ORIGINAL RESEARCH published: 09 October 2019 doi: 10.3389/fphar.2019.01170 Evolutionary Trends in RNA Base Selectivity Within the RNase A Superfamily Guillem Prats-Ejarque, Lu Lu, Vivian A. Salazar, Mohammed Moussaoui and Ester Boix* Department of Biochemistry and Molecular Biology, Faculty of Biosciences, Universitat Autònoma de Barcelona, Barcelona, Spain There is a growing interest in the pharmaceutical industry to design novel tailored drugs for RNA targeting. The vertebrate-specific RNase A superfamily is nowadays one of the best characterized family of enzymes and comprises proteins involved in host defense with specific cytotoxic and immune-modulatory properties. We observe within the family a structural variability at the substrate-binding site associated to a diversification of biological properties. In this work, we have analyzed the enzyme specificity at the secondary Edited by: Olga N. Ilinskaya, base binding site. Towards this end, we have performed a kinetic characterization of Kazan Federal University, the canonical RNase types together with a molecular dynamic simulation of selected Russia representative family members. The RNases’ catalytic activity and binding interactions Reviewed by: have been compared using UpA, UpG and UpI dinucleotides. Our results highlight Arun Malhotra, University of Miami, an evolutionary trend from lower to higher order vertebrates towards an enhanced United States discrimination power of selectivity for adenine respect to guanine at the secondary base Vladimir Alexandrovich Mitkevich, Engelhardt Institute of Molecular binding site (B2). Interestingly, the shift from guanine to adenine preference is achieved in Biology (RAS), Russia all the studied family members by equivalent residues through distinct interaction modes. *Correspondence: We can identify specific polar and charged side chains that selectively interact with donor Ester Boix or acceptor purine groups. Overall, we observe selective bidentate polar and electrostatic Ester.Boix@uab.es interactions: Asn to N1/N6 and N6/N7 adenine groups in mammals versus Glu/Asp and Specialty section: Arg to N1/N2, N1/O6 and O6/N7 guanine groups in non-mammals. In addition, kinetic This article was submitted to and molecular dynamics comparative results on UpG versus UpI emphasize the main Translational Pharmacology, a section of the journal contribution of Glu/Asp interactions to N1/N2 group for guanine selectivity in lower order Frontiers in Pharmacology vertebrates. A close inspection at the B2 binding pocket also highlights the principal Received: 16 April 2019 contribution of the protein β6 and L4 loop regions. Significant differences in the orientation Accepted: 12 September 2019 and extension of the L4 loop could explain how the same residues can participate in Published: 09 October 2019 alternative binding modes. The analysis suggests that within the RNase A superfamily Citation: Prats-Ejarque G, Lu L, Salazar VA, an evolution pressure has taken place at the B2 secondary binding site to provide novel Moussaoui M and Boix E (2019) substrate-recognition patterns. We are confident that a better knowledge of the enzymes’ Evolutionary Trends in RNA Base Selectivity Within the RNase nucleotide recognition pattern would contribute to identify their physiological substrate A Superfamily. and eventually design applied therapies to modulate their biological functions. Front. Pharmacol. 10:1170. doi: 10.3389/fphar.2019.01170 Keywords: RNase, RNA, purine, catalysis, molecular dynamics, evolution, RNase A superfamily Frontiers in Pharmacology | www.frontiersin.org 1 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily INTRODUCTION closely related members suggested a link between shared protein networks and their characteristic biological properties The interest to solve a biological problem is frequently correlated (Narayanan et al., 2018a). The RNase A superfamily includes a to its inherent difficulty. When entering the RNA world we are series of proteins with antimicrobial and immune-modulatory faced with a wide diversity of secondary and tertiary structures. activities and is considered to have emerged with an ancestral An even higher level of complexity is encountered when host-defense role (Boix and Nogués, 2007; Rosenberg, 2008; trying to identify the rules that guide the RNA binding protein Lu et al., 2018). Family members were classified according to recognition process. During the last decades, many efforts have their structural, enzymatic and biological properties into eight been applied to unravel the structural determinants for protein canonical types (Sorrentino and Libonati, 1997; Sorrentino, RNA recognition (Draper, 1999; Allers and Shamoo, 2001; 2010). A better understanding of the structural determinants that Draper, 2015; Terribilini et al., 2007). We are currently witnessing govern the RNases’ substrate specificity can help us to explain the significant advances within the RNA field thanks to the novel divergent functionalities within the family. RNA sequencing methodologies that have laid the path to an Here, we have committed ourselves to undertake a RNA-omics era. Nowadays, we have access to many protein-RNA comprehensive comparative analysis of representative family binding predictors (Miao and Westhof, 2016) and the main basic members and explore the structural drift that has taken place rules that drive the protein–nucleotide interaction process have through evolution to shape the substrate specificity of the been identified (Luscombe, 2001; Denessiouk and Johnson, 2003; secondary base binding site. First, we have performed a kinetic Morozova et al., 2006; Kondo and Westhof, 2011). The study of characterization of the first seven human canonical RNases using RNA cleaving enzymes poses additional complexity. Efficient dinucleotides. Secondly, we have selected representative RNase A RNases should first recognize a specific RNA target, and then superfamily members from lower to higher order vertebrates and provide a proper active site configuration to promote catalysis have performed molecular dynamics simulations of the protein- and ensure the proper cleavage of the substrate. A particular dinucleotide complexes. pharmacological interest relies on the design of tailored enzymes with specific RNA cleavage targets (Tamkovich et al., 2016). Recent work on RNases’ action within a cellular environment is MATERIALS AND METHODS helping to unravel their natural in vivo substrates (Honda et al., 2015; Lyons et al., 2017; Mesitov et al., 2017). A proper knowledge Expression and Purification of the of the RNases’ active site architecture should lead to the design of Recombinant Proteins specific inhibitors of their biological functions (Chatzileontiadou RNase A was purchased from Sigma Aldrich. The cDNA for et al., 2015; Chatzileontiadou et al., 2018). RNase 1 was a gift from Prof. Maria Vilanova (University of In this work, we have explored the nucleotide base preference Girona, Spain) and cDNA for RNase 5 was provided by Prof. within the vertebrate-specific RNase A superfamily. The bovine Demetres Leonidas (University of Thessaly, Greece). RNase 4 pancreatic enzyme RNase A was one of the earliest enzymes to be synthetic gene was purchased from NZYtech (Lisboa, Portugal) studied in the 20th century and is still one of the best characterized and RNase 6 was obtained from DNA 2.0 (Menlo Park, CA, (Cuchillo et al., 2011). All the family members share a common USA). RNase 2, RNase 3 and RNase 7 genes were obtained as three-dimensional fold, catalytic triad and mechanism of action previously described (Torrent et al., 2010). The recombinant on single-stranded RNA. During the last decades, the modular proteins were expressed and purified as previously described subsite arrangement of RNase A for the recognition of bases, (Boix, 2001; Prats-Ejarque et al., 2016). Briefly, the gene was ribose and phosphates has been characterized (Parés et al., 1991; cloned into the pET11c expression vector (Novagen), the protein Nogués et al., 1998). The enzyme cleaves the 3′5′ phosphodiester was expressed in Escherichia coli BL21(DE3) cells (Novagen) bonds with specificity for pyrimidines at the main anchoring and then purified from inclusion bodies. Finally, the protein was site (B1) and preference for purines at the secondary site (B2) purified by cationic exchange FPLC on a Resource S column (GE (Richards and Wyckoff, 1971; Raines, 1998). In a previous work, Healthcare) and lyophilized. Protein purity was confirmed by we analyzed the enzyme residues that were reported to participate SDS-PAGE and mass spectrometry. in the specific binding of adenine (A) and guanine (G) bases at the B2 site among the RNase A superfamily members (Boix et al., Spectrophotometric Kinetic Analysis 2013). A high evolutionary conservation was observed for B1, UpA, UpG and UpI (Biomers, Söflinger, Germany) were used whereas a significant variability was visualized for the secondary as substrates, and the kinetic parameters were determined by a base selectivity. Interestingly, the observed structural differences spectrophotometric method as described (Boix et al., 1999b). at the secondary base site correlate with their substrate specificity Assays were carried out in 50 mM sodium acetate, 1 mM EDTA, and catalytic efficiency (Tarragona-Fiol et al., 1993; Sorrentino, pH 5.5, at 25°C, using 1 cm path length cuvette. Substrate 1998; Boix et al., 2013). Likewise, the analysis of the protein concentration was determined spectrophotometrically using conformational changes induced upon nucleotide binding by the following extinction coefficients: ε260 = 24,600 M−1 cm−1 NMR and molecular dynamics highlighted an evolutionary for UpA, ε261 = 20,600 M−1 cm−1 for UpG and ε260 = 16,400 trend in base interaction selectivity (Gagné and Doucet, 2013; M−1 cm−1 for UpI. The activity was measured by following the Narayanan et al., 2017; Narayanan et al., 2018a). Conserved initial reaction velocities using the difference molar absorbance conformational rearrangements upon ligand binding within coefficients, in relation to cleaved phosphodiester bonds during Frontiers in Pharmacology | www.frontiersin.org 2 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily the transphosphorylation reaction: Δε286 = 570 M−1 cm−1 for constrained with the P-LINCS algorithm (Hess, 2008), with an UpA, Δε280 = 480 M−1 cm−1 for UpG (Imazawa et al., 1968), Δε280 integration time step of 2 fs. = 316 M−1 cm−1 for UpI (experimentally determined). Final The energy of the systems was minimized using the steepest enzyme concentrations were adjusted depending on the RNase descendant algorithm and equilibrated in two steps. First, an activity for each assayed substrate in a range between 0.005 and initial constant volume equilibration (NVT) of 1 ns was performed 10 μM. The reactions were performed in triplicate with 100 with a temperature of 300 K using a modified velocity rescaling μM of substrate and the activity was normalized at an enzyme/ thermostat (Bussi et al., 2007). Then, 1 ns of constant pressure substrate ratio of 1:100. equilibration (NPT) was run at 1 bar with a Berendsen barostat (Berendsen et al., 1984) at 300 K and the same thermostat. Finally, 100 ns production runs were performed under an NPT ensemble Molecular Dynamics Simulations without applying restraints. Three independent simulations in All the molecular dynamics (MD) simulations were performed periodic boundary conditions were conducted for each complex. using GROMACS 2016.2 (Abraham et al., 2015). The force Dinucleotides were generated by modifying the dCpA ligand of field used was a modification of AMBER99SB (Best and an RNase A–d(CpA) complex (Zegers et al., 1994), maintaining Hummer, 2009). Charges of inosine were derived by R.E.D the same initial coordinates. server (Vanquelef et al., 2011). The modifications of the force field to include inosine parametrization are detailed in the Supplemental Materials (Figure S1). All the complexes were RESULTS centred in a dodecahedral cell with a minimum distance box- solute of 1.0 nm. The unit cell was filled with TIP3P (transferable Comparison of Canonical RNases’ intermolecular potential 3P) water (Jorgensen et al., 1983) in Catalytic Activity on Dinucleotide neutral pH conditions supplemented with 150 mM of NaCl. Substrates: A Trend From Guanine to Neighbor search was performed using a Verlet cut-off scheme Adenine Selectivity At the B2 Secondary (Páll and Hess, 2013) with a cut-off of 0.9 nm for both Van der Base Site Waals and coulombic interactions. For long range interactions, In an effort to deepen into our knowledge of the evolutionary smooth particle mesh of Ewald (PME) (Darden et al., 1993; pressure that has guided the nucleotide base preference within Essmann et al., 1995) was used with a fourth-order interpolation the RNase A superfamily, we have compared the catalytic activity scheme and 0.1125 nm grid spacing for FFT. The bonds were of the human canonical members on dinucleotides (Figure 1). FIGURE 1 | Sequence alignment of the eight canonical human RNases together with the RNase A superfamily members analyzed by molecular dynamics (sequences correspond to mature proteins, without the signal peptide). Protein regions identified to participate in B2 site are highlighted in yellow: L4, spanning from b2 to b3, end of β6 (residues 109 and 111) and one of the two catalytic histidines together with a close by residue at β7 (residues 119 and 121). TT indicates the presence of a β-turn. Dots label every 10 residues of the reference protein used (Bt-RNase 1, known as RNase A). The disulphide bonds are labeled with green numbers. Full species names are included in Table S1. Labels are as follows: red box with white character for strict identity; red character for similarity within a group; and character with blue frame for similarity across groups. The alignment was performed using Clustal Omega (Sievers and Higgins, 2018), and the picture was drawn using ESPript (Robert and Gouet, 2014). Frontiers in Pharmacology | www.frontiersin.org 3 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily First, each canonical RNase was expressed using a prokaryote a shift at the substrate secondary base predilection, from lower recombinant expression system. We successfully expressed to higher order vertebrates, from guanine to adenine (Boix et al., and purified with high yield the first seven human canonical 2013). Basically, the characterized fish, amphibian and reptile RNases using the T7 promotor and the pET expression system. RNases show a marked preference for G at B2 site (Hsu et al., 2003; Unfortunately, using the same prokaryote expression system Ardelt et al., 2008), while mammalian prefer A (Richards and we were unable to obtain a properly folded and catalytically Wyckoff, 1971; Zhao et al., 1998; Prats-Ejarque et al., 2016). We active human RNase 8. In fact, inspection of the RNase 8 coding can group the family members, according to their relative activity transcript by Rosenberg and co-workers revealed an unusual on dinucleotide substrates, within three main subcategories by gene organization and protein disulphide pairing, suggesting a their base preference at the B2 site: G > A, G ~ A and A > G significant functional divergence from the canonical characteristic (Figure 2). The results suggest that an evolutionary pressure has structure of the family (Chan et al., 2012). The authors do not taken place to promote selectivity for the adenine base within discard the possibility that RNase 8 is not expressed as a standard the family’s more recently evolved members, coming from an secretory RNase. Therefore, we decided to perform our kinetic ancestral precursor with a marked preference for guanine. study using the first seven human canonical RNases. This is the Last, we have studied the RNases’ activity on UpI dinucleotides. first simultaneous comparison of the catalytic activity of all seven Inosine (I) was selected as an appropriate model to inspect the proteins within a single laboratory. particular effect of the presence of a C=O group at the purine C6 The catalytic activity of the RNases was assayed using atom and the influence of the NH2 group at the C2 position, in dinucleotide substrates, where the first pyrimidine was kept comparison to the other two purine base structures. Detectable invariable as a uridine and the secondary base was substituted activity for the inosine dinucleotide was mainly registered for the by the natural standard purines and the modified base RNases 1, 2, 4 and 5 (Table 1). Overall, kinetic results indicate inosine. Together with the two natural purines incorporated that no important differences are observed between the proteins’ in RNA during transcription, we have also selected inosine, enzymatic activity on UpG and UpI, although a slight preference a modified base frequently present in cellular RNA, as one of for I over G is shown. Interestingly, the family members that have the main post-transcriptional modifications. Kinetic activity a closer relationship to lower order vertebrates (RNases 1, 4 and 5) on UpA, UpG and UpI was measured by a spectrophotometric present a significant activity against dinucleotides with inosine at the assay and the relative preference for the secondary base was B2 position, but no detectable activity in the presence of a guanine. estimated for each protein. Bovine pancreatic RNase A was The results suggest that A/G discrimination within the RNase A taken as a reference control. superfamily relies partly in the recognition of N1/N2 group. Interestingly, the respective catalytic activities of the seven human canonical RNases indicate a shift of the secondary base specificity, from a poor A/G discrimination to a pronounced B2 Base Selectivity Within the RNase preference for A (Table 1). In particular, the human RNase A Superfamily by Molecular Dynamics 5, which is the canonical member more closely related to Following, to complement the enzymatic characterization of the ancestral RNases (Sorrentino, 2010), shows only a mild canonical RNases, we performed a comparative analysis within preference for adenine over guanine. In turn, the pancreatic- the RNase A superfamily by molecular dynamics. To gain insight type RNase 1 shows a significant preference for adenine at B2 into the structural determinants of the protein recognition position. Last, the more evolved RNase subgroups (types 2/3 pattern at the B2 site, we have selected here representative and 6/7) do not have any detectable activity using UpG as a members for each vertebrate family subgroup. Ten representative substrate (Table 1, Figure 2). RNases were chosen and their binding mode to dinucleotides On the other hand, when we analyze the kinetic characterization was predicted by MD simulations. Figure 2 illustrates the of other family members available in the literature, we can infer selected proteins and their evolutionary relationships. When no TABLE 1 | Kinetic activity of RNase A and the human RNases 1–7. V0 (μmol/min) B2 ratio UpA UpG UpI A/G G/I Bt-RNase A 0.783 ± 0.033 1.49·10−2 ± 9.99·10−4 3.62·10−2 ± 2.22 ·10−3 52.65 0.411 Hs-RNase 1 0.108 ± 3.93·10−3 1.15·10−3 ± 6.09·10−6 7.87·10−3 ± 5.54 ·10−4 93.63 0.146 Hs-RNase 2 2.57·10−3 ± 1.32·10−4 n.d. 5.53·10−4 ± 1.56 ·10−4 ∞ 0 Hs-RNase 3 2.39·10−3 ± 4.96·10−5 n.d. n.d. ∞ ∞ Hs-RNase 4 5.92·10−2 ± 3.97·10−3 2.38·10−3 ± 2.86·10−4 6.86·10−3 ± 1.7 ·10−4 24.85 0.348 Hs-RNase 5 2.39·10−4 ± 3.74·10−5 1.94·10−5 ± 2.5·10−6 5.33·10−5 ± 8.19 ·10−6 12.29 0.365 Hs-RNase 6 6.74·10−3 ± 1.90·10−4 n.d. n.d. ∞ ∞ Hs-RNase 7 4.25·10−5 ± 6.21·10−6 n.d. n.d. ∞ ∞ The reactions were performed using 100 μM of substrate. Initial velocity (V0) of dinucleotide phosphodiester bond cleavage is indicated. The average of three replicates is shown. Standard error of the mean is shown. n.d.: not detected at the assayed conditions. Bt, Bos taurus; Hs, Homo sapiens. Frontiers in Pharmacology | www.frontiersin.org 4 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily FIGURE 2 | Phylogenetic tree of the RNases studied in the kinetic and molecular dynamics analysis. Right column indicates the UpA/UpG preference ratio according to kinetic results (see Table 1). Left column includes the A/G preference at B2 according to the literature. The evolutionary history tree was inferred by using the maximum likelihood method and JTT matrix-based model (Jones et al., 1992). The tree with the highest log likelihood (−3745.36) is shown. The percentage of trees in which the associated taxa clustered together is shown next to the branches. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbour-Join and BioNJ algorithms to a matrix of pairwise distances estimated using a JTT model, and then selecting the topology with superior log likelihood value. The tree is scaled, with branch lengths measured as the number of substitutions per site. Evolutionary analyses were conducted in MEGA X (Kumar et al., 2018). A more complete phylogenetic tree of RNase A family members is included in Figure S2. Blue star highlights the selected representative RNases to perform MD simulation runs. solved 3D structure was available (RNase 4 of Ornithorhynchus representative members of avian, reptiles, amphibian and fishes anatinus and RNase 1 of Iguana iguana), a prediction model was were chosen, based on the availability of previously solved 3D generated using the Modeller software by the ModWeb server structures. Chicken RNase 1 was taken (UniProtKB P27043) as (Webb and Sali, 2016). From lower to higher order vertebrates, the only member with a known 3D structure (Lomax et al., 2014). the following organisms were analyzed: Danio rerio (Dr), In turn, reptiles have been represented by turtle (UniProtKB Rana pipiens (Rp), Iguana iguana (Ii), Chelonia mydas (Cm), P84844) and iguana (UniProtKB P80287) (Nitto et al., 2005). Gallus gallus (Gg), Ornithorhynchus anatinus (Oa), Bos taurus Next, we selected the northern leopard frog (R. pipiens) RNase (Bt) and Homo sapiens (Hs). Representative organisms were (also named Onconase, UniProtKB P22069), which has been selected based on the available information on the evolutionary extensively characterized because of its antitumoral properties relationships and previous structure-functional characterization (Boix et al., 1996; Lee and Raines, 2003; Lee et al., 2008). Lastly, studies (Goo and Cho, 2013). We also indicate, when known, the for fish representative sequences, we selected D. rerio RNases canonical type of each selected RNase (Sorrentino and Libonati, (Dr-RNase 1 and Dr-RNase; UniProtKB A5HAK0 and E7FH77), 1997). Within the placental mammals, we have included two also named as zebrafish RNases 3 and 5 respectively. Both RNases representative human members (RNases 2 and 6; UniProtKB were previously reported to display a high catalytic activity in P10153 and Q93091), which are expressed during infection and comparison to other fish homologues (Cho and Zhang, 2007; inflammation and are endowed with a high catalytic activity. Pizzo et al., 2011). In particular, the zebrafish 5 (Dr-RNase) was We have also selected the bovine pancreatic RNase, or RNase classified as one of the most ancestral family members, showing A (UniProtKB P61823), which is the family reference member a high catalytic activity along with both antimicrobial and and one of the best characterized enzymes (Cuchillo et al., 2011). angiogenic properties (Pizzo et al., 2011). In all cases, previously Bovine pancreatic RNase belongs to the RNase 1 type. On the reported 3D structures were taken as a reference, except for the other hand, for early mammals, we selected the platypus (O. platypus RNase, where a prediction model had to be generated. anatinus), an egg-laying animal and precursor to higher order To compare the RNases’ selectivity at the B2 site, the three vertebrates, before divergence of placental RNases. Accordingly, dinucleotides, UpA, UpG and UpI, were selected (see Figure S3 the platypus RNase belongs to type 4 (UniProtKB F6ZXU1), for atom nomenclature). Molecular dynamics were performed and was previously identified as the predecessor of higher order using GROMACS software as detailed in the methodology. mammalian RNase types (Goo and Cho, 2013). Following, Triplicates for each protein complex were carried out at 100 Frontiers in Pharmacology | www.frontiersin.org 5 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily ns. The RMSD between the dinucleotide positioning during by the L4 loop and β6 strand structures (Figure 4). Contribution the simulation is shown in Figure S4. The following common of each interacting residue was monitored as a function of time. criteria were established to confirm at the end of each modelling Each run was subdivided into initial, central and late periods. run that the nucleotide is positioned in a productive orientation, Although some mobility of the substrate positioning is observed favorable for catalysis: phosphate location at the RNase catalytic during the 100 ns MD production runs (Figure S4), overall no triad and pyrimidine proximity to B1 site. Equivalent residues to major significant differences are identified as a function of time. RNase A (H12/K41/H119 at the catalytic triad and T45 at B1 site) The most representative interacting residues and atom types were taken as a reference for each protein. involved in each modelled complex are summarized in Figure 5. The dinucleotides’ RMSD fluctuations during each production Figure 5 illustrates the main residues that contribute to B2 run indicate a reduced substrate mobility, oscillating within a base recognition. The figure indicates the main residues that value range of 0.1–0.4 nm (Figure S4). The total hydrogen bond were found involved in interactions with the purine ligand for interactions per residue were calculated for each simulation and at least one third of the total 100 ns molecular dynamics run. expressed as a fraction of occurrence. Figure S5 illustrates the We observe the contribution of polar and charged residues that interacting residues with the purine moieties. act as acceptors/donors to purine representative groups. We Overall, we observe at the end of each simulation run a similar can identify the protein residues that can provide a bidentate productive positioning of the dinucleotides at the active site cleft anchoring with the purine base and selectively interact with for most of the studied proteins (Figures 3 and S6). However, unique base groups. In particular, we find specific discriminators comparison between all different RNase–nucleotide complexes for adenine (N1/N6 and N6/N7 groups) versus guanine (N1/O6 and among triplicates highlights that most variability is located and N1/N2 groups). Likewise, discrimination between guanine at the purine moiety (Figure 4). Likewise, time course analysis and inosine binding was identified by looking for the residues for each dynamic run shows significantly much higher mobility with specific interactions at the base N1/N2 group, unique for the purine nucleoside in comparison to the pyrimidine main to guanine. nucleoside and phosphate portions. We can confirm that the Each studied family member was analyzed taking bovine protein phosphate p1 and base B1 sites are mostly conserved pancreatic RNase A sequence numbering as a reference (see Table among all the family members and provide stronger and more S2 and Figure 1) (Raines, 1998; Boix et al., 2013). The adenine specific interactions. base is fixed in bovine RNase by residues Asn67, Gln69, Asn71, Following, we have analyzed the specific binding interactions at Glu111 and His119 (see Figure 5). The reliability of the dynamic the B2 purine portion. Specific binding residues at the B2 site were simulation was first evaluated by comparing the obtained results identified. In the majority of complexes, the purine base is fixed for RNase A using UpA with the previous structural work by FIGURE 3 | Schematic illustration of RNase–UpA complexes obtained at the end of the MD 100 ns simulation run. Colors of each vertebrate group are taken according to the family phylogenetic tree selection (see Figure S2). Mammal RNases are shown in shades of red, turtle RNase in magenta, chicken RNase in orange, frog RNase in green and fish RNase in blue. Some parts of the RNase are colored in gray to allow better visualization of the ligand and the interaction regions of the protein. The picture was generated using PyMOL 1.7.2 (Schrödinger, Inc). Frontiers in Pharmacology | www.frontiersin.org 6 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily Glu111 and His119). Specific bidentate interactions for adenine are provided by Asn71/Gln69 at N1/N6, Asn67 at N6/N7 and Glu111 at N6. In particular, our molecular dynamics results corroborate the key contribution of all Asn71 counterparts in mammalian members for adenine specificity. On the other hand, we observe the flexibility of residues such as Glu111, which can offer a bidentate anchoring at either the NH2 group at C6 position in adenine or at N2/N1 groups in guanine (Figure 5). In addition, we observe the contribution of the His119 catalytic residue by π-π interactions with the purine 5-membered ring in all the predicted complexes for any of the three assayed dinucleotides (Figure 5). Previous structural studies have FIGURE 4 | Overlapping of all RNase-UpA and RNase-UpG complexes revealed that the catalytic His119 in the free protein can adopt obtained by molecular dynamics (see Figures 3 and S6). The secondary two conformations (A and B), where only one rotamer (A) is base binding site (B2) is highlighted (loop L4 and strand β6). As shown in compatible with catalysis and purine interaction (Berisio et al., Figure 3, the molecules are colored according to vertebrate groups as 1999; Merlino et al., 2002). Favored stacking interactions of the indicated in the family phylogenetic tree (see Figure S2). Mammal RNases His imidazole with the purine ring are suggested to participate are shown in shades of red, turtle RNase in magenta, chicken RNase in orange, frog RNase in green and fish RNase in blue. in nucleotide discrimination (Gagné and Doucet, 2013). In our molecular dynamics study we cannot find any significant differences between the complexes obtained with any of the three X-ray crystallography on RNase A–dinucleotide complexes dinucleotide types. On the other hand, significant differences (Boix et al., 2013). In particular the adenine binding residues are observed for some particular RNases, where the purine ring identified by molecular dynamics were compared with the RNase is also establishing cation-π interactions with other residues, A–d(CpA) complex (Zegers et al., 1994), where the same residues in particular arginine (such as Arg68 in Hs-RNase 2, Arg66 in for adenine binding had been identified (Asn67, Gln69, Asn71, Gg-RNase, Arg117 in Cm-RNase and Arg8 in Dr-RNase 1 (see FIGURE 5 | Schematic representation of the predominant interactions observed during the 100 ns of each simulation run. Only the interactions present during more than 30% of the total dynamic runs were considered. The interactions observed in more than 50% of the total dynamic runs are highlighted in bold. Protein residues are listed according to RNase A numbering reference (see Table S2 for structural residues overlapping between the analyzed homologues). Polar and electrostatic interactions are detailed for each location. The box labeled with a star includes the shared common van der Waals interactions for the three base types. Frontiers in Pharmacology | www.frontiersin.org 7 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily Figure 5). Overall, we observe that most differences among the adenine binding. Finally, another substitution that is observed to studied family members are located at the L4 Loop. The loop favor guanine binding in non-mammalian RNases is Ala122 to mobility is restricted by a disulphide bridge (Cys65–Cys72 pair Arg. The Arg counterpart residues in fish and turtle RNases can in RNase A), that is conserved in most mammalian RNases interact by bidentate interactions with the O6/N7 group of the (except in RNase type5/angiogenin-like), but absent in all the guanine/inosine bases (Figure 5). non-mammalian vertebrate groups (see Figure 1). Overall, although key residues for purine binding are mostly We can conclude from the analysis of predicted protein– conserved in all the studied members, such as Asn71, His119 and dinucleotide complexes that the main key residues for purine Glu111, our molecular dynamics analysis indicates that distinct interactions (Asn71 and Glu111) are mostly conserved among binding modes could promote a shift from G to A at the B2 site. all the studied family members, although distinct binding modes are identified depending on the nature of the purine base. Asn71 in RNase A, and equivalent residues both in human and platypus An Evolutionary Trend Shaping the B2 proteins, specifically bind by a bidentate interaction at the N1/ Selectivity Within the RNase A N6 of the adenine ring. Likewise, the Asn side chain can establish Superfamily Lineage equivalent interactions for guanine or inosine binding, by shifting To validate the significance of the residues identified by MD to their NH and C˭O amide groups and thereby interacting with the participate in purine recognition, we have supplemented our respective N1/O6 groups. However, these interactions are not so study with the comparative analysis of other family member often observed for guanine/inosine interaction and frequently close homologues. Accordingly, each representative member only the Asn binding to the O6 group is identified. analyzed by MD simulations has been compared within its own When we inspect the non-mammalian vertebrate members, vertebrate subgroup. By close inspection of sequence alignments, we find a similar scenario: an Asn residue (Asn71 RNase A we have identified the counterpart to the key residues for counterpart) can also interact with both N1/N6 groups for A binding of a purine at B2 location. Figures S7–S11 include the and N1/O6 in G/I in turtle, frog and fish proteins. Significant respective sequence alignments within each vertebrate subgroup. differences are found in chicken RNase, where an Arg is located The relationships between all the aligned sequences of family at the same position. On its turn, the nearby residue Gln69 homologues are illustrated in the phylogenetic tree included in would contribute to provide a specificity for adenine. A Gln Figure S2. at this position is only present in the pancreatic RNase 1 type. First, we have analyzed the fish RNase sequences, taking as Substitutions of Gln by an Argin Hs-RNases 2 and 6 and platypus a reference D. rerio RNase 1 (Pizzo et al., 2011), also named RNase favor the bidentate interaction with G/I at the N7/O6 zebrafish RNase 3 (ZF3). Acharya and co-workers solved the group. The equivalent counterpart in fish is an Asn (), which crystal structure of this RNase together with a polymorphism shows a preference for guanine/inosine binding. No equivalent variant (Kazakou et al., 2008). The chosen protein structure residues are found in any other lower order vertebrates, due corresponds to the variant identified as ZF3e. Overall, the to a deletion in the loop L4 region from residues 65 to 71 (see researchers identified five protein variants, with substitutions Figures 1 and S7–S11). In addition, we find another Asn residue at six sequence locations. Among them, we observe that one in mammalian RNases that is also favoring the adenine versus of the residues involved in the purine binding (Arg123) is only guanine binding: Asn67 (Figure 5). In this case, the Asn is present in the ZF3e polymorphism and is substituted by a Lys providing a bidentate interaction to N6/N7 adenine groups. The in the other variant. On the other hand, comparison with the presence of an additional Asn is also found in zebrafish RNase other fish RNase sequences (Figure S7) highlights the presence 5 (Dr-RNase) but is missing in all the other studied lower order of one or two conserved Asn residues at L4 loop region. The loop vertebrates. Interestingly, the shorter L4 loop version in the fish is present in fishes in a short-reduced version in comparison to protein still permits the appropriate Asn positioning. the extended version present in more evolved mammal RNase The molecular dynamics results also highlight two other types: 2/3–6/7/8 (Figure 1). However, the Asn residue at position protein regions, which are also participating in the purine 72/74 (corresponding to positions 67 and 71 in RNase A) can binding: residues 109–111 (β5) and 119–121 (β6). In particular, also participate in the adenine interaction but would preferably we observe the main contribution of Glu111 in Bt-RNase A and interact with the N1-O6 group of a guanine. Noteworthy, several the respective Glu/Asp counterparts in the other studied family fish RNase sequences display an Asp at 74 position, which members (Figure 5 and Table S2). Both the Glu/Asp bidentate according to molecular dynamics results is a suitable binder for anionic side chains are observed to bind at both the NH2/N6 guanine. Two other anionic residues at the protein C-terminus are adenine and the N2/N1 guanine specific groups. However, Glu key for the studied RNase complexes; that is Glu114 and Glu122 substitution by an Asp residue (found in Hs-RNases 2 and 6) (corresponding to Glu111 and Asp121 in RNase A counterparts). prevents, or reduces drastically, the base interactions. Similar While most fish RNases show a Glu at 114 position, we also find interactions at the adenine N6 NH2 group and the guanine N2/ in some cases the presence of an Asp. This is the case of zebrafish N1 group are established by Asp121 at Bt-RNase A and their 5 (Pizzo et al., 2011), which was reported to have a relative much counterparts in mammals and chicken RNases. Although a Glu/ higher catalytic activity than the other characterized fish RNases Asp is present in all the studied proteins, frog RNases show (Pizzo et al., 2011). Likewise, residue Glu122 is either conserved significant differences. Interestingly, the zebrafish 3 counterpart or substituted by an Asp residue. Finally, molecular dynamics (Glu122) interacts with guanine base but is not involved in reveal the presence of an Arg residue at the zebrafish proteins’ Frontiers in Pharmacology | www.frontiersin.org 8 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily N-terminus that shows favored interactions to guanine and highlighted, although distinct conformations are observed among inosine. The Arg is only present in about 50% of the analyzed the bullfrog RNases that could account for the higher catalytic fish sequences. activity displayed by the oocyte RC-RNase. When we overlap the Following, we inspected the residues potentially involved in reported NMR structures with our modelled structures in complex purine binding in amphibians. In this vertebrate group we also with dinucleotides, we also observe that the oocyte RC-RNase is the observed a short version of the L4 loop. However, in comparison only one that has an Asn residue at an equivalent position to Asn71 to fish RNases, the analyzed amphibian members show a less in RNase A, that can establish interactions with the N1-O6 group optimal loop conformation. The loop is orientated to the opposite of the guanine. Therefore, the higher catalytic efficiency of bullfrog direction respect to RNase A, and lacks one of the key Asn found oocyte RNase respect to Onconase could be mostly attributed to in mammalian RNases. In particular, in northern leopard frog R. residue Asn57, which is substituted by a Ser in the latter. When pipiens RNase (Onconase) we can identify Asn56 (counterpart we compare the sequence identities of the distinct frog RNases, of Asn71 in Bt-RNase A) but no other equivalent residues in we observe a high variability at the loop L4, where Asn residues the region (Figure 1). Comparative structural alignment only are mostly substituted by either a Ser or an Asp (Figure S8). reveals the presence of a conserved Glu residue at position 91 Besides, presence of Pro and short amino acid insertions in other (Glu111 counterpart in RNase A). Molecular dynamics results amphibian RNases might also modify significantly the interaction on Onconase interaction with dinucleotides have been compared at this site. Interestingly, whereas most R. catesbeiana RNases show with the previous reported solved crystal structure in complex a particular four amino acid insertion, we found several Xenopus with a tetranucleotide (Lee et al., 2008). Raines and co-workers species that display an alternative loop version, with a slightly studied in detail the enzyme binding to the d(AUGA) substrate extended insert (Figure S8). analogue and observed that while an equivalent binding pocket is In turn, reptiles present a short version of the L4 loop (see conserved for the pyrimidine base at B1, significant differences are Figure S9), with a similar length to the one observed in fishes, found for the B2 site. In particular, specific bidentate interactions although encompassing a higher sequence divergence at the of Glu91 with the guanine base were identified at B2 position. region. In particular, most species include only one Asn within Moreover, the authors confirmed by site-directed mutagenesis the region. In our molecular dynamics study of iguana and turtle that this residue was responsible for the frog RNase preference of RNase-dinucleotide complexes we can identify one Asn (Asn68 guanine over adenine. In addition, the authors also highlighted in turtle RNase and Asn67 in iguana) equivalent to the Asn71 the importance of the nature of the nearby residue located at counterpart in RNase A (Table S2). Close inspection of sequence position 89 (Onconase counterpart of Ala109 in RNase A). alignment identifies few reptile species with two Asn residues Ala109 is conserved in all mammalian and most reptile sequences at 67/71 positions (such as the species of the Micrurus or Boiga but shows a significant variability in fishes and amphibians. genera), whereas other species show an Asn to Asp substitution Interestingly, ZF3 presents an Ala at this position, as observed at position 71. However, we observe an overall higher variability in mammals, whereas other fish RNases have a polar or cationic at L4 loop, which incorporates non-conserved substitutions residue (Thr/Lys or Arg), as observed in frog RNases. Substitution (Figure S8). We have also analyzed within reptiles the other of Thr89 in Onconase by an Asn residue reduced the enzyme’s G > residues that were identified in turtle or iguana to potentially A preference. The authors suggested that long-range electrostatic participate in binding at the B2 site (Figure 5). Rosenberg and interactions were key for the enzyme turnover activity on cellular co-workers characterized the RNase from iguana, which is RNA substrate in physiological conditions (Lee et al., 2008). The mostly expressed in the pancreas and displays a significantly hypothesis was further backed up by recent NMR and molecular high catalytic activity (Nitto et al., 2005). In the present work, dynamics studies by Doucet and collaborators, that emphasized productive binding conformations obtained by molecular the key role of network interactions connecting distant protein dynamics of turtle RNase with dinucleotides highlight the residues (Narayanan et al., 2018a). Interestingly, site-directed contribution at the protein C-terminus of Asp116 and Arg117 mutagenesis in Onconase revealed also the contribution of the (Asp116, counterpart of Asp121 in RNase A, is only present in N-terminus in the B2 base discrimination (Lee et al., 2008). few reptile sequences). In turn, the presence of an Arg at position In particular, insertion of an Arg at position 5 is significantly 117, not shared by all the family homologues, is rare. enhancing the frog RNase catalytic activity. Likewise, in our Avian RNases present the shortest L4 loop version that molecular dynamics analysis we observe equivalent Arg residues incorporates the most significant deviation from the L4 loop at the protein N-terminus of turtle and fish RNases that contribute consensus sequence (Figure S10). Most L4 sequences do not to purine binding (Figure 5 and Table S2). include any Asn residue. In our modelled complex of chicken Molecular dynamics results of Onconase were also compared RNase, Asn65 is equivalent to Asn67 in RNase A. However, with the structural information reported for bullfrog (R. we did not observe any direct participation of Asn65 in purine catesbeiana) RNase purified from oocytes (RC-RNase), the most binding. In turn, the neighboring residue Arg66 is significantly catalytically active frog RNase (Chang et al., 1998; Lee and Raines, participating in B2 binding and was observed to bind to any 2003). A structural study by NMR of bullfrog oocyte RNase of the three purine bases. Arg66 position can be equated to analyzed the enzyme interaction with tetranucleotides (Hsu et al., Bt-RNase A Asn71 counterpart, although the loop conformation 2015). The authors reported a much higher catalytic activity for is very divergent at this region (Table S2). Arg66 is only found in oocyte RC-RNase in comparison to RC-RNase 2 and RC-RNase few bird sequences but is located close to Arg66/70 in some other 4. The contribution of the L4 loop to guanine binding was also mammalian RNases (Hs-RNase 2, Hs-RNase 6 and Oa-RNase Frontiers in Pharmacology | www.frontiersin.org 9 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily in our study; Figure 5 and Table S2). Most strikingly, there is were already present in most ancestral RNases; but the observed no equivalent Glu/Asp counterpart to Bt-RNase A Glu111. On purine selective specificity is modulated in each family member the contrary, a Trp is present at that location in the studied by complementary interactions of environment residues. chicken RNase (Figure 1). Trp is conserved in some chicken and snake (Boiga) RNase sequences (Figures S9, S10). In other avian sequences we find another bulky hydrophobic residue, DISCUSSION followed by Asp, which might substitute the Glu111 function. Interestingly, in our molecular dynamics study we find the The RNase A superfamily is currently a reference model for contribution of stacking interaction of Trp105 with the purine evolutionary and enzymology studies. Although a wealth base. On the other hand, the presence of a residue equivalent of information is available on ruminant evolution and the to Asp121 is only observed in some of the sequences, whereas pancreatic-type RNases (Beintema and Kleineidam, 1998; others show a substitution by an Ala. In any case, a higher Goo and Cho, 2013; Xu et al., 2013; Lomax et al., 2017), a proportion of non-productive dinucleotide binding is obtained comprehensive full understanding of the whole family is still by molecular dynamics (>75% of all run assays) in relation to the missing. Following a pattern characteristic of host-defense other studied members (<30% in fish, frog, turtle or platypus), proteins, the RNase A family has undergone frequent duplication which might be attributed to the chicken RNase’s different and gene sorting events (Rosenberg et al., 1995; Zhang et al., conformations of the L4 loop and the presence of Arg66 and 2000; Zhang et al., 2002; Liu et al., 2015). Many studies have Trp105, that tend to establish stacking interactions with the tried to unveil the structural determinants for the distinct purine base. Noteworthy, Rosenberg and co-workers performed RNases’ biological activities (Lu et al., 2018); however, we find a comparative study of available sequences for chicken RNases much less information on the evolutionary trends that shaped and concluded that the evolution within this group of proteins the family’s enzymatic diversity. Nonetheless, the understanding might not respond to functional constraints directly related to of the evolutionary processes that determined the enzymes’ the enzyme catalytic activity (Nitto et al., 2006). Comparison of substrate selectivity is key to unravel their physiological roles. two chicken leukocyte RNases identified key regions for either Distinct nucleotide specificities should respond to an adaptation antimicrobial or angiogenic activity. By construction of hybrid to their respective biological functions (Narayanan et al., proteins, they concluded that following a duplication event, a 2018b). Undoubtedly, mastering the structural basis for protein selective evolutionary pressure unrelated to the protein enzymatic nucleotide recognition is essential to assist the design of novel activity had taken place. In our study, we have selected the only anti-infective and immunomodulatory drugs. available 3D structure of a chicken RNase. Unfortunately, this Here, we have compared for the first time the catalytic RNase corresponds to the angiogenin-type RNase instead of activity of the human canonical RNases. The analysis of all the the other characterized chicken RNase (leukocyte RNase-A2 or recombinant proteins, obtained by the same expression system RSFR-RNase), which displays a much higher catalytic activity and using the same kinetic characterization methodology, (Nitto et al., 2006). ensures a reliable comparative evaluation of their respective In contrast to lower order vertebrates, we observe that all efficiencies. To unravel the enzyme specificity for the binding of mammalian RNases, except the RNase 5 type, share an extended the purine secondary base, we have tested the respective catalytic L4 loop fixed by a disulphide bridge (Cys65 and Cys72 in RNase activity of the distinct canonical RNases using UpA, UpG and A). In contrast, all the non-mammalian vertebrate RNases have UpI dinucleotides as a substrate. Interestingly, when we compare either a single Cys or none at this location (Figure 1). Mammalian the A/G ratio at B2 site for the studied seven canonical RNases RNases’ extended L4 loop includes in all cases the RNase A (Figure 2), we observe a pronounced evolutionary tendency from Asn71 counterparts and, in the majority of cases, the RNase A guanine to adenine preference. Previous evolutionary studies Asn67 residue. On the other hand, more variability is observed at identified within the RNase A superfamily the phylogenetic 69 position, where either a Gln, Ser or Arg is found (Figure S11). relationship between the eight canonical subtypes (Cho and On its side, Glu in position 111 is mostly conserved but can also Zhang, 2007). By comparative analysis of the family coding be substituted by an Asp or even a Lys residue. Our MD results sequences we can order the different RNase types from ancestral indicate that the presence of the shorter Asp residue is associated to modern as follows: 5, 4, 1 and 6/7/8–2/3 group (Zhang, 2007; in Hs-RNase 2 and Hs-RNase 6 with scarce or null interactions Sorrentino, 2010) (see the family phylogenetic tree in Figure S2). with the purine (Figure 5). Structural crystallographic data on Kinetic results of human canonical RNases follow the same Hs-RNases 2 and 3 complexes (Mohan et al., 2002) also highlighted ordering when considering the UpA/UpG ratio (Table 1). The the different interaction mode of Asp at this position. Hs-RNase present result is in agreement with previously reported kinetic 6 structural studies also indicated that the Glu to Asp substitution data, where lower order vertebrates, such as amphibians and might significantly modify substrate specificity (Prats-Ejarque reptiles, show a preference for G (Liao, 1992; Irie et al., 1998), et al., 2016). On the other hand, it is also interesting to note the while mammalian RNases have a clear preference for A (Boix presence of an Arg at position 122 (Arg132 in RNase 2), which is et al., 2013). In addition, the recent kinetic characterization of shared with some fish and other lower order vertebrates (Table human RNase 6 corroborated the previously reported preference S2) and might provide a significantly differentiated specificity. for adenine at B2 for human RNases 2 and 3 (Boix et al., 1999a; Overall, we can conclude that counterpart residues to Asn71 and Sikriwal et al., 2007; Prats-Ejarque et al., 2016). On its turn, Glu111 in Bt-RNase A, shared by all the mammalian RNases, human RNase type 5 shows a much less pronounced preference Frontiers in Pharmacology | www.frontiersin.org 10 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily for adenine over guanine (Acharya et al., 1994; Shapiro, 1998). tetranucleotides identified the main residues that conformed Early kinetic characterization of human RNase 5 (angiogenin) the RNase A substrate binding subsites (Fontecilla-Camps et al., already reported its poor discriminating ability on the purine 1994; Nogués et al., 1998; Hsu et al., 2015). located at B2 position (Shapiro et al., 1986; Harper and Vallee, Notwithstanding, despite the RNase protein small size and 1989). Vallee and co-workers engineered an RNase 5 hybrid structure stability, that facilitated the pioneer biochemistry works protein by replacing the L4 loop residues 60-70 with the RNase during the first half of the 20th century, the polymeric nature A counterparts, successfully enhancing the enzyme catalytic and structural complexity of the substrate is still challenging activity on UpA dinucleotides (Harper and Vallee, 1989). Further the enzymologists. In this context, it is important to analyze the work by site-directed mutagenesis in angiogenin suggested that protein family members as a whole dynamic entity. The protein an Asn at Gln69 position in RNase A would provide most of has a kidney-shaped structure conformed by two domains that the purine selective binding. On the contrary, replacement of delimitate the catalytic active site groove. The open and closed the Glu111 RNase A counterpart (Glu108 in RNase 5) did not conformation of enzymes were compared in the presence of significantly alter the enzyme B2 selectivity (Curran et al., 1993). nucleotide ligands (Watt et al., 2011; Gagné and Doucet, 2013). Indeed, structural and molecular dynamics simulation studies Key residues involved in the RNase protein motion would have indicate that Glu111 in RNase A could contribute to either co-evolved to shape the enzyme catalytic efficiencies, as described adenine or guanine binding by alternative modes, by direct or for other enzyme families (Maguid et al., 2006; Ramanathan water-mediated interactions. The nature of the nearby residue and Agarwal, 2011; Narayanan et al., 2018a). Within the RNase (109 in RNase A) could determine the potential participation A superfamily we observed the conservation of key domains of the corresponding Glu residue (Glu111 in RNase A) in the involved the protein motion (Merlino et al., 2003; Gagné and purine binding. The hypothesis was elegantly confirmed by site- Doucet, 2013). Notwithstanding, comparative studies from lower directed mutagenesis studies in Onconase (Lee et al., 2008). to higher order family members infer an inverse relationship Raines and co-workers demonstrated that the B2 site specificity between the protein’s structural rigidity and its catalytic efficiency could be shifted from guanine to adenine preference by impeding (Merlino et al., 2005; Holloway et al., 2011). the long distance network interactions that Glu establishes for Although our molecular dynamics runs using dinucleotides the purine recognition (Lee et al., 2008). Likewise, substitution of are overall in agreement with the reported crystal complex Glu111 by the shorter Asp side chain in human RNases 2 and 6 structures (Fontecilla-Camps et al., 1994; Zegers et al., 1994; could enhance the adenine versus guanine discriminating power Leonidas et al., 2001; Mohan et al., 2002; Lee et al., 2008), we in relation to ancestral RNases, such as RNase-type 4 and 5, as do observe some significant differences. This might be due to observed in our kinetic comparison studies (Table 1, Figure 2). the allowed protein flexibility during the molecular dynamics In an effort to unravel the structural determinants underlying simulations, a fact that could enable a better accommodation the observed differentiated kinetic behaviors, we have carried out of the nucleotide substrates. Besides, MD studies permitted a molecular dynamics analysis of RNase-dinucleotide complexes. us to work with the natural enzyme substrates, rather than the Representative family members were chosen from lower order analogues, commonly used in crystallographic studies. On the vertebrates to placental mammals. Overall, molecular dynamics other hand, NMR titration studies using mononucleotides could corroborate the observed shift from guanine to adenine preference only mimic the enzyme interactions that were to take place by kinetic analysis (Table 1, Figure 2). Notwithstanding, results with the enzyme reaction product (Narayanan et al., 2018a). also highlight that conservative sequence identities are frequently Interestingly, when we analyze the results of our molecular not accompanied by equivalent substrate binding. A similar simulation, we can observe significant differences among the conclusion was reached by NMR analysis of RNases’ nucleotide residues that participate in the distinct periods of the reaction. binding (Narayanan et al., 2018a). Therefore, no straightforward Mostly, interactions with the purine base are frequently lost at conclusions can be directly inferred from the identification the end of the production run. Interestingly, in our modelling of individual interactions to nucleotides by mere structural studies we observe how the substitution of Glu111 by an Asp overlapping analysis. On the other hand, although molecular residue in human RNases 2 and 6 is only participating in the dynamics considers the protein–ligand complex as an entity in purine binding at the initial step of the reaction. In contrast, the motion and provides the equivalent freedom and flexibility that positioning of the pyrimidine base, located at the main B1 site, could be found in experimental conditions, the methodology and the phosphate are mostly retained during all the simulation has also its own limitations when trying to simulate the enzyme run, as reported in previous molecular dynamics using RNase A behavior. Fortunately, Bt-RNase A, the family reference member, or angiogenin (Madhusudhan and Vishveshwara, 2001). Indeed, has been one of the best enzymes ever characterized (Raines, 1998; Raines and co-workers’ kinetic studies indicated that the RNase Cuchillo et al., 2011). RNase A was classified by earlier studies A catalytic mechanism relies mostly on the substrate association as an almost “perfect” enzyme, where the transphosphorylation step (delCardayre and Raines, 1994). A high catalytic efficiency state is not limited by the transition state (Albery and Knowles, would mostly be associated to the enzyme facility to throw away 1976). Raines and coworkers analyzed the behaviour of RNase the product from the catalytic site. In this context, previous A on UpA substrate by experimental kinetics and concluded studies emphasized the importance of the active site flexibility for that the cleavage efficiency is mostly limited by the substrate substrate recognition, catalysis and product release (Sanjeev and desolvation (Thompson et al., 1995). Early crystallographic Vishveshwara, 2005; Gagné et al., 2012; Gagné and Doucet, 2013). and NMR studies of RNase A in complex with mono-, di- and The authors identified two main clusters involved in the protein Frontiers in Pharmacology | www.frontiersin.org 11 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily motion that participate in substrate recognition and product First, bidentate interactions can mainly discriminate between release. In particular, L4 loop plays a key role in the protein motion binding to either adenine or guanine at N1/N6 or N1/O6 groups (Gagné et al., 2012). In addition, a distant residue, Ala109, was respectively. In addition, we observe specific interactions at N7/ identified in RNase A to work as a hinge and promote the active N6 for adenine versus N7/O6 for guanine; and eventually specific cleft opening and product release (Gagné et al., 2015). Ala109 is binding at guanine N1/N2 group. A summary of the most conserved in almost all the studied vertebrate members, except representative residues that provide selectivity for each base is in frog and chicken RNases. To note, chicken family members are shown in Figure 6. Although no universal rules can be written characterized by a much lower catalytic efficiency. On the other for protein-nucleotide base binding, the residues identified in hand, comparison of zebrafish proteins indicates that presence our study for RNase A superfamily members match most of the of an Ala or Gly at this position is associated to high catalytic previously reported in the literature (Luscombe, 2001; Kondo efficiency (Kazakou et al., 2008). On the other hand, a network and Westhof, 2011). Our previous statistical analysis of protein– of sequential hydrogen bond interactions was found mostly nucleotide complexes available at the Protein Data Bank also dependent on His48 protonation state, where deprotonation is highlighted the main contribution of Asn/Gln, Arg and Glu/ associated to product release (Doucet et al., 2009; Watt et al., Asp that provide bidentate interactions at N1/N6 and N1/O6 or 2011). Interestingly, His48 is close to the protein family signature N1/N2 groups, respectively (Boix et al., 2013). Other polar or CKXXNTF and is conserved in most members, except in fish and charged secondary residues, such as Thr, Ser or Lys could also be amphibian sequences (Figures S7–S11). identified (Luscombe, 2001; Boix et al., 2013). Complementarily, Early dynamic predictions could also clearly differentiate stacking interactions are also significantly influencing the protein between the main B2 residue (Asn71), which directly interacted binding mode (Luscombe, 2001; Boix et al., 2013). Interestingly, with the adenine base, and other contributing residues, such as another structural feature reported by Westhof and co-workers as residues Gln69 and Glu111, which participated through water- characteristic for adenine binding is the combined contribution mediated interactions (Seshadri et al., 1995; Madhusudhan of amino acid side chain and the peptide backbone atoms (Kondo and Vishveshwara, 2001). The results helped to interpret and Westhof, 2011). Our molecular dynamics analysis highlights previous results obtained by site-directed mutagenesis and the conserved binding mode for adenine of Asn71 in RNase A, kinetic characterization (Tarragona-Fiol et al., 1993). Likewise, and counterparts, together with L4 loop main chain atoms. This the NMR analysis of several frog RNases in complex with a emphasizes the importance of Asn and loop L4 conformation deoxytetranucleotide also highlighted the key role of Asn71 in RNase A superfamily to favor adenine binding in mammals counterparts for guanine binding, even if the respective L4 loops (Figure 4). On their turn, lower order vertebrates tend to present are significantly shortened in contrast to the bovine RNase A an Arg that facilitates the interactions at N7/O6 for guanine structure (Hsu et al., 2015). Moreover, the studies by Hsu and recognition, as reported for other nucleotide-binding proteins Chen corroborated the importance of Glu111 counterpart in (Luscombe, 2001). specific guanine recognition at the N1/N2 group (Hsu et al., 2015). Last, together with the two natural purine bases found in RNA On the other hand, significant divergence is evidenced at we decided here to analyse the modified base inosine. Inosine the guanine-binding mode between the present molecular molecular structure was used as a purine binding model that dynamics analysis and previous structural characterization by served to visualize unique interactions at N7/O6 and N1/O6, in X-ray crystallography. Mostly, although our data emphasizes relation to guanine. Comparison of kinetic and MD results on UpG the preference for adenine at B2 site in mammal RNases, we and UpI highlights the importance of specific Glu/Asp residues do not observe any impediment for guanine positioning at the enzyme base secondary site, nor any tendency of guanine to bind at the main B1 base site. Surprisingly, RNase A crystallographic studies using 2′5′-UpG and d(UpG), both in soaking and co-crystallization conditions, showed an unusual binding mode (Lisgarten et al., 1995). Specifically, the guanine was located at B1 instead of B2 site; this peculiar non-productive positioning was classified as a “retrobinding” mode (Aguilar et al., 1992). In addition, not only was “retrobinding” reported by independent researchers for RNase A for both d(CpG) and d(UpG) (Aguilar et al., 1992; Lisgarten et al., 1995; Vitagliano et al., 2000), but also for bullfrog RNase binding to d(CpA) (Chang et al., 1998). Noteworthy, the present kinetic results are also emphasizing a FIGURE 6 | Schematic depiction of the main interactions identified by molecular dynamics according to the two main family classes (mammalian much more pronounced substrate selectivity at B2 site than the versus non-mammalian RNases). In dark and light red, the main and MD data reveal (Table 1, Figure 2). secondary specific interactions characteristic for either mammalian or non- Overall, our molecular dynamics study using UpA and mammalian RNases and in grey, the other interactions observed in all the UpG enabled us to outline the main residues involved in the RNases analyzed. Representative icon labels for humans and fishes illustrate RNases’ distinct specificities for B2. Figure 5 illustrates the the predominant interaction mode for each RNase class. Asn1 corresponds to Asn71 and Asn2 to Asn 67 in RNase A. main interactions that participate in the purine recognition. Frontiers in Pharmacology | www.frontiersin.org 12 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily in non-mammalian RNases involved in guanine N1/N2 group to adenine preference. Interestingly, a close inspection of the recognition. Interestingly, we find in the literature an inosine- residues potentially involved in the enzyme B2 site reveals that specific RNase that can accommodate the base in its active site the main contributors (Asn71 and Glu111 in RNase A and groove and provides specificity by discriminating the modified equivalent counterparts) are present in all the family members. base against the two natural purines (Versées et al., 2002). To Notwithstanding, significant differences in L4 loop extension note, the contribution of Trp side chain in packing the inosine and contribution of complementary residues can facilitate a base by stacking interactions is observed. Inosine represents distinct binding mode that confers discrimination between one of the main posttranscriptional modifications in cellular both purine bases. Overall, Asn, Glu/Asp and Arg bidentate side transcripts. RNA modifications not only contribute to regulate chains provide selective binding to adenine N1/N6 and N6/N7 the translation pathway, they are also involved in the generation versus guanine N1/O6, O6/N7 and N1/N2 groups. of regulatory tRNA fragments (Lyons et al., 2018). It is important to highlight that specific tRNA cleavage participates in the host response in stress conditions (Thompson et al., 2008) and RNA DATA AVAILABILITY STATEMENT posttranscriptional modification can alter the target specificity for All datasets generated for this study are included in the cellular endonucleases. For example, base methylation can protect manuscript/Supplementary Files. tRNA from cleavage by human RNase 5 (angiogenin) (Lyons et al., 2017). Overall, RNA modifications not only alter their own processing rate but also influence their association to selective binding proteins, participating in the cellular metabolism and AUTHOR CONTRIBUTIONS physiology (Boccaletto et al., 2017). Besides, the complexity of EB and GP-E conceived and designed the experiments. GP-E, LL, cellular RNA structure and its organization into supramolecular VS and MM performed the experimental work. EB, GP-E and complexes within the cell further difficult our understanding of LL analysed the data. EB and GP-E drafted the manuscript. EB, the cellular RNA metabolism (Van Treeck et al., 2018). Definitely, GP-E, LL, VS and MM revised the final manuscript. All authors we are still facing important methodological limitations to approved the final manuscript version. interpret the RNases’ behavior in physiological conditions. On the other hand, a comprehensive analysis of the protein nucleotide recognition pattern cannot disregard the existence of FUNDING an extended substrate binding site architecture as demonstrated by many structural and kinetic studies (Boix et al., 1994; Research work was supported by the Ministerio de Economía y Fontecilla-Camps et al., 1994; Irie et al., 1998; Nogués et al., Competitividad (SAF2015-66007P) and by AGAUR, Generalitat 1998; Raines, 1998; Hsu et al., 2015; Prats-Ejarque et al., 2019). de Catalunya (2016PROD00060; 2017SGR1010), co-financed by Interestingly, recent work on the protein motion and ligand FEDER funds and by Fundació La Marató de TV3 (20180310). binding energies using a pentanucleotide suggests that induced GP-E is a recipient of a PIF (UAB) predoctoral fellowship. LL is a conformational changes take place upon RNA interaction with recipient of a CSC predoctoral fellowship. secondary binding sites and can eventually provide a synergistic addition effect (Narayanan et al., 2017). The cooperative participation of secondary substrate binding sites could explain ACKNOWLEDGMENTS the enzyme low binding affinity for mono- and dinucleotides and is also significantly limiting the potency of molecular dynamics The authors wish to thank Helena Carbó for laboratory predictions, when working with such probes. However, our technical support, Clara Villalba for her careful revision of the present results, together with previously reported data, are manuscript and the Laboratori d’Anàlisi i Fotodocumentació, definitely indicating an evolutionary trend in B2 base selectivity Universitat Autònoma de Barcelona for providing the necessary within the vertebrate-specific RNase A superfamily that should infrastructure. We also wish to thank Dr. Marc Torrent for his respond to changing environmental conditions and adaptation help in molecular dynamic studies. to novel physiological needs. There is still a long path to walk to unveil the RNases’ substrate selectivity in vivo. We are confident that the identification of the structural patterns for nucleotide SUPPLEMENTARY MATERIAL recognition in host defense RNases would provide valuable tools for structure-based drug design. The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphar.2019.01170/ full#supplementary-material CONCLUSIONS SUPPLEMENTARY FIGURE 1 | Modifications of the force field to include In this work, we have analysed the base selectivity at B2 site inosine parametrization. within the RNase A superfamily by kinetic assays and molecular SUPPLEMENTARY FIGURE 2 | Phylogenetic tree of representative sequences dynamics simulations using dinucleotide substrates. Our of pancreatic ribonucleases. The evolutionary history was inferred by using the results indicate an evolutionary drift tendency from guanine Maximum Likelihood method and JTT matrix-based model (Jones et al., 1992). Frontiers in Pharmacology | www.frontiersin.org 13 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily The tree with the highest log likelihood (−27534.43) is shown. The percentage as follows: red box, white character for strict identity; red character for similarity of trees in which the associated taxa clustered together is shown next to the within a group; and character with blue frame for similarity across groups. branches. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbour-Join and BioNJ algorithms to a matrix of pairwise distances SUPPLEMENTARY FIGURE 9 | Sequence alignment of representative estimated using a JTT model, and then selecting the topology with superior sequences of reptilian RNases. Protein regions identified to participate in B2 site log likelihood value. The tree is drawn to scale, with branch lengths measured are highlighted in yellow (L4, spanning from b2 to b3, end of β6 and one of the in the number of substitutions per site. This analysis involved 160 amino acid two catalytic histidines together with a close by residue at β7). Main conserved sequences. There were a total of 212 positions in the final dataset. Evolutionary key residues are: Asn68 and Asp116/Arg117. TT indicates the presence of a analyses were conducted in MEGA X (Kumar et al., 2018). RNases are labeled β-turn. Dots label every 10 residues of the reference protein used (Cm-RNase with the species abbreviation (see Table S1) and its UNIPROT code, or, in its 1). The disulphide bonds are labelled with green numbers. The alignment was absence, with its NCBI code. performed using Clustal Omega (Sievers and Higgins, 2018), and the picture was drawn using ESPript (Robert and Gouet, 2014). Labels are as follows: red box, SUPPLEMENTARY FIGURE 3 | Atom nomenclature of the three dinucleotides white character for strict identity; red character for similarity within a group; and used in the molecular dynamics simulations. character with blue frame for similarity across groups. SUPPLEMENTARY FIGURE 10 | Sequence alignment of representative SUPPLEMENTARY FIGURE 4 | Mobility of the dinucleotides, calculated sequences of bird RNases. Protein regions identified to participate in B2 site in RMSD (nm), during each 100 ns simulation run. Each color represents a are highlighted in yellow (L4, spanning from b2 to b3, end of β6 and one of the different replicate. two catalytic histidines together with a close by residue at β7). Main conserved key residues are: Asn65, Arg66 and Trp105. TT indicates the presence of a SUPPLEMENTARY FIGURE 5 | Fraction of hydrogen bond interaction β-turn. Dots label every 10 residues of the reference protein used (Gg-RNase occurrence of the key protein residues involved in the binding to the purine base 1). The disulphide bonds are labelled with green numbers. The alignment was during each MD simulation run. performed using Clustal Omega (Sievers and Higgins, 2018), and the picture was drawn using ESPript (Robert and Gouet, 2014). Labels are as follows: red box, SUPPLEMENTARY FIGURE 6 | Schematic illustration of RNase-UpG white character for strict identity; red character for similarity within a group; and complexes obtained by molecular dynamics simulations using GROMACS. The character with blue frame for similarity across groups. picture was generated using PyMOL 1.7.2 (Schrödinger, Inc). SUPPLEMENTARY FIGURE 11 | Sequence alignment of representative SUPPLEMENTARY FIGURE 7 |Sequence alignment of representative sequences of mammalian RNases. Protein regions identified to participate in B2 sequences of fish RNases. Protein regions identified to participate in B2 site site are highlighted in yellow (L4, spanning from b2 to b3, end of β6 and one of the are highlighted in yellow (L4, spanning from b2 to b3, end of β6 and one of the two catalytic histidines together with a close by residue at β7). Main conserved two catalytic histidines together with a close by residue at β7). Main conserved key residues are: Asn67/Gln69/Asn71, Ala109, Glu111 and Arg122. TT indicates key residues are: Asn 72/74, Glu 114 and Glu 122/Arg123. TT indicates the the presence of a β-turn. Dots label every 10 residues of the reference protein presence of a β-turn. Dots label every 10 residues of the reference protein used (Bt-RNase 1). The disulphide bonds are labelled with green numbers. The used (Dr-RNase 1). The disulphide bonds are labeled with green numbers. The alignment was performed using Clustal Omega (Sievers and Higgins, 2018), alignment was performed using Clustal Omega (Sievers and Higgins, 2018), and the picture was drawn using ESPript (Robert and Gouet, 2014). Labels are and the picture was drawn using ESPript (Robert and Gouet, 2014). Labels are as follows: red box, white character for strict identity; red character for similarity as follows: red box, white character for strict identity; red character for similarity within a group; and character with blue frame for similarity across groups. within a group; and character with blue frame for similarity across groups. SUPPLEMENTARY TABLE 1 | Phylogenetic classification and abbreviations SUPPLEMENTARY FIGURE 8 | Sequence alignment of representative used for the analysed species. sequences of amphibian RNases. Protein regions identified to participate in B2 site are highlighted in yellow (L4, spanning from b2 to b3, end of β6 and SUPPLEMENTARY TABLE 2 | Main key residues identified by molecular one of the two catalytic histidines together with a close by residue at β7). Main dynamic simulations of the studied RNase complexes with UpA, UpG and UpI. conserved key residues are: Arg5, Asn56 and Thr89/Glu91. TT indicates the Equivalent residues by structural superposition are indicated. Cationic residues presence of a β-turn. Dots label every 10 residues of the reference protein are indicated in blue, anionic residues in red and polar residues in black. Histidine used (Rp-RNase). The disulphide bonds are labelled with green numbers. The residues involved in stacking interactions with the purine base are coloured alignment was performed using Clustal Omega (Sievers and Higgins, 2018), in orange. Only specific interactions with purine atoms are included. Specific and the picture was drawn using ESPript (Robert and Gouet, 2014). Labels are interaction to purine atoms are illustrated in Figure 5. REFERENCES Allers, J., and Shamoo, Y. (2001). Structure-based analysis of protein–RNA Abraham, M. J., Murtola, T., Schulz, R., Páll, S., Smith, J. C., Hess, B. et al. (2015). interactions using the program ENTANGLE. J. Mol. Biol. 311, 75–86. doi: GROMACS: High performance molecular simulations through multi-level 10.1006/jmbi.2001.4857 parallelism from laptops to supercomputers, SoftwareX, 1–2. 19–25. doi: Ardelt, W., Shogen, K., and Darzynkiewicz, Z. (2008). Onconase and amphinase, 10.1016/j.softx.2015.06.001 the antitumor ribonucleases from Rana pipiens oocytes. Curr. Pharm. Acharya, K. R., Allen, S. C., Riordan, J. F., Valleeo, B. L., Shapiro, R., Allen, S. C., et al. Biotechnol. 9, 215–225. doi: 10.2174/138920108784567245 (1994). Crystal structure of human angiogenin reveals the structural basis for Beintema, J. J., and Kleineidam, R. G. (1998). The ribonuclease A superfamily: its functional divergence from ribonuclease. Proc. Natl. Acad. Sci. U. S. A. 91, general discussion. Cell. Mol. Life Sci. 54, 825–832. doi: 10.1007/ 2915–2919. doi: 10.1073/pnas.91.8.2915 s000180050211 Aguilar, C. F., Thomas, P. J., Mills, A., Moss, D. S., and Palmer, R. A. (1992). Newly Berendsen, H. J. C., Postma, J. P. M., Van Gunsteren, W. F., Dinola, A., and Haak, observed binding mode in pancreatic ribonuclease. J. Mol. Biol. 224, 265–267. J. R. (1984). Molecular dynamics with coupling to an external bath. J. Chem. doi: 10.1016/0022-2836(92)90589-C Phys. 81, 3684–3690. doi: 10.1063/1.448118 Albery, W. J., and Knowles, J. R. (1976). Evolution of enzyme function and the Berisio, R., Lamzin, V. S., Sica, F., Wilson, K. S., Zagari, A., and Mazzarella, L. development of catalytic efficiency. Biochemistry 15, 5631–5640. doi: 10.1021/ (1999). Protein titration in the crystal state. J. Mol. Biol. 292, 845–854. doi: bi00670a032 10.1006/jmbi.1999.3093 Frontiers in Pharmacology | www.frontiersin.org 14 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily Best, R. B., and Hummer, G. (2009). Optimized molecular dynamics force fields Doucet, N., Watt, E. D., and Loria, J. P. (2009). The flexibility of a distant loop applied to the helix-coil transition of polypeptides. J. Phys. Chem. B. 113, 9004– modulates active site motion and product release in ribonuclease A. 9015. doi: 10.1021/jp901540t Biochemistry 48, 7160–7168. doi: 10.1021/bi900830g Boccaletto, P., Magnus, M., Almeida, C., Żyła, A., Astha, A., Pluta, R., et al. (2017). Draper, D. E. (1999). Themes in RNA-protein recognition. J. Mol. Biol. 293, 255– RNArchitecture: a database and a classification system of RNA families, with 270. doi: 10.1006/jmbi.1999.2991 a focus on structural information. Nucleic Acids Res. 46, D202–D205. doi: Draper, D. E. (2015). Reflections on 20 years of RNA. RNA 21, 601–602. doi: 10.1093/nar/gkx966 10.1261/rna.050930.115 Boix, E. (2001). “Eosinophil cationic protein,” in Methods in Enzymology CA, USA: Essmann, U., Perera, L., Berkowitz, M. L., Darden, T., Lee, H., and Pedersen, L. G. Academic Press, 287–305. doi: 10.1016/S0076-6879(01)41159-1 (1995). A smooth particle mesh Ewald method. J. Chem. Phys. 103, 8577. doi: Boix, E., Blanco, J., Nogués, M. V., and Moussaoui, M. (2013). Nucleotide binding 10.1063/1.470117 architecture for secreted cytotoxic endoribonucleases. Biochimie 95, 1087– Fontecilla-Camps, J. C., De Llorens, R., Le Du, M. H., and Cuchillo, C. M. (1994). 1097. doi: 10.1016/j.biochi.2012.12.015 Crystal structure of ribonuclease A·d(ApTpApApG) complex. Direct evidence Boix, E., Leonidas, D. D., Nikolovski, Z., Nogués, M. V., Cuchillo, C. M., and for extended substrate recognition. J. Biol. Chem. 269, 21526–21531. doi: Acharya, K. R. (1999a). Crystal structure of eosinophil cationic protein at 2.4 Å 10.2210/pdb1rcn/pdb resolution. Biochemistry 38, 16794–16801. doi: 10.1021/bi9919145 Gagné, D., Charest, L. A., Morin, S., Kovrigin, E. L., and Doucet, N. (2012). Boix, E., Nikolovski, Z., Moiseyev, G., Rosenberg, H. F., Cuchillo, C. M., and Conservation of flexible residue clusters among structural and functional enzyme Nogues, M. V. (1999b). Kinetic and product distribution analysis of human homologues. J. Biol. Chem. 287, 44289–44300. doi: 10.1074/jbc.M112.394866 eosinophil cationic protein indicates a subsite arrangement that favors Gagné, D., and Doucet, N. (2013). Structural and functional importance of local exonuclease-type activity. J. Biol. Chem. 274, 15605–15614. doi: 10.1074/ and global conformational fluctuations in the RNase A superfamily. FEBS J. jbc.274.22.15605 280, 5596–5607. doi: 10.1111/febs.12371 Boix, E., and Nogués, M. V. (2007). Mammalian antimicrobial proteins and Gagné, D., French, R. L., Narayanan, C., Simonović, M., Agarwal, P. K., and Doucet, N. peptides: overview on the RNase A superfamily members involved in innate (2015). Perturbation of the conformational dynamics of an active-site loop alters host defence. Mol. Biosyst. 3, 317–335. doi: 10.1039/b617527a enzyme activity. Structure 23, 2256–2266. doi: 10.1016/j.str.2015.10.011 Boix, E., Nogués, M. V., Schein, C. H., Benner, S. A., and Cuchillo, C. M. (1994). Goo, S. M., and Cho, S. (2013). The expansion and functional diversification of the Reverse transphosphorylation by ribonuclease a needs an intact p2-binding mammalian ribonuclease a superfamily epitomizes the efficiency of multigene site: point mutations at LYS-7 and ARG-10 alter the catalytic properties of the families at generating biological novelty. Genome Biol. Evol. 5, 2124–2140. doi: enzyme. J. Biol. Chem. 269, 2529–2534. 10.1093/gbe/evt161 Boix, E., Wu, Y., Vasandani, V. M., Saxena, S. K., Ardelt, W., Ladner, J., et al. (1996). Harper, J. W., and Vallee, B. L. (1989). A covalent angiogenin/ribonuclease hybrid Role of the N terminus in RNase A homologues: differences in catalytic activity, with a fourth disulfide bond generated by regional mutagenesis. Biochemistry ribonuclease inhibitor interaction and cytotoxicity. J. Mol. Biol. 257, 992–1007. 28, 1875–1884. doi: 10.1021/bi00430a067 doi: 10.1006/jmbi.1996.0218 Hess, B. (2008). P-LINCS: a parallel linear constraint solver for molecular Bussi, G., Donadio, D., and Parrinello, M. (2007). Canonical sampling through simulation. J. Chem. Theory Comput. 4, 116–122. doi: 10.1021/ct700200b velocity rescaling. J. Chem. Phys. 126, 1–7. doi: 10.1063/1.2408420 Holloway, D. E., Singh, U. P., Shogen, K., and Acharya, K. R. (2011). Crystal structure Chan, C. C., Moser, J. M., Dyer, K. D., Percopo, C. M., and Rosenberg, H. F. of onconase at 1.1 ?? resolution—insights into substrate binding and collective (2012). Genetic diversity of human RNase 8. BMC Genomics 13, 40. doi: motion. FEBS J. 278, 4136–4149. doi: 10.1111/j.1742-4658.2011.08320.x 10.1186/1471-2164-13-40 Honda, S., Loher, P., Shigematsu, M., Palazzo, J. P., Suzuki, R., Imoto, I., et al. Chang, C. F., Chen, C., Chen, Y. C., Hom, K., Huang, R. F., and Huang, T. H. (2015). Sex hormone-dependent tRNA halves enhance cell proliferation in (1998). The solution structure of a cytotoxic ribonuclease from the oocytes of I breast and prostate cancers. Proc. Natl. Acad. Sci. 112, E3816–E3825. doi: (bullfrog). J. Mol. Biol. 283, 231–244. doi: 10.1006/jmbi.1998.2082 10.1073/pnas.1510077112 Chatzileontiadou, D. S. M., Parmenopoulou, V., Manta, S., Kantsadi, A. L., Kylindri, Hsu, C.-H., Liao, Y.-D., Pan, Y.-R., Chen, L.-W., Wu, S.-H., Leu, Y.-J., et al. (2003). P., Griniezaki, M., et al. (2015). Triazole double-headed ribonucleosides as Solution structure of the cytotoxic RNase 4 from oocytes of bullfrog Rana inhibitors of eosinophil derived neurotoxin. Bioorg. Chem. 63, 152–165. doi: catesbeiana. J. Mol. Biol. 326, 1189–1201. doi: 10.1016/S0022-2836(02)01472-9 10.1016/j.bioorg.2015.10.007 Hsu, C. H., Chang, C. F., Liao, Y., Wu, S. H., and Chen, C. (2015). Solution Chatzileontiadou, D. S. M., Tsika, A. C., Diamantopoulou, Z., Delbé, J., structure and base specificity of cytotoxic RC-RNase 2 from Rana catesbeiana. Badet, J., Courty, J., et al. (2018). Evidence for novel action at the cell- Arch. Biochem. Biophys. 584, 70–78. doi: 10.1016/j.abb.2015.08.010 binding site of human angiogenin revealed by heteronuclear NMR Imazawa, M., Irie, M., and Ukita, T. (1968). Substrate specificity of ribonuclease spectroscopy, in silico and in vivo studies. Chem. Med. Chem. 13, 259–269. from Aspergillus saitoi. J. Biochem. 64, 595–602. doi: 10.1093/oxfordjournals. doi: 10.1002/cmdc.201700688 jbchem.a128936 Cho, S., and Zhang, J. (2007). Zebrafish ribonucleases are bactericidal: implications Irie, M., Nitta, K., and Nonaka, T. (1998). Biochemistry of frog ribonucleases. Cell for the origin of the vertebrate RNase a superfamily. Mol. Biol. Evol. 24, 1259– Mol. Life Sci. 54, 775–784. doi: 10.1007/s000180050206 1268. doi: 10.1093/molbev/msm047 Jones, D. T., Taylor, W. R., and Thornton, J. M. (1992). The rapid generation Cuchillo, C. M., Nogués, M. V., and Raines, R. T. (2011). Bovine pancreatic of mutation data matrices. Bioinformatics 8, 275–282. doi: 10.1093/ ribonuclease: fifty years of the first enzymatic reaction mechanism. Biochemistry bioinformatics/8.3.275 50, 7835–7841. doi: 10.1021/bi201075b Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., and Klein, M. L. Curran, T. P., Shapiro, R., and Riordan, J. F. (1993). Alteration of the enzymatic (1983). Comparison of simple potential functions for simulating liquid water. J. specificity of human angiogenin by site-directed mutagenesis. Biochemistry 32, Chem. Phys. 79, 926. doi: 10.1063/1.445869 2307–2313. doi: 10.1021/bi00060a023 Kazakou, K., Holloway, D. E., Prior, S. H., Subramanian, V., and Acharya, K. R. Darden, T., York, D., and Pedersen, L. (1993). Particle mesh Ewald: an N·log(N) (2008). Ribonuclease A homologues of the zebrafish: polymorphism, crystal method for Ewald sums in large systems. J. Chem. Phys. 98, 10089. doi: structures of two representatives and their evolutionary implications. J. Mol. 10.1063/1.464397 Biol. 380, 206–222. doi: 10.1016/j.jmb.2008.04.070 delCardayre, S. B., and Raines, R. T. (1994). Structural determinants of enzymic Kondo, J., and Westhof, E. (2011). Classification of pseudo pairs between processivity. Biochemistry 33, 6031–6037. doi: 10.1021/bi00186a001 nucleotide bases and amino acids by analysis of nucleotide-protein complexes. Denessiouk, K. A., and Johnson, M. S. (2003). Acceptor–donor–acceptormotifs Nucleic Acids Res. 39, 8628–8637. doi: 10.1093/nar/gkr452 recognize the Watson–Crick, Hoogsteen and sugardonor–acceptor– Kumar, S., Stecher, G., Li, M., Knyaz, C., and Tamura, K. (2018). MEGA X: donoredges of adenine and adenosine-containing ligands. J. Mol. Biol. 333, Molecular evolutionary genetics analysis across computing platforms. Mol. 1025–1043. doi: 10.1016/j.jmb.2003.09.017 Biol. Evol. 35, 1547–1549. doi: 10.1093/molbev/msy096 Frontiers in Pharmacology | www.frontiersin.org 15 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily Lee, J. E., Bae, E., Bingman, C. A., Phillips, G. N., and Raines, R. T. (2008). of high resolution structures. Bioinformatics 22, 2746–2752. doi: 10.1093/ Structural basis for catalysis by onconase. J. Mol. Biol. 375, 165–177. doi: bioinformatics/btl470 10.1016/j.jmb.2007.09.089 Narayanan, C., Bernard, D. N., Bafna, K., Gagné, D., Agarwal, P. K., and Doucet, N. Lee, J. E., and Raines, R. T. (2003). Contribution of active-site residues to the (2018a). Ligand-induced variations in structural and dynamical properties function of onconase, a ribonuclease with antitumoral activity. Biochemistry within an enzyme superfamily. Front. Mol. Biosci. 5, 1–12. doi: 10.3389/ 42, 11443–11450. doi: 10.1021/bi035147s fmolb.2018.00054 Leonidas, D. D., Boix, E., Prill, R., Suzuki, M., Turton, R., Minson, K., et al. (2001). Narayanan, C., Bernard, D. N., Bafna, K., Gagné, D., Chennubhotla, C. S., Mapping the ribonucleolytic active site of eosinophil-derived neurotoxin Doucet, N., et al. (2018b). Conservation of dynamics associated with biological (EDN): High resolution crystal structures of EDN complexes with adenylic function in an enzyme superfamily. Structure 26, 426–436.e3. doi: 10.1016/j. nucleotide inhibitors. J. Biol. Chem. 276, 15009–15017. doi: 10.1074/jbc. str.2018.01.015 M010585200 Narayanan, C., Gagné, D., Reynolds, K. A., and Doucet, N. (2017). Conserved Liao, Y.-D. (1992). A pyrimidine-guanine sequence-specific ribonuclease from amino acid networks modulate discrete functional properties in an enzyme Rana catesbeiana (bullfrog) oocytes. Nucleic Acids Res. 20, 1371–1377. doi: superfamily. Sci. Rep. 7, 3207. doi: 10.1038/s41598-017-03298-4 10.1093/nar/20.6.1371 Nitto, T., Dyer, K. D., Czapiga, M., and Rosenberg, H. F. (2006). Evolution and Lisgarten, J. N., Maes, D., Wyns, L., Aguilar, C. F., and Palmer, R. A. (1995). Structure function of leukocyte RNase A ribonucleases of the avian species, Gallus gallus. of the crystalline complex of deoxycytidylyl-3’,5’-guanosine (3’,5’-dCpdG) J. Biol. Chem. 281, 25622–25634. doi: 10.1074/jbc.M604313200 cocrystallized with ribonuclease at 1.9 Angstrom resolution. Acta Crystallogr.– Nitto, T., Lin, C., Dyer, K. D., Wagner, R. A., and Rosenberg, H. F. (2005). Sect. D Biol. Crystallogr. 51, 767–771. doi: 10.1107/S0907444995001570 Characterization of a ribonuclease gene and encoded protein from the reptile, Liu, J., Wang, X., Cho, S., Lim, B. K., Irwin, D. M., Ryder, O. A., et al. (2015). Iguana iguana. Gene 352, 36–44. doi: 10.1016/j.gene.2005.03.002 Evolutionary and functional novelty of pancreatic ribonuclease: a study of Nogués, M. V., Moussaoui, M., Boix, E., Vilanova, M., Ribó, M., and Cuchillo, C. M. Musteloidea (order Carnivora). Sci. Rep. 4, 5070. doi: 10.1038/srep05070 (1998). The contribution of noncatalytic phosphate-binding subsites to the Lomax, J. E., Bianchetti, C. M., Chang, A., Phillips, G. N., Fox, B. G., and Raines, mechanism of bovine pancreatic ribonuclease A. Cell. Mol. Life Sci. 54, 766– R. T. (2014). Functional evolution of ribonuclease inhibitor: insights from birds 774. doi: 10.1007/s000180050205 and reptiles. J. Mol. Biol. 426, 3041–3056. doi: 10.1016/j.jmb.2014.06.007 Páll, S., and Hess, B. (2013). A flexible algorithm for calculating pair interactions on Lomax, J. E., Eller, C. H., and Raines, R. T. (2017). Comparative functional analysis SIMD architectures. Comput. Phys. Commun. 184, 2641–2650. doi: 10.1016/j. of ribonuclease 1 homologs: molecular insights into evolving vertebrate cpc.2013.06.003 physiology. Biochem. J. 2219–2233. doi: 10.1042/BCJ20170173 Parés, X., Nogués, M. V., de Llorens, R., and Cuchillo, C. M. (1991). Structure and Lu, L., Li, J., Moussaoui, M., and Boix, E. (2018). Immune modulation by human function of ribonuclease A binding subsites. Essays Biochem. 26, 89–103. secreted RNases at the extracellular space. Front. Immunol. 9, 1–20. doi: Pizzo, E., Merlino, A., Turano, M., Russo Krauss, I., Coscia, F., Zanfardino, A., et al. 10.3389/fimmu.2018.01012 (2011). A new RNase sheds light on the RNase/angiogenin subfamily from Luscombe, N. M. (2001). Amino acid-base interactions: a three-dimensional zebrafish. Biochem. J. 433, 345–355. doi: 10.1042/BJ20100892 analysis of protein-DNA interactions at an atomic level. Nucleic Acids Res. 29, Prats-Ejarque, G., Arranz-Trullén, J., Blanco, J. A., Pulido, D., Nogués, M. V., 2860–2874. doi: 10.1093/nar/29.13.2860 Moussaoui, M., et al. (2016). The first crystal structure of human RNase 6 Lyons, S. M., Fay, M. M., Akiyama, Y., Anderson, P. J., and Ivanov, P. (2017). RNA reveals a novel substrate-binding and cleavage site arrangement. Biochem. J. biology of angiogenin: current state and perspectives. RNA Biol. 14, 171–178. 473, 1523–1536. doi: 10.1042/BCJ20160245 doi: 10.1080/15476286.2016.1272746 Prats-Ejarque, G., Blanco, J. A., Salazar, V. A., Nogués, V. M., Moussaoui, M., Lyons, S. M., Fay, M. M., and Ivanov, P. (2018). The role of RNA modifications and Boix, E. (2019). Characterization of an RNase with two catalytic centers. in the regulation of tRNA cleavage. FEBS Lett. 592, 2828–2844. doi: Human RNase6 catalytic and phosphate-binding site arrangement favors the 10.1002/1873-3468.13205 endonuclease cleavage of polymeric substrates. Biochim. Biophys. Acta–Gen. Madhusudhan, M. S., and Vishveshwara, S. (2001). Computer modeling of Subj. 1863, 105–117. doi: 10.1016/j.bbagen.2018.09.021 human angiogenin-dinucleotide substrate interaction. Proteins Struct. Raines, R. T. (1998). Ribonuclease A. Chem. Rev. 98, 1045–1066. doi: 10.1021/ Funct. Genet. 42, 125–135. doi: 10.1002/1097-0134(20010101)42:1<125:: cr960427h AID-PROT120>3.0.CO;2-K Ramanathan, A., and Agarwal, P. K. (2011). Evolutionarily conserved linkage Maguid, S., Fernández-Alberti, S., Parisi, G., and Echave, J. (2006). Evolutionary between enzyme fold, flexibility, and catalysis. PLoS Biol. 9, e1001193. doi: conservation of protein backbone flexibility. J. Mol. Evol. 63, 448–457. doi: 10.1371/journal.pbio.1001193 10.1007/s00239-005-0209-x Richards, F. M., and Wyckoff, H. W. (1971). “Bovine pancreatic ribonuclease,” in Merlino, A., Mazzarella, L., Carannante, A., Di Fiore, A., Di Donato, , A., et al., Hydrolysis The Enzymes. Ed. P. D. Boyer NY, USA: Academic Press, 647–806. et al. (2005). The importance of dynamic effects on the enzyme activity. J. Biol. doi: 10.1016/S1874-6047(08)60384-4 Chem. 280, 17953–17960. doi: 10.1074/jbc.M501339200 Robert, X., and Gouet, P. (2014). Deciphering key features in protein structures Merlino, A., Vitagliano, L., Antoine Ceruso, M., Di Nola, A., and Mazzarella, L. with the new ENDscript server. Nucleic Acids Res. 42, 320–324. doi: 10.1093/ (2002). Global and local motions in ribonuclease A: a molecular dynamics nar/gku316 study. Biopolymers 65, 274–283. doi: 10.1002/bip.10225 Rosenberg, H. F. (2008). RNase A ribonucleases and host defense: an evolving Merlino, A., Vitagliano, L., Ceruso, M. A., and Mazzarella, L. (2003). Subtle story. J. Leukoc. Biol. 83, 1079–1087. doi: 10.1189/jlb.1107725 functional collective motions in pancreatic-like ribonucleases: from Rosenberg, H. F., Dyer, K. D., Tiffany, H. L., and Gonzalez, M. (1995). Rapid ribonuclease A to angiogenin. Proteins Struct. Funct. Bioinform. 53, 101–110. evolution of a unique family of primate ribonuclease genes. Nat. Genet. 10, doi: 10.1002/prot.10466 219–223. doi: 10.1038/ng0695-219 Mesitov, M. V., Soldatov, R. A., Zaichenko, D. M., Malakho, S. G., Klementyeva, T. S., Sanjeev, B. S., and Vishveshwara, S. (2005). Dynamics of the native and Sokolovskaya, A. A., et al. (2017). Differential processing of small RNAs during the ligand-bound structures of eosinophil cationic protein: network of endoplasmic reticulum stress. Sci. Rep. 7, 1–14. doi: 10.1038/srep46080 hydrogen bonds at the catalytic site. J. Biomol. Struct. Dyn. 22, 657–671. doi: Miao, Z., and Westhof, E. (2016). RBscore&NBench : a high-level web server for 10.1080/07391102.2005.10507033 nucleic acid binding residues prediction with a large-scale benchmarking Seshadri, K., Rao, V. S., and Vishveshwara, S. (1995). Interaction of substrate uridyl database. Nucleic Acids Res. 44, W562–W567. doi: 10.1093/nar/gkw251 3’,5’-adenosine with ribonuclease A: a molecular dynamics study. Biophys. J. 69, Mohan, C. G., Boix, E., Evans, H. R., Nikolovski, Z., Nogués, M. V., Cuchillo, C. M., 2185–2194. doi: 10.1016/S0006-3495(95)80094-9 et al. (2002). The crystal structure of eosinophil cationic protein in complex Shapiro, R. (1998). Structural features that determine the enzymatic potency and with 2’,5’-ADP at 2.0 Å resolution reveals the details of the ribonucleolytic specificity of human angiogenin: threonine-80 and residues 58-70 and 116- active site †, ‡. Biochemistry 41, 12100–12106. doi: 10.1021/bi0264521 123. Biochemistry 37, 6847–6856. doi: 10.1021/bi9800146 Morozova, N., Allers, J., Myers, J., and Shamoo, Y. (2006). Protein-RNA interactions: Shapiro, R., Fett, J. W., Strydom, D. J., and Vallee, B. L. (1986). Isolation and exploring binding patterns with a three-dimensional superposition analysis characterization of a human colon carcinoma-secreted enzyme with pancreatic Frontiers in Pharmacology | www.frontiersin.org 16 October 2019 | Volume 10 | Article 1170 Prats-Ejarque et al. B2 Specificity in RNaseA Superfamily ribonuclease-like activity. Biochemistry 25, 7255–7264. doi: 10.1021/ and building force field libraries for new molecules and molecular fragments. bi00371a002 Nucleic Acids Res. 39, 511–517. doi: 10.1093/nar/gkr288 Sievers, F., and Higgins, D. G. (2018). Clustal Omega for making accurate alignments Versées, W., Decanniere, K., Van Holsbeke, E., Devroede, N., and Steyaert, J. of many protein sequences. Protein Sci. 27, 135–145. doi: 10.1002/pro.3290 (2002). Enzyme-substrate interactions in the purine-specific nucleoside Sikriwal, D., Seth, D., Dey, P., and Batra, J. K. (2007). Human eosinophil-derived hydrolase from Trypanosoma vivax. J. Biol. Chem. 277, 15938–15946. doi: neurotoxin: involvement of a putative non-catalytic phosphate-binding 10.1074/jbc.M111735200 subsite in its catalysis. Mol. Cell. Biochem. 303, 175–181. doi: 10.1007/ Vitagliano, L., Merlino, A., Zagari, A., and Mazzarella, L. (2000). Productive and s11010-007-9471-0 nonproductive binding to ribonuclease A: X-ray structure of two complexes with Sorrentino, S. (1998). Human extracellular ribonucleases: multiplicity, molecular uridylyl(2′,5′)guanosine. Protein Sci. 9, 1217–1225. doi: 10.1110/ps.9.6.1217 diversity and catalytic properties of the major RNase types. Cell Mol. Life Sci. Watt, E. D., Rivalta, I., Whittier, S. K., Batista, V. S., and Loria, J. P. (2011). 54, 785–794. doi: 10.1007/s000180050207 Reengineering rate-limiting, millisecond enzyme motions by introduction of an Sorrentino, S. (2010). The eight human “canonical” ribonucleases: molecular unnatural amino acid. Biophys. J. 101, 411–420. doi: 10.1016/j.bpj.2011.05.039 diversity, catalytic properties, and special biological actions of the enzyme Webb, B., and Sali, A. (2016). Comparative protein structure modeling using proteins. FEBS Lett. 584, 2194–2200. doi: 10.1016/j.febslet.2010.04.018 MODELLER. Curr. Protoc. Bioinform. 54, 5.6.1-5.6.37. doi: 10.1002/cpbi.3 Sorrentino, S., and Libonati, M. (1997). Structure-function relationships in human Xu, H., Liu, Y., Meng, F., He, B., Han, N., Li, G., et al. (2013). Multiple bursts of ribonucleases: main distinctive features of the major RNase types. FEBS Lett. pancreatic ribonuclease gene duplication in insect-eating bats. Gene 526, 112– 404, 1–5. doi: 10.1016/S0014-5793(97)00086-0 117. doi: 10.1016/j.gene.2013.04.035 Tamkovich, N., Koroleva, L., Kovpak, M., Goncharova, E., Silnikov, V., Vlassov, V., Zegers, I., Maes, D., Dao-Thi, M. H., Wyns, L., Poortmans, F., and Palmer, R. et al. (2016). Design, RNA cleavage and antiviral activity of new artificial (1994). The structures of rnase a complexed with 3′-CMP and d(CpA): active ribonucleases derived from mono-, di- and tripeptides connected by linkers of site conformation and conserved water molecules. Protein Sci. 3, 2322–2339. different hydrophobicity. Bioorganic Med. Chem. 24, 1346–1355. doi: 10.1016/j. doi: 10.1002/pro.5560031217 bmc.2016.02.007 Zhang, J. (2007). Disulfide-bond reshuffling in the evolution of an ape placental Tarragona-Fiol, A., Eggelte, H. J., Harbron, S., Sanchez, E., Taylorson, C. J., Ward, J. M., ribonuclease. Mol. Biol. Evol. 24, 505–512. doi: 10.1093/molbev/msl177 et al. (1993). Identification by site-directed mutagenesis of amino acids in the Zhang, J., Dyer, K. D., and Rosenberg, H. F. (2000). Evolution of the rodent subsite of bovine pancreatic ribonuclease A. Protein Eng. Des. Sel. 6, 901–906. doi: eosinophil-associated RNase gene family by rapid gene sorting and positive 10.1093/protein/6.8.901 selection. Proc. Natl. Acad. Sci. 97, 4701–4706. doi: 10.1073/pnas.080071397 Terribilini, M., Sander, J. D., Lee, J. H., Zaback, P., Jernigan, R. L., Honavar, V., et al. Zhang, J., Zhang, Y., and Rosenberg, H. F. (2002). Adaptive evolution of a (2007). RNABindR: a server for analyzing and predicting RNA-binding sites in duplicated pancreatic ribonuclease gene in a leaf-eating monkey. Nat. Genet. proteins. Nucleic Acids Res. 35, 578–584. doi: 10.1093/nar/gkm294 30, 411–415. doi: 10.1038/ng852 Thompson, D. M., Lu, C., Green, P. J., and Parker, R. (2008). tRNA cleavage is a Zhao, W., Kote-Jarai, Z., van Santen, Y., Hofsteenge, J., and Beintema, J. J. conserved response to oxidative stress in eukaryotes. RNA 14, 2095–2103. doi: (1998). Ribonucleases from rat and bovine liver: purification, specificity and 10.1261/rna.1232808 structural characterization. Biochim. Biophys. Acta 1384, 55–65. doi: 10.1016/ Thompson, J. E., Kutateladze, T. G., Schuster, M. C., Venegas, F. D., Messmore, S0167-4838(97)00213-6 J. M., and Raines, R. T. (1995). Limits to catalysis by ribonuclease A. Bioorg. Chem. 23, 471–481. doi: 10.1006/bioo.1995.1033 Conflict of Interest: The authors declare that the research was conducted in the Torrent, M., Badia, M., Moussaoui, M., Sanchez, D., Nogués, M. V., and Boix, E. absence of any commercial or financial relationships that could be construed as a (2010). Comparison of human RNase 3 and RNase 7 bactericidal action at the potential conflict of interest. Gram-negative and Gram-positive bacterial cell wall. FEBS J. 277, 1713–1725. doi: 10.1111/j.1742-4658.2010.07595.x Copyright © 2019 Prats-Ejarque, Lu, Salazar, Moussaoui and Boix. This is an open- Van Treeck, B., Protter, D. S. W., Matheny, T., Khong, A., Link, C. D., and Parker, R. access article distributed under the terms of the Creative Commons Attribution (2018). RNA self-assembly contributes to stress granule formation and defining License (CC BY). The use, distribution or reproduction in other forums is permitted, the stress granule transcriptome. Proc. Natl. Acad. Sci. 115, 2734–2739. doi: provided the original author(s) and the copyright owner(s) are credited and that the 10.1073/pnas.1800038115 original publication in this journal is cited, in accordance with accepted academic Vanquelef, E., Simon, S., Marquant, G., Garcia, E., Klimerak, G., Delepine, J. C., practice. No use, distribution or reproduction is permitted which does not comply et al. (2011). R.E.D. Server: a web service for deriving RESP and ESP charges with these terms. Frontiers in Pharmacology | www.frontiersin.org 17 October 2019 | Volume 10 | Article 1170