Artículo No 4 Evidence of RNA editing in Leishmania braziliensis

Evidence of RNA editing in

Leishmania braziliensis

promastigotes

César Ramírez&Concepción Puerta&Jose M. Requena

Received: 14 August 2010 / Accepted: 6 October 2010 #Springer-Verlag 2010

Abstract RNA editing in trypanosomatids is an elaborate form of post-transcriptional processing that inserts and deletes uridines in many mitochondrial pre-mRNAs, pro- viding the genetic information needed to create functional transcripts. The process has been extensively analyzed in

Trypanosoma brucei,Crithidia fasciculata, andLeishmania tarentolae. However, few data exist on this mechanism in pathogenic Leishmania species. Here, we show evidence that this process also operates in Leishmania braziliensis, being the first time that RNA editing has been described in a species of the Viannia subgenus. A partially edited transcript corresponding to the NADH dehydrogenase subunit 8 (ND8) gene was identified in L. braziliensis

promastigotes. Sequence analysis allowed the identification of the maxicircle-encoded cryptogene, which shows a high degree of sequence conservation with the corresponding cryptogenes in other Leishmania species. Although an edition pattern could be postulated for the ND8 transcripts in L. braziliensis, attempts to isolate completely edited transcripts by RT-PCR were not fruitful; instead, many transcripts with partial and unexpected editing patterns were isolated. This data, together with our inability to

detect full-size transcripts by Northern blotting in promas- tigotes of L. braziliensis, led us to the suggestion that the strain used in this study (M2904) lacks of critical RNA guides for a complete edition ofND8transcripts.

Introduction

Protozoan parasites of the generaLeishmaniaare causative agents of a group of diseases, known as leishmaniases. These range from the spontaneously healing cutaneous lesions arising from Leishmania major infection to muco- cutaneous leishmaniasis usually associated withLeishmania braziliensisand the often fatal visceralising disease, caused by Leishmania donovani, in the Indian sub-continent, and

Leishmania infantum (Leishmania chagasi), in Latin America and the Mediterranean basin (Bañuls et al. 2007). World Health Organization (WHO) epidemiologi- cal data indicate that there are over two million new cases of leishmaniasis each year in 88 countries, with 367 million people at risk (http://www.who.int/health-topics/ leishmaniasis.htm). Leishmaniaparasites undergo a dige- netic life cycle, differentiating from the promastigote form in the insect vector, the phlebotomine sand fly, to the amastigote form in the lysosomal compartment of the macrophages of mammals.

Leishmania, and related trypanosomatids of the genera

Trypanosoma, derive from one of the earlier evolutionary lines diverging from the common trunk of eukaryotes (Moreira et al.2004). As a reflection of their evolutionary distance to most of eukaryotes, they possess several distinctive features and surprising molecular mechanisms (Donelson et al.1999). Perhaps one of the most fascinating molecular mechanisms operating in these organisms is RNA editing of their mitochondrial transcripts (Estevez and

Electronic supplementary materialThe online version of this article (doi:10.1007/s00436-010-2190-6) contains supplementary material, which is available to authorized users.

C. Ramírez:C. Puerta

Laboratorio de Parasitología Molecular,

Departamento de Microbiología, Pontificia Universidad Javeriana, Bogotá, Colombia

C. Ramírez:J. M. Requena (*)

Centro de Biología Molecular Severo Ochoa (CSIC-UAM), Universidad Autónoma de Madrid,

c/ Nicolás Cabrera 1, 28049 Madrid, Spain

the addition or removal of uridine (U) residues at precise sites of mitochondrial transcripts. This process creates initiation and termination codon, corrects frameshifts, and builds entire open-reading frames from nonsense DNA sequences. The mitochondrial genome of trypanosomatids is also peculiar; it consists of a network of two classes of topologically interlocked circular DNA molecules: maxi- circles and minicircles (Schneider2001; Lukes et al. 2005). There exist about 50 copies of the maxicircle DNA (20–

40 kb in size, depending on the species), they are similar in composition and contain the complement of genes typically found in the mitochondrial genomes. Minicircles are smaller molecules (0.65–2.5 kb), heterogeneous in sequence, that are found in 5,000–10,000 copies per organelle. Some of the maxicircle genes (termed cryptogenes) encode transcripts that need to be remodeled by RNA editing in order to convert them into translatable mRNAs. The genetic infor- mation dictating RNA editing of particular transcripts is specified by short transcripts called guide RNAs (gRNAs), most of them encoded in the minicircle DNAs (Arts and Benne1996).

The first description of RNA editing was provided by Benne et al. (1986) and was based on the finding of a major transcript of the cytochrome oxidase subunit II (coxII) containing four nucleotides non-encoded in the DNA and that corrected the frameshift in the Trypanosoma brucei coxII gene. Soon thereafter, additional examples of RNA editing were reported in T. brucei and in two other trypanosomatids, the saurian parasiteLeishmania tarentolae

and the insect parasiteCrithidia fasciculata, which became model organisms for studying RNA editing (Arts and Benne 1996; Simpson and Shaw1989). Twelve of the 17 pre-mRNAs encoded in the mitochondrial genomes of these trypanosomatids were shown to be edited (Stuart et al. 1997). Some pre-mRNAs are extensively edited (pan-edited) and this remodeling may affect more than 50% of the final sequence. However, the extent of editing within a given gene may be different between species (Seiwert 1995; Simpson et al.2000). Another remarkable feature is the co-existence in the mitochondrial steady- state RNA population of completely unedited, partially edited, and completely edited forms for each gene transcript (Simpson and Shaw 1989). RNA editing is believed to proceed in the 3′ to 5′ direction; however, partially edited RNAs with aberrant patterns of editing are frequently observed at the transition of unedited to edited sequences, in which editing sites contain wrong number of Us, non-edited sites are edited, and 5′ sites are edited before complete editing of 3′ sites. The frequency of this class of aberrant transcripts is both transcript and species dependent (Benne 1994).

extensively, there are few reports of editing in other species of the class Kinetoplastea (Simpson et al. 2000). Until recently, mammalian-infecting Leishmania species were not analyzed for the presence of RNA editing. Ibrahim and co-workers (Ibrahim et al. 2008) described that L. donovani coxII gene is edited in an identical manner as the homologous gene in L. tarentolae. Recently, a more complete picture of the RNA editing process in this species has been reported (Nebohacova et al.2009). InL. major, after analysis of the maxicircle sequence and its comparison with theL. tarentolae maxicircle sequence, it has been suggested that the RNA editing process should be equivalent to that described inL. tarentolae(Yatawara et al.2008). More recently, a study on the editing process inLeishmania mexicana amazonensishas been published (Maslov 2010). However, no data have yet been reported on the existence of RNA editing in any Leishmania

species of the subgenusViannia. Thus, this study provides the first experimental evidence of active RNA editing inL. braziliensis, a species of the subgenus Viannia, which cause cutaneous and mucocutaneous leishmaniasis in Latin America (Martinez et al. 2010).

Materials and methods

Parasites

Promastigotes of L. braziliensis (MHOM/BR/75/M2904) were grown at 26°C in Schneider's medium supplemented with 20% heat-inactivated fetal calf serum and 100 ng/ml of 6-biopterin (Sigma–Aldrich). This strain was provided by the Centro Internacional de Entrenamiento e Inves- tigaciones Médicas (Cali, Colombia).

RNA extraction and RT-PCR

Total cell RNA was isolated from L. braziliensis promas- tigotes using Total Quick RNA extraction kit (Talent, Italy). First strand of cDNA synthesis was carried out on L. braziliensistotal RNA using oligo-dT primer and a cDNA synthesis kit (Pharmacia LKB). Amplification of edited transcripts was performed by RT-PCR on poly(T)-primed

cDNA using specific primers: LbND8-D, 5′-TTTTA

GTGTT TTTAA TAATT ATATG-3′; and LbND8-R, 5′-

TACTC GTATA ATAAT TAAAC AAAAC-3′. RT-PCR

products were separated on agarose gels, visualized under UV transilluminator and purified using the Favorgen gel purification kit (Biotech Corp., Taiwan). Finally, the amplification product was cloned into the pCR2.1 T-vector (Invitrogen).

Sequences were determined using the Big Dye Terminators v3.1 kit (Applied Biosystem) by automatic sequencing at the Servicio de Genómica (Parque Tecnológico de Madrid, Universidad Autónoma de Madrid). Sequence of clone 600 D has been deposited in the European Nucleotide Archive (EMBL-EBI) with the accession number FR686353.

Sequence analyses were performed using the BioEdit program (Hall 1999). Maxicircle DNA sequences were obtained from the L. braziliensis database at GeneDB (www.genedb.org). The assembled sequence used in this study (see Supplementary material) was derived from the following shotgun reads: brazil20h04.q1k, brazil27a07.q1k, brazil81e11.p1k, and brazil802h07.q1k.

Northern blot analysis

L. braziliensis total RNA (4 μg) was fractionated by electrophoresis in formaldehyde (6%)-agarose (1.5%) gel as previously described (Folgueira and Requena 2007). After electrophoresis, gels were blotted onto positively charged Nylon membranes (Roche Diagnostics, Germany). The filters were hybridized with the 600D probe (see below), which was labeled with digoxigenin-dUTP using the DIG High Prime DNA Labeling Kit (Roche).

Results and discussion

In the process of cloning the 3′-UTRs of L. braziliensis HSP70 genes, we isolated a cDNA clone (named 600D) containing a sequence that did not correspond to the searched regions. An analysis for sequence homology in the L. braziliensis database (GeneDB) yielded no results when the 600D sequence was compared with either predicted genes or chromosome contigs, pointing to a cloning artifact. However, when a BLAST search againstL. braziliensis unassembled shotgun reads was performed, a disconcerting result was obtained. Of the 349 nucleotides composing the 600D cDNA sequence, only the first 155 nucleotides were found to be present (100% identical) in a large number of unassembled entries. However, none of the database entries showed significant homology with the rest of the sequence. An important clue about the nature of this cDNA clone was obtained after searching in the GenBank database: the sequence gave a high score of homology with the FJ416603 entry, which contains a partial sequence of the L. donovani maxicircle (Nebohacova et al. 2009). Again, the homology was restricted to the 5′-end of the 600D sequence. This finding prompted us to analyze whether the clone 600D might be a cDNA copy of an edited transcript from the L. braziliensis mitochondria.

analyze further the clone. Thus far, within the Leishmania

genus, experimental demonstration of active RNA editing of mitochondrial transcripts has been obtained forL. tarentolae

(Simpson and Shaw1989)L. donovani(Ibrahim et al.2008) andL. m. amazonensis(Maslov2010).

In order to check the hypothesis that the 600D cDNA might represent an edited transcript, we searched in theL. braziliensis database for the putative cryptogene. A 2083- nt-long sequence, centered on the 155 conserved nucleo- tides of the 600D clone, was assembled from shotgun reads found in the L. braziliensis database (see Supplementary material). Afterwards, a manual alignment between the 600D sequence and the assembled region was carried out considering the fact that the GAC-sequences of both cryptogene and cDNA must be identical, since RNA editing affects only the U residues. As shown in Fig.1, a perfect co-linearity was found between both sequences, suggesting that indeed the clone 600D was derived from an edited transcript. Moreover, the edition was found to be extensive in the edited region of transcript 600D (from nucleotide 156 to 301; Fig. 1). The editing entails addition of 81 uridines and removal of 20 uridines at 40 total sites, resulting in an edited region that is 52% larger than the pre-edited region. These data suggested that the 600D transcript was a result of massive editing, comparable to extreme cases of editing such as those affecting to COIII (Feagin et al. 1988) and MURF3 transcripts (Koslowsky et al. 1990) of T. brucei. The 600D cDNA ended with a non-encoded poly(A) tail, a typical characteristic of mitochondrial mRNAs in trypano- somatids (Van der Spek et al.1990). Another typical feature of maxicircle genes encoding extensively edited mRNAs is the existence of a guanidine (G) versus cystine (C) strand bias, and the coding sequence located invariably in the G- rich strand (Stuart1991). This bias exists also in the 600D cDNA sequence, which has a G%−C% value of 11.8.

In order to identify whether a protein is encoded in the 600D cDNA, we analyzed possible ORFs. The results were rather disappointing, since stop codons were observed in all three frames. Thus, we assumed that the 600D cDNA might correspond to a partial edited transcript, and searches against the UniProtKB database (http://www.uniprot.org/) were carried out using the amino acid sequences deduced in the three frames. Remarkably, short stretches from frames 1 and 3 yielded significant sequence homology with the Q7M3U1 entry, which corresponds to the NADH dehydro- genase subunit 8 (ND8,G1) ofL. tarentolae. In this species, the protein is encoded by the G1 cryptogene that needs to be extensively edited to yield a translatable mRNA (Gao et al.2001; Thiemann et al.1994). Similarly, inT. brucei, the homologous cryptogene, named CR1, is edited by the addition of 259 uridines and the removal of 46 uridines

(Souza et al.1992). In both species, only short stretches at both ends of the ND8 transcripts are not edited. However, despite these tremendous editions, the final amino acid sequences are highly conserved (Thiemann et al.1994).

Assuming that the 600D cDNA was probably an editing intermediate of the ND8 transcript, we analyzed sequence conservation between L. tarentolae G1cryptogene and the putative homologous region inL. braziliensis. In addition, we included in the comparison analysis the sequences for

600D cDNA andL. braziliensis maxicircle DNA sequence (LbMax). Edited positions (due to uridine insertions) in the 600D sequence are shaded in green. Positions in the genomic sequences, removed during editing (uridine deletions), are shaded inred

Fig. 2 Alignment of theND8 cryptogene ofL. tarentolae (LtND8c; (Thiemann et al. 1994)), DNA sequence forL. braziliensismaxicircle (LbMax; this work),L. donovani ND8 cryptogene (LdND8c; positions 1501–1920 from the GenBank entry with accession number FJ416603), andL. amazonensis ND8cryptogene (LaND8c; Maslov2010). Nucleotide positions with sequence conservation greater than or equal to 75% areshaded. Dashesdenote gaps introduced to maximize alignment. The nucleotides that form part of the initiation and termination codons, which are created by editing, areunderlined

determined. Figure 2 shows the alignment of ND8 cryptogenes from these fourLeishmaniaspecies, evidencing remarkable sequence conservation, which suggests both a selective pressure to maintain the DNA sequence and, probably, a functional importance for these cryptogenes. This high sequence conservation prompted us to deduce a theoretical edited transcript forL. braziliensisND8 based on the edited ND8 transcript sequence in L. tarentolae (Gao et al.2001). As shown in Fig.3, it was possible to deduce a hypothetical edition pattern that fits perfectly to the cryptogene sequence and, more importantly, codes for an amino acid sequence sharing 93% of identity (97% of

probable frameshift (denoted by“+”in Fig.3) in the edited region of the 600D cDNA, suggesting that a misedition took place in the original transcript. RNA editing intermediates with unexpected patterns of editing in which uridines are inserted at sites not normally edited are frequently observed (Sturm and Simpson 1990). Another noticeable finding of the deduced ND8 sequence, also occurring in the L. tarentolaesequence, is that the tryptophan residue is coded by the UGA stop codon. Reassignation of the UGA codon to tryptophan occurs in many mitochondria including the ones from trypanosomatids (Schneider 2001). In summary, these data suggest that ND8 transcript inL. braziliensisneeds to be

Fig. 3 Editing ofL. braziliensis ND8mRNA (LbND8h) and comparison with the homolo- gous transcript ofL. tarentolae (LtND8). The sequence for theL. braziliensis ND8crypto- gene (LbMax) is also shown. Genomically encoded nucleoti- des are shown inupper case, inserted uridines withlower case(u), and deleted residues are indicated withasterisks. Experimentally deduced editing events, according to the 600D cDNA, are shaded ingreen, whereas those inferred by sequence comparison with LtND8transcript (Gao et al. 2001; Thiemann et al.1994) are shaded inyellow. Gaps (shown withdashes) were introduced to maximize alignment. The initiation (AuG) and the stop (uAA) of theL. braziliensis ND8transcript, created by editing, are indicated inbold. The predicted amino acid sequence forL. braziliensis ND8 (LbND8p) is also shown. A putative mis-edited position in the 600D cDNA sequence is denoted byplus sign,“+”. The position of the oligonucleotides, designed for RT-PCR amplifica- tion of editedND8mRNAs, is shown byunderliningin the LbND8h sequence

pan-edited to become a functional mRNA and the 600D cDNA represents only a partial, misedited intermediate.

Based on the edition pattern deduced for theL. braziliensis

ND8 transcript (Fig.3), we designed a forward oligonucleotide (5′TTTTA GTGTT TTTAA TAATT ATATG 3′), located in the 5′-end, non-edited sequence of the cryptogene, and a reverse

oligonucleotide (5′ TACTC GTATA ATAAT TAAAC

AAAAC 3′), located in the edited region of the 600D cDNA. These oligonucleotides, and oligo(dT)-primed cDNA fromL. braziliensis, were used to specifically PCR-amplify edited ND8 sequences. After cloning of the RT-PCR product, ten cDNA clones were analyzed by restriction analysis. The inserts

GenBank accesión number AAA91499),L. tarentolae (LtND8; (Thiemann et al.1994)), L. braziliensis(LbND8), andL. amazonensis(Maslov2010). Amino acid positions with sequence conservation greater than or equal to 75% areshaded

Fig. 5 Alignment of cDNA sequences derived from partial- ly editedL. braziliensis ND8 transcripts. In the sequence of the clone 600D, nucleotides expected to be deleted during editing are shaded inblue, those added during editing are shaded ingreen, and a misedited nucle- otide present in the cDNA 600D is shaded inred. In the sequence of clone Cl-F, an adenine seems to have been deleted during editing (position shaded in yellow). The position of the oligonucleotides, designed for RT-PCR amplification of edited ND8mRNAs, is shown by underliningin the sequence of clone Cl-B

(Fig.3). Four of them (clones B, C, F, and H) were sequenced (Fig. 5). As expected from their insert size, all clones correspond to partial edited intermediates. The extent of edition found in these clones was lower than that observed in the clone 600 D. Nevertheless, the 3′-ends of the sequences showed an edition pattern identical to that found in clone 600 D; however, in upstream regions, the sequences were either unedited or misedited. The existence of“unexpected”editing patterns, which are inconsistent with strictly progressive 3′to 5′editing has been previously documented (Sturm et al.1992; Sturm and Simpson 1990). In two of the clones (F and H; Fig. 5), the uridine predicted to be misedited in the 600D transcript (Fig. 3; see above), was missing. This finding reinforced the conclusion that this uridine residue was misadded by the editing process in the 600D transcript.

Since all ND8-derived cDNA clones had a different sequence and, more importantly, they contained many miseditions, it can be postulated that incomplete editing intermediates must be abundantly generated during the editing process of the ND8 transcript inL. braziliensis; it is likely that the editing of this transcript is very prone to error. Abundant, imprecisely edited transcripts have been found in the editing process of many trypanosomatid mitochondrial RNAs (Decker and Sollner-Webb 1990). In fact, partially edited mRNAs are substantially more abundant than mature mRNAs for extensively edited transcripts (Koslowsky et al. 1991). Remarkably, for all the ND8-derived cDNAs (Fig. 5), the 3′ portion was identical and correctly edited, the central region was partially edited (and misedited in some positions) and the 5′ portion was unedited. These features suggest the existence of several editing domains in theL. braziliensis ND8 transcript; additionally, these findings are consistent with a 3′ to 5′ global direction of the editing. However, within a given domain (i.e., the central region of ND8 cDNAs), editing appeared to be both indiscriminate and disorganized. Thus, in this region, there are incompletely edited sites that contain either too many or too few uridines relative to the edited sequences. Furthermore, some sites were found to be incompletely edited in more than one way in different clones, suggesting that there is no single way to edit a site during the editing process. For example, clone Cl-B (Fig.5) contained an insertion of 11 uridines in a site where only one uridine must be inserted. Also, this clone contained insertions at sites where edition is not expected to occur. Perhaps more striking was the observation that an encoded adenine, present in the cryptogene, seems to have been deleted during editing process, as deduced from the sequence of the Cl-F cDNA (Fig. 5). However, clones displaying unexpected patterns in which purine residues are deleted have been found in partially edited mRNAs for

observations suggest that the editing process, at least for the