Characterization of the small RNA content of Trypanosoma cruzi extracellular vesicles

Abstract

A growing body of evidence in mammalian cells indicates that secreted vesicles can be used to mediate intercellular communication processes by transferring various bioactive molecules, including mRNAs and microRNAs. Based on these findings, we decided to analyze whether Trypanosoma cruzi-derived extracellular vesicles contain RNA molecules and performed a deep sequencing and genome-wide analysis of a size-fractioned cDNA library (16–40 nt) from extracellular vesicles secreted by noninfective epimastigote and infective metacyclic trypomastigote forms. Our data show that the small RNAs contained in these extracellular vesicles originate from multiple sources, including tRNAs. In addition, our results reveal that the variety and expression of small RNAs are different between parasite stages, suggesting diverse functions. Taken together, these observations call attention to the potential regulatory functions that these RNAs might play once transferred between parasites and/or to mammalian host cells.

Recent reports have described that the intercellular exchange of genetic material also occur in eukaryotes and multicellular organisms through several distinct pathways such as those dependent on nanotubes, extracellular vesicles (EVs), and nucleic acid-binding proteins [1]. The phenomenon of microRNA (miRNA) secretion and its role in intercellular crosstalk has recently gained increasing attention as some studies have provided evidence that membrane-bound vesicles released by several cell types can transfer miRNAs to neighboring or distant cells, playing a previously unrecognized role in modulating cell functions. Several mechanisms have been hypothesized describing the interactions of membrane-bound vesicles and recipient cells, these include receptor–ligand interactions, fusion of vesicles with target-cell membrane and vesicle endocytosis [2].

There are currently three types of EVs that are named according to the mechanisms by which they are released into the extracellular environment: (i) exosomes, which are small cup-shaped vesicles ranging around 20–100 ηm in diameter released by the exocytic fusion of multivesicular bodies (MVBs); (ii) ectosomes, which are large heterogeneous vesicles ranging around 100–1000 ηm in diameter released by the budding of the plasma membrane; and (iii) apoptotic bodies, which are larger vesicles (>1 μm in diameter) that originate from apoptotic cells [3]. In a previous report, we demonstrated that T. cruzi releases at least two types of membrane vesicles (i.e., ectosomes and exosomes), generated by distinct pathways. We also showed that infective metacyclic forms release vesicles carrying virulence factors such as GP82 glycoproteins and mucins, while in contact with HeLa cells, thus suggesting that parasite-derived EVs could be used as carriers to deliver virulence and modulatory factors into host cells [4]. In the light of recent findings showing that mammalian cells are able to use membrane vesicles to transfer mRNAs and microRNAs to neighboring cells [5], together with the identification of a great number of proteins containing nucleic acid-binding domains in the proteomic analysis of T. cruzi-derived vesicles [4], we decided to analyze whether T. cruzi-secreted EVs also contain RNA molecules and perform their characterization.

To obtain the T. cruzi EVs, epimastigotes and metacyclic trypomastigotes (clone Dm28c) were cultivated as described previously with minor modifications [4]. Epimastigotes forms were collected at exponential growth phase (~2.5 × 10⁷ cells/mL), while metacyclic trypomastigotes were collected from the culture supernatant of TAU3AAG medium after 72 h of differentiation (5 × 10⁶ cells/mL). To obtain the sufficient amount of EVs-derived small RNAs for library construction, we combined the total RNA extracted from two distinct biological preparations. Briefly, two distinct preparations of 3 × 10⁹ epimastigotes or metacyclic trypomastigotes (1 × 10⁸ parasites/mL) were incubated for 12 h in DMEM without fetal bovine serum (FBS) or in TAU3AAG medium, respectively. Parasite viability was assessed by propidium iodide incorporation and showed that more than 98% of cells were viable. After the incubation, cells were removed by centrifugation at 3,000 × g for 10 min (1^st pellet), and the supernatant was filtered in 0.45-μm syringe filters and ultracentrifuged at 100,000 × g for 2 h (2^nd pellet). The pellets containing cells (1^st pellet) and EVs (2^nd pellet) were mixed with 1 mL of Tri Reagent (Sigma) and RNA extraction was performed as described by the manufacturer with minor modifications. To improve the recovery of small RNAs, 30 μg of glycogen (Invitrogen) was added followed by isopropanol precipitation for 16 h at −20 °C.

Total RNAs extracted from two distinct biological replicates were mixed 1:1 (1 μg each) before small RNA isolation and library generation. Such procedure was performed for all samples (eVes, mVes and mCell) so each one of them was composed of two distinct biological replicates, thus results presented in the manuscript are an average of independent biological replicates. Small RNAs were sequenced by LC Sciences (Houston, TX). Briefly, the small RNA fraction of 16–40 nt was isolated from total RNA of epimastigote- and metacyclic-derived vesicles (eVes and mVes, respectively), and metacyclic trypomastigote parental cells (mCell) in a 15% Tris-borate-EDTA-urea polyacrylamide gel. Figure 1A shows a denaturing formaldehyde agarose gel displaying the differences observed between total RNA of eVes and total RNA of parental eCell from which vesicles were isolated. It is interesting to note that eVes seem to be composed of a wide variety of RNA molecules, including rRNA, mRNAs, and small RNAs. The profile of mVes is quite similar to the eVes in its RNA composition (data not shown). However, as we were unable to perform the characterization of all types of RNA molecules contained in these T. cruzi-derived EVs, we decided to focus on the small noncoding RNAs, which have the potential to be regulatory molecules.

Figure 1. Profile of the small RNAs of T. cruzi extracellular vesicles.

(A) Denaturing formaldehyde agarose gel (1%) showing the distinct RNA band pattern between epimastigote-derived vesicles (eVes) and parental epimastigote cells (eCell). (B) Length distribution of tRNA-derived small RNAs.

A small RNA library was generated using the Illumina TruseqTM Small RNA Preparation kit, which is specifically designed to isolate small RNAs having 5′ phosphate and 3′ hydroxyl ends. The purified cDNA library was used for cluster generation on Illumina’s Cluster Station and sequenced on Illumina GAIIx. Raw sequencing reads were obtained using Illumina’s Pipeline v1.5 software, following sequencing image analysis by pipeline Firecrest module and base-calling by pipeline Bustard module. Sequencing data analysis was performed by a proprietary pipeline script (LC Sciences). After the raw sequence reads were extracted from image data, a series of digital filters was employed to remove unmappable/low quality reads and adaptors. Those remaining filtered reads were grouped and used to map with the reference database that was composed of T. cruzi transcripts from TriTrypDB 4.0 (http://tritrypdb.org/tritrypdb/). Filtered unique reads were aligned against the reference database using Bowtie2 (http://bowtie-bio.sourceforge.net/bowtie2/index.shtml). One mismatch was allowed in a seed alignment, the seed length was set to 20 and the strategy of local alignment was used for mapping. The unique reads were clustered into seven classes (rRNA, tRNA, snRNA, snoRNA, coding-sequences (CDS), pseudogene, and unspecified) based on their products annotated in the NCBI Gene database. Normalization of sequence counts in each sample was performed by the DESeq method.

All unique reads were distributed into seven categories using genome annotations (Table 1) and results revealed some differences between EVs derived from both developmental forms (eVes vs. mVes) as well as between EVs and cells (mVes vs. mCell). The length distribution of all unique reads separated according to their categories and a summary of all data obtained and analyzed in the present work can be found in Tables S1A-I and S2A-G, respectively. rRNA-Derived small RNAs were the most abundant class in the three samples (54-74%), followed by tRNA-derived small RNAs that were the second most abundant class in eVes (34%) and mCell (19%), and present in lower proportion in mVes fraction (6%). Reads that could not be aligned to the genome (no hit) were the second most abundant class in mVes (13%) and the third in eVes (11%) and mCell (15%). Small RNAs derived from coding-sequences were slightly more abundant in metacyclic-derived samples (5%) as compared to epimastigotes (1%), and snoRNA-derived small RNAs were mainly found in the mCell sample (4%). Several studies have shown that RNAs are selectively packed into membrane vesicles as the RNA profiles in these vesicles do not fully reflect the RNA profiles observed in the parental cells [5]. A list of small RNAs found in the EVs of epimastigotes and metacyclic trypomastigotes containing their relative differential expression is shown in Table S3A-G. Thus, in accordance with these results we found several sequences that are differentially expressed in EVs derived from both parasite forms (eVes vs. mVes) as well as differentially expressed in EVs and parental cells (mVes vs. mCell) (Table S3), which supports the differential packaging of these small RNAs into parasite-derived EVs. It is worth mentioning that since there are no replicates, significance could not be assessed (Table S3).

Table 1.

Classes of small RNAs

RNA	eVes		mVes		mCell
	# of reads	%	# of reads	%	# of reads	%
rRNA	2,373,222	55.80	2,717,445	76.46	984,404	57.95
tRNA	1,496,361	35.18	234,005	6.58	337,837	19.89
CDS	56,245	1.32	187,876	5.29	80,085	4.71
snoRNA	3,068	0.07	18,129	0.51	67,930	4.00
pseudogene	14,125	0.33	31,433	0.88	13,970	0.82
snRNA	413	0.01	2,954	0.08	972	0.06
unspecified	48	0.00	419	0.01	77	0.00
nohit	466,011	10.96	473,791	13.33	255,694	15.05
total reads	4,252,955	100.00	3,554,196	100.00	1,698,747	100.00

In accordance with previous results showing that rRNA-derived small RNA fragments are present in almost all small RNA sequence libraries, these molecules constituted the most abundant class in the three libraries analyzed in our work (eVes, mVes, and mCell). Although these results might suggest that these small RNAs are randomly degraded products, more recent studies in mammals and yeast have revealed that rRNA-derived small RNAs are functional molecules processed by special enzymes [6] and they might play important roles in heterochromatin assembly and transcription regulation [7]. Moreover, in a previous work we have shown the presence of a relatively high percentage (~16%) of ribosomal proteins and nucleic acid-binding proteins in the T. cruzi secretome [4]. Thus, the rRNA-derived small RNAs could originate from the turnover of ribosomes and/or they could be also associated with polysomes carrying mRNAs which may be mobilized into secreted vesicles [5].

Similarly to what was obtained in a different study analyzing T. cruzi small RNAs [8], in our work around 11-15% of the sequenced reads could not be aligned to the genome (no hit) using the default alignment procedures. One possible reason is due to genomic differences between the reference clone CL Brener used as database and the clone Dm28c used to construct the cDNA library.

tRNA-Derived small RNA fragments (tsRNA) have been identified in other Trypanosoma spp. and experimental evidence revealed that these fragments were usually synthesized when cells or organisms were under stress conditions [8-11]. In our study, the median length of tsRNA was of 32 nt (Figure 1B), which is consistent with the current view of bisectional cleavage of mature tRNA (~70-80 nt) near the anticodon. Most tRNA isoacceptors were found to be precursors of tsRNA identified in parasite-derived EVs, but specific isoacceptors were detected in higher levels as compared to others, suggesting that tsRNA are differentially packed into membrane vesicles (Table 2 and Table S3). Consistent with results published by Reifur et al. [10], we found that the most abundant tsRNAs in metacyclic forms (mCell) were tRNA^Glu (42%), tRNA^Thr (21%), and tRNA^Val (15%). Moreover, contrasting with tsRNA identified in mCell, the most abundant tsRNAs found in mVes were quite diverse and were composed of tRNA^Glu (33%), tRNA^Thr (17%), tRNA^Gly (10%), tRNA^Val (9%), and tRNA^Phe (9%). In addition, Garcia et al. [9] showed that tRNA^Glu (60%) and tRNA^Asp (35%) were the most abundantly expressed tsRNAs in epimastigote cells (eCell), while the most abundant tsRNAs identified in eVes by the present study comprised tRNA^Glu (49%), tRNA^Val (31%), tRNA^Thr (12%), and tRNA^Asp (3%). Taken together, these data reinforce the differential packaging of small RNAs into parasite-derived EVs and show the difference in vesicle composition between the two parasite developmental forms.

Table 2.

Small RNAs mapped on tRNA isoacceptors

tRNA type	eVes		mVes		mCell
	# of reads*	%	# of reads*	%	# of reads*	%
Ala	4,997	0.0	2,520	2.0	2,190	1.0
Arg	8,009	0.0	2,686	2.0	5,405	1.0
Asn	14	0.0	26	0.0	17	0.0
Asp	67,610	3.0	9,113	7.0	20,542	5.0
Cys	274	0.0	339	0.0	351	0.0
Glu	1,028,208	49.0	43,706	33.0	182,381	42.0
Gln	2,954	0.0	1,988	2.0	2,515	1.0
Gly	39,870	2.0	12,713	10.0	38,617	9.0
His	11,475	1.0	3,287	2.0	11,926	3.0
Ile	80	0.0	107	0.0	62	0.0
Leu	2,490	0.0	4,180	3.0	2,302	1.0
Lys	145	0.0	303	0.0	329	0.0
Met	1,270	0.0	785	1.0	626	0.0
Phe	15,584	1.0	12,129	9.0	3,354	1.0
Pro	9,972	0.0	1,490	1.0	2,656	1.0
Ser	96	0.0	259	0.0	171	0.0
Thr	255,589	12.0	22,855	17.0	93,032	21.0
Trp	1,448	0.0	331	0.0	4,760	1.0
Tyr	257	0.0	320	0.0	314	0.0
Val	650,606	31.0	12,364	9.0	65,375	15.0
	2,100,948	100.0	131,501	100.0	436,925	100.0

^*normalized values derived from Table S3 were used.

In eukaryotes, tRNA cleavage has been proposed as a mechanism to down-regulate or inhibit protein synthesis during differentiation and under oxidative stress, as well as under nutrient starvation. In T. cruzi, the machinery required for RNA interference (RNAi) seems to be either entirely lost or extensively simplified as the parasite is unable to use dsRNA to trigger degradation of target mRNAs [12]. In this context, tsRNAs have gained special attention as it was shown that these molecules can associate with Argonaute/Piwi protein family members, leading to the assembly of specific ribonucleoprotein complexes involved in silencing gene expression [6, 13, 14]. It is interesting to note that RNA-binding proteins constitute an important fraction of T. cruzi secretome [4] and these proteins are known to play an essential role in the control of gene expression in trypanosomes, modulating RNA processing, stability, turnover, and translation [15]. Moreover, additional experimental evidence demonstrated that the tRNA-maturing enzyme tRNase Z was able to down-regulate expression of human genes by degrading mRNAs under the direction of small-guide tsRNA [16, 17]. The tRNase Z belongs to the metallo-lactamase family and it was reported that trypanosomatid genomes encode for at least 4 metallo-lactamases [9]. Furthermore, tsRNAs were reported to be recruited to cytoplasmic granules that partially colocalized with reservosomes [9], which suggests that they could be packed into vesicles for secretion.

Small RNAs derived from CDS (1-5%) and pseudogenes (1%) comprised a minor portion of the libraries and could simply represent degradation products derived from mRNAs due to the constitutive polycistronic transcription and post-transcription gene expression regulation of these organisms. However, these small RNAs might also represent functional molecules that form sense-antisense dsRNAs by complementary bases thus generating natural antisense siRNAs (NAT-siRNAs), which can display regulatory functions as already shown in T. brucei [18]. NATs can be divided into two types: cis-NATs in which the two transcripts derive from the same genomic locus but opposite strands, and trans-NATs in which the transcripts derive from different genomic loci. Small RNAs derived from pseudogenes were assigned to multigene families that are enriched in subtelomeric regions such as the surface glycoprotein GP63 (50% of eVes, 23% of mVes and 17% of mCell), trans-sialidase (18% of eVes, 23% of mVes and 34% of mCell), and retrotransposons hot spot (RHS) (4% of eVes, 10% of mVes and 7% of mCell), suggesting that these subtelomeric pseudogenes might also display additional roles apart from increasing the chance of homologous recombination and antigenic variability.

Although the functional meaning of the presence of these small RNA molecules in T. cruzi-secreted EVs still need to be elucidated, results from mammalian cells [5, 19] and two recent reports from Plasmodium falciparum [20] and T. cruzi [21] showing that these parasites are able to use membrane vesicles to transfer genetic information between different cells, make us intrigued about the possible functions that these small RNAs might play once transferred to mammalian host cells and/or to other parasites. Garcia-Silva et al. [21] showed that epimastigotes submitted to nutrient starvation secrete extracellular vesicles carrying small tRNAs and TcPIWI-tryp proteins as cargo, which could be transferred to other parasites and to mammalian cells, thus increasing metacyclogenesis and susceptibility of mammalian cells to infection [21]. These findings make us speculate whether these vesicles could be used as vehicles to exchange genetic material between parasites, and between parasites and host cells as suggested by Hecht et al. [22]. Further work will be needed to clarify the molecular mechanisms that underlay these vesicles-mediated exchange events and their further implications in host-pathogen interactions and quorum sensing.

Highlights.

• The small RNAs contained in vesicles originate from multiple sources.

• Results revealed differences in the variety of small RNAs between parasite stages.

• The data call attention to the potential regulatory functions that these RNAs might play.