- Email: [email protected]
The ins and outs of editing RNA in kinetoplastids
The Ins and Outs of Editing RNA in Kinetoplastids SD. Seiwert Over 30 million people in fropical regions suffer from Chagas disease, African sleeping sickness or leishmaniasis. The causative agents of these diseases, flagellated protozoa collecrively known as kinetoplastids, represent an ancient lineage of eukayotes. These unusual organisms carry out a large number of unique biochemical processes, one striking example being the sequence editing of mitochondrial messenger RNAs. In this review, Scott Seiwert focuses on recent studies that examine the reaction mechanism, molecular machinery and evolutionary history of this unusual RNA processing reaction. In the mitochondria of kinetoplastids, the precursors of many messenger (m)RNAs are altered by a baroque RNA processing reaction which inserts and deletes uridylate (U) residues at specific sites within their coding regions. This process of RNA editing makes sense of an enigmatic and defining characteristic of these organisms: their mitochondrial (mt)DNA. In kinetoplastid species which cause human disease, mtDNA is a single catenated structure comprising two classes of circular DNAs (for review see Ref. 1). The first class, maxicircles, are identical 20-35 kb molecules which number between 20 and 50 per network. Although mature mRNAs derived from maxicircles encode proteins that are homologous to those specified by the mitochondrial DNA of other organisms, many of the genes for these RNAs contain translational frameshifts, lack initiation codons, or are otherwise ‘encrypted.’ A dizzying series of papers in the late 1980s suggested that the transcripts of encrypted genes are converted to mature mRNAs by RNA editing” (Box 1). Several years later it was discovered that the sequence information required to direct RNA editing is supplied by 40-70 nucleotide RNAs, called guide (g)RNAs’. The genes for gRNAs are often found on the second class of mitochondrial DNAS-12, minicircles, which are -1 kb and heterogeneous enough in sequence to specify all of the observed editing. Thus, RNA editing rationalizes the arrangement of mtDNA in kinetoplastids, but raises the question of how genetic information is transferred from gRNAs to pre-mRNAs. The answer to this question is suggested in turn by the primary structure of gRNAs. All gRNAs have three distinct functional elements: an anchor sequence of 4-14 nucleotides; an information section capable of directing U insertion and deletion; and a 3’ oligo-U tail of 5-20 residues (Fig. 1). Since gRNAs often use G:U base pairing to specify sequence, it is impossible that U insertion and deletion occur Scott Seiwert is at the Seattle Biomedical Research Institute. 4 Nicker-son Street, Seattle, WA 98109. USA and the Department of Molecular Biophysics and Biochemlstr/, Yale Unlvenity School of Medicine, 333 Cedar Street, New Haven, CT 065 IO, USA.
Tel: +I 206 284 8846, Fax: +I [email protected]
206
284
03 13, e-mail:
by template-directed polymerization. Instead, it is thought that the insertion and deletion steps are guided by gRNA /mRNA base pairing’. This process is initiated when the gRNA forms a short intermolecular (anchor) duplex with the mRNA (Fig. 1). Base pairing interactions between the information section of the gRNA and the pre-edited RNA then direct U insertion and deletion to extend the complementarity of the anchor duplex. This simple model frames many of the key questions concerning the mechanism of RNA editing in kinetoplastids &RNA editing). A more-detailed understanding of how gRNAs transfer sequence information to pre-edited RNAs requires a knowledge of how multiple gRNAs are orchestrated to direct extensive editing, what role gRNAs play in editing site selection, and what role they play during individual rounds of insertion and deletion. Linked to these issues of gRNA function is the nature of the macromolecular assembly (molecular machinery) within which kRNA editing is catalyzed. Recent progress has been made in elucidating these three facets of gRNA action and in identifying editing complexes. Taken together with the investigation of the phylogenetic distribution of kRNA editing, these studies paint a broad-brush sketch of editing in kinetoplastid protozoa. Sequential action of gRNAs within domains In some transcripts, kRNA editing supplies so much of the coding information that the true identity of the mRNA is discernible only after modification (Box 1). Such extensive editing over a large domain requires that multiple gRNAs participate in editing (Fig. 2). A means of examining how several gRNAs could be co-ordinated to act on one transcript is provided by partially edited mRNAs isolated from mitochondrial RNA. These molecules invariably have unedited sequence 5’ to edited sequence, and a junction where the two meet. Thus, it seems kRNA editing travels in a general 3’ to 5’ direction along the preedited rnRNAlsl4 and that gRNAs are used in a sequential 3’ to 5’ order (Fig. 2). Such polarity requires that edited RNA sequence base paired to downstream acting gRNAs be made available to anchor upstream acting gRNAs before upstream sequence can be specified. Downstream acting gRNAs may be directly displaced after they function, since anchor duplexes are typically more stable than the information encoding portion of the downstream acting gRNA which they replace’s. Alternatively, downstream gRNAs may be stripped from mRNAs by a helicase activity in kinetoplastid mitochondriale. Further work is needed to distinguish between these two possibilities, and to determine whether or not gRNAs are removed from mRNAs in the same molecular complex(es) in which U insertion and deletion occur (see below).
Reviews Fig. I. (right) Schematic model of guide @RNA-messenger (m)RNA interaction and kRNA editing. Closed black circles indicate 5’ ends of gRNA and pre-edited RNA. For each pairing, gRNA is below and pre-edited RNA above, the anchor duplex is indicated in red, and vertical bars represent Watson-Crick (solid) or G:U (dashed) base pairs. Bulged purines in the gRNA (green) specify U addition (green) and Us bulged in the pre-edited RNA are deleted; both events extend the complementarity of the anchor duplex. Also shown is the non-encoded oligo-U tail of gRNAs which contains 5-20 residues, which may interact (?) with purine-rich sequence upstream of processing sites.
DELETION ADDITION
SITE
SITE
Editing site selection within a gRNA specified block What is the order in which internucleotide sites are selected for processing in a block of sequence specified by a single @WA? The junctions of unedited and edited sequence in partially edited molecules hold clues to this issue as well. Essentially all of the junctions examined in T. bruceP2’J, and many of those examined in L. turentoIaelQ1~~,display unexpected editing patterns where edited sequence converges with unedited
Typanosoma and Leishmania
Box 1. The Extent of Editing Varies Among Species species, along with other members of the trypanosomatid clade of kinetoplastids, share the
maxicircle gene arrangement schematic (above). Many transcripts from these genes are remodeled by RNA editing, some
over a small portion of their length (local editing) and others so extensively that their true identity is discernible only after modification (extensive editing or pan-editing). In T. brucei, an early diverging trypanosomatidm, the majority of transcripts are extensively edited (blue in schematic and in Table). The remaining pre-mRNAs are locally edited (red in schematic and in Table) or not edited at all (white in schematic). In the later-diverging genus of Leishmunia [represented by L. turentolue (LEM 125)], the same number of trans&ipts are subject to editing@, but only six are subject Uridylates inserted - deleted” to extensive remodeling (blue in Table). In all of the trypanosomatids examined, the complexity of the RNAb Trypanosomo Leishmania Ttypanoplasmo gRNA population is large enough to specify all of the observed RNA editing. In T. brucei, several gRNAs often specify identical or extremely overlapping por? 215.41= ND8 259-45 (H) tions of mRNA sequences”, while later-diverging ? 335.4oc ND9 345-20 (H) species such as Leishmuniu display less gRNA redun? 25-o ND7 553-89 (H) dancy@). In T. brucei, editing of certain genes is also N.E. 29-15 colll 547.41 restricted to the vector or mammalian host forms of the 39-o 190.44 CYB 34-o (V) ? parasite (V and H, respectively, in Table). Interestingly, 106-5 ATP6 447-28 ? 35-14= CR3lG3 T. brucei pre-mRNAs that encode components of 14813 (H) 4-o N.E. COII 4-o (V) NADH dehydrogenase are edited in the form of the ? 28-4 MURF2 26-4 (H) organism which inhabits the mammalian host. Thus, ? N.0.c CR4Ki4 32C40 (H) bloodstream form organisms probably utilize mito? N.0.c CR5Ki5 210-13 chondrial NAD+ reducing power even though cyto117.32 131-32 UPS12 132-28 (H) chrome-mediated oxidative phosphorylation is inoperN.E. 247-80 COI N.E. ative in this life cycle stage. Kinetoplast RNA editing has recently been > 825151 ? TOTALS 3030-322 observed in Trypunoplusmu borrelW2, a member of the second clade of kinetoplastids, the bodonids. These -83 ? gRNAs >1200 organisms lack a readily identifiable mitochondrial DNA network and have a mitochondrial DNA gene a Editing known, as of January 1995, with estimated gRNA populations for T. brucei (EATRO 164) (Ref. 24) (Trypanosoma) and L tarentoloe (LEM 125) order unrelated to that of the trypanosomatids. (Ref. 60) (Leishmonio). ND, extensive editing with the precise number of Nonetheless, three of the five mRNAs identified in inserted/deleted uridylates undetermined; N.E., no editing; ?. not known. this organism are extensively edited (see Table left). b ND7, ND8, ND9= NADH dehydrogenase subunits 7,8 and 9: COI, COII. Although not fully characterized, small RNAs with COIII = cytochrome oxidase subunits I, II and Ill, Cfb =cytochrome B, the characteristics of gRNAs also seem to exist in this ATP6; tentatively assigned as ATP synthase subunit 6 (previously MURF4); species. Interestingly, an overlapping but distinct set CR3/G3,G/C strand biased sequence 3, MURFZ = maxicircle unidentified of pre-mRNAs are edited in this organism, suggesting reading frame 2; CR4/G4=G/C strand biased sequence 4; CRS/GS=G/C that extensive editing was a trait of the common strand biased sequence 5; RPS I2 = tentatively assigned as ribosomal protein 512 (previously CR6/G6). ancestor of trypanosomatids and bodonids but that ’ These transcripts are not edited in the UC strain of Leishmonia tarentolae editing of different transcripts was maintained in difdue to gRNA loss during prolonged culture. ferent species.
r Parasrtology Today, vol. I I, no / 0. 1??5
RPsl2
363
Reviews EDITING
duplexes with pre-edited RNAs and direct mis-editing in a strict 3’ to 5’ directio+23. This model predicts that the frequency of unexpected editing patterns (mis-edited molecules) should vary in proportion to the diversity of gRNAs in different species, a correlation that does hold true in L. tarentolae and T. bruceW4. Regardless of the order in which sites are chosen for processing, gRNAs could be involved in U addition and deletion in two principle ways. Sites in the preedited RNA could be (1) identified for processing if bulges in pre-mRNA/gRNA duplexes are recognized for processing’, or (2) sequestered from modification by gRNA hybridization if sites contain the correct sequence due to random U insertion and deletion within a gRNA specified block’*. Seiwert and Stuart have shown that exogenous gRNA specifies the precise number of Us deleted in vitro and that only a single intemucleotide position is subject to processingz5. Thus, this system argues against random insertion/ deletion models for editing.
DOMAIN
gRNA 1 4
gRNA1 Helicase? gRNA 1 +-
gRNA 3 7 Helicase? gRNA 2
5’.--
BLOCK 3
BLOCK 2
BLOCK 1
3’
11111111111111 9~31 Fig. 2. Sequential use of gRNAs within domains. A region specified by overlapping gRNAs is referred to as a domain (black) and the region specified by a single gRNA is referred to as an editing block (shown in white after editing). In the pre-edited RNA, an anchor duplex (red vertical bars) is possible only with the 3’-most acting gRNA. Editing directed by this gRNA creates an anchor for gRNA2, and editing with gRNA2 creates an anchor for gRNA3. Removal of gRNAs could be accomplished by direct strand displacement or through the action of an RNA helicase.
sequence. These junctions are referred to as ‘unexpected’ because, within them, several clustered internucleotide sites match neither the gene nor the mature message. Two models attempt to rationalize these molecules by exploring the order of editing site selection. Common to both is that, if these molecules give rise to mature RNAs, sites undergo multiple rounds of editing to reach the correct sequence. Koslowsky et al. propose that unexpected editing patterns are bona fide intermediates in processing*O. In this view, unexpected patterns occur because processing sites are initially chosen through the most favorable base-pairing interaction of the gRNA with the pre-edited RNA, not the fully edited mRNA. Realignment of the gRNA and partially edited RNA identifies new sites for processing, and this process is repeated until the gRNA /mRNA duplex is continuous (and the final sequence achieved). A contrasting view argues that unexpected editing patterns result when non-cognate gRNAs form spurious anchor 364
Proposed mechanisms for insertion and deletion Two possible sources of Us have led to the suggestion of three models for the process of U addition and deletior?JQ-7 (Fig. 3). In all of these models, gRNAs are presumed to identify processing sites by mismatch recognition. So far, these models are suggested only on the basis of circumstantial evidence, but recently developed in vitro system&28 hold the promise of differentiating among them. Model 1. The original model for kRNA editing suggests that free uridine triphosphate (UTP) in the mitochondrion may serve as the source of Us inserted into pre-mRNA7. As illustrated by the blue arrows in Fig. 3, in this model a round of U addition is initiated when the pre-mRNA is cleaved across from a bulged gRNA purine to generate a 5’ half RNA with a 3’ hydroxyl (OH) and a 3’ half RNA with a 5’ monophosphate (p). UTP could be added to the 3’ end of the 5’ cleavage product, possibly while the 5’ half molecule is held in place through hybridization to the 3’ oligo-U tail of the gRNAB, creating a base-pairing partner for the bulged purine. Subsequent ligation of the two half RNAs produces a transcript with an inserted U. If Us are to be deleted, uridine 5’ monophosphate could be removed from the 5’ cleavage product through the same activity operating in reverse or by an unrelated 3’ exonuclease, and the two half RNAs ligated together. All of the required enzymatic activities for this pathway have been detecteda@ (see below). In addition, the intermediates predicted by this model have been isolated from in vivo pools of RNA*%” and produced in vifroQ33. Most of the putative 5’-half intermediates isolated from mitochondrial RNA contain either (1) completely unedited sequence, or (2) unedited sequence with unexpected editing patterns near the cleavage site. As editing proceeds in a 3’ to 5’ direction, and junctions probably represent regions of active editing, these molecules have characteristics of kRNA editing intermediates. Such products often have Us added to their 3’ ends, which can number up to 25 residuesis,34. Therefore, many residues may be added to the 5’ cleavage product and the excess removed in a way specified by gRNA pairing interactions. Parasitology To&y, vol. I I, no. IO, I995
Reviews
LIG
0
+
I)
norn-1
Fig. 3. Two possible sources of Us suggest three models for insertion and deletion. insertion of a single U in the pre-edited RNA (yellow) in apposition to an adenosine bulged in the gRNA (red) is depicted. Reactive 3’ hydroxyls (OH) and 5’ monophosphates (p) are also indicated in red. RNA cleavage (CL), ligation (LIG), and transesterification (TE) are indicated. Free UTP (blue arrows) or the oligo-U tail of the gRNA (green arrows) could provide the incorporated U. If the oligo-U tail donates residues, reaction could be through cleavage and ligation (solid green arrows) or transesterification (dashed green arrows). Note that some intermediates and enzymatic activities are common to multiple pathways. pathways shown in green result in the loss of a U from the 3’ end of the gRNA and preclude the base pairing of the oligo-U tail of the gRNA with the pre-mRNA during reaction.
Model 2. As illustrated by the solid green arrows in Fig. 3, Us located at the 3’ ends of the-reactant gRNA may directly supply the residues incorporated into the pre-mRNA *7,32,33.As in Model 1, endonucleolytic cleavage of the substrate to generate 5’ and 3’ half molecules is the initiating event in this pathway. Instead of base pairing with the 5’ cleavage product while free UTP is added, in this model the 3’ end of the gRNA is ligated to the 3’ cleavage product to generate a gRNA/mRNA chimera. Cleavage of the chimera 5’ of the U which was originally the 3’-most nucleotide of the gRNA would produce a 3’ cleavage product with an added U and a gRNA shortened by one residue. Subsequent ligation of the two half RNAs would then produce a transcript with an incorporated U (Fig. 3, solid green arrows). A U could be removed in this pathway by resolution of the chimera 3’ to a U derived from the pre-mRNA. Model 3. Based on analogy to RNA splicing, chimeric molecules could also be generated in a reaction pathway consisting of two RNA catalyzed transesterification reaction+27 (Fig. 3, dashed green arrows). Formation of a chimeric intermediate would be coupled to pre-mRNA scission at the processing site by a phosphotransfer reaction in which the 3’ OH of the gRNA acts as the attacking nucleophile at the editing site. Pseudoreversal of this reaction with the 3’ OH of the 5’ cleavage product as the attacking nucleophile would then insert a U residue at the processing site if attack was 5’ of the original gRNA Porositology Today, vo/
/ I no 1I:
‘JQ5
terminus. If attack was 3’ to the original terminus, a U would be deleted. Although transesterification is an attractive proposal because it requires only one activity and a single active site, no direct evidence favoring this model has been obtained. Indirect evidence for Models 2 and 3 is based on the detection of chimeras from in viva pools of mitochondrial RNA27,3+35 and their production in uitr&36,37. Although the existence of chimeras provides strong evidence for a mechanism such as those in Models 2 and 3, a substantial portion of chimeras identified in some studies do not display the characteristics expected for productive editing intermediates. Most puzzling is that some chimeras do not have Us at the linkage site35.36, making addition or deletion impossible in these cases. The low abundance of chimeras relative to gRNA, pre-edited RNA, and edited RNA38 also casts doubt upon whether they are productive intermediates in editing. Nonetheless, chimera abundance parallels the abundance of edited RNA38 and chimera-forming activity co-sediments in glycerol gradients with complexes which display editing-related activities39 (see below). Therefore, chimeras appear to be formed by the same machinery that converts pre-edited molecules to edited ones. Could chimeras be non-productive side products? Kinetoplastid RNA editing, like pre-mRNA splicing in higher eukaryotes, generates a 5’ cleavage product which must be held in position during the 365
Reviews
r
UTP
Fig. 4. A critical view of chimeras. The non-encoded oligo-U tail of the gRNA and/or an unidentified factor (X encircled in blue) must hold the 5’ cleavage product during accurate editing (blue arrows). Attenuation of this interaction (green arrows) could result in loss of the 5’ cleavage product. Placement of the gRNA’s 3’ end into the site normally occupied by the 3’ end of the 5’ cleavage product would juxtapose reactive 3’ OH and 5’ monophosphate residues (red). Chimera formation (green arrows) rather than production of edited RNA (blue arrows) would result upon ligation of these two groups.
course of processing. During splicing, U residues in the conserved loop of the U5 small nuclear RNA and the ~220 protein perform this taskdo. During kRNA editing, the oligo-U tail of the gRNA could perform a similar function29. Conceivably, this interaction could also be aided by an unidentified factor (factor X in Fig. 4). Regardless of the nature of the activity that holds the 5’ cleavage product, the cleavage product could be lost from the editing machinery if this interaction became unstable (Fig. 4, green arrows). The 3’ hydroxyl of the gRNA could then be juxtaposed to the 5’ end of the 3’ cleavage product, as the 5’ cleavage product is normally. Aberrant ligation in this pathway would generate a chimera (green arrows) rather than edited product (blue arrows). Major issues concerning the mechanism of insertion and deletion clearly remain unresolved. Characterization of potential intermediates and enzymatic activities has proven unable to distinguish among models for these chemical steps, since intermediates and activities are shared in the different pathways. Final proof of whether chimeric molecules are intermediates in editing will require probing of the role of the 3’ OH of the gRNA during in vitro processing and determining whether chimeras produced in vitro can be ‘chased’ into edited product. 366
Molecular machinery . . In addition to progress gained wrth molecular genetic and biochemical analyses, advances have been made in the physical characterization of putative editing complexes, or editosomes. These studies have relied most heavily on the analysis of the sedimentation of putative editing-related activities in glycerol gradients 28,39. Although such work does not provide formal proof of the existence of macromolecular editing complexes, co-sedimentation does provide persuasive circumstantial evidence. Pollard et al. have described two putative editing complexes in T. brucei39: one that sediments at -20s contains endogenous gRNAs, terminal uridylyl transferase (TUTase), RNA ligase, chimera-forming activity and, although initially reported otherwise39, pre-edited RNA specific nuclease (B. Adler and S. Hajduk, pers. commun.). A second complex that sediments at -35-40s lacks TUTase, contains lesser amounts of the other activities, but contains pre-edited mRNAs. Using similar gradients, it has been shown that an in vitro gRNA-dependent U deletion activity sediments in the 20 S region (K. Stuart, pers. commun.). This complex may therefore represent the catalytically active editosome, provided subsequent assembly into the 35-40s complex does not occur. The 35-40 S region of the gradient itself does not display Parasrtology
Today, vol. I I, no. IO, I995
Reviews U deletion activity, but this complex may simply be unable to take up exogenous substrates in vitro. Alternatively, the 35-40s complex may be involved in U addition or in recycling kRNA editing components. In Leishmania, putative editing complexes have also been observed28. Provocatively, endogenous gRNAs and U addition activity also sediment at -205 in this system. It is not yet known whether U insertion requires the co-sedimenting gRNAs but, if so, this would suggest that insertion28,41 and deletion25 activities are related. Unlike T. brucei, the majority of the RNA ligase and TUTase activities in Leishmania are found uncomplexed and in 10s fractions, respectively28. Although this represents a significant difference from T. bruc@, it may reflect a difference in complex stability rather than a fundamental difference in the nature of the catalytic machinery. This notion is supported by the ability to alter the sedimentation behavior of editing-related activities in T. brucei by changing ionic conditions (K. Stuart, pers. commun.). The existence of RNA ligase, TUTase and preedited RNA-specific endonuclease activities%33, and their partial co-sedimentation with editing activity in T. brucei (K. Stuart, pers. commun.), seems to argue that the kRNA editing reaction is catalyzed by protein(s) in these complexes. This is further supported by the sensitivity of U deletion in vitro to proteinase treatment (S. Seiwert, unpublished). The sizes of putative editing complexes in T. brucei and L. tarentolue indicate that the editing machinery contains numerous other proteins 39. Additional protein factors could include RNA helicaselh, proofreading factors, factors required for scaffolding of the active site(s) and other undefined functions. RNA-protein crosslinking studies have identified several gRNA specific proteins which may perform some of these function@43. Efforts are now under way in a number of laboratories to purify and characterize these and other proteins associated with editing complexes.
Evolutionary considerations Although a diverse collection of eukaryotes posttranscriptionally alter information in their RNAs (for reviews see Refs 44,45), no organisms besides kinetoplastids are known to transfer sequence information from one RNA to another. However, since kinetoplastids represent one of the oldest eukaryotic lineages with mitochondriaM, the possibility remains that kRNA editing is an ancient trait that has been lost in other organisms. To estimate the antiquity of kRNA editing, its occurrence within various kinetoplastids has been examined. The order Kinetoplastida contains two suborders, the Trypanosomatina and the Bodonina, which are estimated to have diverged 680-900 million years ago4’. Within the Trypanosomatina, extensive editing of transcripts occurs frequently in Typanosornu and (probably) in Herpetomonusm. Several of the homologous transcripts in Leishrnuniu and Crithidiu are subject to editing only in small domains near their 5’ ends. Phylogenies based upon ribosomal RNA suggest that the divergence of the former lineages is more ancient, implying that extensive editing is an ancestral state within trypanosomatids@-50. The recent demonstration of extensive editing in the bodonid TrypunoPorasrtology Today, vol. I I, no. / 0. I995
plasma borreli (family Cryptobiidae)51,52 suggests extensive editing must be older than the split between trypanosomatids and bodonids and could be as old as the kinetoplastid lineage itself. As Maslov and Simpson point out, the restriction of editing to the 5’ portions of mRNAs in Leishmuniu and Crithidiu may have arisen by retroposition of partially edited transcripts in the ancestor of these organism#‘. The discovery of a reverse transcriptase gene in kinetoplastids lends support to this view53 and provides a plausible model for loss of kRNA editing in non-kinetoplastids. Direct support that kRNA editing is an ancestral trait of all mitochondria bearing eukaryotes would require its discovery in other organisms, the most likely candidates possibly being the kinetoplastid sister group of Euglenoids or in the probable bacterial cousins of mitochondria, the ol-ProteobacteriaM. Further investigation of the time of origin of kRNA editing will provide clues to its original function by placing it in a paleohistorical context. At present, one can only speculate about the selective pressures experienced by the ancestors of kinetoplastids that resulted in the development of such a baroque pathway of gene expression. If kRNA editing dates to the ‘RNA world’55, possible functions include an involvement in the replication of RNA56, or in an ‘excision’ repair process required for the maintenance of RNA genomes. Once established, kRNA editing could have been exploited to restrict the expression of certain mitochondrial genes to particular life cycle stages of parasites, a function used today by certain digenic trypanosomatids (for review see Ref. 57). An altemative view suggests that kRNA editing may help to enhance genetic variation by duplication and divergence of gRNA@s, sloppy editing59, or usage of misguiding gRNAs22,U.
Prospects In a relatively short time, kRNA editing has relinquished many of its secrets. The field has progressed from documentation of the phenomenon of kRNA editing to characterization of possible editing intermediates. Direct probing of in vitro systems now holds the promise of answering fundamental questions about the mechanism and the machinery of this unusual RNA processing reaction, but care must be taken to relate these in vitro results to the process of kRNA editing which occurs in viva. The continued search for kRNA-like editing in other organisms will indicate whether RNA directed RNA modification is a trait ancestral to all eukaryotes bearing mitochondria, or a peculiar feature of the kinetoplastid lineage. Acknowledgements I am indebted to Kenneth Stuart for financial support and invaluable discussions, and to Theo deVos, Stefan Heidmann, Robert Hughes, Kym Jacobson and Elisabetta Ullu for comments on the manuscript. References Stuart, K. and Feagin, J.E. (1992) Int. Rev. Cytol. 141,65-87 Benne, R. et ~2. (1986) Cell 46,819-826 Feagin, J.E. et ai. (1987) Cdl 49,337-345 Shaw. I.M. et al. (1988) Cell 53.401-411 van cidr Spek, H.‘et a2.‘(1988) &MB0 1. 7,2509-2514 Feagin, J.E. et al. (1988) Cell 53,4X%-422 Blum, B. et al. (1990) Cell 60,189-198 Sturm, N.R. and Simpson, L. (1990) Cell 61,879-884 367
Reviews 9 Sturm, N. and Simpson, L. (1991) Nucleic Acids Res. 19, 6277-6281 10 Bhat, G.J. et al. (1990) Cell 61, 885-894 11 Pollard, V.W. et al. (1990) Cell 63,783-790 12 van der Spek, H. et al. (1991) EMBO J. 10,1217-1224 13 Abraham, J.M. et al. (1988) Cell 55,267-272 14 Sturm, N.R. and Simpson, L. (1990) Cell 61,871-878 15 Maslov, D.A. and Simpson, L. (1992) Cell 70,459-467 16 Missel, A. and Goringer, U.H. (1994) Nucleic Acids Res. 22, 405m56 17 Souza, A.E. et ~2. (1992) Mol. Cell. Biol. 12,2100-2107 18 Decker, CJ. and Sollner-Webb, B. (1990) Cell 61,1001-1011 19 Riley, G.R. et aZ. (1994) 1. Biol. Chem. 269,61014108 20 Kosiowsky, D.J. et al. (i991) Cell 67,537-546 21 Maslov. D.A. et al. (19921 Mol. Cell. Biol. 12. 56-67 22 Sturm, N.R. et uZ.(i992) ?eZZ70,469-476 23 Maslov, D.A. et al. (1994) Mol. Biochem. Parusitol. 68, 155159 24 Corell, R.A. et al. (1993) Nucleic Acids Res. 21,4313-4320 25 Seiwert, S.D. and Stuart, K. (1994) Science 266,llP117 26 Cech, T.R. (1991) Cell 64,667-669 27 Blum, B. et al. (l&l) Cell 65,543-550 28 Peris. M. et al. 119941 EMBO 1.13.1664-1672 29 Blud, B. and Simpsbn, L. (l&O)CeZZ 62,391-397 30 White, T.C. and Borst, P. (1987) NucZeic Acids Res. 15, 32753289 31 Bakalara, N. et al. (1989) J. Biol. Chem. 264,18679-18686 32 Harris, M.E. and Hajduk, S.L. (1992) Cell 68,1091-1099 33 Harris, M. et al. (1992) Mol. Cell. BioZ. 12,2591-2598 34 Read, L.K. et al. (1992) Nucleic Acids Res. 20, 2341-2347 35 Arts, G.J. et UZ.(1993) EMBO 1. 12,1523-1532 36 Koslowsky, D.J. et al. (1992) Nature 356,807-809 37 Blum, B. and Simpson, L. (1992) Proc. NutZ Acud. Sci. USA 89 11944-11948
38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Riley, G.R. et al. (1995) Nucleic Acids Res. 23,708-712 Pollard, V.W. et al. (1992) EMBO 1. 11,4429-4438 Sontheimer, E.J. and Steitz, J.A. (1994) Science 262,1989-1996 Harris, M.E. et al. (1990) 1. BioZ. Chem. 265,11368-11376 Read, L.K. et al. (1994) MoZ. CeZZ.Bid. 14,2629-2639 Keller, J. et uZ.(1994) Ir]ucZeicAcids Res. 22, 1988-1995 Bass. B. (1993) in The RNA World (Gesteland, R.F. and Atkins, J.F., eds),‘pp 3318, Cold Spring Harbor Laboratory Press Adler, B.K. and Hajduk, S.L. (1994) Curr. @in. Genet. Dev. 4, 316-322 Sogin, M.L. (1994) in Early Life on Earth (Bengtson, S., ed.), pp 181-192, Columbia University Press Femandes, AI’. et UI. (1993) Proc. Nut2 Acud. Sci. USA 90, 11608-11612 Landweber, L.F. and Gilbert, W. (1994) Proc. NutZ Ad. Sci. USA 91,918-921 Landweber, L.F. and Gilbert, W. (1993) Nature 363,179-182 Maslov, D.A. et ~2. (1994) Nature 368,345-348 Lukes, J. et a2. (1994) EMBO ].13,5086-5098 Maslov, D. and Simpson, L. (1994) Mol. CeZZ. BioZ. 14, 8174-8182 Gabriel, A. and Boeke, J-D. (1991) Proc. NutZ Acud. Sci. USA 88, 9794-979s Gray, M.W. (1994) Nature 368,288 Gilbert, W. (1986) Nature 319,618 Benne, R. (1990) Trends Genet. 6,177-181 Priest, J.W. and Hajduk, S.L. (1994) 1. Bioenerg. Biomembrunes 26, 179-191 Landweber, L.F. et al. (1993) Proc. NufZ Acud. Sci. USA 90, 9242-9246 Read, L.K. et al. (1994) Nucleic Acids Res. 22, 1489-1495 Thiemann, O.H. et al. (1994) EMBO ].13,5689-5700
Polydnaviruses: Potent Mediators of Host Insect Immune Dysfunction M.D. Lavine and N.E. Beckage Endoparasitic insects are used as biological control agents to kill many species of insect pest. One key to the success of parasitoids that develop in the hemocoel of their host is their ability to knock out the host’s immune system, inducing a decline in the responsiveness of a variety of cellular and humoral components so that parasitoid eggs are not encapsulated. In many species parasitized by braconid and ichneumonid wasps, host immunosuppression appears to be mediated by polydnaviruses (PDVs) injected by the female parasitoid into the host hemocoel. The viruses exhibit a complex and intimate genetic relationship with the wasp, since viral sequences are integrated within the wasp’s chromosomal DNA. Here Mark Lavine and Nancy Beckage summarize the current evidence for mechanisms of virally induced host immunosuppression in parasitized insects, as well as the roles of other factors including wasp ovarian proteins and venom components, in suppressing hemocytemediated and humoral immune responses. Interestingly, in some species, the PDV-induced host immunosuppression appears transitory, with older parasitoid larvae probably exploiting other mechanisms to protect themselves from the Mark Lavlne IS at the Department of Biology, and Nancy Beckage IS at the Department of Entomology 5419 Boyce Hall, Univenlty of Califomla-Rivewde, Rivenlde, CA 9252 I-03 14, USA. Tel: + I 909 787 3521, Fax: + I 909 787 3087, e-mail: [email protected]
host’s immune system during thefinal stages of parasitism. During the final stages of parasitism, the parasitoids likely exploit other mechanisms of immunoaasion via antigen masking, antigen mimicry, or production of active inhibitors of the hemocyte-mediated encapsulation response as well as inhibiting melanization. Many members of the insect order Hymenoptera are endoparasitoids, which are parasites that develop within, and always kill, their insect hosts. For endoparasitoids to develop successfully, they must avoid or suppress their host’s natural defense mechanisms, and hence have evolved complex mechanisms for modulating the host insect’s cellular and humoral immune system.G. Parasitoids exhibit sophisticated mechanisms of interaction with their host insect’s immune, endocrine and metabolic pathways, and, frequently, even modify the development and behavior of their host; the outcome is the successful maturation of the parasitoid accompanied simultaneously by the host’s demise. Endoparasitic wasps in the families Braconidae and Ichneumonidae have developed unique associations with a family of polydisperse DNA viruses (polydnaviruses, or PDVs). These viruses replicate exclusively in the female wasp’s ovaries, and each species of wasp is presumed to have a unique PDV; the
Tel: +I 206 284 8846, Fax: +I [email protected]
206
284
03 13, e-mail:
by template-directed polymerization. Instead, it is thought that the insertion and deletion steps are guided by gRNA /mRNA base pairing’. This process is initiated when the gRNA forms a short intermolecular (anchor) duplex with the mRNA (Fig. 1). Base pairing interactions between the information section of the gRNA and the pre-edited RNA then direct U insertion and deletion to extend the complementarity of the anchor duplex. This simple model frames many of the key questions concerning the mechanism of RNA editing in kinetoplastids &RNA editing). A more-detailed understanding of how gRNAs transfer sequence information to pre-edited RNAs requires a knowledge of how multiple gRNAs are orchestrated to direct extensive editing, what role gRNAs play in editing site selection, and what role they play during individual rounds of insertion and deletion. Linked to these issues of gRNA function is the nature of the macromolecular assembly (molecular machinery) within which kRNA editing is catalyzed. Recent progress has been made in elucidating these three facets of gRNA action and in identifying editing complexes. Taken together with the investigation of the phylogenetic distribution of kRNA editing, these studies paint a broad-brush sketch of editing in kinetoplastid protozoa. Sequential action of gRNAs within domains In some transcripts, kRNA editing supplies so much of the coding information that the true identity of the mRNA is discernible only after modification (Box 1). Such extensive editing over a large domain requires that multiple gRNAs participate in editing (Fig. 2). A means of examining how several gRNAs could be co-ordinated to act on one transcript is provided by partially edited mRNAs isolated from mitochondrial RNA. These molecules invariably have unedited sequence 5’ to edited sequence, and a junction where the two meet. Thus, it seems kRNA editing travels in a general 3’ to 5’ direction along the preedited rnRNAlsl4 and that gRNAs are used in a sequential 3’ to 5’ order (Fig. 2). Such polarity requires that edited RNA sequence base paired to downstream acting gRNAs be made available to anchor upstream acting gRNAs before upstream sequence can be specified. Downstream acting gRNAs may be directly displaced after they function, since anchor duplexes are typically more stable than the information encoding portion of the downstream acting gRNA which they replace’s. Alternatively, downstream gRNAs may be stripped from mRNAs by a helicase activity in kinetoplastid mitochondriale. Further work is needed to distinguish between these two possibilities, and to determine whether or not gRNAs are removed from mRNAs in the same molecular complex(es) in which U insertion and deletion occur (see below).
Reviews Fig. I. (right) Schematic model of guide @RNA-messenger (m)RNA interaction and kRNA editing. Closed black circles indicate 5’ ends of gRNA and pre-edited RNA. For each pairing, gRNA is below and pre-edited RNA above, the anchor duplex is indicated in red, and vertical bars represent Watson-Crick (solid) or G:U (dashed) base pairs. Bulged purines in the gRNA (green) specify U addition (green) and Us bulged in the pre-edited RNA are deleted; both events extend the complementarity of the anchor duplex. Also shown is the non-encoded oligo-U tail of gRNAs which contains 5-20 residues, which may interact (?) with purine-rich sequence upstream of processing sites.
DELETION ADDITION
SITE
SITE
Editing site selection within a gRNA specified block What is the order in which internucleotide sites are selected for processing in a block of sequence specified by a single @WA? The junctions of unedited and edited sequence in partially edited molecules hold clues to this issue as well. Essentially all of the junctions examined in T. bruceP2’J, and many of those examined in L. turentoIaelQ1~~,display unexpected editing patterns where edited sequence converges with unedited
Typanosoma and Leishmania
Box 1. The Extent of Editing Varies Among Species species, along with other members of the trypanosomatid clade of kinetoplastids, share the
maxicircle gene arrangement schematic (above). Many transcripts from these genes are remodeled by RNA editing, some
over a small portion of their length (local editing) and others so extensively that their true identity is discernible only after modification (extensive editing or pan-editing). In T. brucei, an early diverging trypanosomatidm, the majority of transcripts are extensively edited (blue in schematic and in Table). The remaining pre-mRNAs are locally edited (red in schematic and in Table) or not edited at all (white in schematic). In the later-diverging genus of Leishmunia [represented by L. turentolue (LEM 125)], the same number of trans&ipts are subject to editing@, but only six are subject Uridylates inserted - deleted” to extensive remodeling (blue in Table). In all of the trypanosomatids examined, the complexity of the RNAb Trypanosomo Leishmania Ttypanoplasmo gRNA population is large enough to specify all of the observed RNA editing. In T. brucei, several gRNAs often specify identical or extremely overlapping por? 215.41= ND8 259-45 (H) tions of mRNA sequences”, while later-diverging ? 335.4oc ND9 345-20 (H) species such as Leishmuniu display less gRNA redun? 25-o ND7 553-89 (H) dancy@). In T. brucei, editing of certain genes is also N.E. 29-15 colll 547.41 restricted to the vector or mammalian host forms of the 39-o 190.44 CYB 34-o (V) ? parasite (V and H, respectively, in Table). Interestingly, 106-5 ATP6 447-28 ? 35-14= CR3lG3 T. brucei pre-mRNAs that encode components of 14813 (H) 4-o N.E. COII 4-o (V) NADH dehydrogenase are edited in the form of the ? 28-4 MURF2 26-4 (H) organism which inhabits the mammalian host. Thus, ? N.0.c CR4Ki4 32C40 (H) bloodstream form organisms probably utilize mito? N.0.c CR5Ki5 210-13 chondrial NAD+ reducing power even though cyto117.32 131-32 UPS12 132-28 (H) chrome-mediated oxidative phosphorylation is inoperN.E. 247-80 COI N.E. ative in this life cycle stage. Kinetoplast RNA editing has recently been > 825151 ? TOTALS 3030-322 observed in Trypunoplusmu borrelW2, a member of the second clade of kinetoplastids, the bodonids. These -83 ? gRNAs >1200 organisms lack a readily identifiable mitochondrial DNA network and have a mitochondrial DNA gene a Editing known, as of January 1995, with estimated gRNA populations for T. brucei (EATRO 164) (Ref. 24) (Trypanosoma) and L tarentoloe (LEM 125) order unrelated to that of the trypanosomatids. (Ref. 60) (Leishmonio). ND, extensive editing with the precise number of Nonetheless, three of the five mRNAs identified in inserted/deleted uridylates undetermined; N.E., no editing; ?. not known. this organism are extensively edited (see Table left). b ND7, ND8, ND9= NADH dehydrogenase subunits 7,8 and 9: COI, COII. Although not fully characterized, small RNAs with COIII = cytochrome oxidase subunits I, II and Ill, Cfb =cytochrome B, the characteristics of gRNAs also seem to exist in this ATP6; tentatively assigned as ATP synthase subunit 6 (previously MURF4); species. Interestingly, an overlapping but distinct set CR3/G3,G/C strand biased sequence 3, MURFZ = maxicircle unidentified of pre-mRNAs are edited in this organism, suggesting reading frame 2; CR4/G4=G/C strand biased sequence 4; CRS/GS=G/C that extensive editing was a trait of the common strand biased sequence 5; RPS I2 = tentatively assigned as ribosomal protein 512 (previously CR6/G6). ancestor of trypanosomatids and bodonids but that ’ These transcripts are not edited in the UC strain of Leishmonia tarentolae editing of different transcripts was maintained in difdue to gRNA loss during prolonged culture. ferent species.
r Parasrtology Today, vol. I I, no / 0. 1??5
RPsl2
363
Reviews EDITING
duplexes with pre-edited RNAs and direct mis-editing in a strict 3’ to 5’ directio+23. This model predicts that the frequency of unexpected editing patterns (mis-edited molecules) should vary in proportion to the diversity of gRNAs in different species, a correlation that does hold true in L. tarentolae and T. bruceW4. Regardless of the order in which sites are chosen for processing, gRNAs could be involved in U addition and deletion in two principle ways. Sites in the preedited RNA could be (1) identified for processing if bulges in pre-mRNA/gRNA duplexes are recognized for processing’, or (2) sequestered from modification by gRNA hybridization if sites contain the correct sequence due to random U insertion and deletion within a gRNA specified block’*. Seiwert and Stuart have shown that exogenous gRNA specifies the precise number of Us deleted in vitro and that only a single intemucleotide position is subject to processingz5. Thus, this system argues against random insertion/ deletion models for editing.
DOMAIN
gRNA 1 4
gRNA1 Helicase? gRNA 1 +-
gRNA 3 7 Helicase? gRNA 2
5’.--
BLOCK 3
BLOCK 2
BLOCK 1
3’
11111111111111 9~31 Fig. 2. Sequential use of gRNAs within domains. A region specified by overlapping gRNAs is referred to as a domain (black) and the region specified by a single gRNA is referred to as an editing block (shown in white after editing). In the pre-edited RNA, an anchor duplex (red vertical bars) is possible only with the 3’-most acting gRNA. Editing directed by this gRNA creates an anchor for gRNA2, and editing with gRNA2 creates an anchor for gRNA3. Removal of gRNAs could be accomplished by direct strand displacement or through the action of an RNA helicase.
sequence. These junctions are referred to as ‘unexpected’ because, within them, several clustered internucleotide sites match neither the gene nor the mature message. Two models attempt to rationalize these molecules by exploring the order of editing site selection. Common to both is that, if these molecules give rise to mature RNAs, sites undergo multiple rounds of editing to reach the correct sequence. Koslowsky et al. propose that unexpected editing patterns are bona fide intermediates in processing*O. In this view, unexpected patterns occur because processing sites are initially chosen through the most favorable base-pairing interaction of the gRNA with the pre-edited RNA, not the fully edited mRNA. Realignment of the gRNA and partially edited RNA identifies new sites for processing, and this process is repeated until the gRNA /mRNA duplex is continuous (and the final sequence achieved). A contrasting view argues that unexpected editing patterns result when non-cognate gRNAs form spurious anchor 364
Proposed mechanisms for insertion and deletion Two possible sources of Us have led to the suggestion of three models for the process of U addition and deletior?JQ-7 (Fig. 3). In all of these models, gRNAs are presumed to identify processing sites by mismatch recognition. So far, these models are suggested only on the basis of circumstantial evidence, but recently developed in vitro system&28 hold the promise of differentiating among them. Model 1. The original model for kRNA editing suggests that free uridine triphosphate (UTP) in the mitochondrion may serve as the source of Us inserted into pre-mRNA7. As illustrated by the blue arrows in Fig. 3, in this model a round of U addition is initiated when the pre-mRNA is cleaved across from a bulged gRNA purine to generate a 5’ half RNA with a 3’ hydroxyl (OH) and a 3’ half RNA with a 5’ monophosphate (p). UTP could be added to the 3’ end of the 5’ cleavage product, possibly while the 5’ half molecule is held in place through hybridization to the 3’ oligo-U tail of the gRNAB, creating a base-pairing partner for the bulged purine. Subsequent ligation of the two half RNAs produces a transcript with an inserted U. If Us are to be deleted, uridine 5’ monophosphate could be removed from the 5’ cleavage product through the same activity operating in reverse or by an unrelated 3’ exonuclease, and the two half RNAs ligated together. All of the required enzymatic activities for this pathway have been detecteda@ (see below). In addition, the intermediates predicted by this model have been isolated from in vivo pools of RNA*%” and produced in vifroQ33. Most of the putative 5’-half intermediates isolated from mitochondrial RNA contain either (1) completely unedited sequence, or (2) unedited sequence with unexpected editing patterns near the cleavage site. As editing proceeds in a 3’ to 5’ direction, and junctions probably represent regions of active editing, these molecules have characteristics of kRNA editing intermediates. Such products often have Us added to their 3’ ends, which can number up to 25 residuesis,34. Therefore, many residues may be added to the 5’ cleavage product and the excess removed in a way specified by gRNA pairing interactions. Parasitology To&y, vol. I I, no. IO, I995
Reviews
LIG
0
+
I)
norn-1
Fig. 3. Two possible sources of Us suggest three models for insertion and deletion. insertion of a single U in the pre-edited RNA (yellow) in apposition to an adenosine bulged in the gRNA (red) is depicted. Reactive 3’ hydroxyls (OH) and 5’ monophosphates (p) are also indicated in red. RNA cleavage (CL), ligation (LIG), and transesterification (TE) are indicated. Free UTP (blue arrows) or the oligo-U tail of the gRNA (green arrows) could provide the incorporated U. If the oligo-U tail donates residues, reaction could be through cleavage and ligation (solid green arrows) or transesterification (dashed green arrows). Note that some intermediates and enzymatic activities are common to multiple pathways. pathways shown in green result in the loss of a U from the 3’ end of the gRNA and preclude the base pairing of the oligo-U tail of the gRNA with the pre-mRNA during reaction.
Model 2. As illustrated by the solid green arrows in Fig. 3, Us located at the 3’ ends of the-reactant gRNA may directly supply the residues incorporated into the pre-mRNA *7,32,33.As in Model 1, endonucleolytic cleavage of the substrate to generate 5’ and 3’ half molecules is the initiating event in this pathway. Instead of base pairing with the 5’ cleavage product while free UTP is added, in this model the 3’ end of the gRNA is ligated to the 3’ cleavage product to generate a gRNA/mRNA chimera. Cleavage of the chimera 5’ of the U which was originally the 3’-most nucleotide of the gRNA would produce a 3’ cleavage product with an added U and a gRNA shortened by one residue. Subsequent ligation of the two half RNAs would then produce a transcript with an incorporated U (Fig. 3, solid green arrows). A U could be removed in this pathway by resolution of the chimera 3’ to a U derived from the pre-mRNA. Model 3. Based on analogy to RNA splicing, chimeric molecules could also be generated in a reaction pathway consisting of two RNA catalyzed transesterification reaction+27 (Fig. 3, dashed green arrows). Formation of a chimeric intermediate would be coupled to pre-mRNA scission at the processing site by a phosphotransfer reaction in which the 3’ OH of the gRNA acts as the attacking nucleophile at the editing site. Pseudoreversal of this reaction with the 3’ OH of the 5’ cleavage product as the attacking nucleophile would then insert a U residue at the processing site if attack was 5’ of the original gRNA Porositology Today, vo/
/ I no 1I:
‘JQ5
terminus. If attack was 3’ to the original terminus, a U would be deleted. Although transesterification is an attractive proposal because it requires only one activity and a single active site, no direct evidence favoring this model has been obtained. Indirect evidence for Models 2 and 3 is based on the detection of chimeras from in viva pools of mitochondrial RNA27,3+35 and their production in uitr&36,37. Although the existence of chimeras provides strong evidence for a mechanism such as those in Models 2 and 3, a substantial portion of chimeras identified in some studies do not display the characteristics expected for productive editing intermediates. Most puzzling is that some chimeras do not have Us at the linkage site35.36, making addition or deletion impossible in these cases. The low abundance of chimeras relative to gRNA, pre-edited RNA, and edited RNA38 also casts doubt upon whether they are productive intermediates in editing. Nonetheless, chimera abundance parallels the abundance of edited RNA38 and chimera-forming activity co-sediments in glycerol gradients with complexes which display editing-related activities39 (see below). Therefore, chimeras appear to be formed by the same machinery that converts pre-edited molecules to edited ones. Could chimeras be non-productive side products? Kinetoplastid RNA editing, like pre-mRNA splicing in higher eukaryotes, generates a 5’ cleavage product which must be held in position during the 365
Reviews
r
UTP
Fig. 4. A critical view of chimeras. The non-encoded oligo-U tail of the gRNA and/or an unidentified factor (X encircled in blue) must hold the 5’ cleavage product during accurate editing (blue arrows). Attenuation of this interaction (green arrows) could result in loss of the 5’ cleavage product. Placement of the gRNA’s 3’ end into the site normally occupied by the 3’ end of the 5’ cleavage product would juxtapose reactive 3’ OH and 5’ monophosphate residues (red). Chimera formation (green arrows) rather than production of edited RNA (blue arrows) would result upon ligation of these two groups.
course of processing. During splicing, U residues in the conserved loop of the U5 small nuclear RNA and the ~220 protein perform this taskdo. During kRNA editing, the oligo-U tail of the gRNA could perform a similar function29. Conceivably, this interaction could also be aided by an unidentified factor (factor X in Fig. 4). Regardless of the nature of the activity that holds the 5’ cleavage product, the cleavage product could be lost from the editing machinery if this interaction became unstable (Fig. 4, green arrows). The 3’ hydroxyl of the gRNA could then be juxtaposed to the 5’ end of the 3’ cleavage product, as the 5’ cleavage product is normally. Aberrant ligation in this pathway would generate a chimera (green arrows) rather than edited product (blue arrows). Major issues concerning the mechanism of insertion and deletion clearly remain unresolved. Characterization of potential intermediates and enzymatic activities has proven unable to distinguish among models for these chemical steps, since intermediates and activities are shared in the different pathways. Final proof of whether chimeric molecules are intermediates in editing will require probing of the role of the 3’ OH of the gRNA during in vitro processing and determining whether chimeras produced in vitro can be ‘chased’ into edited product. 366
Molecular machinery . . In addition to progress gained wrth molecular genetic and biochemical analyses, advances have been made in the physical characterization of putative editing complexes, or editosomes. These studies have relied most heavily on the analysis of the sedimentation of putative editing-related activities in glycerol gradients 28,39. Although such work does not provide formal proof of the existence of macromolecular editing complexes, co-sedimentation does provide persuasive circumstantial evidence. Pollard et al. have described two putative editing complexes in T. brucei39: one that sediments at -20s contains endogenous gRNAs, terminal uridylyl transferase (TUTase), RNA ligase, chimera-forming activity and, although initially reported otherwise39, pre-edited RNA specific nuclease (B. Adler and S. Hajduk, pers. commun.). A second complex that sediments at -35-40s lacks TUTase, contains lesser amounts of the other activities, but contains pre-edited mRNAs. Using similar gradients, it has been shown that an in vitro gRNA-dependent U deletion activity sediments in the 20 S region (K. Stuart, pers. commun.). This complex may therefore represent the catalytically active editosome, provided subsequent assembly into the 35-40s complex does not occur. The 35-40 S region of the gradient itself does not display Parasrtology
Today, vol. I I, no. IO, I995
Reviews U deletion activity, but this complex may simply be unable to take up exogenous substrates in vitro. Alternatively, the 35-40s complex may be involved in U addition or in recycling kRNA editing components. In Leishmania, putative editing complexes have also been observed28. Provocatively, endogenous gRNAs and U addition activity also sediment at -205 in this system. It is not yet known whether U insertion requires the co-sedimenting gRNAs but, if so, this would suggest that insertion28,41 and deletion25 activities are related. Unlike T. brucei, the majority of the RNA ligase and TUTase activities in Leishmania are found uncomplexed and in 10s fractions, respectively28. Although this represents a significant difference from T. bruc@, it may reflect a difference in complex stability rather than a fundamental difference in the nature of the catalytic machinery. This notion is supported by the ability to alter the sedimentation behavior of editing-related activities in T. brucei by changing ionic conditions (K. Stuart, pers. commun.). The existence of RNA ligase, TUTase and preedited RNA-specific endonuclease activities%33, and their partial co-sedimentation with editing activity in T. brucei (K. Stuart, pers. commun.), seems to argue that the kRNA editing reaction is catalyzed by protein(s) in these complexes. This is further supported by the sensitivity of U deletion in vitro to proteinase treatment (S. Seiwert, unpublished). The sizes of putative editing complexes in T. brucei and L. tarentolue indicate that the editing machinery contains numerous other proteins 39. Additional protein factors could include RNA helicaselh, proofreading factors, factors required for scaffolding of the active site(s) and other undefined functions. RNA-protein crosslinking studies have identified several gRNA specific proteins which may perform some of these function@43. Efforts are now under way in a number of laboratories to purify and characterize these and other proteins associated with editing complexes.
Evolutionary considerations Although a diverse collection of eukaryotes posttranscriptionally alter information in their RNAs (for reviews see Refs 44,45), no organisms besides kinetoplastids are known to transfer sequence information from one RNA to another. However, since kinetoplastids represent one of the oldest eukaryotic lineages with mitochondriaM, the possibility remains that kRNA editing is an ancient trait that has been lost in other organisms. To estimate the antiquity of kRNA editing, its occurrence within various kinetoplastids has been examined. The order Kinetoplastida contains two suborders, the Trypanosomatina and the Bodonina, which are estimated to have diverged 680-900 million years ago4’. Within the Trypanosomatina, extensive editing of transcripts occurs frequently in Typanosornu and (probably) in Herpetomonusm. Several of the homologous transcripts in Leishrnuniu and Crithidiu are subject to editing only in small domains near their 5’ ends. Phylogenies based upon ribosomal RNA suggest that the divergence of the former lineages is more ancient, implying that extensive editing is an ancestral state within trypanosomatids@-50. The recent demonstration of extensive editing in the bodonid TrypunoPorasrtology Today, vol. I I, no. / 0. I995
plasma borreli (family Cryptobiidae)51,52 suggests extensive editing must be older than the split between trypanosomatids and bodonids and could be as old as the kinetoplastid lineage itself. As Maslov and Simpson point out, the restriction of editing to the 5’ portions of mRNAs in Leishmuniu and Crithidiu may have arisen by retroposition of partially edited transcripts in the ancestor of these organism#‘. The discovery of a reverse transcriptase gene in kinetoplastids lends support to this view53 and provides a plausible model for loss of kRNA editing in non-kinetoplastids. Direct support that kRNA editing is an ancestral trait of all mitochondria bearing eukaryotes would require its discovery in other organisms, the most likely candidates possibly being the kinetoplastid sister group of Euglenoids or in the probable bacterial cousins of mitochondria, the ol-ProteobacteriaM. Further investigation of the time of origin of kRNA editing will provide clues to its original function by placing it in a paleohistorical context. At present, one can only speculate about the selective pressures experienced by the ancestors of kinetoplastids that resulted in the development of such a baroque pathway of gene expression. If kRNA editing dates to the ‘RNA world’55, possible functions include an involvement in the replication of RNA56, or in an ‘excision’ repair process required for the maintenance of RNA genomes. Once established, kRNA editing could have been exploited to restrict the expression of certain mitochondrial genes to particular life cycle stages of parasites, a function used today by certain digenic trypanosomatids (for review see Ref. 57). An altemative view suggests that kRNA editing may help to enhance genetic variation by duplication and divergence of gRNA@s, sloppy editing59, or usage of misguiding gRNAs22,U.
Prospects In a relatively short time, kRNA editing has relinquished many of its secrets. The field has progressed from documentation of the phenomenon of kRNA editing to characterization of possible editing intermediates. Direct probing of in vitro systems now holds the promise of answering fundamental questions about the mechanism and the machinery of this unusual RNA processing reaction, but care must be taken to relate these in vitro results to the process of kRNA editing which occurs in viva. The continued search for kRNA-like editing in other organisms will indicate whether RNA directed RNA modification is a trait ancestral to all eukaryotes bearing mitochondria, or a peculiar feature of the kinetoplastid lineage. Acknowledgements I am indebted to Kenneth Stuart for financial support and invaluable discussions, and to Theo deVos, Stefan Heidmann, Robert Hughes, Kym Jacobson and Elisabetta Ullu for comments on the manuscript. References Stuart, K. and Feagin, J.E. (1992) Int. Rev. Cytol. 141,65-87 Benne, R. et ~2. (1986) Cell 46,819-826 Feagin, J.E. et ai. (1987) Cdl 49,337-345 Shaw. I.M. et al. (1988) Cell 53.401-411 van cidr Spek, H.‘et a2.‘(1988) &MB0 1. 7,2509-2514 Feagin, J.E. et al. (1988) Cell 53,4X%-422 Blum, B. et al. (1990) Cell 60,189-198 Sturm, N.R. and Simpson, L. (1990) Cell 61,879-884 367
Reviews 9 Sturm, N. and Simpson, L. (1991) Nucleic Acids Res. 19, 6277-6281 10 Bhat, G.J. et al. (1990) Cell 61, 885-894 11 Pollard, V.W. et al. (1990) Cell 63,783-790 12 van der Spek, H. et al. (1991) EMBO J. 10,1217-1224 13 Abraham, J.M. et al. (1988) Cell 55,267-272 14 Sturm, N.R. and Simpson, L. (1990) Cell 61,871-878 15 Maslov, D.A. and Simpson, L. (1992) Cell 70,459-467 16 Missel, A. and Goringer, U.H. (1994) Nucleic Acids Res. 22, 405m56 17 Souza, A.E. et ~2. (1992) Mol. Cell. Biol. 12,2100-2107 18 Decker, CJ. and Sollner-Webb, B. (1990) Cell 61,1001-1011 19 Riley, G.R. et aZ. (1994) 1. Biol. Chem. 269,61014108 20 Kosiowsky, D.J. et al. (i991) Cell 67,537-546 21 Maslov. D.A. et al. (19921 Mol. Cell. Biol. 12. 56-67 22 Sturm, N.R. et uZ.(i992) ?eZZ70,469-476 23 Maslov, D.A. et al. (1994) Mol. Biochem. Parusitol. 68, 155159 24 Corell, R.A. et al. (1993) Nucleic Acids Res. 21,4313-4320 25 Seiwert, S.D. and Stuart, K. (1994) Science 266,llP117 26 Cech, T.R. (1991) Cell 64,667-669 27 Blum, B. et al. (l&l) Cell 65,543-550 28 Peris. M. et al. 119941 EMBO 1.13.1664-1672 29 Blud, B. and Simpsbn, L. (l&O)CeZZ 62,391-397 30 White, T.C. and Borst, P. (1987) NucZeic Acids Res. 15, 32753289 31 Bakalara, N. et al. (1989) J. Biol. Chem. 264,18679-18686 32 Harris, M.E. and Hajduk, S.L. (1992) Cell 68,1091-1099 33 Harris, M. et al. (1992) Mol. Cell. BioZ. 12,2591-2598 34 Read, L.K. et al. (1992) Nucleic Acids Res. 20, 2341-2347 35 Arts, G.J. et UZ.(1993) EMBO 1. 12,1523-1532 36 Koslowsky, D.J. et al. (1992) Nature 356,807-809 37 Blum, B. and Simpson, L. (1992) Proc. NutZ Acud. Sci. USA 89 11944-11948
38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Riley, G.R. et al. (1995) Nucleic Acids Res. 23,708-712 Pollard, V.W. et al. (1992) EMBO 1. 11,4429-4438 Sontheimer, E.J. and Steitz, J.A. (1994) Science 262,1989-1996 Harris, M.E. et al. (1990) 1. BioZ. Chem. 265,11368-11376 Read, L.K. et al. (1994) MoZ. CeZZ.Bid. 14,2629-2639 Keller, J. et uZ.(1994) Ir]ucZeicAcids Res. 22, 1988-1995 Bass. B. (1993) in The RNA World (Gesteland, R.F. and Atkins, J.F., eds),‘pp 3318, Cold Spring Harbor Laboratory Press Adler, B.K. and Hajduk, S.L. (1994) Curr. @in. Genet. Dev. 4, 316-322 Sogin, M.L. (1994) in Early Life on Earth (Bengtson, S., ed.), pp 181-192, Columbia University Press Femandes, AI’. et UI. (1993) Proc. Nut2 Acud. Sci. USA 90, 11608-11612 Landweber, L.F. and Gilbert, W. (1994) Proc. NutZ Ad. Sci. USA 91,918-921 Landweber, L.F. and Gilbert, W. (1993) Nature 363,179-182 Maslov, D.A. et ~2. (1994) Nature 368,345-348 Lukes, J. et a2. (1994) EMBO ].13,5086-5098 Maslov, D. and Simpson, L. (1994) Mol. CeZZ. BioZ. 14, 8174-8182 Gabriel, A. and Boeke, J-D. (1991) Proc. NutZ Acud. Sci. USA 88, 9794-979s Gray, M.W. (1994) Nature 368,288 Gilbert, W. (1986) Nature 319,618 Benne, R. (1990) Trends Genet. 6,177-181 Priest, J.W. and Hajduk, S.L. (1994) 1. Bioenerg. Biomembrunes 26, 179-191 Landweber, L.F. et al. (1993) Proc. NufZ Acud. Sci. USA 90, 9242-9246 Read, L.K. et al. (1994) Nucleic Acids Res. 22, 1489-1495 Thiemann, O.H. et al. (1994) EMBO ].13,5689-5700
Polydnaviruses: Potent Mediators of Host Insect Immune Dysfunction M.D. Lavine and N.E. Beckage Endoparasitic insects are used as biological control agents to kill many species of insect pest. One key to the success of parasitoids that develop in the hemocoel of their host is their ability to knock out the host’s immune system, inducing a decline in the responsiveness of a variety of cellular and humoral components so that parasitoid eggs are not encapsulated. In many species parasitized by braconid and ichneumonid wasps, host immunosuppression appears to be mediated by polydnaviruses (PDVs) injected by the female parasitoid into the host hemocoel. The viruses exhibit a complex and intimate genetic relationship with the wasp, since viral sequences are integrated within the wasp’s chromosomal DNA. Here Mark Lavine and Nancy Beckage summarize the current evidence for mechanisms of virally induced host immunosuppression in parasitized insects, as well as the roles of other factors including wasp ovarian proteins and venom components, in suppressing hemocytemediated and humoral immune responses. Interestingly, in some species, the PDV-induced host immunosuppression appears transitory, with older parasitoid larvae probably exploiting other mechanisms to protect themselves from the Mark Lavlne IS at the Department of Biology, and Nancy Beckage IS at the Department of Entomology 5419 Boyce Hall, Univenlty of Califomla-Rivewde, Rivenlde, CA 9252 I-03 14, USA. Tel: + I 909 787 3521, Fax: + I 909 787 3087, e-mail: [email protected]
host’s immune system during thefinal stages of parasitism. During the final stages of parasitism, the parasitoids likely exploit other mechanisms of immunoaasion via antigen masking, antigen mimicry, or production of active inhibitors of the hemocyte-mediated encapsulation response as well as inhibiting melanization. Many members of the insect order Hymenoptera are endoparasitoids, which are parasites that develop within, and always kill, their insect hosts. For endoparasitoids to develop successfully, they must avoid or suppress their host’s natural defense mechanisms, and hence have evolved complex mechanisms for modulating the host insect’s cellular and humoral immune system.G. Parasitoids exhibit sophisticated mechanisms of interaction with their host insect’s immune, endocrine and metabolic pathways, and, frequently, even modify the development and behavior of their host; the outcome is the successful maturation of the parasitoid accompanied simultaneously by the host’s demise. Endoparasitic wasps in the families Braconidae and Ichneumonidae have developed unique associations with a family of polydisperse DNA viruses (polydnaviruses, or PDVs). These viruses replicate exclusively in the female wasp’s ovaries, and each species of wasp is presumed to have a unique PDV; the