Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction to Small RNAs in Protozoa

Small RNAs are defined by their length (20–30 nucleotides) and play important roles in post-transcriptional regulation (Ha and Kim 2014). There are two main types of well-studied small RNAs: miRNAs (microRNAs ) and endo-siRNAs (endogenous small interfering RNAs). Although, these two types of small RNAs are all involved in RNAi (RNA interference) pathways, their source, processing mechanisms and functions are very distinct.

MicroRNAs have been well studied in the past two decades since their first identification in Caenorhabditis elegens . They were later found in a diverse range of animals and these microRNAs appear to be widely expressed in animal somatic cells. They probably drive many of the epigenetic developmental processes that transform embryonic stem cells to terminally differentiated cells. Different tissues or cell lines contain their own specific profile of miRNAs, which regulate the expression of genes that are essential for cell-fate determination in cell differentiation (Sood et al. 2006; Li and Belmonte 2015) and hematopoietic development (Aravin and Hannon 2008). In the genome , microRNAs are self-governed genes which exist in inter-gene and intron regions. In animals, microRNA genes are typically transcribed by RNA polymerase II into long precursors pri-miRNAs (Li and Rana 2014). They are then processed by the Drosha enzyme into stem loop structure pre-miRNAs in the nucleus. The pre-miRNAs are then transported into cytoplasm, where pre-miRNAs are cleaved by the protein Dicer to generate a pair of 18–22 nt small RNAs—mature miRNA and miRNA star. The mature miRNA are loaded into the Argonaute complex and guide the complex to the 3′ UTR of mRNAs through base pair complementarity in the seed region (2–8 nt) of the microRNA. The expression of protein-coding genes are then suppressed at the post-transcriptional level. In general, the miRNA star is degraded in miRNA biogenesis, however, in some cases, these miRNA stars can also target mRNAs and have biological functions in development of diseases such as cancer (Yang et al. 2013; Shan et al. 2013; Wu and Arora 2014).

Endo-siRNAs are derived from long double-strand RNAs (dsRNAs), and then processed into 21–26 nt small RNA clusters . These genes were first found in Caenorhabditis elegans and can silence the expression of genes by means of perfect complementarity with the transcripts (Wightman et al. 1993). Endo-siRNAs are mainly derived from Transponsable Elements (TEs) and are believed to be the main mechanism for retrotransposon element silencing. There are many types of endo-siRNAs in plants (Nilsen 2008), however, it is difficult to ascertain their characteristics in animals (Okamura and Lai 2008). The pioneer work in the investigation of endo-siRNAs was performed by two groups and both of them identified a substantial number of endo-siRNAs in mouse oocytes (Tam et al. 2008; Watanabe et al. 2008). A range of studies, conducted at a similar time, used deep-sequencing technology to reveal that there were also an abundant repertoire of endo-siRNAs in Drosophila melanogaster (Ghildiyal et al. 2008; Okamura et al. 2008a, b; Kawamura et al. 2008; Czech et al. 2008; Chung et al. 2008).

MicroRNAs have been discovered in sponges and in Chlamydomonas reinhardtii, which are the lowest multicellular animals and plants, respectively. But homologous sequences and the characteristic microRNAs were not found between these two taxa (Wheeler et al. 2009). This suggested that microRNAs in plants and animals seem to have originated independently from different ancestors. A few studies have attempted to analyze the small RNAs in protozoans such as trypanosomes (Wen et al. 2011; Mallick et al. 2008; Tschudi et al. 2012a; Zheng et al. 2013), trichomonads (Chen et al. 2009; Lin et al. 2009; Huang et al. 2012), giardias (Chen et al. 2009; Liao et al. 2014; Saraiya and Wang 2008; Huang et al. 2012), amoeba (De et al. 2006) and toxoplasma (Braun et al. 2010). Some of them have reported that microRNAs demonstrate homology with those found in higher species. This implied that microRNAs might have originated from a common ancestor in the eukaryotes . Later other studies questioned the microRNAs identified in protists . Tarver et al. used critical criteria to re-evaluate these microRNAs and concluded that most of the identified microRNAs in protists were not bona fide microRNA (Tarver et al. 2012). These results bring more questions about the origin of microRNAs. Do microRNAs exist in ancient eukaryotes ? Whether the features of these so called “microRNAs” are in common with/or specific from those that have been demonstrated in the plants and animals? If the molecules previously reported are not microRNAs, do they perform other specific functions? To answer these questions, a systematic analysis of small RNAs in protists is required. However, it is difficult work because of the following problems.

2 Current Problems Associated with the Identification of Small RNAs, and Particularly Authentic microRNAs , from Protozoan Organisms

There are at least four difficulties in small RNA identification in protozoan species:

  1. 1.

    The protozoans are a very diverse group of unicellular eukaryotic organisms that may well have had extensive polyphyletic origins (Cavalier-Smith 2003). The characteristics of these organisms are very broad, some are animal-like, some are plant-like and others also like fungi . Thus, there are a wide range of potential model protozoa organisms but may not be representative of all (Montagnes et al. 2012). Therefore, there are no common characteristics of small RNAs in protozoan organisms which makes it very difficult to investigate them systematically.

  2. 2.

    Most of the parasitic or symbiotic protozoan species have complicated life-cycles, due to the fact that they can infect or inhabit multiple types of hosts (including a wide range of plants and animals). To adapt to the very different environments within different hosts, protozoans have to change their morphology frequently. These changes are dependent on the expression of genes (Lun et al. 2015). Gene expression and regulation in protozoan organisms is, therefore, potentially very complicated.

  3. 3.

    The current analysis software for microRNA identification has been specifically designed for use with higher animals and plants. However, the small RNAs in protozoan species are very different from those found in higher organisms. Thus, this software cannot be used directly for the investigation of protozoan small RNAs.

  4. 4.

    The next generation sequencing (deep-sequencing) technology has improved the convenience and accuracy of microRNA identification. However, compared with higher organisms, there is very limited small RNA transcriptome data for protozoan species lodged in databases. This is especially true for comparative data between different life cycle stages.

3 Computational Methods for Genome -Wide Analysis of Small RNAs in Animal-Like Protozoans

To specifically investigate small RNAs in protozoans, we developed two computational methods to analyze endo-siRNAs and microRNAs from deep-sequencing results which could be used for animal-like protozoans. The origins of siRNAs have been found to be from longer double stranded RNA which have been cleaved by the protein Dicer . Thus siRNAs can be distinguished from random degradation or small RNA production by pretreating extracted protozoan RNA with Dicer (Fig. 11.1). The families of fragments are then subjected to deep sequencing using standard approaches. Next the 3′ adaptors were removed from the illumina-generated sequence reads and then the remaining 18–30 nt reads were mapped to the desired protozoan genome database(s) . For reads that mapped to multiple loci, they were excluded if there were greater than nine hits on the genome. Figure 11.1 shows the flowchart for siRNA cluster identification. We first counted the numbers (N) of small RNAs that mapped to both strands of the genome to generate two numbers: one for the sense and the other for the antisense strand. These small RNAs are linked into a cluster if they have at least one base that is overlapping.The length of the longest cluster is then calculated (L). These numbers were then divided by the length of the cluster (L) to obtain the densities (N/L) of small RNA enrichment. We chose the density of 0.04 as a cutoff in order to select candidate NATs with high potential to produce NAT derived siRNAs. This threshold is the same as used in the study by Guo et al. on trans-NATs in rice (Guo et al. 2009). Identification of microRNA was carried out as shown in Fig. 11.2. These are usually found in a single location and have no adjacent clusters . They are recognized by regions of reverse complementarity enabling the formation of hairpin secondary structure.

Fig. 11.1
figure 1

siRNA cluster identification

Full size image
Fig. 11.2
figure 2

microRNA identification

Full size image

4 Small RNA Investigation in Trypanosoma brucei

We combined computational analysis and high-throughput sequencing to study siRNAs and microRNAs in T. brucei , produced during the two main life cycle stages, bloodstream form (BF) and procyclic form (PF). We focused specifically on Natural Antisense Transcript (NAT) pairs derived siRNAs in order to gain further insight into the extent and functional role of these RNAs in protozoan organisms. We identified hundreds of NATs on the annotated transcript-scale and whole genome-wide analyses (Tschudi et al. 2012a). We identified the majority of these small RNAs as siRNAs, clearly indicating the prevalence of this mechanism and its potential importance in protozoan parasites.

4.1 cis- and trans-NATs Were Identified Within the T. brucei Genome

Putative cis and trans-NATs identification was based on the annotation of T. brucei transcripts (Fig. 11.3). To identify the potential cis-NAT pairs, we compared the genomic loci of all annotated transcripts to search for gene pairs that in the same loci but from opposite strands. Eleven enclosed pairs of transcript elements were identified by overlapping reverse-complementary sequences and all of which were formed by protein coding genes and non-coding genes (snoRNAs and snRNA).

Fig. 11.3
figure 3

Genome-wide identification of cis- and trans-NATs

Full size image

We carried out a pairwise alignment on all annotated transcript sequences to find trans-NAT pairs and identified 209 pairs of trans-NAT candidates. Specially, two tRNAs in adjacent locations could form into hairpin structures. Both of the two tRNAs were glutamine tRNA , locating in adjacent loci in plus strand of Tb927_08_v4 (Fig. 11.4a). The two inverted tRNAs were composed of perfect palindromes with an 89 nt internal sequence and were able to form highly stable hairpin loops (Fig. 11.4b).

Fig. 11.4
figure 4

Location and structure of two adjacent tRNAs which can form into NAT pairs. a Two Glutamine tRNAs (Tb927.8.6566 and Tb927.8.6568) at the adjacent locus of Tb927_08_v4 from 1892288 to 1892520. b Predicted structures that would be formed by base-pair interactions of the region. Secondary structures are folded using Mfold (Zuker 1989)

Full size image

We did not find any NATs in the mitochondrial genome. This was in accordance with the results in P. falciparum (Gunasekera et al. 2004) but inconsistant with the results in humans, mice and worms in which the abundance of antisense transcripts were low in the nuclear genome but high in abundance in the mitochondrial genome (Zhang et al. 2006). This might be an interesting subject for further study.

4.2 NAT-Derived Small RNAs Show Features of siRNAs

The NATs in both life stages were distributed across all of the 12 chromosomes of T. brucei TREU927 strain (Fig. 11.5). Chromosome sequences of T. brucei were downloaded from the TriTrypDB (release-2.1) component of EuPathDB (http://tritrypdb.org/tritrypdb/) (Aslett et al. 2010). Some NATs exhibit specific clustering across particular regions of the chromosomes (Fig. 11.5). Furthermore, we realized that these clusters were the regions where duplication events occurred frequently. For example, the longest NAT extended 8436 bp and spanned a cluster of transcripts which were mainly transcribed from retrotransposon hot spot protein (RHS) pseudogenes . As shown in Fig. 11.6a, the NAT on Tb927_01_v4 from 126434 to 134869, where 6 transcripts were crossed, which included four RHS pseudogenes and two protein-coding genes with unknown function.

Fig. 11.5
figure 5

Chromosome map of NATs. All of the NATs are mapped by there location on the chromosome in BF stage (a) and PF stage (b)

Full size image
Fig. 11.6
figure 6

Example of small RNAs derived from NAT. a is the screenshot of a region of T. brucei genome browser from TriTrypDB (Aslett et al. 2010) which show transcripts position information across the genome. Arrow indicates the transcription direction (5′ to 3′). Color also indicates the transcription direction: blue for sense transcripts and red for antisense transcripts. b Is the screenshot of a region of T. brucei genome browser from IGV (Thorvaldsdottir et al. 2013) The small lines with different color indicate the small RNAs originated position the color indicate their transcription direction (blue for sense red for antisense). Top is BF stage and bottom is PF stage

Full size image

For each NAT type (i.e. cis-NATs or trans-NATs), we computed the densities of small RNA loci in the overlapping region. Small RNAs which derived from sense and antisense of the genome region are widely distributed in the bloodstream form and procyclic form (Fig. 11.6b). We then calculated the features of these small RNAs. In total, we got 259433 and 84339 NAT-derived small RNAs in bloodstream form and procyclic form, respectively. Most of these small RNAs were in the length region of 23–26 nt and had a characteristic “U” at the 5′-terminal (Fig. 11.7). These features correspond to the known siRNAs in T. brucei (Djikeng et al. 2001). Our further experiments showed that these small RNAs are indeed siRNAs and dependent on dicer processing (Zheng et al. 2013).

Fig. 11.7
figure 7

Feature distribution of BTR-derived small RNAs. The length distribution of small RNAs derived from BTRs in BF stage (a) and PF stage (b). The 5′ end nucleotide distribution of small RNAs derived from BTRs in BF stage (c) and PF stage (d)

Full size image

4.3 Endo-siRNAs Have Multiple Sources and Differential Expression in Two Life Cycle Stages

In addition to NATs, it has been shown that there are many regions in the genome that could generate siRNAs (Djikeng et al. 2001; Zheng et al. 2013; Tschudi et al. 2012a). These sources are listed in Table 11.1. Stage-specific siRNAs have been proved to be induced to respond to special environmental stress (Borsani et al. 2005; Hao et al. 2010). But what about the siRNAs in T. brucei ? In our results, 78.6 % of siRNAs were only identified in the bloodsteam slender form, whereas only 9.2 % were specifically expressed in the insect procyclic form. Interestingly, the stage-specific siRNAs tended to be transcribed from clustered regions. For example, the region of Tb927_07_v4 (317344–621335) is a source of siRNAs. Surprisingly, although plenty of siRNAs were derived from this region in the slender form, little was found in the procyclic form (Fig. 11.8). Next, we analyzed the expression patterns of siRNAs in both stages and found 534 siRNAs exhibited significant expression differences (p < 0.001), 79 % are expressed significantly higher in the BF while only 21 % are expressed significantly higher in the PF stage.

Table 11.1 The sources of siRNAs in T. brucei
Full size table
Fig. 11.8
figure 8

Is the screenshot of a region of T. brucei genome browser from IGV (Thorvaldsdottir et al. 2013). The small lines with different color indicate the small RNAs original position and the color indicates their transcriptional direction (blue for sense red for antisense). Top is BF stage and bottom is PF stage

Full size image

4.4 Non-existence of Canonical microRNA in Selected Stages of T. brucei

In 2008, Mallick et al. reported that they used a bioinformatic approach to identify microRNA genes in Trypanosoma brucei and to predict their functions (Mallick et al. 2008). In total, they identified 1162 microRNA candidates and most of them targeted to the variant surface glycoprotein (VSG) in this parasite protozoan. They suggested that the microRNAs in T. brucei might play roles as genetic switches in modulating host-parasite interaction.

However, our results from the high-throughput sequencing on T. brucei did not identify any microRNA molecules from this parasite. Combined with the previous studies (Tschudi et al. 2012b; Kolev et al. 2011) we did not support the existence of microRNA in this parasite, at least in these two stages. We do not know the reasons behind this different result but it might be related to the microRNA genes which are expressed only in different stages of the life cycle of T. brucei . More studies need to be carried out to investigate this interesting topic.

No matter whether microRNAs exist in T. brucei or not, our results have shown that siRNAs as well as other small RNAs, like tRNA -derived small RNAs and rRNA derived small RNAs, are important regulators in the differentiation of T. brucei (Zheng et al. 2013). By comparison with the procyclic form, we found that some genes in the VSG family were specifically regulated by siRNAs in the slender form of T. brucei. VSG is an important antigen in the slender form of T. brucei. The extensive variety of this antigen allows the parasite to escape from the immunological defense of the mammalian hosts. Therefore, we propose that the siRNAs correspondingly evolve to eliminate the obsolete VSG transcripts as rapidly as possible, so that the novel VSG can be effective in a timely way. Taken together, our results provide genome-wide identification of small RNAs in T. brucei. These results will facilitate functional studies of siRNAs in this model protozoan, as well as in other protozoan species, and help to unravel complex gene and non-coding RNA regulatory networks in eukaryotes .

5 Investigation into Small RNAs in Giardia lamblia

Giardia lamblia is one of the most primitive eukaryotes, and the study of its sRNA transcriptome could shed light on the origin and evolution of various sRNAs. We found that the G. lamblia sRNA transcriptome mainly consists of two kinds of sRNAs: endo-siRNA and tRNA derived sRNA (Liao et al. 2014). Both of these two kinds of sRNAs are also found in higher eukaryotic genomes, such as humans, suggesting they might be involved in the regulation of basic biological processes in eukaryotes.

5.1 Retrotransposon and eSGR Derived Endo-siRNAs

G. lamblia endo-siRNAs were primarily derived from two sources, retrotransposon and endo-siRNA generating regions (eSGR). The G. lamblia genome has three retrotransposons: two of these (GilT and GilM) are located in the telomeric regions, while the third retrotransposon (GilD) is dead (Arkhipova and Morrison 2001). Both functional retrotransposons, GilT and GilM, generate large number of endo-siRNAs. The dead GilD does not appear to generate any sRNAs. The endo-siRNA generating region (eSGR) is a region in the G. lamblia genome that seems to be specifically used to produce endo-siRNAs (Liao et al. 2014) and encodes three eSGRs (Fig. 11.9), named eSGRI , eSGRII and eSGRIII respectively. The eSGRI region is the longest and generates most of the eSGR derived endo-siRNAs (87.1 %). It spans 40.55 kb in length and contains 34 protein-coding genes. Intriguingly, all these 34 protein-coding genes in eSGRI are not expressed. It is possible, therefore, that their expression is blocked by eSGRI derived endo-siRNA . In contrast, eSGRII and eSGRIII produce much less endo-siRNAs than eSGRI and they occupy shorter regions, 2.55 kb for eSGRII and 4.32 kb for eSGRIII (Fig. 11.9). These two regions also contain some protein coding genes, but these mRNAs are being expressed (Fig. 11.9). This might be due to the low expression levels of eSGRII and eSGRIII derived endo-siRNAs. The eSGR derived endo-siRNAs have the same characteristics as retrotransposon derived endo-siRNAs, which may be caused by them sharing an identical biogenesis pathway.

Fig. 11.9
figure 9

The mRNA and sRNA expression statuses of three SGRs

Full size image

G. lamblia endo-siRNAs share very similar characteristics with endo-siRNAs of other eukaryotes including T. brucei . For instance, like endo-siRNAs in other eukaryotes, the G. lamblia endo-siRNAs have a length of between 20–30 nt, they are derived from both strands of the endo-siRNA generating regions and have one or more post-transcriptionally added 3′ untemplated nucleotides. The conservation of these characteristics suggests that they may be important for the function of endo-siRNAs. Unexpectedly, the 5′ end nucleotide of G. lamblia endo-siRNAs does not appear to be “U” like the endo-siRNAs of higher eukaryotes. It’s unclear yet how and when endo-siRNA are processed to acquire their 5′ end “U”. This is an interesting question worth pursuing in the future.

5.2 The Function of the G. lamblia Endo-siRNA Pathway

It’s hard to clarify the detailed function of each endo-siRNA , because of the huge number of endo-siRNAs. We have identified more than a hundred thousand endo-siRNAs in the G. lamblia trophozoite , and the expression of each of these endo-siRNAs is low. A good way to investigate endo-siRNA function is to block the key protein in the endo-siRNA biogenesis pathway. Through knock-down of the G. lamblia Dicer gene (GLDICER), which is responsible for the slicing of the endo-siRNA precursor to generate mature endo-siRNAs (Macrae et al. 2006), we significantly reduced the expression level of all endo-siRNAs and observed that the ability of G. lamblia to differentiate was impaired. This result indicated that the endo-siRNA pathway might be involved in the regulation of G. lamblia differentiation. Another function of the G. lamblia endo-siRNA is in regulating antigenic variation (Prucca et al. 2008). This was found by Prucca et al. through knock-down of two proteins GLDICER and G. lamblia RNA dependent RNA Polymerase (GLRDRP). GLRDRP is a protein used to generate appropriate RNA for GLDICER slicing. Both the silencing of GLDICER or GLRDRP lead to the failure of mechanisms of G. lamblia antigenic variation. It’s unclear yet how endo-siRNAs regulates differentiation and antigenic variation. The detailed study of the regulation mechanisms of endo-siRNAs is key to an understanding of the functional evolution of endo-siRNA in eukaryotes .

5.3 The G. lamblia Genome Encodes Six Kinds of tRNA Derived sRNAs

Recent studies have shown that tRNA derived sRNA (tsRNAs) is involved in the regulation of translation and various pathological processes like cancer (Goodarzi et al. 2015) and sperm tsRNAs from father may mediate intergenerational inheritance of diet-induced metabolic disorders (Chen et al. 2016; Sharma et al. 2016). We found that tRNA derived sRNAs are conserved in G. lamblia and exhibited changes in expression profiles between different life stages (Liao et al. 2014; Li et al. 2008), suggesting that the role of tRNA derived sRNA in cells might be more important than previous expected. In total, six kinds of tRNA derived sRNAs were found in G. lamblia , they were named as 5tasRNA, actasRNA, 3tasRNA, 5sitRNA, 3sitRNA and 5EsRNA respectively. The 5tasRNA, actasRNA, 3tasRNA members were derived from the 5′ end, middle region (most of the anticodon stem loop) and 3′ end of the mature tRNA, with lengths of around 20–22, 24–26, 28–30 nt respectively. Similarly 5tasRNA and 3tasRNA, 5sitRNA and 3sitRNA were also processed from the 5′ and 3′ ends of mature tRNA respectively. However these had much longer lengths, 50 nt for 5sitRNA and 46 nt for 3sitRNA. 5EsRNA was derived from the 5′ end of pseudotRNA-Gln (TTT), with a length of 36 nt.

5.4 The Biogenesis and Potential Function of G. lamblia tRNA Derived sRNAs

With the exception of 5EsRNA, all five remaining kinds of tRNA derived sRNAs were generated from various mature tRNAs. We identified two cleavage sites, Cleavage Site I and II (CSI and CSII), in the clover leaf structure of various tRNAs (Fig. 11.10). We found that 5tasRNA, actasRNA, 3tasRNA, and 5sitRNA, 3sitRNA might be all generated from cleavage at these two sites by an, as yet, unknown RNA endonuclease . Considering that the 5tasRNA and 3sitRNA can be integrated into an intact tRNA , as does 3tasRNA and 5sitRNA, and also 5tasRNA, actasRNA and 3tasRNA, it’s possible that the cleavage of mature tRNAs may generate one of these three groups of products (Fig. 11.11). The sequence of 5EsRNA covers the whole 5′-end exon of pseudotRNA-Gln(TTG), its biogenesis might borrow the pecursor process machinery from pseudotRNA-Gln(TTT).

Fig. 11.10
figure 10

Five kinds of sRNAs derived from mature tRNAs. CS refers to cleavage site

Full size image
Fig. 11.11
figure 11

The potential biogenesis pathway of six kinds of G. lamblia tRNA derived sRNAs

Full size image

The detailed functions of G. lamblia tRNA derived sRNAs are as yet unclear. 3sitRNA was the first identified G. lamblia tRNA derived sitRNA (Li et al. 2008). It is not expressed in trophozoites , but can be induced by several stress factors, thus connecting them with a potential stress resistant related function. Similarly, 3sitRNA, 5sitRNA, 5tasRNA, actasRNA and 3tasRNA are not expressed in the trophozoite stage either. However, all of these five kinds of sRNAs are expressed when the trophozoite starts to differentiate into cysts. They are also expressed at a high level in cysts, indicating that they might be involved in the differentiation process in Giardia. It’s difficult to investigate the function of tRNA derived sRNAs as yet, because we currently lack appropriate methods to manipulate their gene expression. However, the identification of endonucleases responsible for the biogenesis of these sRNAs may overcome this obstacle.

5.5 Does the G. lamblia Genome Possess microRNA-Encoding Genes?

Because miRNAs share the same biogenesis machinery and effectors with endo-siRNA , it allows the possibility that the genome of G. lamblia may encode miRNA genes. Indeed, many studies have identified 166 potential miRNA in G. lamblia (Saraiya et al. 2011; Huang et al. 2012; Zhang et al. 2009; Chen et al. 2009; Saraiya and Wang 2008; Li et al. 2011, 2012). However, 76 % (126) of these miRNAs could not be detected by our very high depth sRNA sequencing. Among the 40 miRNAs that were revealed in our sequencing data, 12.5 % (5) were derived from ncRNAs, 50 % (20) were derived from an eSGR region, and the expression level of the remaining 37.5 % (15) miRNA is very low and only accounts for 0.0048 % of all reads in the trophozoite sRNA library. Therefore, there is no conclusive evidence, yet, that the G. lamblia genome encodes any miRNAs.

6 Concluding Remarks

In this review, we summarize the research on the analysis of small RNA transcriptome data on Trypanosoma brucei and Giardia lamblia . The results show that most of the protozoans do not have canonical miRNAs. However, there are many types of endogenous small interfering RNAs (endo-siRNAs) derived from pseudogenes , transcriptional elements, bidirectional transcription and NATs (natural antisense transcripts). In addition, widespread variation among the tRNA derived small RNAs were also observed in different stages of life cycle in these parasites. These results shed light on questions regarding the potential origins and evolution of mechanisms of small RNA-based control of gene expression.