Multiple Guide RNAs for Identical Editing of lbypanosoma brucei Apocytochrome b mRNA Have an Unusual Minicircle Location and &e Developmentally Regulated*

We identified four different guide RNAs (gRNAs) that specify identical editing of lZypanosoma brucei apocytochrome b (CYb) mFtNA, which indicates gRNA redun- dancy in T. brucei. All four gRNAs appear functional since they occur in chimeras, some of which contain an interesting gRNA 3’ “extension.” The gRNAs are encoded in different minicircles, rather than maxicircles as in other species. However, these gRNA genes are not be- tween 18-base pair repeats as are the other minicircle gRNA genes in T. brucei. The three minicircles cloned contain the same gRNA genes, one of which is substantially diverged, all in the same order, indicating that they are related. CYb gRNA is less abundant in procyclic than bloodstream forms. Procyclic forms contain abun- dant edited CYb mRNA unlike bloodstream forms thus suggesting that CYb mRNA editing may be regulated at the level of gRNA utilization. Most initiation Refs. Over 45% of the nucleotides in the mRNAs of ATPase subunit 6 (A6),’ NADH-dehydrogenase subunit 7 (ND7), and C-rich gene (CR3) are of editing (3). In contrast, limited editing 4 uridines to

Accumulation of edited mRNAs is developmentally regulated in lkypanosoma brucei, which is especially evident in the case of CYb. Edited CYb mRNA is abundant in procyclic forms, which have a functional cytochrome system. However, edited CYb mRNA is nearly undetectable in slender blood form cells (7) and is present in low abundance in stumpy blood form (8), which contain oxidoreduction activities absent in slender blood form but lack cytochromes (9,10). The abundance of edited COII mRNAparallels that of edited CYb mRNA (8). In contrast, ND7 mRNA that is edited in its 3' domain is primarily present in blood form cells (11).
* This work was supported by Public Health Service Grants 1 F32 GM14126-01A1 (to R. A. C.) and GM42188 (to K. S.) from the National Institutes of Health. 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. The nucleotide sequence(s) reported in this paper has been submitted L25.589, L25590.
Guide RNAs (gRNAs) are small RNAs that apparently specify the edited sequences of mitochondrial mRNAs. The gRNAs share common features: they are slightly less than 60 nucleotides in length, many have an RYAYA sequence at their 5' end, they complement edited mRNA by a combination of Watson-Crick and G-U base pairing, and they have 10-15 posttranscriptionally added uridines at their 3' end (12-15). Some gRNA genes are in the maxicircle in Leishmania and Crithidia (12, 16) but most have a minicircle location. Many gRNA genes have been identified in I: brucei, and all of these are encoded in minicircles (14, [17][18][19]. Interestingly, all the ' I : brucei minicircle gRNA genes identified to date are on the same DNA strand and are located in "cassettes" flanked by a pair of inverted 18-bp repeat sequences (14,17,20). The I: brucei maxicircle sequence potentially encodes maxicircle unidentified reading frame 2 (MURF2) and COII gRNA genes similar to those found in Leishmania and Crithidia (16). However, unlike Leishmania and Crithidia, the II brucei maxicircle sequence can not encode CYb gRNAs.
We report here the identification of a t least three CYb gRNA genes in Z! brucei. These gRNAs are encoded in minicircles, but they are not encoded between 18-bp repeats as are all other minicircle gRNA genes found in ' I : brucei. These gRNAs specify identical editing of the 3' portion of the CYb mRNA-editing domain, and their presence in gRNNmRNA chimeras suggests that all of them are functional, Two of the gRNAs in chimeras have an unusual 20 nt non-guiding 3' extension. Each of the three minicircles that encode the CYb gRNA genes also encode A6 and CR3 gRNAs in identical locations indicating that the minicircles are related. The gRNA genes diverge among the minicircles, so substantially in one case that its guiding capacity may be reduced. The CYb gRNAs are present in both life cycle stages but are lower in abundance in procyclic form where edited CYb mRNA accumulates. This may suggest preferential CYb gRNA utilization in this life cycle stage.

GCGCGTAATACGACTCACTATAGGGGTTGGTGTAATACAACA~;
AS-MC-CS-T3, GCGCGCAATTAACCCTCACTAAAGGGCC'MTCGAA-   1. CYb gRNA gene complementarity with edited CYb -A . The three gRNA genes are aligned below a portion of the edited CYb mRNA sequence with Watson-Crick base pairs indicated by vertical lines and G-U base pairs by colons. Sequence identity between gCYb(560A and B) is indicated by dots. Uridines added to CYb mRNA by editing are shown in lowercase. The gRNA 5' ends were determined by RNA sequencing and gRNA 3' ends observed in chimeric gRNA/mRNA cDNA clones are marked by arrowheads. The single nucleotide difference between the gCYb(560A and C) sequences, determined from a single chimera clone shown in Fig. 4, is indicated by the G under gCYb(560B).

TAAA.
RNA sequencing (17) was done using 180 and 4 p! dideoxy-and deoxynucleotide triphosphate terminating nucleotides, respectively. gRNA/mRNA homologies were identified with DNAStar, allowing for G-U base pairing. Minicircles were compared with the MULTALIN program (28). DNA sequence was determined from both strands.
Selective PCR Amplification of Minicircle gRNA Genes-The presence of G-U base pairing in the complementarity between a gRNA and its corresponding mRNA prevent conventional screening for gRNA genes. We therefore devised a method of selective PCR amplification of gRNA genes from minicircles. It uses primers specific for the "conserved region" present in all I: brucei minicircles (indicated in Fig. 6 A ) along with gRNA-specific primers predicted for the anchor duplex between gRNA and mRNA (15 nt of CYb mRNA sequence downstream of the most 3"inserted U). The gRNA sequence in the anchor duplex can be predicted since it is almost invariably composed of Watson-Crick base pairs. Conserved region primers for both orientations of the minicircle were used to provide a negative control. Touchdown PCR (29), starting above the calculated T, of the primer, was used since the length of the anchor duplex cannot be predicted.
Partial minicircles containing CYb guide RNA genes were amplified from 5 ng of decatenated MNA using primers gCYb-ANCl and AS-MC-CS-T3: denaturation was for 1 min at 94 "C, annealing for 30 s, and extension for 2.5 min at 72 "C. Twenty cycles of amplification were performed during which the annealing temperature was decreased from 40 to 30 "C by 1 "C every second cycle. The annealing temperature was then raised to 51 "C for an additional 15 cycles. PCR-amplified DNA was cloned into the EcoRV site of pBluescript I1 SK(-) (Stratagene) (30) and transformed into SURE cells (Stratagene).
A library of full-length minicircles was produced by inverse PCR amplification (31) of 5 ng of decatenated MNAusing AS-MC-CS-T3 and SA-MC-CS-T7 primers. Denaturation was for 1 min at 94 "C, annealing for 30 s, and extension for 2.5 min at 72 "C. Two cycles were performed at an annealing temperature of 40 "C, followed by 25 cycles at 48 "C. A library representing 4 x lo4 clones was screened with oligonucleotides 3'-CYb(558) and 3'-CYb(560A).
Chimeric and Partially Edited CYb RNAs-CYb gRNA/mRNA chimeras were amplified from cDNA synthesized from 2.5 pg of procyclic form mitochondrial RNA or 5 pg of blood form total RNA using CYb-CS4 (complementary to CYb mRNA 173 nt 3' to the editing domain) as the primer and Superscript (Life Technologies, Inc.). One-fifth of the cDNA was amplified by PCR using CYb-CS3 (complementary to CYb mRNA 65 nt 3' to the editing domain) and 5'-CYb(558) or 5'-CYb(560). Denaturation was for 1 min at 94 "C, annealing for 30 s, and extension for 30 s at 72 "C. Two cycles were performed at an annealing temperature of 41 "C, followed by 25 cycles at 53 "C. cDNA clones from partially edited CYb mRNA were prepared from mitochondrial RNA as described above, using CYb-CS2 as a primer. This primer is complementary to CYb mRNA 14 nt 3' to the editing domain. One-fifth of the cDNA was amplified by PCR using CYb-CS2 and 5'-CYbN as primers. The latter primer matches the 5' end of CYb &A.
Denaturation was for 1 min at 92 "C, annealing for 1 min, and extension for 1 min at 72 "C. Two cycles were performed at an annealing temperature of 27 "C, followed by 25 cycles at 37 "C. The chimeras and partially edited CYb RNA PCR products were cloned into the EcoRIIBamHI and KpnIIBamHI sites of Bluescript I1 SK(-) (Stratagene), respectively, and transformed into DH5-a (Life Technologies, Inc.) and SURE cells (Stratagene).

Identification of CYb gRNA Genes-Analysis of the T brucei
IsTaRl maxicircle, which we have completely sequenced (34,35), shows that it cannot encode CYb gRNA genes, unlike the maxicircles of Leishmania tarentolae and Crithidia fasciculata (12, 16). We therefore devised a strategy to selectively amplify CYb gRNA genes from minicircles by PCR (see "Experimental Procedures"). A 0.6-kb PCR product was obtained using a conserved minicircle sequence primer AS-MC-CS-T3 that allows amplification of a gRNA gene with the same orientation as all minicircle gRNA genes previously identified in T brucei. No product was obtained with the conserved minicircle sequence primer SA-MC-CS-T7 for the opposite orientation (data not shown). Direct sequencing of the PCR product confirmed the presence of CYb gRNA gene sequence but revealed sequence degeneracy at several positions (data not shown), suggesting multiple gRNA genes. Five partial minicircles cloned from the PCR product were sequenced and found to encode three different gRNAs. Five full-length minicircles isolated by screening a minicircle library with oligonucleotides 3'-gCYb(558) and 3'-gCYb(560A) specific for the gRNA genes were sequenced (see below) and contain gRNA genes 558 and 560A; the full-length 560B minicircle was not isolated although oligonucleotide 3'-gCYb(560A) can hybridize with both 560A and B minicircles. A fourth possible gRNA was detected in a chimera (see below), but its gene was not found in the cloned minicircles.
All three minicircle-encoded gRNA genes are perfectly complementary to edited CYb mRNA over the 3' region of its editing domain (Fig. 1). The gCYb(558) gene is complementary to 42 nt of edited CYb mRNA and predicts gRNA with a 10-bp Watson-Crick anchor duplex with unedited CYb mRNA and the ability to specify 32 nt of edited sequence. gCYb(560A) and gCYb(560B) genes predict gRNAs that differ from each other by a single nucleotide and which are both complementary to 39 nt of edited CYb mRNA with an 8-bp anchor duplex and 31 nt of edited sequence. The gCYb(558) gene has 75% identity to gCYb(560A) and gCYb(560B) over the region of gRNA homology to mRNA and the gCYb(560A and B) minicircles have five differences over 483 base pairs, including the mismatch in the CYb gRNAs.
CYb Guide RNAs-The minicircle-encoded CYb g R N h were detected in cellular RNA by Northern blot analysis using DNA oligonucleotide probes 3'-gCYb(558), 3'-gCYb(560A) and 3'-gCYb(560B). All three probes hybridized to an approximately 60-nt RNA in both bloodstream form and procyclic form total RNA but not to RNA from the dyskinetoplastic mutant that lacks kDNA (Fig. 2, panels 13). The small size heterogeneity that is characteristic of gRNA is evident and presumably reflects the 3' U tail heterogeneity (see below). The 5' ends of the gRNAs were determined by sequencing RNA (Fig. 3) and match those predicted from the gRNA genes. The closely related gCYb(560A) and gCYb(560B) could not be distinguished since the sequencing primers covered the single mismatch region. However, both gRNAs are present in chimeras (see below), indicating that both gRNA sequences are present in cellular RNA.
Developmental Regulation-The Northern blot in Fig. 2 shows that edited CYb mRNA is primarily present in procyclic form compared to blood form RNA (panel 5). Unedited CYb mRNA is also more abundant in the procyclic form RNA (panel 4 ). All three oligonucleotide probes not only show that gRNA is present in the same RNA preparations from both life cycle stages but that the gRNA is more abundant in blood form than in procyclic form RNA. Densitometric analysis indicates that blood form RNA contains about two to four times more CYb gRNAs than procyclic form RNA (Table I). In contrast, there is about five times more edited mRNA in procyclic form than blood form RNA. The procyclic form RNA also contains about twice as much unedited mRNA than blood form. Thus, there is a greater relative proportion of the gRNAs to edited and unedited mRNAs in blood form compared with procyclic form RNA.
CYb gRNAImRNA Chimeras-CYb gRNNmRNA chimeras, thought to be intermediates in the editing process, were cloned and sequenced to assess the participation of the different gRNAs in RNA editing. PCR products were produced from procyclic form and blood form RNA with primers specific for gCYb(558) or gCYb(560A, B, and C) and a primer specific for CYb mRNA. CYb gRNNmRNA chimeras for four gRNAs were obtained (Fig. 4). Of 19 clones sequenced, eight contained gCYb(558), eight contained gCYb(560A), two contained gCYb(560B1, and one contained gCYb(560C), which differs by a single base from gCYb(560A) and by two bases from  gCYb(560B). The gRNAs in the chimeras are linked by their non-encoded U-tail to various sites that are normally edited in CYb mRNA. The size of the gRNA portions and the length of their U-tails varies among the different chimeras. These are characteristics that have been reported previously for chimeras (15,36). In general, the gCYb(558) chimeras have more Us linking the gRNA and mRNA than do the gCYb(560A) chimeras. In addition, the gRNA in six of the eight gCYb(560A) chimeras are linked to the first editing site (562), while in six of the eight gCYb(558) chimeras, the gRNA is linked to a further 5' editing site. The mRNA sequence just downstream of the site of gRNA linkage is incompletely edited in clone 34 (AGtttG rather than AtttGtG), similar to incompletely edited sites in previous reports (15,36). Surprisingly, two gCYb(558) chimeras, that are independent based on their different oligo(U) sequences, have a non-encoded G residue at the site of gRNA linkage. Furthermore, the encoded 3' portion of the gRNA in two independent chimeras, 27 and 28, extends 20 nt beyond the end of gRNNmRNA complementarity. These two chimeras predict 70-and 79-nt gRNAs, including U-tails, which is larger than the bulk of the CYb gRNAs detected in the Northern blots, suggesting that they are present at low levels in the gRNA population in steady-state RNA. Partially Edited CYb cDNA Clones-We sequenced cDNA clones corresponding to partially edited CYb mRNA to examine molecules in the process of editing (Fig. 5). They were prepared using primers for sequences which flank the editing domain to avoid selection of any subset of molecules. They are compared gCYbC5581 Chimeras e 7 AACACAATGTGAATTTTTAGGTGATAAAGGGAAT~~~T GTCTTTTAATGTCAGGTTGT A T~~~~~~~~~~~, , , , , , , , , , T~~~~~~~. , , , , , , . , , , , . GTCTTTTAATGTCAGGTTGT   with partially edited CYb cDNAs from Z! brucei TREU 667 which were prepared using a 5' primer for unedited sequence within the editing domain (37). The two sets of cDNAs have similar characteristics, and clones c162 and c95 are identical to 1-22 and 2-16, respectively. However, the TREU 667 cDNAs appear to be biased toward molecules that are incompletely edited in the 3' region of the CYb-editing domain. All 43 partially edited clones are edited in a 3' to 5' direction; no cases of clones edited in the 5' but not the 3' region of the CYb-editing domain were observed. Two cDNAs (c122 and c160) go directly from edited to unedited sequence. However, most cDNAs have an incompletely edited "junction sequence" between the edited and unedited sequence (38); sites within the junction that require further editing to match the mature mRNA are interspersed among those that do not. Twelve IsTaRl and one TREU 667 cDNAs are fully edited through the region specified by the identified CYb gRNAs. The 3' limit of the junction of six of these clones is immediately upstream of the region specified by the CYb gRNAs. This suggests that the editing by the first gRNA is completed and editing by a subsequent gRNA has begun. Another seven cDNAs (a3-cl21) may have been at a similar stage since their 3' junction limits are one or two editing sites within the region specified by the CYb gRNAs and four of these (c181 and a8) differ from completely edited RNA by only a single T. gRNAs for the 5' region may form an anchor duplex within the region specified by the identified gRNA andor there may be multiple gRNAs for the 5' portion of the domain. Seventeen of the remaining cDNAs are incompletely edited entirely within the region specified by the identified CYb gRNAs, suggesting that they were being edited by these or related gRNAs when iso-lated. Six other cDNAs have junctions that span both the region specified by the identified CYb gRNAs and the region specified by subsequent guide(s).

21C G T A A M G A C M T G T A G A T T T C T G A G T A A T A G G G A G G A T M C~~~~~~~~~~~A t t t t t t t t A t t A t t t A G A A A t t t G t G t t G T C T T T T M T G T C A G G T T G T 28C G T A A M G A C M T G T A G A T T T C T G A G T A A T A G G G A G G A T A C T~~~~~~~~~~A t t t t t t t t A t t A t t t A G A M t t t G t G t t G T C T T T T A A T G T C A G G T T G T
Relationships Among CYb Minicircles-Multiple clones of the same minicircle show low levels of base substitutions. The full-length minicircle clone MCP-5 that encodes gCYb(558) (Fig. a) has a sequence that is identical to that of partial minicircle clones TA-3 and TA-25 but has a total of six bp differences from full-length minicircles MCP-2 and -7 (three each, not shown). Similarly, full-length minicircle MCP-23 that encodes gCYb(560A) is identical to partial minicircle clones TA-7 and -35 but has a single bp difference from full-length minicircle MCP16 and 3 bp differences from partial minicircle clone TA-27 (not shown). This frequency of base substitution is low but higher than expected from PCR and cloning (39), suggesting sequence microheterogeneity among minicircles that encode the same gRNAs. Only a single partial minicircle clone (TA-4) was obtained for gCYb (560B) suggesting that minicircles encoding this gRNA are low in abundance in kDNA. No minicircle clones were obtained for gCYb(560C), which suggests that its minicircle may be in low abundance or that the single nucleotide difference is a PCR artifact.
Conservation of sequence and gene structure indicates that the minicircles encoding three of the different CYb gRNAs are related. Those encoding gCYb(560A) and gCYb(560B) are closely related since they differ by only five base pairs in the region cloned for both. The differences include three transitions, a base insertion, and a base deletion, indicated by s pbols in the last four lines of MCP23 sequence (Fig. a). One transition is in the CYb gRNAgene (*, in the MCP23 sequence) and a base deletion is in the CR3 gRNA gene (&). The

! ! % GTMCAGAAAAT TAGAGGG 5 '
-::IIIIIIIIII II II  minicircles encoding gCYb(558) and gCYb(560A) are more divergent, with 70% nucleotide sequence identity (Fig. 6A). The gRNA gene coding sequences are more conserved between these minicircles than are the other sequences except for the approximately 130-bp minicircle conserved region and part of the "adjacent region" (last line of Fig. 6A), which lies between the conserved region and cassette 3 (40). The conserved region plus the 5' adjoining 45 bp of the adjacent region have 90% identity while the remainder of the adjacent region has 50% identity. The 18-bp repeats of the gRNA coding cassettes are consellred and the -100-bp sequence sequences within the cassettes have 66-72% identity while sequences between the cassettes have 4658% identity. The gRNA genes are also more conserved than the flanking region within their cassettes that is not complementary to the mRNA.

t G t t G T C T T T T
The gRNA genes in these minicircles are located in the same positions, as diagrammed in Fig. 6B, but the CYb gRNAs are not encoded in cassettes. The coding sequences for gCYb(558) and gCYb(560A) are not flanked by 18-bp inverted repeat sequences, unlike all other gRNA coding sequences found thus far in l? brucei. Interestingly, the 3'-gRNA sequence that is present in gRNA/mRNA chimeras 27 and 28 (Fig. 4) is encoded by sequence that extends into the 18-bp repeat of the adjacent gRNA gene (Fig. 6, A and B). The different CYb minicircles encode A6 gRNAs in the second cassettes and CR3 gRNAs in the third cassettes (Fig. 6, A and C ) . The first cassettes may encode gRNAs since the sequences at the center of the cassettes are conserved between the minicircles, but the gRNAs have not been identified. Interestingly, the A6(138B) gRNA gene is substantially diverged from the A6(138A) gene. It encodes a gRNA that can form a 7 rather than 8 bp anchor duplex and it has five mismatches with A6 mRNA over the 31-nt region (exclusive of the anchor) with which gA6(138A) can form a perfect duplex (Fig. 6C).
Thus, while gA6(138B) is clearly related to gA6(138A), its divergence would prevent its recognition by homology search (38) and probably affects its function as a gRNA. The gCR3(73A, B, and C) genes are very similar although gCR3(73B and C ) have a nucleotide substitution 3' to the anchor duplex region that causes a mismatch with the mRNA, and gCR3(73B) has a one-and gCR3(73C) has a two-nucleotide deletion at the (gRNA) 5' end of the anchor duplex. Since A6 and CR3 gRNAs do not specify the initial editing of these mRNAs, the mRNA sequence that forms the anchor duplex with A6 and CR3 gRNAs is created by previous editing events.
As a result of G-U base pairing and few cytidines in mRNA and gRNA sequence, most transitions in gRNA genes do not affect the sequence specified by gRNAs. The gCR3(73B and C) gRNAs have eight transitions and two or three transversions relative to gCR3(73A), resulting in an internal mismatch and a smaller anchor duplex. The gA6(138B) gRNA has seven transitions and six transversions relative to gA6( 138A). CYb gRNAs gCYb(558A and B) have 11 transitions and two transversions relative to gCYb(5581, reducing the length of guide homology by one nt at the 3' end and the length of the anchor duplex by two nt. Both transversions in CYb gRNA and the one in CR3 gRNA do not affect the sequence they specify but five of the six transversions in A6 gRNAs create mismatches with mRNA. DISCUSSION We have identified three different gRNAs that specify identical editing of the 3 .. A.........G.TT.CAC........................A...GG.......................................'...........    The minicircles encoding the three CYb gRNAs appear to have diverged from a common molecule based on substantial conservation of sequence homology and gRNA gene order. Additional minicircle sequence diversity is also suggested by the sequence microheterogeneity among minicircle clones. Except for the minicircle conserved region, the gRNA coding sequences are more conserved than intergenic sequences (7241% versus 4658%). The 14 nt differences over 41 positions between the CYb gRNA genes do not alter gRNA complementarity to edited CYb mRNA sequence, as a consequence of G-U base pairing. This is also true for CR3 gRNAs, except for a single mismatch. However, the divergence of 12 nts over 41 positions in an A6 gRNA results in five mismatches with the edited mRNA, probably preventing it from directing editing to the mature mRNA sequence, although it could direct editing of a few sites. This A6 gRNA could not be identified by computer analysis (38) that we routinely use for this purpose. It was identified by its minicircle location and homology to a n identified gRNA. Divergence of this type may account for some of the 33% of gRNA genes we have been unable to identify in minicircle sequences.' Thus, selective pressure appears to conserve gRNA genes that specify mature mRNA sequence, as reflected in the bias toward transitions, which are less likely to alter the editing, rather than transversions (85% of mutations in gCYb and gCR3 are transitions, versus 55% for gA6 and about 25% for intergenic regions). The redundancy resulting from several different gRNAs for the same or overlapping sequences and the potential tolerance of gRNA/mRNA mismatch (16, 17) suggests that some diverged gRNAs may not edit directly to the final mRNA sequence in I: brucei; the mRNA could be re-edited to the mature sequence with a n overlapping gRNA. The minicircle sequence diversity and gRNA redundancy in I: brucei may reflect the presence of three or four gRNA genes per minicircle in I: brucei versus a single gRNA gene in minicircles of L. tarentolae, Pairs of 18-bp inverted repeats might serve as foci of recombination, further increasing the potential sequence diversity of the minicircles.

MCP5 T A A T A G A T~A A G C A T A G M T A A A --T T T A A A A T T M T A T T A T A T
CYb chimeras have the general characteristics of other chimeras including variation in gRNA length, U-tail length, and site of linkage to mRNA( 15,36). The differences in gRNAU-tail length and sites of mRNA linkage between chimeras of gCYb(558) and (560A) suggest that while both gRNAs are functional, they may not function equivalently. This may reflect specific differences in the gRNA interactions with CYb mRNA, perhaps driven by thermodynamic stability as previously proposed (38). The occurrence of two independent chimeras with a n extended 20-nt 3'-gRNA sequence that is not complementary to mRNA indicates that the gRNA coding sequence is unusual not only in its location but also in its extension into an 18-bp inverted repeat. Although Northern blots reveal that only a small fraction of CYb gRNAs are the larger size predicted for these gRNAs, their presence in chimeras suggests that they function in editing despite this extended sequence. The larger gRNAs may result from alternative transcription termination or gRNA processing. Occurrence of a non-encoded G at a n identical position in two independent chimeras seems unlikely to be an artifact of PCR and cloning or even addition in vivo by terminal uridylyltransferase. Non-encoded G residues have been previously found in A6 chimeras (15). It is possible that they might represent a step in editing considering that GTP is involved in group I intron splicing (41).
Edited CYb mRNA is abundant in procyclic forms but not (slender) blood forms (7). However, free CYb gRNAs are present in lower abundance in procyclic form than in blood form, and the proportion of CYb gRNA to unedited CYb mRNA is lower still. This indicates that the presence of edited CYb mRNA is not regulated by the presence or absence of CYb gRNAs during development. These levels suggest the possibility that accumulation of edited mRNA may be regulated at the level of gRNA utilization, perhaps at the level of the editing complex. n Y o ND8 gRNAs are also less abundant when the mRNA is preferentially edited, similar to the CYb gRNAs (19). The lack of apparent gRNA abundance differences of 3'-ND7 gRNAs previously reported (17) may be because these gRNAs do not edit the most 3', and hence initial, editing sites as do the CYb gRNAs. In addition, the edited ND7 mRNAs are preferentially present in blood forms which can contain partially differentiated stumpy and intermediate stage cells that may exhibit some characteristics of procyclic forms (81, and there is a smaller difference in abundance of these edited mRNAs between life cycle stages compared with CYb mRNA. The lower abundance of CYb gRNA in procyclic form could also indicate that gRNA is consumed during the editing process. The length of gRNAs seen in chimeras is variable (independent of the U-tail length), with truncations of up to half of the guiding region of the gRNA (15,36). This variation of gRNA size in CYb and other gRNNmRNA chimeras may also be a manifestation of gRNAconsumption. The finding of 20 nt of 3'-gRNA sequence that does not complement edited mRNA is unique to CYb mRNA to date.
While most partially edited CYb cDNAs were probably being edited in the 3' region of the editing domain under the direction of the gRNAs described here or in the 5' region with a second set of gRNAs when isolated, a few cDNAs are incompletely edited over a sequence spanning these two regions. This may reflect editing by a gRNA outside its homology to mature mRNA, as previously proposed (38). Alternatively, there may be a set of gRNAs that specify the editing of this overlapping region, since several gRNAs for substantially overlapping regions have been found recently (17,42). Another possibility is that these cDNAs reflect inappropriate editing by gRNAs that would specify complete editing of another site (43).