KREX2 Is Not Essential for Either Procyclic or Bloodstream Form Trypanosoma brucei

Background Most mitochondrial mRNAs in Trypanosoma brucei require RNA editing for maturation and translation. The edited RNAs primarily encode proteins of the oxidative phosphorylation system. These parasites undergo extensive changes in energy metabolism between the insect and bloodstream stages which are mirrored by alterations in RNA editing. Two U-specific exonucleases, KREX1 and KREX2, are both present in protein complexes (editosomes) that catalyze RNA editing but the relative roles of each protein are not known. Methodology/Principal Findings The requirement for KREX2 for RNA editing in vivo was assessed in both procyclic (insect) and bloodstream form parasites by methods that use homologous recombination for gene elimination. These studies resulted in null mutant cells in which both alleles were eliminated. The viability of these cells demonstrates that KREX2 is not essential in either life cycle stage, despite certain defects in RNA editing in vivo. Furthermore, editosomes isolated from KREX2 null cells require KREX1 for in vitro U-specific exonuclease activity. Conclusions KREX2 is a U-specific exonuclease that is dispensable for RNA editing in vivo in T. brucei BFs and PFs. This result suggests that the U deletion activity, which is required for RNA editing, is primarily mediated in vivo by KREX1 which is normally found associated with only one type of editosome. The retention of the KREX2 gene implies a non-essential role or a role that is essential in other life cycle stages or conditions.


Introduction
The mitochondrial genome of Trypanosoma brucei encodes 12 genes whose mRNAs undergo post-transcriptional editing that dramatically changes their protein coding sequences [1][2][3]. Using information provided by guide RNA (gRNA) templates, uridine (U) nucleotides are either inserted or deleted at specific editing sites within these RNAs. The extent of editing varies between RNAs, with some RNAs undergoing insertion and deletion of hundreds and tens of Us, respectively. Multiple editing sites are specified by a single gRNA and multiple gRNAs are used in the editing of most mRNAs. This RNA editing is catalyzed by protein complexes called editosomes that contain endoribonuclease, 39 Terminal Uridylyl-Transferase (TUTase), 39 U-specific exoribonuclease (exoUase), and RNA ligase activities.
Two components of the ,20S editosome have been shown to have U-specific exoribonuclease activity: KREX1 and KREX2 [9][10][11]. A third editosome protein, KREPA3, was also reported to possess U-specific exoribonuclease activity [12][13][14] but it contains no recognizable catalytic motif and deletion editing activity persists after KREPA3 knockdown [15,16]. Thus, whether KREPA3 performs such a role in vivo is unresolved. RNAi-mediated knockdown of KREX2 produced no defect in either growth or editing, but prevented normal association of KREL1 and KREPA2 with the ,20S editosomes. In contrast, knockdown of KREX1 resulted in defects in both growth and editing, and prevented normal association of KREN1 with the ,20S editosomes. Simultaneous RNAi knockdown of both KREX1 and KREX2 produced greater defects in both growth and editing than observed by knockdown of KREX1 alone, suggesting that KREX2 can play a role in RNA editing in vivo. In T. brucei, both KREX1 and KREX2 have a C-terminal endonuclease/exonuclease/phosphatase (EEP) domain (Pfam 03372) with characteristic conserved catalytic amino acids that indicate that these exonucleases likely have the same catalytic mechanism [17,18]. Notably, the Leishmania tarentolae KREX2 lacks the EEP domain and does not have exonuclease activity [10]. Thus, the nature of KREX2 function in RNA editing remains unclear.
RNAi is a useful albeit unpredictable tool in T. brucei. We have shown that RNAi targeting another editosome protein (KREN3) produces an insufficient knockdown to reveal its essentiality; something subsequently demonstrated using the conditional double knockout approach [4,19]. To unambiguously determine whether KREX2 is an essential gene, we eliminated both endogenous KREX2 gene coding sequences by homologous recombination. We show here using these null mutants that the KREX2 gene is dispensable in both bloodstream and procyclic form T. brucei. Procyclic KREX2 null cells exhibit defects in growth, RNA editing, and editosome sedimentation on glycerol gradients. Bloodstream form KREX2 null cells exhibit normal growth, but have defects in RNA editing in vivo as well as editosome sedimentation on glycerol gradients. Curiously, only a subset of the phenotypes observed in KREX2 null cells are rescued by reintroduction of an ectopic KREX2 allele. Purification of editosomes from KREX2 null cells using TAP-tag fused to either KREN1 or KREN2 reveals that only editosomes with KREX1 retain exoUase activity in vitro. These results demonstrate that KREX1 can be sufficient for editing and perhaps functionally compensate for the loss of KREX2. These results also suggest that most editing exoUase activity in vivo is catalyzed by KREX1 and KREX2 has a limited function.

Creation of KREX2 null cells
To create cell lines without KREX2, the endogenous KREX2 alleles were eliminated by homologous recombination in both bloodstream (BF) and procyclic form (PF) cells. PCR analyses detect the KREX2 open reading frame in genomic DNA isolated from parental cells, BF 427 wild-type (wt) or PF 29.13, but not in derived BF-KREX2-null and PF-KREX2-null cell lines ( Figure 1). In complementary PCR analyses, products corresponding to the junction of the transgenic knockout constructs in the KREX2 locus are detected in BF-KREX2-null and PF-KREX2-null but not parental cells. The elimination of KREX2 is also demonstrated by Southern analysis of BF-KREX2-null cells and Western analysis of PF-KREX2-null cells ( Figure S1).

Growth of KREX2 null cells
Both BF and PF cells grow in the absence of KREX2 expression ( Figure 2). Growth of BF-KREX2-null cells is indistinguishable from parental 427wt cells in both in vitro culture and in mice. Growth of PF-KREX2-null cells is slightly slower than parental 29.13 cells in SDM-79 media; however, there is no difference in growth between these cell lines in the absence of glucose. Growth after addition of glucose to glucose-free media demonstrates that the presence of glucose is responsible for the slower growth phenotype of PF-KREX2-null cells. Curiously, growth of the PF-KREX2-rDKO cell line (PF-KREX2-null cells transformed by addition of tetracycline-regulated expression of an ectopic KREX2 allele) is unaltered by tetracycline-induced expression of an ectopic KREX2 allele, despite evidence that the KREX2 protein is produced (data not shown, Figure S1).

Editing in vivo
The effect of the loss of KREX2 on editing in vivo was assessed by quantitative real-time PCR (Figure 3). Comparisons of mRNAs isolated from BF-KREX2-null and parental BF 427wt cells revealed that both KREX2 mRNA and edited CYb mRNA were essentially absent (i.e. not detected) while pre-edited CYb mRNA accumulated in null cells relative to wild type ( Figure 3A). The abundance of edited COIII and CR3 transcripts was significantly reduced in null cells relative to wild type, while edited ND3 was reduced to a lesser extent. Other never-edited, edited and pre-edited transcripts were similar in abundance in BF-KREX2-null and wild type cells. Comparisons of mRNAs from PF-KREX2-null and parental 29.13 cells revealed a profile that was distinct from BF-KREX2-null cells ( Figure 3B). Again, KREX2 mRNA was not detected in PF-KREX2-null cells. However, the relative abundance of edited CYb and COIII mRNAs was essentially the same in null and wild type cells, while edited CR3, edited ND3, and both edited and pre-edited RPS12 had substantially reduced relative abundance in the null cells, and edited ND8 was increased. As with BF-KREX2-null cells, the levels of other never-edited, edited and pre-edited transcripts are essentially the same in PF-KREX2-null and wild type cells. KREX1 mRNA abundance in both BF and PF KREX2 null cells is essentially the same as in wild type cells. Expression of ectopic KREX2 in the BF-KREX2-rDKO cell line resulted in edited COIII mRNA levels at normal BF 427wt levels, but the relative amount of CYb edited mRNA remained unaltered from that in null cells; KREX2 expression in PF-KREX2-rDKO cells also did not return the amount of edited ND3 mRNA to wild type levels (data not shown).  Editosomes from KREX2 null cells are shifted to lower S values on glycerol gradients compared to parental cells ( Figure 4). Western analysis shows that ,20S editosomes isolated from BF 427wt peak in fractions 9-11, and that editosomes isolated from BF-KREX2-null cells are shifted towards fraction 9. The most obvious evidence of this shift is the increase in KREL1 at the top of the gradient, consistent with its disrupted association with the deletion subcomplex. KREL1 is much more prominent in fraction 3 in samples from KREX2-null cells, as observed by both Western blot and adenylation assay. A more significant shift is observed with editosomes isolated from PF-KREX2-null cells. The ,20S editosome peaks in fraction 11 in control 29.13 cells, and shifts to fraction 9 in PF cells lacking KREX2. As with BF-KREX2-null cells, PF-KREX2-null cells have a notable increase in KREL1 at the top of the gradient.

In vitro editing activities in KREX2 null cells
In vitro editing activities are observed in gradient fractions isolated from both KREX2 null and parental cell lines ( Figures 5  and 6). KREX2 null cells maintain the ability to cleave both insertion and deletion editing site substrates in vitro ( Figure 5). While cleavage activity is restricted to the ,20S peak in fractions from BF cells, activity extends from ,20S to higher S values in PF, presumably due to the larger amount of editosome isolated. For fractions from both BF and PF KREX2 null cells, the amount of symbols) and derived KREX2 null cells (open symbols) in vitro. In normal SDM-79 media (squares), KREX2 null cells grow more slowly than parental 29.13 cells. In media lacking glucose (triangles) KREX2 null cells grow indistinguishably from parental 29.13 cells. Addition of 6 mM glucose to glucose-free media re-establishes growth defect of KREX2 null cells (circles). doi:10.1371/journal.pone.0033405.g002 Figure 3. Real-time PCR analysis of KREX2 null cells. RNA abundance for nuclear mRNAs KREX1 and KREX2 (black bars), never-edited mRNAs COI and ND4 (light grey bars), pre-edited mRNAs (white bars), and edited mRNAs (dark grey bars) is calculated relative to parental cell line for both BF (427wt) and PF (29.13). Analysis was performed in triplicate. For each target amplicon, the relative change in RNA abundance was determined by using either 18S rRNA (left bar in each pair) or b-tubulin (right bar in each pair) as an internal control. Asterisks denote mRNAs that were not detected in KREX2 null cells. A. A significant loss of CYb editing is shown in BF KREX2 null cells. B. PF KREX2 null cells have predominant decreases in CR3 and ND3 edited mRNAs. doi:10.1371/journal.pone.0033405.g003 cleavage activity appears slightly reduced compared to control fractions. The observed cleavage activities are present in the same fractions for control parental and derived KREX2 null cells. KREX2 null cells also maintain U addition, U removal, and ligase activities as measured by in vitro pre-cleaved editing assays ( Figure 6). In contrast to cleavage activities, pre-cleaved editing peaks in different fractions in KREX2 null cells, mirroring the observed shift in editosome proteins towards lower S values. For both BF and PF control fractions pre-cleaved insertion editing peaks at fraction 13, while KREX2 null editing peaks in fraction 11 ( Figures 6A & B). The shift in pre-cleaved deletion editing for BF-KREX2-null cells is subtle, most notable by the decrease in edited product and U-deletion intermediates in fraction 15 compared to parental control ( Figure 6C). For PF-KREX2-null cells, the peak of pre-cleaved deletion is shifted to fractions 9-11 from fraction 13 in control cells, and the overall amount of editing is decreased in the absence of KREX2 ( Figure 6D).

KREX1 removes Us in the absence of KREX2
In vitro U deletion activity is restricted to editosomes that contain KREX1 in KREX2 null cells ( Figure 7). While KREX2 and KREPA3 are present in all types of editosomes, KREX1 is restricted to KREN1 editosomes. We therefore isolated KREN1 and KREN2 editosomes from KREX2 null cells to determine  whether U removal is performed solely by KREX1 in the absence of KREX2. As expected, KREX1 is found in KREN1 but not KREN2 editosomes ( Figure 7A). The cleavage activity of these editosomes confirms the expected specificity and functional capacity of these samples ( Figure 7B). To examine U-specific exonuclease activity of the isolated KREN1 and KREN2 editosomes, samples were tested using a modified pre-cleaved deletion substrate with a single U replaced by an A (Figure 7C, top). U-specific exonuclease activity is observed with KREN1 editosomes, but not with KREN2 editosomes in KREX2 null cells from both BF and PF ( Figure 7C, left and right, respectively). In contrast, both KREN1 and KREN2 editosomes show U-specific deletion when isolated from PF 29.13 cells. The absence of the -2U intermediate product is particularly noticeable for KREN2 editosomes isolated from KREX2 null cells.

Discussion
The data we present here demonstrate that KREX2 is not required for RNA editing, suggesting that U deletion in vivo is primarily mediated by the subset of editosomes that contain KREX1. In vitro deletion assays reveal that KREX1 is necessary for the U-specific exonuclease activity of editosomes isolated from KREX2 null cells. The loss of KREX2 leads to a smaller editosome size, as determined by glycerol gradient sedimentation, and a disruption to the deletion subcomplex, which is apparent by the shift of KREL1 to lower S value fractions. The abundance of most edited, pre-edited, and never edited transcripts is unaltered in KREX2 null cells compared to controls, with distinct exceptions. Edited CYb is essentially eliminated in BF-KREX2-null cells, while edited ND3 and edited CR3 are severely decreased in PF-KREX2-null cells. Regardless of these editing defects, the persistence of cell growth in the absence of KREX2 in both BF and PF stages shows that it is not required catalytically or structurally for editing to occur.
Several lines of evidence indicate that KREX2 has been eliminated from both BF-KREX2-null and PF-KREX2-null cells.
First, the open reading frame of KREX2 was not detected by two distinct sets of primers by PCR, while PCR products consistent with the integrated knockout constructs were found ( Figure 1). Second, Southern analysis of genomic DNA from BF-KREX2null cells demonstrates the loss of the KREX2 gene. Third, Western analysis using anti-KREX2 antibody shows that PF-KREX2-null cells lack KREX2 protein. Fourth, real-time PCR analysis shows that KREX2 mRNA is gone in both BF-KREX2null and PF-KREX2-null cells. Fifth, the shift in editosome sedimentation matches independently derived KREX2 RNAi results. Finally, the loss of U-specific exonuclease activity in editosomes lacking KREX1 shows the loss of activity consistent with KREX2. Together these data show that KREX2 has been eliminated from both BF and PF cells, and thus KREX2 is dispensable for growth in T. brucei.
While the growth of BF-KREX2-null cells was indistinguishable from parental control cells both in vivo and in vitro, PF-KREX2-null cells grew slower than parental 29.13 cells in media containing glucose (Figure 2). Unlike PF cells BF cells are normally dependent on glucose. Ectopic expression of KREX2 in the PF null cells, which was confirmed by Western analysis and resulted in normal editosome sedimentation but did not rescue normal growth in PF (data not shown and Figure S1). A trivial possibility is that the glucose-dependent growth defect in the null cells is coincidental and unrelated to the loss of KREX2. Alternatively, loss of KREX2 may have resulted in anomalies in mitochondrial function such as disrupting the balance between oxidative phosphorylation and glycolysis as implied by the growth deficiency in glucose [20][21][22]. In this case, differences between ectopic and endogenous expression may have resulted in the different growth rates. For example, different rates of protein expression may have altered editosome assembly, turnover or function.
KREX2 mRNA is absent in null mutant cells but the levels of KREX1 mRNA and of most edited transcripts are unaltered compared to parental control cells. However, there are distinct defects in RNA editing in vivo in BF and PF KREX2 null cells with effects primarily on Cytochrome mRNAs in BFs and on complex I mRNAs in PFs. The loss of KREX2 appears to result in transcriptspecific defects in editing. The lack of edited CYb, increased relative abundance of pre-edited CYb, and ,85% relative reductions of edited CR3, COIII and ,70% reduction of edited ND3 mRNA in BF cells is consistent with the normal levels of these edited mRNAs and their protein products in this life cycle stage. In particular, edited CYb mRNA, which only has insertion sites, is normally abundant in PFs and is essentially absent in slender BFs [23]. In previous experiments, edited CYb was not detected in BF cells in which transcription of KREN2 or KREPA3 was expressed using a similar conditional knockout system, suggesting that CYb editing is particularly sensitive to changes to the editing machinery [15,24]. The rescue in BFs of editing of COIII but not CYb mRNA upon expression of an ectopic KREX2 (data not shown) may also reflect the process that controls the differential editing between life-cycle stages. The substantial decreases in edited ND3 and CR3 mRNAs and large increase in edited ND8 mRNAs in PF-KREX2-null cells relative to the parental cells implies that alterations to complex I are tolerated in this life cycle stage, which is consistent with previous observations [25]. The knockdown of both pre-edited and edited RPS12 suggests that aberrant editing of this transcript increases its turnover rate. The lack of rescue of particular editing defects upon expression of an ectopic KREX2 gene could reflect a change unrelated to loss of the KREX2 gene; however, it seems unlikely that such coincidental defects would specifically disrupt editing of different mRNAs. Unknown quantitative or qualitative differences between KREX2 expressed ectopically and KREX2 expressed from the endogenous locus might compromise its ability to incorporate into editosomes or function at wild type levels.
The shift of editosomes from both BF and PF null mutants to lower S values compared to control parental cell lines is consistent with the loss of KREX2 resulting in structural disruption of the deletion subcomplex (KREX2/KREPA2/KREL1). The greater shift of KREL1 compared to KREPA2 may reflect the association of the latter with KREPA3 and KREPA6 editosome components, and suggests that KREX2 and KREL1 directly interact [8]. RNAi knockdown of KREX2 in PF resulted in a similar shift of KREL1, but not KREPA2, to lower S values [11]. The structural disruption caused by the loss of KREX2 apparently does not prevent KREL1 function, since KREL1 is essential [26]. Whether KREL1 can perform its essential function in trans is unknown.
KREX2 null cells maintain the in vitro editing activities present in parental control cells, but assays of glycerol gradient fractions reveal clues concerning both KREX2 and editosome function. While pre-cleaved editing activities are present in KREX2 null extracts, they are uniformly shifted up toward smaller S values, mimicking the profile of editosome proteins observed by Western blot (Figures 4 and 6). In contrast, cleavage activity is observed in the same fractions from both KREX2 null and parental cells, namely ,20S and higher ( Figure 5). These results suggest that cleavage activity requires more intact editosomes, i.e. interactions among multiple proteins, while pre-cleaved activities can be performed by smaller partial complexes. The somewhat diminished editing activities in KREX2 null extracts compared to parental cells appears largely due to decreased post-cleavage activities and is consistent with the loss of KREX2 U removal activity. Thus KREX2 has exonuclease activity in vitro, but whether it functions as an exonuclease in vivo is unknown given the functional redundancy of KREX1. Nevertheless, in the absence of KREX2 U removal in vitro is restricted to editosomes that contain KREX1 (Figure 7). The presence of KREX1 in editosomes with deletion site cleavage specificity, combined with the dispensability of KREX2, suggests that U removal in vivo is primarily done by KREX1.
What role might KREX2 play in RNA editing in vivo? It could be an evolutionary relic whose role in U removal activity has been supplanted by KREX1. The smaller Leishmania KREX2 is consistent with this possibility, as it lacks the catalytic domain and hence exoUase activity. KREX2 may have a role in the structural architecture of the editosome. While this structural role may not be essential, at least in the laboratory, it may influence subtle aspects of editosome function such as differential editing during the life cycle. For example, KREX2 may influence the interactions of accessory factors with the editosome (Lerch, et al unpublished). It might also provide a U removal activity in specific circumstances such as for unusual editing sites from which numerous Us are removed, or a proofreading activity that removes excess Us that have been added in error. Hence, KREX2 absence may only be deleterious under conditions where a substantial decrease in editing fidelity cannot be compensated.

PCR analysis of genomic DNA
PCR was used to determine if the KREX2 coding sequence had been eliminated by the intended homologous recombinations. Primer sequences are in Table S1. Two sets of primers were used to detect two distinct portions of the KREX2 coding sequence; primers X2for4571 and X2rev3135 amplify nucleotides 320-1092 (773 bp product), while primers X2for3127 and X2rev3177 amplify nucleotides 1872-2585 (714 bp product). Primers that flank the junction between the inserted knockout construct and the genomic region 39 of KREX2 were used to amplify products that demonstrate the intended homologous recombination. Primers pLEW13for5154 (anneals to two places in pLEW13 insert) and X2rev3utr5952 generate products of 473 bp and 1811 bp that indicate integration of the first BF knockout construct. Primers pLEW90for5209 and X2rev3utr5952 generate a product of 702 bp that indicates integration of the second BF knockout construct. Primers BSDfor7258 and X2rev3utr5952 generate a product of 468 bp that indicates integration of the first PF knockout construct. Primers PACfor3748 and X2rev3utr5952 generate a product of 456 bp that indicates integration of the second PF knockout construct.
Wild-type KREX2 gene was cloned into the pLEW79 plasmid [27], creating the pReg-KREX2 plasmid as follows. A 2746 bp PCR product containing the KREX2 open reading frame was PCR amplified from 427 genomic DNA with Pfu polymerase using 5orf-3281 and 3orf-3309 primers. The KREX2 ORF was cloned into the HindIII/BamHI sites of pLEW79. This plasmid was used to create cell lines with tetracycline-induced expression of an ectopic KREX2 allele from the rDNA locus.

Cell lines
BF-KREX2-null cell line was generated by a series of transfections to introduce the pSKO-KREX2 and pDKO-KREX2 plasmids sequentially into 427 strain cells. First, BF427 cells were transfected with 10 mg NotI-linearized pSKO-KREX2, and recombinants were selected by G418 resistance. The second endogenous KREX2 allele was eliminated by transfection with 10 mg NotI-linearized pDKO-KREX2 and subsequent hygromycin selection. Cells were grown in HMI-9 media containing 2.5 mg/ml G418 and 5 mg/ml hygromycin. PF-KREX2-null cell line was generated by a series of transfections to introduce the pBSD-KREX2 and pPAC-KREX2 plasmids sequentially into PF 29.13 strain cells [27,31]. First, 29.13 cells were transfected with 10 mg NotI-linearized pBSD-KREX2, and recombinants were selected by blasticidin resistance. The second endogenous KREX2 allele was eliminated by transfection with 10 mg NotI-linearized pPAC-KREX2 and subsequent puromycin selection. BF-KREX2null cells were grown in HMI-9 media containing 2.5 mg/ml G418 and 5 mg/ml hygromycin. PF-KREX2-null cells were grown in HMI-9 media containing 15 mg/ml G418, 25 mg/ml hygromycin, 1 mg/ml puromycin, and 10 mg/ml blasticidin. Proper integration of each plasmid was confirmed by PCR in both BF and PF, as well as Southern analysis (BF only; Figure S1). TAP-tagged versions of KREN1 or KREN2 were introduced into both BF and PF KREX2 null cells using plasmids previously described [5]. A regulatable ectopic KREX2 allele was subsequently introduced into the rDNA intergenic locus of both BF-KREX2-null and PF-KREX2-null cells by transfection with NotI-linearized pReg-KREX2 and selection in 2.5 mg/ml phleomycin, generating BF-KREX2-rDKO and PF-KREX2-rDKO cell lines, respectively. Induction of pReg-KREX2 used 1 mg/ml tetracycline.

Growth of cells in vitro
BF cells were grown in HMI-9 with 10% FBS. PF cells were grown in SDM-79 with 10% FBS, or in SDM-79 modified by the removal of glucose and glucosamine with 10% FBS that had been dialyzed to remove glucose; as noted glucose was returned to depleted media at final concentration of 6 mM [32]. For each cell line, cell density was measured by Coulter counter, and subsequently each culture was reseeded at 2610 5 cells/mL in 10 mL (BF) or at 1610 6 cells/mL in 5 mL (PF).
Growth of cells in vivo. For both BF-427wt and BF-KREX2-null cell lines, 2.5610 7 total cells from log-phase cultures grown in HMI-9 with 10% FBS were centrifuged at 1300 g for 10 minutes at room temperature, washed once with 20 mL 16 PBS-G, and then resuspended in 1 mL 16 PBS-G so that the 200 mL injection volume contained 5610 6 cells. For each cell line, two ,20 g BALB/c mice were infected via intraperitoneal injection with 5610 6 trypanosomes. Parasitemia was measured at various time points by tail prick to draw 2 ml of blood that was preserved in 200 ml of Fixing Solution (3.7% formaldehyde, 16 SSC). The number of trypanosomes was then counted using a hemocytometer. Mice were killed by CO 2 asphyxiation when their parasitemia neared 1610 9 trypanosomes/mL.

Fractionation of cell lysates on glycerol gradients
Fractionation of BF whole cell lysates on 10-30% glycerol gradients was performed as previously described [24] with the following differences: ,1.7610 9 cells were lysed and fractionated on each gradient, and the gradients were centrifuged at 38,000 rpm in a Beckman SW40 Ti rotor for 8 hours at 4uC. Briefly, cells were resuspended in lysis buffer (10 mM Tris pH 7.2, 10 mM MgCl2, 100 mM KCl, 1 mM Pefabloc, 2 mg/mL leupeptin, 1 mg/ml pepstatin, 1 mM DTT) to final volume of 900 mL, and 100 mL 10% Triton X-100 was added. Samples were mixed by inversion for 15 minutes at 4uC and cleared by two centrifugation steps of 17,0006 g for 15 minutes at 4uC. After fractionation, glycerol gradients were divided into 0.5 mL fractions from the top, flash frozen on liquid nitrogen, and stored at 280uC. For each sample within an experiment, equivalent cell numbers were lysed. Fractionation of PF whole cell lysates was done as for bloodform, except ,4610 9 cells were used. Positive control ,20S samples from purified PF mitochondria (IsTaR 1.7a strain) were generated as previously described [24,33].

TAP-tag purifications
Isolation of PF control KREN1 and KREN2 ,20S editosomes have been previously described [5,6]. Tandem affinity purification (TAP) was used to isolate KREN1 and KREN2 editosomes from the background of KREX2 null cells that had been induced to express tagged protein by addition of 500 ng/mL tetracycline. Briefly, equivalent cell numbers (,2.8610 9 cells for BF; ,2610 10 cells for PF;) were harvested and lysed in 20 mL of IPP150, 1% Triton X-100, and Complete protease inhibitors (Roche) at 4uC, and then clarified by centrifugation at 10,0006 g. Purification of editosomes via TAP-tagged KREN1 or KREN2 used sequential IgG and Calmodulin affinity chromatography as previously described [34].

In vitro enzymatic assays
For standard pre-cleaved editing and endonuclease cleavage assays, 15 or 10 mL of glycerol gradient fractions were used, respectively. Reactions were incubated at 28uC for 3 hours. For all endonuclease and precleaved assays, RNAs were ethanol precipitated, resolved on 11% polyacrylamide 7M urea gels, and analyzed by PhosphorImager (Molecular Dynamics). A6-derived substrate assays follow standard protocols described in detail elsewhere [24,[36][37][38].
A6-derived pre-cleaved editing. Standard pre-cleaved deletion and insertion editing were assayed as previously described using 59-labeled U5 59CL and U5 39CL with gA6 [14]PC-del and 59-labeled 59CL18 and 39CL13pp with gPCA6-2A RNAs, respectively [39,40]. For assays with precleaved deletion substrate U5 59CL3A that is modified to test U specificity, either 5 or 10 mL of calmodulin eluate from BF samples or 7 mL from PF samples was assayed as previously described [41].

Real-time PCR analysis
Real-time PCR was performed as previously described, with values normalized to either 18S rRNA or b-tubulin and an internal control [4,11,24]. Primers for CR3 and ND8 targets are in Table S1. Amplicons for these primer sets were sequenced to confirm they amplified the specified target. For each RNA measured, the average of three cycle threshold (C T ) values was used in calculations. Relative changes in target amplicons were determined by using the Pfaffl method, with PCR efficiencies calculated by linear regression using LinRegPCR [42,43]. Figure S1 Southern and Western analyses demonstrate elimination of KREX2 in BF-KREX2-null and PF-KREX2-null cells. A. Genomic DNA from parental BF-427wt and derived KREX2 single knockout (SKO) and KREX2-null (X2null) cell lines was subjected to Southern analysis using a radiolabeled probe to detect the KREX2 open reading frame (KREX2 ORF). The band corresponding to KREX2 is present in DNA from either BF-427wt or BF-KREX2-SKO cell lines, but completely absent in DNA from BF-KREX2-null cells. B. The same genomic DNAs used in panel A were also analzyed using a radiolabeled probe to detect the 39 intergenic region of KREX2. Hybridization with this probe permits simultaneous detection of the endogenous KREX2 alleles, the first allele knockout (1 st allele KO) with T7 RNA polymerase and Neo R , and the second allele knockout (2 nd allele KO) with tetracycline regulator and Hyg R . While parental BF-427wt cells have only endogenous KREX2, the BF-KREX2-null cells lack endogenous KREX2 and only display hybridization consistent with the knockout constructs that replaced each KREX2 allele. C. Western analysis of ,20S peak glycerol gradient fractions from parental PF 29.13 or PF-KREX2-null (X2null) and derived cells. Anti-KREX2 antibody reveals presence of KREX2 in 29.13 cells, and absence in PF-KREX2-null cells. Expression of the tetracycline (tet) regulatable ectopic KREX2 allele in the PF-KREX2-null+KREX2Reg cell line (X2null+X2reg) was demonstrated in the presence of tet (+tet), but not in its absence (2tet). The sizes of proximate marker bands are as indicated. The amount of KREX2 in extracts from BF cells was below the limit of detection with this antibody (data not shown). Southern analysis. Genomic DNA was isolated from each cell line (ACS protocol). For each cell line, 20 mg of genomic DNA was digested with either EcoRI or a combination of BamHI and KpnI and then fractionated by electrophoresis on 0.8% agarose gel. EcoRI digestion of genomic DNA generates a 1141 bp fragment (WT), a 5550 bp fragment (first knockout), and and a 2049 bp fragment (second knockout) detected by NotI probe. BamHI and KpnI digestion of genomic DNA generates a 4104 bp fragment (WT) detected by EcoRI probe. Two different DNAs were used as templates for making radiolabeled probes: an 823 bp EcoRI fragment of pReg-KREX2 plasmid was used to hybridize to the KREX2 open reading frame, and a 358 bp XbaI/NotI fragment of pSKO-KREX2 plasmid was used to hybridize to the 39 intergenic region of KREX2. Radiolabeled DNA probes were generated using Ready-To-Go DNA Labeling Beads (GE Healthcare) and a 32 P dCTP according to manufacturer's protocol. After labeling, DNA probes were purified using MicroSpin G-25 spin columns (GE Healthcare) according to manufacturer's protocol to remove unincorporated dCTP. KREX2 Western analysis. For each sample, 30 mL of glycerol gradient fraction 11, corresponding to ,20S peak, was separated by electrophoresis on TGX 10% SDS-PAGE gels (BioRad), transferred to Immobilon-P membranes (Fisher), and probed using 1:5 diluted affinity-purified rabbit polyclonal antibody against KREX2 as previously described [11]. Blot was then probed with 1:2000 goat anti-rabbit conjugated HRP secondary antibody and visualized by chemiluminescence (Pierce) detected by x-ray film. PageRuler ladder (Fermentas) was used as a size marker. (TIF)