Substitutional and insertional RNA editing of the cytochrome c oxidase subunit 1 mRNA of Physarum polycephalum.

The term RNA editing encompasses two types of specific alterations in the coding potential of RNA molecules: base substitution and the insertion (or deletion) of nucleotides. Such changes in RNA sequence can have profound effects on gene expression, and, indeed, most genes in the mitochondria of plants, trypanosomatids, and Physarum appear to require editing for their expression. We describe here the first instance of the utilization of both types of RNA editing in the processing of a single mRNA, that of the mitochondrially encoded cytochrome oxidase subunit I of the acellular slime mold, Physarum polycephalum. Editing of this mRNA includes the insertion of cytidine, guanosine, and uridine residues, as well as the apparent conversion of cytidines to uridines. No edited version of this gene was detected in Physarum DNA, and amino acid alignments suggest that both types of RNA editing are required to produce a functional protein.

RNA editing via the insertion of nucleotides has been observed in the mitochondria of trypanosomatids (1-4) and Physarum polycephalurn (5, 6). In Physarurn, the mRNA encoding the a subunit of the ATP synthetase (a ATPase) is altered via the specific addition of 54 single cytidine (C) residues (6). The source of the added information is unknown. Although it has been proposed that the multiple uridine (U) insertions that occur in trypanosomatid mitochondria are mediated by guide RNAs (71, there is currently no evidence that guide RNAs exist in Physarurn mitochondria. The identity and spacing of added nucleotides and the lack of nucleotide deletions also argue that a unique editing mechanism may be utilized by I? polycephalurn. We demonstrate here that this system is quite complex, as multiple types of RNA editing events are taking place in Physarurn mitochondria. EXPERIMENTAL PROCEDURES DNAs and RNAs-A cloned 5.5-kb' XbaI fragment of Physarum mitochondrial DNA (pQm1202, 8) was generously provided by Dennis Miller. A subcloned 2.1-kb Hind111 fragment derived from this plasmid hybridized to mitochondrial RNA on Northern blots and was sequenced in its entirety. Mitochondrial DNA and RNA were isolated from Physa-*This work was supported by the John D. and Catherine T. MacArthur Foundation and the American Cancer Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Oligonucleotides-For restriction digestions of end-labeled PCR products, the upstream primer PP-labeled B'-CTGGAACTGGTTG-GACTG-3') was 5'-end-labeled using [32P1ATP and polynucleotide kinase (Boehringer Mannheim), and the downstream primer (5'-TAAGACCACCTAATGTAAAC-3') was unlabeled. Oligonucleotides  We have focused our studies on a previously uncharacterized 2.1-kb region of the genome that is known to be actively transcribed (Ref. 8; data not shown). Sequencing of mitochondrial DNA (mtDNA) revealed that this region is highly homologous to genes for subunit I of cytochrome oxidase (col) from a variety of mitochondria. However, protein alignments indicated that numerous frameshifts would be required to produce a functional CoI protein, suggesting that the Physarum CoI mRNA is edited. Using oligonucleotide primers that hybridize to regions that are not predicted to be edited (nonedited), we generated cDNAs by reverse transcription of mitochondrial RNA (mtRNA), then amplified these cDNAs, as well as fragments of mtDNA, and total Physarum DNA using PCR.
As anticipated from our preliminary sequence data, mtDNA and mtRNA (cDNA) coZ sequences differ by the addition of nucleotides at predicted frameshift sites (Fig. 1). Most, but not all, are single C insertions (59/64 editing sites; Fig. 2). The distribution and context of C insertions are similar to those of the a ATPase mRNA (6), with irregular spacing (13-125 nucleotides) between added C nucleotides and a strong bias for inser- DNA sequence from cloned PCR products derived from mtDNA and cDNAs made by reverse transcription of mtRNA. A region that includes four C insertions (*), a U insertion b ) , and three C -+ U substitutions (0) is shown. Right, primer extension sequencing of total mtRNA using reverse transcriptase and an endlabeled col-specific oligonucleotide primer. Shown is the region that includes the four C + U changes (0) and three C insertions (*). Brackets indicate equivalent regions of each panel.

and GU. Specific insertion of nucleotides other than C into
Physarum mitochondrial RNAs has also been observed by Miller and colleagues (5).
Surprisingly, we also observed four apparent C + U changes within the coZ cDNAs. This was unexpected, as there are no reported instances of both substitutional and insertional editing within the same organism. The existence of these alterations was confirmed via primer extension sequencing of total mtRNA. As shown in Fig. 1, the four C + U changes are clearly present in the bulk coZ mRNA. Similar base substitutions in viral (9, 10) and mammalian RNAs (11,12) appear to be the result of transaminatioddeamination reactions. However, it has been suggested that C + U changes in plant mitochondria may actually occur via a deletion-insertion mechanism (13), and we cannot exclude the specific deletion of C and insertion of U at these four sites.
Both the base substitutions and the insertion of C, G, and U residues into the Physarum coZ mRNA are likely to be func-   tional, as the deduced protein sequence is highly homologous to coZ proteins from a variety of species (Fig. 3). Interestingly, one of the codons created by a C + U substitution has also been suggested to be a site of C + U change in the coZ mRNA from primrose mitochondria (14).

ACAGCUGGUAUCWACAWUCAUUACUACCWCAWGGUGWMUCWACCWCUUCCCU
To exclude the possibility that a second, edited gene exists, we employed both Southen hybridization and a sensitive PCR b, HindIII + PuuII; c, DraI +PuuII 2 HindIII; d , DraI + PuuII + MspI = HindIII. B, restriction digestions of end-labeled PCR products. Cloned mtDNA (unedited), cloned cDNA (edited), bulk cDNA synthesized from mtRNA, mtDNA, and total Physarum DNA were used as templates for PCR using as primer oligodeoxynucleotides that should anneal to both edited and unedited sequences. End-labeled PCR products were digested with either DraI or MspI and separated via electrophoresis through a 5% acrylamide, 7 M urea gel.

Insertional and Substitutional Editing in Physarum
assay. Based on the sequence of the mtDNA and cDNA clones, an edited version of the gene would have a restriction map that differs significantly from the unedited copy. Two of these differences, sites for DraI and MspI, were exploited in experiments designed to detect a second coZ gene. To further increase sensitivity, we utilized three different oligonucleotides as hybridization probes (indicated schematically at the top of Fig.   4A 1. The first probe hybridizes to a region of the coZ gene that is not edited (nonedited, N), the second oligonucleotide is complementary to the unedited gene (U), and the third probe is specific for edited sequences (E). As shown in the Southern blots presented in Fig. 4 A , no edited version of the coZ gene was detected. An edited gene should give a pattern identical to the cDNA(edited) clone (lanes 1 (lanes 3 and 4 ). Only the unedited pattern was observed with either mtDNA (lanes 5-8) or total Physarum DNA (lanes 9-12) with probes hybridizing to nonedited or unedited regions. In addition, no hybridization of the probe specific for edited sequence was observed to either mtDNA or total DNA under conditions that give a strong signal with the cDNA clone. Similar results have been obtained with other restriction enzymes and other oligonucleotide probes that are specific for edited or unedited sequences (data not shown).
A related strategy was employed for a PCR assay in which one of the primers was end-labeled and the radiolabeled PCR fragments were digested with either DraI or MspI (Fig. 4B). Digestions of PCR products derived from mtDNA and total DNA resembled that from an unedited mtDNA clone, yielding a 153-base pair DraI fragment (lanes 4, 7, and 8) and lacking restriction sites for MspI (lanes 9,12, and 13 ). Conversely, PCR products derived from bulk or cloned cDNAs yielded a 303-base pair MspI fragment (lunes 10 and 11) and were undigested by DruI (lanes 5 and 6). These experiments provide strong evidence that no edited version of the coZ gene is present in l?

polycephalum.
Three lines of evidence support the idea that editing of the Physarum coZ mRNA is a highly efficient and accurate process. 1) Primer extension sequencing of total mtRNA (Fig. 1) gives an unambiguous sequence, indicating that the bulk of the RNA is homogeneous and edited. 2) Restriction digests of PCR products derived from bulk coZ cDNA (Fig. 4B, lanes 6 and 1 1 ) suggest that virtually all of the mRNA is edited. 3) The oligonucleotide probes used in Fig. 4A have also been used to probe Northern blots of total mitochondrial RNA. Probes complementary to nonedited and edited RNAs each hybridized to an RNA of approximately 1900 nucleotides, while the oligonucleotide specific for unedited sequences did not hybridize to RNA (data not shown). Taken together, these data suggest that Physarum editing is either a co-transcriptional process or that transcription and editing are tightly coupled in Physarum mitochondria. Further work will be needed to distinguish between these mechanistic alternatives. Nevertheless, the existence of both substitutional and insertional RNA editing in a system that is amenable to biochemical, developmental, and genetic analysis (15-17) should facilitate the study of multiple editing mechanisms.