A Conserved Glutamate Residue in the C-terminal Deaminase Domain of Pentatricopeptide Repeat Proteins Is Required for RNA Editing Activity*

Background: Pentatricopeptide repeat (PPR) proteins that are required for RNA editing frequently include a C-terminal DYW deaminase domain. Results: Mutagenesis of a glutamate residue in the conserved deaminase HXE motif results in loss of editing activity. Conclusion: The glutamate residue is required for editing. Significance: The DYW deaminase domain of PPR proteins has the molecular characteristics of a deaminase. Many transcripts expressed from plant organelle genomes are modified by C-to-U RNA editing. Nuclear encoded pentatricopeptide repeat (PPR) proteins include an RNA binding domain that provides site specificity. In addition, many PPR proteins include a C-terminal DYW deaminase domain with characteristic zinc binding motifs (CXXC, HXE) and has recently been shown to bind zinc ions. The glutamate residue of the HXE motif is catalytically required in the reaction catalyzed by cytidine deaminase. In this work, we examine the activity of the DYW deaminase domain through truncation or mutagenesis of the HXE motif. OTP84 is required for editing three chloroplast sites, and transgenes expressing OTP84 with C-terminal truncations were capable of editing only one of the three cognate sites at high efficiency. These results suggest that the deaminase domain of OTP84 is required for editing two of the sites, but another deaminase is able to supply the deamination activity for the third site. OTP84 and CREF7 transgenes were mutagenized to replace the glutamate residue of the HXE motif, and transgenic plants expressing OTP84-E824A and CREF7-E554A were unable to efficiently edit the cognate editing sites for these genes. In addition, plants expressing CREF7-E554A exhibited substantially reduced capacity to edit a non-cognate site, rpoA C200. These results indicate that the DYW deaminase domains of PPR proteins are involved in editing their cognate editing sites, and in some cases may participate in editing additional sites in the chloroplast.

Editing in higher plants and in Physcomitrella patens is known to require nuclear proteins (4 -8).
Pentatricopeptide repeat (PPR) 2 genes have been shown to be required for RNA editing (9), and form a large family of protein-coding genes in higher plants with over 400 members in Arabidopsis (10). The known editing factors are members of the PLS subfamily of PPR proteins, which are composed of characteristic P, L (long), and S (short) repeats (11). Amino acid residues located in specific locations within the repeats have been shown to specify the base recognized in the cis-element (12)(13)(14), and the PLS repeat domain interacts with specific nucleotides within the cis-element to provide site specificity for RNA editing (12,15).
The PLS subfamily of PPR proteins also includes characteristic C-terminal domains known as the E, Eϩ, and DYW domains (10). Bioinformatics analysis identified characteristic structural motifs present in part of the E domain, the entire Eϩ domain, and most of the DYW domain that place the protein in the deaminase superfamily (16). This region has been identified as the "DYW family of nucleic acid deaminases" (Pfam 14432) (17) and is referred to as the "DYW deaminase domain" in this work.
There is mounting evidence that supports the role of the DYW deaminase domain as the catalytic component of the editing reaction. This region has canonical zinc binding motifs (HXE, CXXC) (18 -20), which are conserved in deaminases that act on nucleotides, RNA, and DNA (16,(21)(22)(23)(24)(25)(26). The DYW deaminase domain has recently been shown to bind zinc ions (19,20). In addition, mutagenesis of the zinc binding motifs has been shown to interfere with editing in transgenic plants (20) and through transient expression in protoplasts (27). The glutamate residue of the HXE motif has been shown to be directly involved in the E. coli cytidine deaminase mechanism through deprotonation of the substrate water molecule and transfer of the proton to the product ammonia (21). Thus, the requirement of this conserved glutamate residue is a key characteristic expected for an editing deaminase.
The editing of the ndhD C2 site in Arabidopsis chloroplasts requires two PPR proteins, CRR4 and DYW1 (28). CRR4 lacks the entire DYW deaminase domain but has a canonical PLS repeat region and is apparently required for editing as a site specificity factor (29). DYW1 has an intact DYW deaminase domain; however, the N-terminal PLS-like region is small and composed of degenerate repeats. Furthermore, DYW1 has canonical zinc binding motifs and has been shown to bind zinc ions (19,20). Mutagenesis of the zinc binding motifs and the catalytic glutamate residue of DYW1 ablated that ability to edit the ndhD site (20). Thus, the editing of the ndhD site requires both CRR4 and DYW1, and the functions of site recognition and deamination appear to be separated into two proteins in this case.
Many PPR proteins that are required for editing lack a DYW deaminase domain. In addition, there are several examples in which the DYW deaminase domain is present in a PPR, but may be eliminated by truncation and still support editing of the cognate sites (19,30,31). Finally, PPR genes have been shown to undergo truncation in evolution (19), indicating that the DYW deaminase domain is dispensable in an evolutionary context. Thus, the role of the DYW deaminase domain in these editing reactions has remained elusive.
In this work, we have investigated the role of the DYW deaminase domain of two PPR proteins, OTP84 and CREF7. In contrast to DYW1 and CRR4, both of these PPR proteins have canonical PLS repeat regions and an intact DYW deaminase domain. In this work, we examine the function of the DYW deaminase domain by gene truncation and mutagenesis of the catalytic glutamate residue and show that the DYW deaminase domain is required for editing the cognate editing sites.
Gene Cloning and Plant Transformation-Gene sequences from OTP84, OTP84trcPG (Met-1-Lys-754), and OTP84trcDYW (Met-1-Phe-770) were amplified by PCR to introduce 5Ј BamHI and 3Ј SalI restriction sites. Sequences for CREF7 were amplified with flanking BglII and SalI restriction sites. Mutations were introduced by amplification with primers that altered codon Glu-824 of OTP84 and Glu-554 of CREF7 to alanine codons. All gene fragments were cloned into pCHF1 using BamHI and SalI restriction sites (32). Binary vectors were electroporated into Agrobacterium strain ASE (33), and plants were transformed by floral dip (34). Seedlings were selected using 100 mg/liter gentamicin.
RNA Editing Analysis through Bulk Sequencing-Total RNA was isolated from green leaves using RiboZol from AMRESCO (Solon, OH). The GoScript reverse transcription system (Madison, WI) was used with random hexamers from Thermo Fisher to create a cDNA pool. Gene-specific primers were used to amplify sequences including editing sites from the cDNA templates with DreamTaq Green DNA polymerase from Thermo Fisher. Bulk sequencing of RT-PCR products was carried out at the University of California, Berkeley DNA Sequencing Facility. Peak heights were measured from electropherograms using BioEdit V7.0.9.0 to estimate the extent of RNA editing in RNA templates isolated from leaves.
Poisoned Primer Extension Editing Analysis-Poisoned primer extension assays were performed as described previously (35) with one modification. Purified PCR products were treated with FastAP from Thermo Scientific and incubated for 10 min at 37°C followed by 5 min at 75°C. This is similar to the Exo-SAP-IT step of the poisoned primer extension assay from the Hanson laboratory (36). The addition of the FastAP step consistently reduced read-through to less than 3%. Oligonucleotide primers contained a 5Ј-hexachlorofluorescein tag from Eurofins (Ebersberg, Germany). Vent DNA polymerase and acyNTPs from New England Biolabs (Ipswich, MA) were utilized for chain termination reactions. Primer extension products were separated by electrophoresis on 12% polyacrylamide gels with 6 M urea. Gels were covered in plastic wrap and immediately scanned using a Typhoon Trio imager from GE Healthcare (Little Chalfont, UK). The percentage of conversion was calculated from the intensity of bands measured from a gel image using GELQUANT.NET V1.8.2.
Mass Spectrometry-The DYW deaminase domain from ELI1 (At4g37380: Asp-478 -Trp-632) was amplified by PCR to add flanking BamHI and SalI restriction sites. A single nucleotide mutation (GAG to GCG) was introduced using a PCR primer to create the E566A mutation. The ELI1-E566A gene was cloned into pET28a using BamHI and SalI restriction sites and introduced into E. coli strain Rosetta 2 (DE3) pLysS from Novozymes (Bagsvaerd, Denmark). Recombinant ELI1-E566A was expressed as described previously (19) and purified. Native protein samples were dialyzed in 20 mM ammonium acetate prior to mass spectrometry. Denatured protein samples were treated with 50% acetonitrile and 1% formic acid prior to mass spectrometry. Protein samples were ionized with an electrospray ionization voltage of 3.6 kV, a cone voltage of 40 V, and a desolvation temperature of 120°C. Mass determinations were made with a Waters QTOF2 mass spectrometer.  (19,30,37,38). Because OTP84 has three editing site targets, the role of the DYW deaminase domain was examined on these editing sites.

Plants
The OTP84 gene has an intact DYW deaminase domain ( Fig.  1) and is required for editing of ndhF C290, psbZ C50, and ndhB C1481 (39). Two truncated variants of OTP84 were each prepared and tested for editing activity in three independent transgenic plants (Fig. 1). Truncation immediately C-terminal to the PG box eliminated most of the DYW deaminase domain (OTP84trcDYW), and truncation immediately N-terminal of the PG box eliminated all of the PG box and downstream DYW deaminase domain (OTP84trcPG). The truncated variants were introduced into an OTP84 knock-out background (39). The effects of OTP84 truncation are shown in Fig. 2. Expression of the OTP84trcDYW transgene did not restore editing of ndhF and psbZ transcripts to wild type levels (Fig. 2, A and B, respectively), suggesting that the DYW deaminase domain is required for editing these two cognate editing sites. In contrast, the editing of ndhB C1481 remained at wild type levels in transgenic plants expressing OTP84trcDYW (Fig. 2C), which suggests that editing of the ndhB site does not require the OTP84 DYW deaminase domain. Truncations that eliminate the PG box (OTP84trcPG) resulted in highly reduced editing for all three sites. This result is consistent with earlier studies performed with ELI1 that indicated that the PG box is critical for editing a site in ndhB transcripts (19). Although OTP84 is required for editing three sites, transgenes lacking a large portion of the DYW deaminase domain were capable of restoring editing to wild type levels in ndhB transcripts, suggesting the participation of an unknown DYW deaminase(s) for that site.
Mutagenesis of the HXE Motif of OTP84 Decreases Editing of the Three Cognate Sites-The HXE motif of OTP84 was mutagenized to HXA to produce OTP84-E824A, which was introduced into OTP84 knock-out line (otp84-3) and into the wild type Col-0 background. The conversion of the ndhF, psbZ, and ndhB editing sites is represented in Fig. 3, A, B, and C, respectively. Each editing site was converted at 90% or greater in the wild type plants and remained unedited in otp84-3, the OTP84 knock-out line. Complementation of otp84-3 with a wild type transgene restored editing to ϳ90% or greater. In contrast, the OTP84-E824A transgene was unable to restore editing of the three editing sites to wild type levels (Fig. 3). Plants expressing OTP84-E824A did not edit the ndhF and psbZ sites. In contrast, 23% of transcripts were edited at ndhB C1481 in plants expressing OTP84-E824A.
Because the OTP84-E824A transgene was unable to restore editing to the cognate sites, the mutant transgene might act in a dominant negative manner to suppress editing in wild type plants by formation of inactive editing complexes. Therefore, we tested whether expression of the OTP84-E824A transgene could disrupt editing in wild type plants. In three independent transgenic lines, the mutant transgene did not reduce editing levels in the wild type background and therefore did not act in a dominant negative manner over the native gene (Fig. 3).
Mutagenesis of the HXE Motif of CREF7 Decreases Editing of the Cognate Site-The CREF7 gene has an intact DYW deaminase domain (Fig. 1) and a single cognate editing site, ndhB C1255 (40). Transgenes for wild type CREF7 and a mutant variant with the HXE motif changed to HXA (CREF7-E554A) were introduced into a CREF7 knock-out line (cref7-1); in addition, the CREF7-E554A transgene was introduced into wild type plants. Transgenic plants were analyzed for editing of the cognate site (Fig. 4). The ndhB C1255 site was 100% edited in wild type plants, but remained unedited in the CREF7 knock-out plants. Editing of ndhB C1255 in CREF7 knock-out plants expressing the CREF7-E554A transgene was ϳ10% based on Sanger sequencing. In contrast, plants with wild type background expressing CREF7-E554A edited ndhB transcripts at wild type levels, and the transgene did not behave in a dominant negative manner. These results demonstrate that the Glu-554 plays a crucial role in the activity of CREF7 in editing ndhB transcripts.
Expression of CREF7-E554A Reduces Editing of rpoA C200, a Non-cognate Site-Editing site conversion was determined for the 34 major chloroplast editing sites in plants expressing OPT84-E824A and CREF7-E554A in their respective gene knock-out backgrounds (data not shown). In several instances, small changes in editing site conversion were detected at noncognate editing sites.
The strongest effect of the mutant transgene was observed on rpoA C200 editing in plants expressing CREF7-E554A (Fig. 5). The rpoA editing site is about 70% converted in wild type Col-0 plants, in cref7-1 knock-out plants, and in cref7-1 plants complemented with the wild type CREF7 gene. The CREF7-E554A transgene had no detectable effect when expressed in wild type plants; however, in cref7-1 knock-out plants, the E554A substitution caused a 25% reduction of rpoA editing to about 45%. Thus, expression of CREF7-E554A caused a significant reduction of rpoA editing, suggesting that CREF7 may participate in editing rpoA transcripts.
The HXA Mutation of the DYW Deaminase Domain Maintains Zinc Binding-Mutagenesis may cause a protein to lose function because of improper folding, or through a direct effect on catalysis. Previous investigations of the structure and function of E. coli cytidine deaminase used circular dichroism and zinc stoichiometry to demonstrate that the polypeptide with the Glu-to-Ala substitution retained native structure (41). The ELI1 DYW deaminase domain was selected for zinc analysis to determine whether the DYW deaminase domain was able to fold into a native structure. A Glu-to-Ala mutation was introduced at position 566 in the ELI1 DYW deaminase domain (19). The recombinant protein was expressed in E. coli, purified, and analyzed by mass spectrometry as described previously (19). The purified recombinant ELI1-E566A protein had a native mass of 19,519 atomic mass units, and the denatured protein had a mass of 19,395 atomic mass units. The mass difference of 124 atomic mass unit is consistent with the native DYW deaminase domain binding two zinc molecules of 65.4 atomic mass unit and the loss of several protons from the cysteine ligands that are predicted to coordinate the zinc atoms (19). This result demonstrates that the ELI1-E566A mutant DYW deaminase domain binds two zinc ions and retains native structure.

DISCUSSION
The enzyme responsible for RNA editing in plants has been controversial. Early biochemical analyses indicated that the enzymatic reaction was probably a deamination reaction (42,43), and in vitro editing assays established that the reaction occurred through a cytidine deamination mechanism (44 -46). After discovery of the role of PPR proteins in RNA processing reactions in plant organelles (10), deaminase-like zinc binding motifs were recognized in the DYW domain (18). More recently, the DYW deaminase domain was shown to bind zinc as a prosthetic group (19,20), further supporting the hypothesis that the DYW deaminase functions as the deaminase in plant editing. Mutagenesis of amino acid residues in the zinc binding motifs and of the catalytic glutamate (HXE, CXXC) of DYW1 resulted in the loss of the ability to edit ndhD C2 in transgenic plants (20). Thus, the DYW deaminase domain possesses several features that would be expected for an editing deaminase.
In this work, we explored the role of the glutamate residue of the HXE motif in full-length PPR proteins that have a canonical PLS repeat region and an intact DYW deaminase domain. Mutation of the putative catalytic glutamate residue caused a dramatic decrease in the editing of the cognate sites for OTP84 and CREF7. The requirement of the glutamate residue of the HXE domain and zinc binding by the DYW deaminase domain are two key features that are expected for an editing deaminase. They strongly support the hypothesis that the DYW deaminase domain provides the catalytic activity for the editing reaction.
There are several examples in which the DYW deaminase is dispensable for complementation of a knock-out mutant phenotype (19,30,31,47). The truncation of OTP84 had markedly different effects on the three editing site targets. The OTP84trcDYW transgene restored editing of ndhB C1481 to wild type levels, and this site responded similarly to previously characterized editing sites for ELI1, CRR22, CRR28, OTP82, and MEF11 (19,30,31,47). Editing of the psbZ and ndhF sites was highly reduced in the presence of the OTP84trcDYW transgene, although the ndhF site may have been somewhat higher than gene knock-out levels. The ndhF C290 editing site is also affected by disruption of the PPR gene, VAC1, which results in a partial loss of editing of the ndhF site (48). The DYW deami- nase domain of VAC1 might participate with OTP84trcDYW to incompletely edit the ndhF C290 site. These results suggest that a set of editing sites can share a single specificity factor, but might rely on different deaminases in trans to complete the editing reaction.
Editing was highly reduced at the three cognate sites of OTP84 in plants expressing the OTP84-E824A transgene. The editing of the psbZ and ndhF sites was near gene knock-out levels, and the OTP84-E554A protein may form non-functional editing complexes for these sites. The editing of the ndhB C1481 site responded somewhat differently, and editing of the ndhB site increased well above knock-out levels to about 23%. The ndhB site also responded differently from the other two cognate sites in plants expressing OTP84trcDYW, and the increase in ndhB editing might result from the participation of other deaminases in the editing of this site.
Expression of CREF7-E554A substantially decreased editing of rpoA C200, a cognate editing site for CLB19 (49). Because CLB19 is truncated at the end of the E domain and lacks an intact DYW deaminase domain (49), it could recruit a deaminase domain from another PPR protein. The reduction in editing of a non-cognate site suggests that CREF7 may participate as a deaminase in rpoA C200 editing. Because editing of rpoA C200 is not reduced in the CREF7 knock-out line (Fig. 5), there may be other DYW deaminases that participate in rpoA editing.
Nineteen PPR proteins have been shown to be required for editing in Arabidopsis chloroplasts. Four of these PPRs (CLB19, CRR4, CRR21, and OTP80) have truncated DYW deaminase domains (37,39,49) and could not function in C deamination. A complex network of PPR proteins could be involved in RNA editing. The PLS repeat domain may be sufficient for specifying an editing site; however, one or more deaminases could be recruited in trans as enzymatic components. Table 1 summarizes the relationships between several chloroplast editing sites, site specificity factors, and possible PPR proteins involved in the deamination reaction. PPR proteins that lack complete DYW deaminase domains may acquire the activity from other proteins as in the case of CRR4 and DYW1. Based on these results, OTP84 acts as a site specificity factor for three cognate sites and participates as a deaminase along with other PPR pro-FIGURE 3. OTP84-E824A does not restore editing to wild type levels in OTP84 knock-out plants. A-C show the response of editing ndhF C290, psbZ C50, and ndhB C1481, respectively. Wild type OTP84 complemented the knock-out phenotype in the OTP84 knock-out line; however, OTP84-E824A did not increase editing of the ndhF or psbZ editing sites. Editing of the ndhB editing site was significantly greater (23%) over the level observed in the knock-out line (5%). A single leaf sample was analyzed from three independent wild type (WT) and otp84-3 knock-out (KO) plants. A single leaf sample was analyzed from each of three independent integration events for these backgrounds and transgenes: wild type plants with OTP84 transgene (WT ϩ OTP84); otp84-3 with OTP84E824A (KO ϩ OTP84 E824A); and wild type expressing OTP84E824A (WT ϩ OTP84 E824A). Error bars show the S.D. of the three leaf samples from three independent plants in the case of WT and KO. Wild type CREF7 complemented the knock-out phenotype when expressed in the CREF7 knock-out line; however, CREF7-E554A only increased editing of ndhB C1255 to about 10%. A single leaf sample was analyzed from three independent wild type (WT) and cref7-1 knock-out (KO) plants. A single leaf sample was analyzed from each of three independent integration events for these backgrounds and transgenes: wild type plants with CREF7 transgene (WT ϩ CREF7); cref7 with CREF7-E554A (KO ϩ CREF7-E554A); and wild type expressing CREF7-E554A (WT ϩ CREF7-E554A).
teins in the editing of these sites. CREF7 appears to function as both a specificity factor and a deaminase in the editing of its cognate site, but also contributes to rpoA editing with the CLB19 as a specificity factor. VAC1 is a PPR protein with a DYW domain, and the gene knock-out results in a strong lethal phenotype and partial editing of ndhF C290 (a cognate site of OTP84) and accD C794 (a cognate site of RARE1) (48). Because the VAC1 knock-out exhibits partial editing, VAC1 is not essential for each site. If other PPR proteins participate in these editing reactions, these proteins have partially redundancy. The partial reduction in editing of rpoA C200 by the CREF7-E554A mutant suggests that CREF7 is partially redundant with other deaminases in rpoA editing.
Expression of catalytically deficient PPR transgenes by mutation of the HXE motif has proven to be a powerful way to distinguish the DYW deaminase function for a PPR protein from the site specificity role. Our results provide direct support for the model that PPR proteins have dual and separable functions as site specificity factors and as deaminases (13,19,47). a Gene knockout exhibits reduced editing. b Glu-to-Ala mutant transgene incapable of restoring editing in knockout line. c Transgene with truncation removing DYW deaminase incapable of restoring editing in knockout line.