’; gray), and of the phosphate cap (‘P-cap’; purple) are highlighted.

(A) Superposition of Orn with REXO2 (left) and the substrate-free state of Orn from Xanthomonas campestris (PDB 2gbz; Chin et al., 2006) (right). Rmsd values are reported for the alignment of a monomer. (B) Surface properties of Orn enzymes. An alignment of Orn orthologs was used to map conservation scores onto the solvent accessible surface of Orn (left). High degree of conservation overlaps with the acidic active site observed in the electrostatic potential map calculated with the APBS software (Baker et al., 2001) and based on Orn (right). (C) Active-site view of E. coli RNase T bound to dAAC (PDB 3v9z; Hsiao et al., 2012). (D) Active-site view of E. coli Exo I bound to dT13 (PDB 4jrp; Korada et al., 2013).

(A) Crystal structures of REXO2 bound to pApA and pApG. The structural overview (left) shows a C-terminal helix that is unique to Orn orthologs in higher eukaryotes and in the current crystal forms is stabilized by crystal packing contacts. (B) Crystal structures of Orn bound to di-purine (pApA, pApG, pGpA), purine-pyrimidine (pGpC, pCpG), and di-pyrimidine (pCpU) substrates. Views are identical to those shown in Figure 2 for the REXO2:pGpG and Orn:pGpG complexes.

Figure 2—figure supplement 1.

Structural comparison of Orn orthologs, RNase T and Exo I.

(A) Superposition of Orn with REXO2 (left) and the substrate-free state of Orn from Xanthomonas campestris (PDB 2gbz; Chin et al., 2006) (right). Rmsd values are reported for the alignment of a monomer. (B) Surface properties of Orn enzymes. An alignment of Orn orthologs was used to map conservation scores onto the solvent accessible surface of Orn (left). High degree of conservation overlaps with the acidic active site observed in the electrostatic potential map calculated with the APBS software (Baker et al., 2001) and based on Orn (right). (C) Active-site view of E. coli RNase T bound to dAAC (PDB 3v9z; Hsiao et al., 2012). (D) Active-site view of E. coli Exo I bound to dT13 (PDB 4jrp; Korada et al., 2013).

Figure 2—figure supplement 2.

Structural comparison of Orn and REXO2 bound to diverse diribonucleotides.

(A) Crystal structures of REXO2 bound to pApA and pApG. The structural overview (left) shows a C-terminal helix that is unique to Orn orthologs in higher eukaryotes and in the current crystal forms is stabilized by crystal packing contacts. (B) Crystal structures of Orn bound to di-purine (pApA, pApG, pGpA), purine-pyrimidine (pGpC, pCpG), and di-pyrimidine (pCpU) substrates. Views are identical to those shown in Figure 2 for the REXO2:pGpG and Orn:pGpG complexes.

Any attempts to obtain structures of Orn with longer RNA substrates (3–5 bases in length) only resolved diribonucleotides at the active site (data not shown), further suggesting a narrow active site that is selective for dinucleotide species. These results are consistent with two recent reports of structural data for Colwellia psychrerythraea Orn and REXO2 bound to substrates (Lee et al., 2019; Chu et al., 2019). In the case of Orn, crystallization was attempted with a 5-mer RNA; however, only two bases were resolved at the active site (Lee et al., 2019). A similar observation was made with REXO2 for RNA substrates, but a complex with a longer DNA-based ligand was obtained (Chu et al., 2019), which may not allow conclusions for RNA substrates. In the same report, RNase activity against longer RNA was observed with REXO2, but required a 20x molar excess of enzyme over RNA, likely not representing physiological conditions (Chu et al., 2019).

The phosphate cap is required for diribonuclease activity

To assess the impact on catalysis of the phosphate cap in comparison to other active site residues identified in the structural analysis, we introduced specific single-point mutations into Orn (Figure 3). All protein variants expressed stably in P. aeruginosa (Figure 3—figure supplement 2). For two representative point mutants with alanine substitutions at a phosphate-cap or active-site/DEDD-motif residue, Orn-R130A or Orn-D12A, respectively, we showed that the purified proteins form stable dimers in solution, comparable to wild-type Orn (Figure 3—figure supplement 3). Purified protein variants were evaluated for pGpG degradation. Mutations of the central phosphate cap residue R130 to alanine led to complete loss of catalytic activity, comparable to mutants in the DEDD active site motif (Figure 3A and Figure 3—figure supplement 4) (Orr et al., 2015). These protein variants were similarly unable to cleave the 7-mer (Figure 1—figure supplement 1). We next asked whether Orn with specific point mutations could complement the deletion of orn in P. aeruginosa. P. aeruginosa ∆orn accumulates pGpG that in turn inhibits c-di-GMP-specific phosphodiesterases (Cohen et al., 2015; Orr et al., 2015). As a net result, c-di-GMP accumulates in these cells, an effect that is associated with a hyper-biofilm and cell aggregation phenotype. While expression of wild-type Orn complements the P. aeruginosa ∆orn resulting in a dispersed culture, complementation with variants that carry mutations in either the active site or phosphate cap residues results in cell aggregation indistinguishable from the ∆orn phenotype (Figure 3B). Together, these experiments demonstrate that an intact phosphate cap is required for enzyme function.

Figure 3.

Functional importance of active-site and phosphate-cap residues for Orn function.

(A) In vitro enzyme activity. Degradation of 32P-pGpG (1 μM) by purified wild-type Orn or variants with alanine substitutions (5 nM) at the indicated sites was assessed. Samples were stopped at the indicated times (min) and analyzed by denaturing 20% PAGE. Representative gel images are shown with the indicated RNA size. Figure 3—figure supplement 4 shows the graphs of the means and SD of three independent experiments. (B) In vivo activity of alanine substituted orn alleles to complement P. aeruginosa Δorn. Overnight cultures of the indicated strains were allowed to stand for 10 min without agitation to allow bacterial aggregates to sediment. Representative images of the cultures of triplicated assays are shown.

The dissociation constants (K) of indicated purified (A) His10-MBP-Orn variants and (B) untagged Orn to pGpG were measured by DRaCALA. Fraction bound from data was plotted and determined by nonlinear regression. All data shown represent average and SD of three independent experiments.

Proteins in whole cell lysates of strains expressing the indicated HA-tagged orn allele were separated by 12% SDS-PAGE and probed with anti-HA antibodies. All images were performed in duplicate independently. A representative image of two independent experiments is shown.

(A) Absolute molecular weights (black data points across elution peaks are plotted on the right axis; theoretical monomer and dimer molecular weights, horizontal dashed lines) of wild-type Orn and point mutants were determined using SEC-MALS (90°-light scattering: colored, solid lines; refractive index signal: black, dashed lines; plotted on the left axis).

1 µM pGpG was treated with 5 nM of the indicated Orn protein and sampled at indicated time points. The graphs show quantification of three independent data sets, of which representative radiographs are shown in Figure 3A.

Figure 3—figure supplement 2.

Immunoblot analysis of Orn in P. aeruginosa ∆orn expressing indicated orn alleles.

Proteins in whole cell lysates of strains expressing the indicated HA-tagged orn allele were separated by 12% SDS-PAGE and probed with anti-HA antibodies. All images were performed in duplicate independently. A representative image of two independent experiments is shown.

Figure 3—figure supplement 3.

Molecular weight determination indicates that Orn variants with point mutations at the phosphate cap and active site remain dimeric.

(A) Absolute molecular weights (black data points across elution peaks are plotted on the right axis; theoretical monomer and dimer molecular weights, horizontal dashed lines) of wild-type Orn and point mutants were determined using SEC-MALS (90°-light scattering: colored, solid lines; refractive index signal: black, dashed lines; plotted on the left axis).

Figure 3—figure supplement 4.

pGpG degradation by purified wild-type Orn or variants with indicated alanine substitutions.

1 µM pGpG was treated with 5 nM of the indicated Orn protein and sampled at indicated time points. The graphs show quantification of three independent data sets, of which representative radiographs are shown in Figure 3A.

Interaction of Orn with substrates

Structures of Orn (and REXO2) with different diribonucleotides, including di-purine, di-pyrimidine and mixed substrates, revealed identical binding poses (Figure 2—figure supplement 2). Additional hydrogen bonding between purine residues and Orn in the 3' or 5' position correlates with a small, but detectable preference of Orn for purine-containing substrates (Figure 4). Together, the structural analysis uncovered Orn’s mode of substrate binding, which is conserved from bacteria to humans and indicates a unique selection for linear diribonucleotides with a 5’ phosphate. The requirement for the 5’ phosphate was tested with GpG as a competitor for Orn binding and cleavage of pGpG. Even in excess GpG, Orn bound pGpG and cleaved pGpG similar to untreated control (Figure 4—figure supplement 1A B) supporting the importance of the 5’ phosphate cap.

Figure 4.

Orn cleaves all diribonucleotides.

(A) Orn (5 nM) was incubated with di-purine (pRpR), purine-pyrimidine (pRpY or pYpR), or di-pyrimidine (pYpY) diribonucleotides (1 μM) containing the corresponding 32P-labeled RNA tracer. Aliquots of each reaction were stopped at the indicated times (min) and assessed by denaturing 20% PAGE. (B) Quantification of the intensities of bands corresponding to the amount of diribonucleotide cleaved at the 10 min time point. Results are the average and SD of duplicate independent experiments. Orn cleaves all diribonucleotides to nucleoside monosphosphates, albeit to varying extents. Diribonucleotides consisting of two purines (pRpR) were hydrolyzed most efficiently, with over 90% of starting RNAs processed by 10 min. A majority (>75%) of diribonucleotides with a 5' purine (pRpY) were also processed by the 10 min endpoint. However, diribonucleotides with a 5' pyrimidines exhibited moderately reduced levels of cleavage. Di-pyrimidine (pYpY) substrates, and in particular pUpC and pCpC, showed the slowest turnover from the substrates tested. These results demonstrate that while all diribonucleotides are acceptable substrates for Orn, the enzyme is likely to exhibit moderate preferences for diribonucleotides that contain a 5' purine.

(A) Cleavage of 130 nM 32P-pGpG by 5 nM Orn in the absence or presence of unlabeled pAp, GpG or GMP (100 µM). (B) Fraction bound of 32P-pGpG to 200 nM Orn in presence of no competitor, 100 µM pGpG or 100 µM GpG. (C) Fraction bound of 32P-pAp to different concentration of Orn was determined by DRaCALA. K could not be measured as 32P-pAp binding to Orn was not saturated. (B) (C) All data shown represent duplicate-independent experiments.

Figure 4—figure supplement 1.

Effect of GpG and pAp on Orn activity.

(A) Cleavage of 130 nM 32P-pGpG by 5 nM Orn in the absence or presence of unlabeled pAp, GpG or GMP (100 µM). (B) Fraction bound of 32P-pGpG to 200 nM Orn in presence of no competitor, 100 µM pGpG or 100 µM GpG. (C) Fraction bound of 32P-pAp to different concentration of Orn was determined by DRaCALA. K could not be measured as 32P-pAp binding to Orn was not saturated. (B) (C) All data shown represent duplicate-independent experiments.

Another distinguishing factor between Orn and other exoribonucleases with similar activities is the apparent inhibition of Orn by 3′-phosphoadenosine 5′-phosphate (pAp), a metabolic byproduct of sulfate assimilation that accumulates upon lithium poisoning of pAp-phosphatase (Mechold et al., 2007). DHH/DHHA-type oligoribonucleases such as NrnA from Bacillus subtilis, Mycobacterium tuberculosis and Mycoplasma pneumoniae are capable of dephosphorylating the pAp mononucleotide and their activity against oligoribonucleotides is unaffected by pAp (Mechold et al., 2007; Postic et al., 2012). In contrast, pAp has been described as a competitive inhibitor for E. coli Orn and the human REXO2 (Mechold et al., 2006). However, the oligo-cytosine RNAs used in these prior studies were labeled at their 5’ end with a bulky fluorescent dye moiety (usually a cyanine or Cy-dye) (Cohen et al., 2015; Mechold et al., 2006); our aggregate data now indicate that these RNAs represent suboptimal substrates for Orn, given its stark requirement for a simple 5’ phosphate (Figure 2). This labeling strategy is therefore likely to incompletely assess native Orn activity by underestimating diribonuclease activity while also overestimating the effect of competitive inhibitors. Revisiting pAp binding to Orn, we show here that Orn does not interact with radiolabeled pAp in the binding assay that was used to quantify oligoribonucleotide interactions with Orn (Figure 4—figure supplement 1C). Furthermore, unlabeled pAp failed to competitively inhibit degradation of radiolabeled pGpG to GMP (Figure 4—figure supplement 1A).

Orn is the only diribonuclease in P. aeruginosa grown under laboratory conditions

All available literature assumes Orn is a 3’−5’ exoribonuclease that is responsible for processing of short (between 2 and 7) oligoribonucleotides. Yet our structural and biochemical data suggest that the enzyme exhibits such a striking preference for diribonucleotide substrates that the in vivo function of Orn as a general exoribonuclease should be reconsidered. Therefore, we developed experimental conditions to measure Orn’s activity in cellular extracts. The orn gene is required for viability in most γ-proteobacteria, including V. cholerae; however, it is not essential for growth of P. aeruginosa under most conditions (Cohen et al., 2015; Orr et al., 2015). Therefore, lysates were generated from P. aeruginosa strains, including parental PA14, ∆orn, and ∆orn complemented with orn. 5'-32P-radiolabeled 2-mer or 7-mer RNA was then added to each of these lysates (Figure 5). Aliquots of the mixtures were removed and analyzed by urea denaturing 20% PAGE at varying time intervals. Extracts from parental PA14 digested the entire radiolabeled diribonucleotide in less than 5 min (Figure 5A). In contrast, lysates from strains lacking Orn failed to show any signs of diribonuclease activity even at longer time points. Similarly, complementation with catalytically inactive alleles of orn (D12A, L18A and H158A) also failed to clear the diribonucleotide (Figure 5 and Figure 5—figure supplement 1). Diribonuclease activity in the lysates could be restored by ectopic expression of orn from a self-replicating plasmid or by addition of purified Orn (Figure 5A). These data confirm that cellular Orn is required for degrading diribonucleotides and no other cellular RNase of P. aeruginosa can substitute for Orn activity under these laboratory growth conditions.

Figure 5.

Orn acts as a diribonuclease in cell lysates.

Degradation of 32P-GG (A) and 32P-AAAAAGG (B) by whole cell lysates of wild-type, orn mutant, orn mutant complemented with orn or orn. For ∆orn and ∆orn complemented with orn,100 nM of purified Orn was added at 30 min time point and incubated for an additional 10 min. Samples were stopped at the indicated time and analyzed by 20% denaturing PAGE. Representative gel images of triplicated assays are shown with the indicated RNA size. Graphs show quantitation of triplicate data for indicated RNA species over time.

Degradation of 32P-AAAAAGG by whole-cell lysates of the indicated strain (PA14, ∆orn mutant complemented with pMMB-orn or pMMB-orn. For ∆orn expressing orn or orn, 100 nM of purified, wild-type Orn protein was added at the 30 min time point and reactions were stopped 10 min thereafter. Samples were stopped at indicated time and analyzed by denaturing 20% PAGE. Representative gel images of two independent experiments are shown.

Figure 5—figure supplement 1.

Catalytic residues of Orn are required for degradation for 32P-AAAAAGG in whole-cell lysates.

Degradation of 32P-AAAAAGG by whole-cell lysates of the indicated strain (PA14, ∆orn mutant complemented with pMMB-orn or pMMB-orn. For ∆orn expressing orn or orn, 100 nM of purified, wild-type Orn protein was added at the 30 min time point and reactions were stopped 10 min thereafter. Samples were stopped at indicated time and analyzed by denaturing 20% PAGE. Representative gel images of two independent experiments are shown.

When the 32P-7-mer RNA substrate was incubated in extracts from parental PA14 it was digested to a ladder of degradative intermediates including 6-mer, 5-mer, 4-mer and 3-mer RNAs (Figure 5B). Of note, while these intermediates and the final mononucleotide product accumulated over time, the 2-mer intermediate was never observed over the time course. In contrast, lysates from the ∆orn mutant specifically accrued the diribonucleotide intermediate with no apparent production of its mononucleotide products. Ectopic expression of plasmid-borne orn, but not orn D, restored the degradation of the 7-mer to mononucleotides and the diribonucleotide intermediate could no longer be observed. Furthermore, the diribonucleotide intermediate that accumulated in the ∆orn lysate was fully processed upon addition of purified Orn protein (Figure 5B). Together, these results show that P. aeruginosa accumulates a bottleneck of diribonucleotide intermediates in a ∆orn background, which is only resolved upon addition of Orn. Degradation of RNA fragments with three or more residues by Orn is negligible in a cellular context, considering that ∆orn lysates preserve nuclease activities for the processing of RNAs down to diribonucleotides. From these aggregate data, we propose that Orn functions not as an oligoribonuclease as stated in the literature but instead functions as a specialized ribonuclease of diribonucleotide substrates (i.e. a ‘diribonuclease’).

The diribonucleotide pool affects P. aeruginosa growth in addition to c-di-GMP signaling

We noticed that PA14 ∆orn had reduced growth on agar plates that is reminiscent of a small colony variant (SCV) phenotype (Figure 6) (D'Argenio et al., 2002). The SCV phenotype has been attributed previously to increased c-di-GMP levels (Hickman et al., 2005). C-di-GMP binds to FleQ and activates transcription of the pel operon (Baraquet et al., 2012; Matsuyama et al., 2016). In addition, c-di-GMP binds to the c-di-GMP-receptor PelD to increase the biosynthesis of the pel exopolysaccharide (Lee et al., 2007), which enhances cell aggregation leading to a compact SCV morphology. Previous reports had shown that P. aeruginosa ∆orn is unable to clear pGpG, which results in the elevation of c-di-GMP signaling and pel-dependent cell aggregation and biofilm mass (Cohen et al., 2015; Orr et al., 2015).

Figure 6.

Small-colony phenotype of ∆orn is independent of c-di-GMP signaling.

Bacterial cultures were diluted and dripped on LB agar plates with the indicated concentration of IPTG. After overnight incubation, representative images of the plates were taken and shown for (A) PA14 and PA14 ∆pelA harboring pMMB vector with the indicated diguanylate cyclase gene and (B) PA14 ∆orn and PA14 ∆orn ∆pelA-G with pMMB and pMMB-orn. Experiments were performed in triplicate.

Bacterial cultures were diluted, 10 µL was spotted on LB agar plates with the indicated concentration of IPTG and drip to form parallel lines of bacteria. After overnight incubation, representative images of the plates were taken of PA14 ∆orn and PA14 ∆orn ∆pelA-G with pMMB alone or containing the indicated allele of orn. Experiments were performed in triplicate and one representative image of the plates is shown.

Indicated bacterial strains were diluted and dripped on (A) LB agar plates or (B) LB agar plates containing 50 µg/mL carbenicillin with indicated concentration of IPTG and incubated overnight at 37°C. All experiments were performed in triplicate and one representative image of the plates is shown.

To determine whether the functional impact of diribonucleotide build-up is due specifically to an increased pGpG level and its effect on c-di-GMP, we asked whether the SCV formation in the ∆orn mutant is dependent on c-di-GMP processes. First, we confirmed that SCV formation in PA14 can be induced by elevated c-di-GMP levels through overexpression of diguanylate cyclases, such as PA1107 (roeA), PA1120 (yfiN/tpbB) and PA3702 (wspR) (Figure 6A) (Hickman et al., 2005; Kulasakara et al., 2006; Ueda and Wood, 2009; Malone et al., 2010). As expected, the effect of c-di-GMP is due to increased production of the pel exopolysaccharide and biofilm formation since the PA14 ∆pelA mutant maintained normal colony morphology even when these diguanylate cyclases were overexpressed (Figure 6A). To determine whether the increase in the pool of pGpG and c-di-GMP is the only reason for small-colony growth in the ∆orn mutant, a ∆orn ∆pel double mutant was tested. When ∆orn ∆pel was grown on agar, it also had an apparent SCV phenotype (Figure 6B). Complementation of ∆orn and ∆orn ∆pel with active alleles of orn restored normal colony morphology, whereas inactive orn alleles failed to complement (Figure 6—figure supplement 1). In every case, the colony morphology was the same between ∆orn and ∆orn ∆pel indicating a second pathway that can restrict colony growth, but in this case independent of pel and c-di-GMP. Similar results were obtained for P. aeruginosa PAK strain (Figure 6—figure supplement 2). These results indicate that increased pools of one or more of the diribonucleotides function to cause small-colony growth in ∆orn in addition to the actions of pGpG on c-di-GMP signaling.

Figure 6—figure supplement 1.

Small-colony phenotype of ∆orn is complemented by functional orn alleles.

Bacterial cultures were diluted, 10 µL was spotted on LB agar plates with the indicated concentration of IPTG and drip to form parallel lines of bacteria. After overnight incubation, representative images of the plates were taken of PA14 ∆orn and PA14 ∆orn ∆pelA-G with pMMB alone or containing the indicated allele of orn. Experiments were performed in triplicate and one representative image of the plates is shown.

Figure 6—figure supplement 2.

P. aeruginosa PAK ∆orn mutants show a small-colony phenotype.

Indicated bacterial strains were diluted and dripped on (A) LB agar plates or (B) LB agar plates containing 50 µg/mL carbenicillin with indicated concentration of IPTG and incubated overnight at 37°C. All experiments were performed in triplicate and one representative image of the plates is shown.

Conclusion

Orn is unique amongst exoribonucleases because it is essential in some γ-proteobacteria and is required to degrade the pGpG intermediate in c-di-GMP signaling. Yet the molecular and structural requirements for Orn were unknown. Our studies reveal that Orn is a dedicated diribonuclease in cells. This appears to be driven by a catalytic site that is restricted by a cap that mediates multiple interactions with the 5’ phosphate of diribonucleotide substrates. This constriction prevents longer substrates from binding with high affinity, rendering them poor substrates for catalytic cleavage. The discovery that Orn acts as a dedicated diribonuclease indicates that this activity clears a specific diribonucleotide bottleneck in global RNA degradation (Figure 7). Prior studies have shown that orn depletion can lead to accumulation of diribonucleotides and longer oligoribonucleotides in some cellular backgrounds (Ghosh and Deutscher, 1999). The increase in oligoribonucleotides longer than dimers in cells may occur through feedback inhibition of other enzymes in the RNA degradation pathway, in analogy to the impact of pGpG on c-di-GMP-degrading phosphodiesterases and c-di-GMP signaling (Figure 7) (Cohen et al., 2015; Orr et al., 2015). Whether the effect of diribonucleotides on essentiality is due to general processes such as RNA degradation, altered transcription (Goldman et al., 2011; Druzhinin et al., 2015; Vvedenskaya et al., 2012), specific interactions with essential proteins, or a combination of these remains to be evaluated in this context. Our studies therefore reveal a key step in RNA degradation, the enzymatic cleavage of diribonucleotides into mononucleotides, and set the stage to address how diribonucleotide accumulation is detrimental to cell survival.

Figure 7.

Model for Orn’s cellular function as a diribonuclease.

RNA degradation is initiated by fragmentation via endoribonucleases (e.g. RNase E and RNase III) that cleave unstructured or structured RNA sequences. RNA fragments are processed further at their 3’ termini by 3’−5’ exoribonucleases (e.g. PNPase, RNase R, and RNase II). Their combined activity produces mononucleotides and various terminal diribonucleotides from the original RNA fragments. The pGpG (GG) linear diribonucleotide is also produced by linearization of c-di-GMP by specific phosphodiesterases, EAL- and HD-GYP-domain-containing enzymes, which terminate c-di-GMP signaling. In Pseudomonas aeruginosa and likely other organisms that rely on Orn for growth, Orn is the only diribonuclease that cleaves diribonucleotides to mononucleotides. In the absence of Orn, diribonucleotides accumulate with a drastic impact on cellular physiology, ranging from transcriptional changes, small-colony phenotypes and growth arrest, depending on the organism. Orn is also unique because it acts as the second phosphodiesterase in the degradation of c-di-GMP by clearing the pGpG intermediate. In an orn mutant, c-di-GMP levels are elevated through feedback inhibition of the c-di-GMP-degrading phosphodiesterases by pGpG, leading to the associated biofilm phenotypes.

Materials and methods

Cloning, protein expression and purification

orn genes from Vibrio cholerae O1 El Tor VC0341 (residues 1–181) and Homo sapiens REXO2 (residues 33–237) were synthesized by Geneart (Life Technologies). Genes were cloned by ligation between BamHI and NotI sites of a modified pET28a vector (Novagen) yielding N-terminally His6-tagged small ubiquitin-like modifier (SUMO) fusion proteins cleavable by recombinant Ulp-1 protease.

Orn proteins were overexpressed in E. coli BL21 T7 Express cells (New England Biolabs). Fresh transformants were grown in Terrific Broth (TB) supplemented with 50 ug/mL kanamycin at 37°C to an OD600 ~1.0, at which point the temperature was reduced to 18°C and expression was induced by addition of 0.5 mM IPTG. Cells were harvested after 16 hr of expression by centrifugation and resuspended in a minimal volume of Ni-NTA binding buffer (25 mM Tris-Cl, 500 mM NaCl, 20 mM imidazole, pH 8.5) followed by flash freezing in liquid nitrogen.

Cells were thawed and lysed by sonication. Cell debris was removed by centrifugation and clarified soluble lysate was incubated with Ni-NTA resin (Qiagen) pre-equilibrated with Ni-NTA binding buffer. Following one hour of binding, the resin was washed three times with 10 column volumes of Ni-NTA binding buffer, and then eluted with six column volumes of Ni-NTA elution buffer (25 mM Tris-Cl, 500 mM NaCl, 350 mM imidazole, pH 8.5). Eluates were buffer exchanged into gel filtration buffer (25 mM Tris-Cl, 150 mM NaCl, pH 7.5) via a HiPrep 26/10 desalting column (GE Healthcare), followed by overnight incubation with Ulp-1 to cleave off the His6-tagged SUMO moiety. Untagged Orn proteins were recovered in the flow-through of a HisTrap Ni-NTA column (GE Healthcare), separated from His6-SUMO, uncleaved proteins and His6-tagged Ulp-1. Orn was concentrated via Amicon Ultra 10K concentrator prior to loading onto a HiLoad 16/60 Superdex 200 gel filtration column (GE Healthcare) equilibrated in gel filtration buffer. Fractions containing Orn were pooled and concentrated to 100 mg/mL, frozen in liquid nitrogen, and stored at −80°C.

Bacterial strains, plasmids and growth conditions

Strains and plasmids are listed in Key Resources table. P. aeruginosa is routinely grown in LB supplemented with the appropriate antibiotic (50 µg/mL carbenicillin or 75 µg/mL gentamicin) at 37°C.

Protein crystallography

REXO2-RNA and Orn-RNA complexes (pGpG and pApA from BioLog Life Science Institute, other nucleotides from GE Healthcare Dharmacon) were formed prior to crystallization by mixing a 1:2 molar ratio of Orn:RNA in gel filtration buffer, followed by incubation for 30 min at the crystallization temperature. Orn-RNA complexes (10–30 mg/ml) were crystallized via hanging-drop vapor diffusion by mixing equal volumes (0.8 μl) of sample with reservoir solution. Orn crystals grew at 20°C over a reservoir solution that was composed of 0.1 M BisTris (pH 5.5), 17% polyethylene glycol 3350, and 20% xylitol. Crystals were flash-frozen in liquid nitrogen. REXO2 crystals grew at 4°C, using a reservoir comprised of 0.2 M sodium malonate (pH 5.5) and 15–20% polyethylene glycol 3350. REXO2 crystals were soaked in cryoprotectant of reservoir solution supplemented with 25% glycerol prior to flash freezing with liquid nitrogen. All crystals were stored in liquid nitrogen. Data were collected by synchrotron radiation at 0.977 Å on frozen crystals at 100 K at beamline F1 of the Cornell High Energy Synchrotron Source (CHESS). Diffraction data sets were processed using XDS, Pointless, and Scala (Evans, 2006; Kabsch, 2010). The initial structures were solved by Molecular Replacement using the software package Phenix (Adams et al., 2010) and the unpublished coordinates of E. coli Orn (PDB: 2igi) as the search model. Manual model building and refinement were carried out with Coot (Emsley et al., 2010) and Phenix, respectively. Illustrations were prepared in Pymol (Version 2.2.0, Schrodinger, LLC). All software packages were accessed through SBGrid (Morin et al., 2013). Crystallographic statistics are shown in Figure 2—source data 1.

Site-directed mutagenesis

To create the point mutants of orn, mutations were generated by using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs). All mutations were verified by sequencing.

Western blot analysis

Overnight cultures were diluted to OD600 of ~0.02 in media with 0.2% arabinose with 75 µg/mL gentamicin. Culture density was monitored by OD600. At OD600 ~1.5, cells were collected by centrifugation and resuspended in 1/15 vol of 1x PBS. Proteins were immediately precipitated by a modified MeOH/CHCl3 procedure (sample/MeOH/CHCl3: 1/1/0.25 [Wessel and Flügge, 1984]). Samples were separated on 12% SDS-PAGE, transferred to PVDF membranes and blocked with Tris buffered saline (50 mM Tris, pH7.4, 200 mM NaCl) with 5% non-fat milk. HA-tags were detected using mouse anti-HA (Sigma H9658) and goat anti-mouse IgG-HRP conjugate (Sigma A9917). HRP signal was developed using chemiluminescence per manufacture protocol (Sigma).

Labeling of RNAs

5’ un-phosphorylated RNAs were purchased from TriLink Biotechnologies or Sigma. Each RNA was subjected to radioactive end-labeling or non-radioactive phosphorylation by T4 Polynucleotide Kinase (New England Biolabs). Each RNA was subjected to phosphorylation with equimolar concentrations of either 32P-γ-ATP or ATP, T4 PNK, and 1X T4 PNK Reaction Buffer. Reactions comprising a final concentration of either 0.5 µM radiolabeled RNA or 2.0 µM phosphorylated RNA were incubated at 37°C for 40 min, followed by heat inactivation of T4 PNK at 65°C for 20 min.

Protein expression and purification for biochemical assays

E. coli T7Iq strains harboring expression vector pVL847 expressing an His10-MBP-Orn and His10-MBP-Orn mutants from V. cholerae were grown overnight, subcultured in LB M9 fresh media supplemented with 15 μg/ml gentamicin and grown to approximately OD6000.5 ~ 1.0 at 30°C. Expression was induced with 1 mM IPTG for 4 hr. Induced bacteria were collected by centrifugation and resuspended in 10 mM Tris, pH 8, 100 mM NaCl, and 25 mM imidazole. After addition of 10 µg/mL DNase, 25 µg/mL lysozyme, and 1 mM PMSF, bacteria were lysed by sonication. Insoluble material was removed by centrifugation. The His-fusion protein was purified by separation over a Ni-NTA column. Purified proteins were pooled and dialyzed for 1 hr and overnight against 10 mM Tris, pH 8, 100 mM NaCl. The proteins were dialyzed for 3 hr in 10 mM Tris, pH 8, 100 mM NaCl, and 50% (vol/vol) glycerol, aliquoted, and flash frozen with liquid nitrogen and stored at −80° ˚C.

Size-exclusion chromatography coupled multiangle light scattering (SEC-MALS)

5 mg/ml (0.24 mM) of purified protein (wild-type Orn, Orn-D12A, or Orn-R130A) was injected onto a Superdex 200 Increase 10/300 column (GE Healthcare) equilibrated with gel filtration buffer (25 mM Tris-Cl, pH 7.5, 150 mM NaCl). Samples were run continuously at a flow rate of 0.75 ml/min through the gel filtration column coupled to a static 18-angle light scattering detector (DAWN HELIOS-II) and a refractive index detector (Optilab T-rEX), with data being collected every second. Data analysis was performed with ASTRA VI, yielding the molar mass and mass distribution (polydispersity) of the sample, using monomeric BSA (Sigma; 5 mg/ml) to normalize the light scattering detectors and as a control sample.

Preparation of whole cell lysates

Overnight cultures of P. aeruginosa PA14 WT, Δorn mutant, or complemented strains were subcultured into fresh media with antibiotic and 1 mM IPTG, grown at 37°C with shaking to OD600 ~ 2.5. All bacteria samples were collected by centrifuge and resuspended in 1/10 vol of 100 mM NaCl, and 100 mM Tris, pH 8, also supplemented with 25 µg/mL lysozyme, 10 µg/mL DNase, and 1 mM PMSF and stored at −80°C.

Oligoribonuclease cleavage reactions

Phosphorylated RNA (1.0 µM), including trace amounts of radiolabeled substrate, was subjected to cleavage by either 5.0 nM or 1.0 µM purified Orn at room temperature. These reactions were in the presence of 10 mM Tris, pH 8.0, 100 mM NaCl, and 5 mM MgCl2. At the appropriate times, aliquots of the reaction were removed and quenched in the presence of 150 mM EDTA on ice and heat inactivated at 95°C for 5 min. Activity of whole cell lysates against 32P-labeled oligoribonucleotide substrates was performed at room temperature in reaction buffer (10 mM Tris, pH 8, 100 mM NaCl, and 5 mM MgCl2). At the indicated times, the reaction was stopped by the addition of 0.2 M EDTA and heated at 98°C for 5 min. Samples were separated on denaturing 20% PAGE containing 1x TBE and 4 M urea. The gels were imaged using Fujifilm FLA-7000 phosphorimager (GE) and analyzed for the appearance of truncated 32P-labeled products. The intensity of the radiolabeled nucleotides was quantified using Fujifilm Multi Gauge software v3.0.

DRaCALA measurement of dissociation constants

To measure K, serial dilutions of purified His10-MBP-Orn, His10-MBP Orn mutants, or untagged Orn in binding buffer (10 mM Tris, pH 8, 100 mM NaCl, and 5 mM CaCl2) were mixed with radiolabeled nucleotides, applied to nitrocellulose sheets, dried, imaged and K values were calculated as described previously (Roelofs et al., 2011; Patel et al., 2014).

Aggregation assay

A colony of each strain of P. aeruginosa grown in LB agar plates supplemented with 50 µg/mL carbenicillin was inoculated into borosilicate glass tubes containing 2.5 mL of LB supplemented with 0.1 mM IPTG. The cultures were placed in a fly-wheel in 37°C incubator to spin for 18 ~ 22 hr. Culture tubes were allowed to settle at room temperature for 10 min and photographed.

Colony morphology

Indicated strains are grown overnight at 37°C in LB agar with either 50 µg/mL of carbenicillin or 15 µg/mL of gentamicin, as appropriate. Three independent colonies are inoculated into LB with the appropriate antibiotics at 37°C with shaking. The bacteria were subcultured and grown until OD600 between 1.0 and 1.5. All cultures were diluted to ~10,000 CFU per mL. To observe colony size of multiple colonies of multiple strains, 10 µL of each strain were dripped in parallel on the same LB plates with the appropriate antibiotic and indicated IPTG conditions to ensure all strains were grown with the same media and conditions. For all strains tested, the control strains were always included on the same plate.

Data deposition

The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.rcsb.org (PDB ID codes 6N6A, 6N6C, 6N6D, 6N6E, 6N6F, 6N6G, 6N6H, 6N6I, 6N6J, and 6N6K).

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "A dedicated diribonuclease resolves a key bottleneck as the terminal step of RNA degradation" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Gisela Storz as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Simon Dove (Reviewer #1); Ben Luisi (Reviewer #2); Arun Malhotra (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

The manuscript from the Lee, Sondermann, and Winkler labs describes results obtained from biochemical, genetic, and structural studies of the cellular mechanism of Oligoribonuclease (Orn), a 3'-5' exonuclease from γ-proteobacteria. The authors provide several lines of evidence indicating that Orn's cellular function is to carry out the terminal step of RNA degradation, thus serving as a "dedicated diribonuclease." These findings challenge the existing view that Orn is involved in degradation not only of diribonucleotides, but also other short RNA products up to ~5-nt in length. The substrate specificity of Orn for dinucleotides is illuminated by high resolution crystal structures of Vibrio cholerae Orn and the human homolog REXO2 bound to dinucleotide substrates. The authors also present evidence that Orn is the only diribonuclease in Pseudomonas aeruginosa, and that Orn's activity as a diribonuclease influences cell growth and exopolysaccharide synthesis.

Each reviewer judged the work to be significant because it challenges the existing view of Orn's cellular function as involved in the degradation of short RNA species up to ~5-nt in length. They also felt the experiments are well designed and the results are convincing, making a strong case that Orn, which is essential in some bacteria, acts at the final step of RNA degradation by digesting dinucleotides rather than larger oligonucleotide substrates.

The reviewers recommended the manuscript be accepted for publication after the text is modified to address the points raised below. In addition, each of the reviewers agreed that the title was not ideal for generating interest in the work and should be changed. One recommendation for a title that may generate more interest in the work is: "Structural basis for the terminal step of RNA degradation".

Essential revisions:

1) In the biochemical and cellular tests of the Orn mutants with substitutions at active site and phosphate cap residues (Figure 3) it is possible that the effects of some of these mutations are due to misfolding. In relation to the biochemical experiments (Figure 3A), do the authors have evidence these mutants still form dimers or are not misfolded? In relation to the cellular assays (Figure 3B), do the authors know that the mutants tested are as abundant as the wild-type protein? If additional data are not provided that address these issues, the authors should acknowledge these limitations in the text.

2) Based on data presented in Figure 5 the authors claim that Orn is the only diribonuclease in P. aeruginosa. The data are certainly consistent with this claim. However, it is difficult to exclude the possibility that there are other dinucleotidases that might be made under conditions that are different to those used to make the cell lysates (e.g. different media or growth phase). The data presented also cannot rule out the possibility that there may be other dinucleotidases with specific substrate specificities that are missed in their assay (only pGpG is used as substrate with lysates in Figure 5). To address these potential caveats, the authors should reword the claim in the subheading “Orn is the only diribonuclease in P. aeruginosa”.

3) In the subsection “Structure of Orn complexes with pGpG and other linear diribonucleotides”, please add text explaining how catalysis was prevented in the crystals.

4) Much of the prior biochemical characterization of this family of enzymes has been done with Orn from E. coli. The authors should comment on why Vibrio cholerae Orn was used for the studies presented here.

5) The binding studies and the structures lack metal ions. Orn uses divalent metal ions at its active site and this can influence substrate binding. Was EDTA added to the binding and crystallization buffers to suppress activity? Could weak residual Orn activity affect the binding analysis from the DRaCALA experiments? One approach may be to use active site mutants for the binding studies.

6) The binding assays (Table 1) also used a His-MBP-Orn fusion, while the activity and in vivo studies used non-tagged (or tag removed) Orn. Were any controls performed to ensure that MBP did not interfere with the K measurements?

7) While the structural results show a tight binding of a dinucleotide at the Orn active-site pocket, does this really preclude activity on larger oligomers? In the electrostatic map of Orn's surface shown in Figure 2—figure supplement 1B, one prominent feature is a conserved basic patch just below the active site which could potentially help bind larger oligomers and direct them into the active site pocket. Was crystallization attempted with larger oligos?

8) The manuscript should have a more thorough discussion of other factors that may explain the low binding/activity seen here on longer oligos. Some prior studies (Cohen et al., 2015; Mechold et al., 2006) have used bulky 5'-Cy5 labeling as pointed in the manuscript. However, others (Ghosh and Deutscher, 1999; Datta and Niyogi, 1975) used substrates that were more natural. Could the discrepancy in the results be due to different buffer conditions?

9) The authors’ results would imply that there is another unknown exonuclease responsible for degrading RNA oligomers into dinucleotides. Limit products from RNase R and RNase II are larger (2-3 nts for RNR; 3-4 or more nts for RNB), as are RNA fragments generated by abortive transcription initiation. In this context, results on degradation of the 7mer even in the orn deleted cell extracts (Figure 5B) are interesting. Is it possible that the added DNase (subsection “Preparation of whole cell lysates”) in the cell extracts is responsible for this activity? Figure 7 suggests that RNA degradation gives rise to only dinucleotides and should be revised.

10) Change the heading from "The phosphate cap is required for diribonuclease activity" to "Phosphate-cap binding residues are required for diribonuclease activity." The title implies that the presence or absence of a 5' phosphate was tested, which it was not. The statement in the subsection “The phosphate cap is required for diribonuclease activity”, should also be rewritten as I do not believe that the results "demonstrate that an intact phosphate cap is required for enzyme function" as stated in the text.

Essential revisions:

1) In the biochemical and cellular tests of the Orn mutants with substitutions at active site and phosphate cap residues (Figure 3) it is possible that the effects of some of these mutations are due to misfolding. In relation to the biochemical experiments (Figure 3A), do the authors have evidence these mutants still form dimers or are not misfolded? In relation to the cellular assays (Figure 3B), do the authors know that the mutants tested are as abundant as the wild-type protein? If additional data are not provided that address these issues, the authors should acknowledge these limitations in the text.

We have tagged wild-type and mutant alleles with HA tag and induced expression in PA14 ∆orn as previous done for our biological assays. Proteins were precipitated and analyzed by Western blot using anti-HA. These new data are presented in Figure 3—figure supplement 1 and suggest that all of the proteins are expressed at similar levels in PA14 ∆orn. In addition, we expressed and purified Orn wild-type alongside point-mutants R130A (phosphate cap) and D12A (DEDD motif), and subjected proteins to size exclusion chromatography-coupled multi-angle light scattering (SEC-MALS) experiments to determine their oligomeric state. All three proteins are stable dimers as shown in Figure 3—figure supplement 2. Together, these results support our previous conclusion that mutations in these key residues ablate Orn function as a diribonuclease without affecting quaternary structure or protein expression. Discussion of these points have been added to the subsection “The phosphate cap is required for diribonuclease activity”).

2) Based on data presented in Figure 5 the authors claim that Orn is the only diribonuclease in P. aeruginosa. The data are certainly consistent with this claim. However, it is difficult to exclude the possibility that there are other dinucleotidases that might be made under conditions that are different to those used to make the cell lysates (e.g. different media or growth phase). The data presented also cannot rule out the possibility that there may be other dinucleotidases with specific substrate specificities that are missed in their assay (only pGpG is used as substrate with lysates in Figure 5). To address these potential caveats, the authors should reword the claim in the subheading “Orn is the only diribonuclease in P. aeruginosa”.

We agree that other dinucleotidases could be expressed under different conditions. To further support the statement that Orn is the only diribonuclease in P. aeruginosa under the assay condition, we have tested the lysates of catalytically inactive orn mutant alleles. We found that L18A and H158A alleles do not support the cleavage of the accumulated dinucleotide product from the degradation of the input 7-mer substrate. These results are presented in Figure 5—figure supplement 1 and indicate that Orn is the only dinucleotidase under the assay condition (subsection “Orn is the only diribonuclease in P. aeruginosa grown under laboratory conditions”, first paragraph). We have altered the subheading title and the text to indicate that our findings are specific for our specific assay conditions.

3) In the subsection “Structure of Orn complexes with pGpG and other linear diribonucleotides”, please add text explaining how catalysis was prevented in the crystals.

The wild-type protein appears to purify free of divalent cations. Catalysis is only observed when divalent cations are added to the reaction. As requested by the reviewer, we added a clarifying statement to the main text (subsection “Structure of Orn complexes with pGpG and other linear diribonucleotides”, first paragraph).

4) Much of the prior biochemical characterization of this family of enzymes has been done with Orn from E. coli. The authors should comment on why Vibrio cholerae Orn was used for the studies presented here.

We initially crystallized Pseudomonas aeruginosa Orn, consistent with the bacterial host used in this study, but the crystal packing prevented substrate binding. Vibrio cholerae was the next protein we tried in order to obtain alternative crystal forms that supported substrate binding. We clarified this point in text by adding the following sentence: “We initially determined the crystal structure of P. aeruginosa Orn (Orn) with diribonucleotide substrate, however crystal packing contacts prevented substrate binding (data not shown).”.

5) The binding studies and the structures lack metal ions. Orn uses divalent metal ions at its active site and this can influence substrate binding. Was EDTA added to the binding and crystallization buffers to suppress activity? Could weak residual Orn activity affect the binding analysis from the DRaCALA experiments? One approach may be to use active site mutants for the binding studies.

See above for response to question #3. The wild-type protein appears to purify free of divalent cations, even without any special treatment (e.g. EDTA). The wild-type protein can bind substrates and crystallizes as a complex, but remains catalytically inactive. Nevertheless, we performed the suggested experiment and repeated the binding studies using the different catalytically dead mutant Orn proteins in the presence of magnesium. The catalytically inactive proteins bind with an affinity that is within an order of magnitude of wild-type in the absence of divalent cations (see Figure 3—figure supplement 1). These new results indicate that our results from our previous studies are correctly interpreted.

In addition, we initially crystallized dinucleotide-bound Orn with a D12A mutation in the conserved DEDD motif. Structures are identical to the wild-type Orn structures presented in the manuscript. We chose not to include structures obtained with active-site mutants in the manuscript since they did not add new information.

6) The binding assays (Table 1) also used a His-MBP-Orn fusion, while the activity and in vivo studies used non-tagged (or tag removed) Orn. Were any controls performed to ensure that MBP did not interfere with the K

We have tested OrnVc that was cleaved from its fusion tag (corresponding to the crystallized protein) and the K is 80 nM. This Kvalue for cleaved Orn is very similar to His-MBP-Orn (90 nM). These results indicate that the His-MBP tag did not significantly alter pGpG binding affinity to OrnVc (see Figure 3—figure supplement 4B). A sentence has been added to the text to indicate that the affinity tag does not affect protein-dinucleotide affinity (subsection “Orn functions as a diribonuclease in vitro”, first paragraph).

7) While the structural results show a tight binding of a dinucleotide at the Orn active-site pocket, does this really preclude activity on larger oligomers? In the electrostatic map of Orn's surface shown in Figure 2—figure supplement 1B, one prominent feature is a conserved basic patch just below the active site which could potentially help bind larger oligomers and direct them into the active site pocket. Was crystallization attempted with larger oligos?

We also attempted crystallization of inactive Orn-D12A with 3-, 4- and 5-mer RNAs and of wild-type protein with a 3-mer RNA. All resulting structures resolved only dinucleotides at the active site. These results are consistent with two recent reports of structural data for Colwellia psychrerythraea Orn and REXO2 bound to substrates (Lee et al., 2019; Chu et al., 2019). In the case of Orn, crystallization was attempted with a 5-mer RNA, however only two bases were resolved at the active site. In the report about REXO2, a similar observation was made for RNA substrates. A complex with a longer ligand was obtained with DNA, which may not allow conclusions for RNA substrates. In the same report, RNase activity against longer RNA was observed with a 20x molar excess of enzyme (1 µM) over RNA (50 nM), likely not representing physiological conditions. We discuss this work in our revised manuscript (subsection “Structure of Orn complexes with pGpG and other linear diribonucleotides”, last paragraph). Our structural analysis in combination with our biochemical data make a strong case for Orn acting as a dedicated dinucleotidase.

8) The manuscript should have a more thorough discussion of other factors that may explain the low binding/activity seen here on longer oligos. Some prior studies (Cohen et al., 2015; Mechold et al., 2006) have used bulky 5'-Cy5 labeling as pointed in the manuscript. However, others Ghosh and Deutscher, 1999; Datta and Niyogi, 1975) used substrates that were more natural. Could the discrepancy in the results be due to different buffer conditions?

The reviewer is correct to ask for a careful comparative review of these papers to provide a rationale for the vast differences between our current observations and those from previous studies. Through studying the Materials and methods sections of the publications, we found that it was challenging to compare the studies as the details within the Materials and methods sections written over a four-decade span have distinct styles and use of language that changed significantly based on the knowledge available at the time of publication. Based on our best estimation, we found large differences in the concentrations of enzyme, substrate, and enzyme to substrate ratios. Notably, we identified a consensus difference in the reaction buffer used in all other studies from the ones performed by us. In all manuscripts, the reaction buffer contains a buffer (Tris, HEPES or bis-tris propane) at pH 8.0 with 5 mM divalent cation (Mg2+ or Mn2+). However, all of the previously published studies perform Orn activity assays in reaction buffer lacking NaCl/KCl instead of the 100 mM NaCl that is used in the current manuscript and manuscripts published earlier by us. The difference in ionic strength of the reaction buffer is significant and likely contributes at least in part to the disparate results from similarly performed experiment. We argue that performing experiments in buffer with near-physiological ionic strength approximates the condition in cells. We have added two sentences to clearly indicate these differences in reaction buffer conditions (subsection “Orn functions as a diribonuclease in vitro”, last paragraph).

9) The authors’ results would imply that there is another unknown exonuclease responsible for degrading RNA oligomers into dinucleotides. Limit products from RNase R and RNase II are larger (2-3 nts for RNR; 3-4 or more nts for RNB), as are RNA fragments generated by abortive transcription initiation. In this context, results on degradation of the 7mer even in the orn deleted cell extracts (Figure 5B) are interesting. Is it possible that the added DNase (subsection “Preparation of whole cell lysates”) in the cell extracts is responsible for this activity? Figure 7 suggests that RNA degradation gives rise to only dinucleotides and should be revised.

We tested the ability of DNase, lysozyme and PMSF individually or in combination to hydrolyze the 7-mer RNA. Our results indicate that none of these additives contributed to the hydrolysis of 7 base oligonucleotide (see Author response image 1). Since this is not particularly relevant to the manuscript, we opt to present the data only in the Response to Reviewer document.

Author response image 1.

Additives to cell lystaes do not contribute to cleavage of 32P-AAAAAGG oligoribonucleotide.

10 µg/ml DNase, 25 µg/ml lysozyme, 1 mM PMSF, and the combination of all three additives were test for cleavage of 32P-AAAAAGG. The samples were stopped at indicated times and analyzed by 20% Urea gel. All data shown represent duplicate independent experiments.

The identity of enzyme(s) responsible for degrading the 3-4 mer RNA fragments to dinucleotides is currently unknown. Future studies will be required to identify these enzymes. We modified the model in Figure 7 as requested.

10) Change the heading from "The phosphate cap is required for diribonuclease activity" to "Phosphate-cap binding residues are required for diribonuclease activity." The title implies that the presence or absence of a 5' phosphate was tested, which it was not. The statement in the subsection “The phosphate cap is required for diribonuclease activity”, should also be rewritten as I do not believe that the results "demonstrate that an intact phosphate cap is required for enzyme function" as stated in the text.

We included new data in the amended manuscript showing that GpG added in vast (1000x) excess over radiolabed pGpG has no effect on pGpG binding (or cleavage), whereas 1000x unlabeled pGpG abolishes radioligand binding completely (Figure 2—figure supplement 4). Attempts to co-crystallize Orn with GpG were also unsuccessful so far, further suggesting that GpG has significantly lower affinity to Orn than pGpG. With this, we argue that the chosen subheading accurately describes our data. Reference to the new figure supplement can be found in the first paragraph of the subsection “Interaction of Orn with substrates”.

Key resources table

Reagent type or resource	Designation	Source or reference	Identifiers	Additional information
Strain (P. aeruginosa)	PA14	Rahme et al., 1995; PMID 7604262
Strain (P. aeruginosa)	PA14 ∆pelA	Lee et al., 2007; PMID: 17824927
Strain (P. aeruginosa)	PA14 ∆orn	Orr et al., 2015; PMID: 26305945
Strain (P. aeruginosa)	PA14 ∆orn ∆pelA-G	Orr et al., 2015; PMID: 26305945
Strain (P. aeruginosa)	PAK	Bradley, 1974; PMID: 4206974
Strain (P. aeruginosa)	PAK ∆orn	This study		Generated using pEX-Gn-∆orn (PAK)
Strain (E. coli)	XL10-Gold	Agilent
Strain (E. coli)	Stellar cells	Takara/Clontech
Strain (E. coli)	BL21(DE3)	New England Biolabs
Strain (E. coli)	NEB T7Iq	New England Biolabs
Genetic reagent (plasmid)	pET28-His6-SUMO-OrnVc	This study		cloned from custom DNA fragment (see below) for purification of His6-SUMO-OrnVc, OrnVc
Genetic reagent (plasmid)	pET28-His6-SUMO-Rexo2	This study		cloned from custom DNA fragment (see below) for purification of His6-SUMO-Rexo2, Rexo2
Genetic reagent (plasmid)	pDONR221-VC0341 (ornVc)	Rolfs et al., 2008; PMID: 18337508
Genetic reagent (plasmid)	pDONR221-ornVc D12A	This study		site-directed mutagenesis to introduce alanine at D12 position
Genetic reagent (plasmid)	pDONR221-ornVc L18A	This study		site-directed mutagenesis to introduce alanine at L18 position
Genetic reagent (plasmid)	pDONR221-ornVc D59A	This study		site-directed mutagenesis to introduce alanine at D59 position
Genetic reagent (plasmid)	pDONR221-ornVc Q111A	This study		site-directed mutagenesis to introduce alanine at Q111 position
Genetic reagent (plasmid)	pDONR221-ornVc R130A	This study		site-directed mutagenesis to introduce alanine at R130 position
Genetic reagent (plasmid)	pDONR221-ornVc Y129A	This study		site-directed mutagenesis to introduce alanine at Y129 position
Genetic reagent (plasmid)	pDONR221-ornVc Y129W	This study		site-directed mutagenesis to introduce tryptophan at Y129 position
Genetic reagent (plasmid)	pDONR221-ornVc W61A	This study		site-directed mutagenesis to introduce alanine at W61 position
Genetic reagent (plasmid)	pDONR221-ornVc W61Y	This study		site-directed mutagenesis to introduce tyrosine at W61 position
Genetic reagent (plasmid)	pDONR221-ornVc H66A	This study		site-directed mutagenesis to introduce alanine at H66 position
Genetic reagent (plasmid)	pDONR221-ornVc H158A	This study		site-directed mutagenesis to introduce alanine at H158 position
Genetic reagent (plasmid)	pVL847-ornVc	Orr et al., 2015; PMID: 26305945		OrnVc from pDONR cloned into pVL847 for purification of His-MBP-OrnVc, OnrVc
Genetic reagent (plasmid)	pVL847-ornVc D12A	This study		D12A from pDONR cloned into pVL847 for purification of His-MBP-OrnVc D12A, OrnVc D12A
Genetic reagent (plasmid)	pVL847-ornVc L18A	This study		L18A from pDONR cloned into pVL847 for purification of His-MBP-OrnVc L18A, OrnVc L18A
Genetic reagent (plasmid)	pVL847-ornVc D59A	This study		L18A from pDONR cloned into pVL847 for purification of His-MBP-OrnVc D59A, OrnVc D59A
Genetic reagent (plasmid)	pVL847-ornVc Q111A	This study		Q111A from pDONR cloned into pVL847 for purification of His-MBP-OrnVc Q111A, OrnVc Q111A
Genetic reagent (plasmid)	pVL847-ornVc R130A	This study		R130A from pDONR cloned into pVL847 for purification of His-MBP-OrnVc R130A, OrnVc R130A
Genetic reagent (plasmid)	pVL847-ornVc Y129A	This study		Y129A from pDONR cloned into pVL847 for purification of His-MBP-OrnVc Y129A, OrnVc Y129A
Genetic reagent (plasmid)	pVL847-ornVc W61A	This study		W61A from pDONR cloned into pVL847 for purification of His-MBP-OrnVc W61A, OrnVc W61A
Genetic reagent (plasmid)	pVL847-ornVc H66A	This study		H66A from pDONR cloned into pVL847 for purification of His-MBP-OrnVc H66A, OrnVc H66A
Genetic reagent (plasmid)	pVL847-ornVc H158A	This study		H158A from pDONR cloned into pVL847 for purification of His-MBP-OrnVc H158A, OrnVc H158A
Genetic reagent (plasmid)	pMMB-Gn	Fürste et al., 1986; PMID: 3549457
Genetic reagent (plasmid)	pMMB-Gn-PA1107	Kulasakara et al., 2006; PMID: 16477007
Genetic reagent (plasmid)	pMMB-Gn-PA1120	Kulasakara et al., 2006; PMID: 16477007
Genetic reagent (plasmid)	pMMB-Gn-PA3702 (wspR)	Kulasakara et al., 2006; PMID: 16477007
Genetic reagent (plasmid)	pMMB-Ap	Fürste et al., 1986; PMID: 3549457
Genetic reagent (plasmid)	pMMB-Ap-VC0341 (ornVc)	This study		Orn from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pMMB-Ap-ornVc D12A	This study		D12A from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pMMB-Ap-ornVc L18A	This study		L18A from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pMMB-Ap-ornVc Q111A	This study		Q111A from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pMMB-Ap-ornVc R130A	This study		R130A from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pMMB-Ap-ornVc Y129A	This study		Y129A from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pMMB-Ap-ornVc Y129W	This study		Y129W from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pMMB-Ap-ornVc W61A	This study		W61A from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pMMB-Ap-ornVc W61Y	This study		W61Y from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pMMB-Ap-ornVc H66A	This study		H66A from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pMMB-Ap-ornVc H158A	This study		H158A from pDONR cloned into pMMB for expressing in P. aeruginosa
Genetic reagent (plasmid)	pEX-Gn-∆orn (PAK)	This study
Genetic reagent (plasmid)	pJHA-Gn	This study
Genetic reagent (plasmid)	pJHA-Gn-ornPa	This study		OrnPA (Orr, et al, PNAS 2015) cloned into pJHA for expressing in P. aeruginosa
Genetic reagent (plasmid)	pJHA-Gn-ornVc	This study		Orn from pDONR cloned into pJHA for expressing in P. aeruginosa
Genetic reagent (plasmid)	pJHA-Gn-ornVc D12A	This study		D12A from pDONR cloned into pJHA for expressing in P. aeruginosa
Genetic reagent (plasmid)	pJHA-Gn-ornVc L18A	This study		L18A from pDONR cloned into pJHA for expressing in P. aeruginosa
Genetic reagent (plasmid)	pJHA-Gn-ornVc D59A	This study		Q111A from pDONR cloned into pJHA for expressing in P. aeruginosa
Genetic reagent (plasmid)	pJHA-Gn-ornVc W61A	This study		R130A from pDONR cloned into pJHA for expressing in P. aeruginosa
Genetic reagent (plasmid)	pJHA-Gn-ornVc H66A	This study		Y129A from pDONR cloned into pJHA for expressing in P. aeruginosa
Genetic reagent (plasmid)	pJHA-Gn-ornVc Q111A	This study		W61A from pDONR cloned into pJHA for expressing in P. aeruginosa
Genetic reagent (plasmid)	pJHA-Gn-ornVc R130A	This study		H66A from pDONR cloned into pJHA for expressing in P. aeruginosa
Genetic reagent (plasmid)	pJHA-Gn-ornVc H158A	This study		H158A from pDONR cloned into pJHA for expressing in P. aeruginosa
Antibody	Mouse anti-HA monoclonal antibody	Sigma; catalog number H9658		1:5000
Antibody	Goat Anti-Mouse IgG-HRP conjugate	Sigma; catalog number A9917		1:5000
DNA fragment	5'-GGATCCATGGCAGCCGGTGAAAG CATGGCACAGCGTATGGTTTGGGTT GATCTGGAAATGACCGGTCTGGATA TTGAAAAAGATCAGATTATTGAAATG GCCTGCCTGATTACCGATAGCGATCT GAATATTCTGGCAGAAGGTCCGAAT CTGATTATCAAACAGCCGGATGAACT GCTGGATAGCATGAGCGATTGGTGT AAAGAACATCATGGTAAAAGCGGTCT GACCAAAGCAGTTAAAGAAAGCACCA TTACACTGCAGCAGGCCGAATATGAAT TTCTGAGCTTTGTTCGTCAGCAGACC CCTCCGGGTCTGTGTCCGCTGGCAGG TAATAGCGTTCATGAAGATAAAAAGTT TCTGGATAAGTATATGCCGCAGTTTAT GAAGCATCTGCATTATCGCATTATTGAT GTGAGCACCGTTAAAGAACTGTGTCGT CGTTGGTATCCGGAAGAATATGAGTTT GCACCGAAAAAAGCAGCAAGCCATCGT GCACTGGATGATATTAGCGAAAGCATC AAAGAGCTGCAGTTTTATCGCAACAAC ATCTTCAAAAAGAAAATCGACGAGAAAA AACGCAAAATCATCGAAAACGGCGAAAA CGAAAAAACCGTTAGCTAAGCGGCCGC-3'	GeneArt		custom DNA fragment for REXO2 that is codon-optimized for E. coli
DNA fragment	5'-GGATCCATGAGCTTTAGCGATCAGAAT CTGATTTGGATTGATCTGGAAATGACCG GTCTGGACCCGGAAATGCATAAAATCAT TGAAATGGCAACCATCGTGACCGATAGCG AACTGAATATTCTGGCAGAAGGTCCGGTT ATTGCAATTCATCAGCCGGAAAGCGAACT GGCAAAAATGGATGAATGGTGTACCACCA CCCATACCGCAAGCGGTCTGGTTGCACGT GTTCGTCAGAGCCAGGTTAGCGAAGAAGA AGCAATTGATCAGACCCTGGCATTTCTGAA ACAGTGGGTTCCGGAAGGTAAAAGCCCGAT TTGTGGTAATAGCATTGGTCAGGATCGTCG CTTTCTGTATAAACATATGCCTCGTCTGGAA GCCTATTTCCATTATCGTTATATTGATGTGAG CACCATCAAAGAACTGACCCGTCGTTGGCAG CCGGAAGTTCTGAAAGAATTTAGCAAAACCG GTAGCCATCTGGCACTGGATGATATTCGTGA AAGCATTGCAGAGCTGCAGTTTTATCGTAAA GCCGTGTTTAAAATCTAAGCGGCCGC-3'	GeneArt		custom DNA fragment for OrnVc that is codon-optimized for E. coli
RNA primer	5'-GG-3'	Sigma
RNA primer	5'-AGG-3'	Sigma
RNA primer	5'-AAGG-3'	Sigma
RNA primer	5'-AAAGG-3'	Sigma
RNA primer	5'-AAAAGG-3'	Sigma
RNA primer	5'-AAAAAGG-3'	Sigma
RNA primer	5'-pGG-3'	Biolog; catalog number P023-01
RNA primer	5'-pAA-3'	Biolog; catalog number P033-01
RNA primer	5'-pAG-3'	GE Healthcare Dharmacon
RNA primer	5'-pGA-3'	GE Healthcare Dharmacon
RNA primer	5'-pGC-3'	GE Healthcare Dharmacon
RNA primer	5'-pCG-3'	GE Healthcare Dharmacon
RNA primer	5'-pCU-3'	GE Healthcare Dharmacon
Software	Prism	GraphPad
Software	XDS	Kabsch, 2010; PMID: 20124693		Distributed through SBGrid
Software	Pointless	Evans, 2006; PMID: 16369096		Distributed through SBGrid
Software	Scala	Evans, 2006; PMID: 16369096		Distributed through SBGrid
Software	Phenix	Adams et al., 2010; PMID: 20124702		Distributed through SBGrid
Software	Coot	Emsley et al., 2010; PMID: 20383002		Distributed through SBGrid
Software	Pymol	Schrödinger
Software	Fujifilm Multi Gauge software v3.0	Fujifilm