Extracellular Vesicle-Mediated RNA Release in Histoplasma capsulatum.

Extracellular vesicles (EVs) play important roles in cellular communication and pathogenesis. The RNA molecules in EVs have been implicated in a variety of processes. EV-associated RNA classes have recently been described in pathogenic fungi; however, only a few reports of studies describing the RNAs in fungal EVs are available. Improved knowledge of EV-associated RNA will contribute to the understanding of their role during infection. In this study, we described the RNA content in EVs produced by two isolates of Histoplasma capsulatum. Our results add this important pathogen to the current short list of fungal species with the ability to use EVs for the extracellular release of RNA.

factor alpha [TNF-␣] blockers). In the United States, it is the most common cause of fungal pneumonia (1). H. capsulatum is of particular concern in certain developing regions (2), especially in Latin American countries, including Brazil (3,4), Guatemala (5), and French Guiana, where it is considered the "first cause of AIDS-related death" (6). Despite its clear importance, enormous gaps exist in our understanding of the pathogenesis of histoplasmosis, the disease caused by H. capsulatum. An interesting facet of the biology of H. capsulatum is its ability to release extracellular vesicles (EVs) (7,8).
EVs are bilayered lipid structures released by remarkably diverse cells across all kingdoms (9). We have demonstrated that EVs are present in both ascomycetes and basidiomycetes (7,(10)(11)(12)(13)(14). This observation implies that mechanisms for EV production and release are truly ancient, as they appear to predate the divergence of these branches 0.5-1.0 billion years ago. Fungal EVs can carry biologically active proteins, carbohydrates, lipids, pigments and nucleic acids (15,16), many of which are constituents of the fungal cell wall and diverse others are associated with stress response and pathogenesis.
EV-mediated transport of fungal RNA was recently shown in both commensal and opportunistic fungi. EV RNA molecules, mostly smaller than 250 nucleotides (nt), were identified in Cryptococcus neoformans, Paracoccidioides brasiliensis, Candida albicans, Saccharomyces cerevisiae, and Malassezia sympodialis (17,18). Since H. capsulatum packages diverse compounds within EVs, we postulated that it too would use these compartments to export RNA. In this study, the EV-associated RNA components were characterized in two different isolates of H. capsulatum. As described in other fungi, H. capsulatum EVs carry both mRNAs and noncoding RNAs (ncRNAs). In addition, proteomic data allowed the identification of 139 RNA-binding proteins (RBPs) in the EVs, suggesting that proteins involved in RNA metabolism might play an important role in cell communication through the EVs. Our results add this important pathogen to the list of fungal species with the ability to use EVs for the extracellular release of RNA.

RESULTS
Histoplasma capsulatum EVs contain RNA. We characterized the RNA molecules contained in EVs isolated from culture supernatant samples of H. capsulatum strains G186AR and G217B. These strains belong to distinct clades, and G217B has been shown to be more virulent than G186AR in experimental models (19,20). The best-known difference between these two strains is that G217B lacks alpha-1,3-glucan on the yeast form cell wall (19,20).
The reads obtained from the mRNA libraries (reads of Ͼ200 nt) were aligned with each strain-specific genome available at the NCBI (G186AR ABBS02 and G217B ABBT01). For data validation, we considered only sequences with expression values of transcripts per million (TPM) of Ն100 in all biological replicates and transcripts with reads covering at least 50% of the coding DNA sequence (CDS). The small RNA (sRNA) fraction was analyzed for the presence of different species of noncoding RNAs (ncRNAs) by aligning the sRNA fraction (reads of Ͻ200 nt) with the H. capsulatum G186AR strain. These RNA molecules were compared between the strains in order to gain insights into the role of the EV RNA in this fungus and also to determine if there were differences with respect to composition between the two strains with distinct phenotypes.
Strain-specific content of EV RNA in H. capsulatum. We identified a total of 124 mRNA sequences in EV samples from the two strains and carried out paired comparisons between the G186AR and G217B samples. We applied the statistical negative binomial test with filters corresponding to TPM values of Ն100, log2 values of Ն2, and false-discovery-rate (FDR) values of Յ0.05. We observed 93 transcripts enriched in EVs derived from the G217B strain, while 31 transcripts were enriched in the G186AR strain (see Table S1 in the supplemental material). In the G217B-associated transcripts, we observed enrichment in biological processes for vesicle-mediated transport (18%), oxidation-reduction mechanisms (12%), transmembrane transport (11%), and translation (8%) (Fig. 1). In the G186AR strain, the mRNA sequences were enriched only in general cellular and metabolic processes (59%). These results suggest that there are important differences with respect to the mRNA composition of EVs derived from these two strains of H. capsulatum.
H. capsulatum EVs contain mRNA fragments and microRNA (miRNA)-like molecules. In addition to the identification of full-length transcripts in EVs, we also detected short reads of averages of 25 to 40 nt in length that aligned consistently in the CDS but at specific positions of the mRNAs (3= end, 5= end, or middle sequence); about 50% of these short fragments aligned to the reverse strand, including 172 (G217B) and 80 (G186AR) sequences of this type (Table 1). A total of 172 fragments were represented in the G217B sample compared to only 80 in the G186AR EVs (Table 1). About 47% of the reference mRNA translate proteins of unknown biological processes; this could be explained by the fact that around 33% of the genes annotated in H. capsulatum genome code hypothetical proteins and/or do not present a conserved domain, which impedes our current ability to determine specific biological activities. Those associated with DNA metabolism/biogenesis were the second most abundant for both EV samples (22 for G217B versus 16 for G186AR), followed by transport for G217B and by protein modification for both strain EVs. Other processes related to short RNAs identified in both strain EVs were oxidation-reduction, signaling, and carbohydrate and lipid metabolism (Table 1). RNA fragments associated with translation were highly enriched in G217B (n ϭ 11) but not in G186AR (n ϭ 2) EVs, while those related to response to stress were found exclusively in the G217B sample. The corresponding proteins are stress response protein whi2, DNA repair protein rad5, and a thermotolerance protein ( Table 1). Analysis of translation-related sequences allowed identification of mRNA fragments associated with distinct steps of the translation process, such as ribosome biogenesis and processing. Other metabolic pathways identified in both strains were protein modification, carbohydrate, and lipid metabolism, signaling, oxidation-reduction, and transmembrane transport, among others (Table 1).    To gain further insight into the role of EV RNAs, to determine if they could be derived from a miRNA-like pathway, and to assess if they could play a biological role in the recipient cell, we searched for RNA secondary structures, since they are fundamental for gene expression regulation (21). A broad study of RNA structures in distinct cells revealed regulatory effects of the RNA structure throughout mRNA life cycle such as polyadenylation, splicing, translation, and turnover (22,23). Using the entire range of EV RNA sequencing (RNA-seq) data, a total of 33 RNAs with putative structures were generated by a probability distribution, using a free energy (ΔG) value of less than or equal to Ϫ7.0 (Table S2). On the basis of this parameter, we identified transcripts for U3 small nucleolar RNA-associated protein, L-isoaspartate O-methyltransferase, serine/threonine-protein kinase, proteasome component C5, pre-rRNA processing protein Utp22, C-x8-C-x5-C-x3-H zinc finger protein, fungus-specific transcription factor domain-containing protein, and DNA damage-responsive transcriptional repressor RPH1 ( Fig. 2; see also Table S2).
Comparison of EV ncRNA classes in H. capsulatum EVs. We used the ncRNA database from H. capsulatum to identify the classes of ncRNA present in EV RNAs. The data analysis revealed 73 different sequences of ncRNA in H. capsulatum EVs from the G186AR strain and 38 from the G217B isolate. A total of 33 molecular species were common to both strains, 40 were exclusively identified in the G186AR strain, and the most abundant class of ncRNA found in H. capsulatum EVs consisted of tRNAs ( Table 2).
Analysis of proteins putatively associated with RNA metabolism in the EVs. As a rule, cellular RNAs are covered with proteins and exist as ribonucleoprotein (RNP) complexes. The proteins associated with RNAs are named RNA-binding proteins (RBPs). These proteins participate in several biological processes, ranging from transcription to RNA decay (24). In this context, we investigated the presence of RBPs in the H. capsulatum EVs. We analyzed the proteomic EV data available for the G217B strain (25), and we identified 139 proteins related to RNA metabolism (8) ( Table 3; see also  Table S3). We found many RBPs, such as poly(A) binding protein (PABP), Nrd1, Prp24, and Snd1; splicing factors, exosome complex components, and ribosomal proteins ( Table 3; see also Table S3) were identified. In addition, we also found quelling-deficient protein 2 (QDE2), an Argonaute protein important in the RNA machinery in fungi. Because we identified the QDE2 in EVs, we searched for the components of the RNA interference (RNAi) machinery in H. capsulatum and compared them with the proteins from Neurospora crassa and Schizosaccharomyces pombe, which are the fungal species for which the RNAi machinery was best described previously (26,27). H. capsulatum EVs contained one Argonaute protein (QDE2), two Dicer-like proteins, the QIP (quelling interaction protein), and the RNA-dependent RNA polymerase (QDE1) ( Table 4). a For some transcripts, there was an alignment in specific positions of the mRNA, not covering the entire sequence. 5=, 3=, or M (middle of the mRNA) followed by an "F" or an "R" represents forward (F) or reverse (R) orientation. GO, gene ontology; GPI, glycosylphosphatidylinositol; ID, identifier; mtDNA, mitochondrial DNA.
Comparisons of cellular RNA versus EV RNA showed a distinct enrichment of molecules in the vesicles. We next assessed the composition of cellular RNA from H. capsulatum yeast cells (28) and compared this information to that obtained from analyses of EV-associated RNA composition under the same conditions. There was no correlation between the transcripts with highest expression levels and their presence in the EVs (Table S4). Examples of highly expressed cellular transcripts included histones 4, 2B, and 2A, allergen Aspf4, chaperones, and translation factors, among others (Table S4). In contrast, zinc knuckle domain-containing protein, vacuolar ATP synthase subunit C, G 1/S regulator, thermotolerance protein, histone variant H2A.Z, and proteasome component C5 had an enrichment value of greater than 7,000 in the EVs, while they showed low expression values in the cell (Table S4). The differences in composition between cells and EVs were also evaluated by grouping the transcripts into biological (Continued on next page) processes (Fig. 3). For the yeast cells, the main pathways were associated with transport, translation, and general metabolic processes (Fig. 3). For the EVs, the enriched pathways were transmembrane transport, protein phosphorylation, and transcription regulation (Fig. 3). This result demonstrates the low levels of correlation between the most highly expressed cellular mRNAs and EV cargo, providing evidence that there might be a mechanism directing the RNA molecules to the EVs.

DISCUSSION
As previously described (17,18), RNA molecules associated with fungal EVs are remarkably diverse. For instance, mRNAs, tRNA fragments, snoRNAs, small nucleolar RNAs (snRNAs), and miRNA-like molecules were characterized in EVs from C. albicans, C. neoformans, P. brasiliensis, and S. cerevisiae (17). We observed similar distributions of RNA molecules in H. capsulatum EVs. The comparison between the G186AR and G217B EVs revealed important differences in the variety of mRNAs identified. When the mRNA composition was compared to what was described for other fungi, important similarities were observed. For example, the most abundant biological process identified in G217B EVs was vesicle-mediated transport, which was also the most abundant process in C. albicans EVs (17). Molecules required for ribosome biogenesis, which were observed in G217B EVs, belonged to the most highly enriched process in S. cerevisiae EVs (17). However, in the comparisons of the ncRNA molecules, different profiles were observed. Most of the ncRNAs in H. capsulatum strains derived from tRNAs; a similar profile was obtained with C. albicans (17). In addition, almost no snoRNAs were identified in H. capsulatum, but this class of ncRNAs was one of the most abundant in the EVs of other fungi (17). Differences in EV composition were observed previously in C. neoformans; the EV-associated RNA produced by mutant cells with defective unconventional secretion differed considerably from similar samples produced by wild-type cells (29).
In our study, we identified short reads that aligned specifically to exons; however, these sequences did not correspond to complete mRNAs in the EVs. They instead corresponded to 25-nt-long fragments that were enriched in specific exons of the transcript. These fragments of mRNAs were previously described in human cells (30), where most of the transcripts identified in the EVs corresponded to a fraction of the mRNA with an enrichment of the 3= UTR of the transcript (30). The results of that human study led to the hypothesis that the mRNA fragments had a role in gene expression regulation in the recipient cells as the secreted mRNA could act as competitors to    Glutaminyl-tRNA synthetase HCBG_08668 C0NKS5 Glutamyl-tRNA synthetase HCBG_03755 C0NE28 Glycyl-tRNA synthetase HCBG_02121 C0NN35 Histidyl-tRNA synthetase HCBG_04162 C0NL66 Isoleucyl-tRNA synthetase, cytoplasmic HCBG_03896 C0NZR4 Leucyl-tRNA synthetase HCBG_08644 C0NH95 Leucyl-tRNA synthetase HCBG_02717 C0NI62 Lysine-tRNA ligase (EC 6.1.1.6) (lysyl-tRNA synthetase) HCBG_03034 C0NMS8 Mitotic control protein dis3 HCBG_04055 C0NBJ8 mRNA splicing protein PRP8 HCBG_00494 C0NY83 NAM9 ϩ protein HCBG_07877 C0NG69 Nucleic acid-binding protein HCBG_01885 C0NUD1 Phenylalanyl-tRNA synthetase subunit beta HCBG_06962 C0NBD1 Phenylalanyl-tRNA synthetase subunit beta cytoplasmic HCBG_00427 C0NUP1 Polymerase II polypeptide D HCBG_06655 C0NNC4 Pre-mRNA-processing factor 39 HCBG_04251 C0NJB4 Pre-mRNA-processing protein prp40 HCBG_03244 C0NXM8 Pre-mRNA-splicing factor HCBG_08220 C0NLW7 Prolyl-tRNA synthetase HCBG_04497 C0NW72 Ribonuclease regulate stability, localization, and translation of mRNAs in target cells (30). In Mucor circinelloides cells, the presence of the RNA silencing pathway (sRNA) resulted in the production of both sense and antisense sRNAs (31)(32)(33). Sequencing analysis of the sRNA content of this fungus showed the existence of exonic small interfering RNAs (exo-siRNAs) as a new type of sRNA. They were produced from exons of the same genes that are later regulated through the repression of the corresponding mRNA (34). This result agrees with our observation of short reads in the exonic regions of the transcripts. We therefore hypothesize that, similarly to what was described for M. circinelloides cells, H. capsulatum EV fragments can regulate expression of their own mRNAs.
Of note, we also found a highly represented population of putative exonic RNA in As H. capsulatum EVs contain different RNA molecules, it is reasonable to hypothesize that proteins that regulate RNA metabolism are also present in the EVs, probably associated with RNA. If validated, this hypothesis could indicate how the RNAs in a specific subset are directed to the vesicles and exported. RNA-binding proteins (RBPs) participate in several biological processes, from RNA transcription to decay (24). We detected a number of RNA-binding proteins in H. capsulatum EVs (25). These proteins were also identified in association with EVs in other systems. For example, in the EVs produced by human epithelial cells, 30 RBPs were identified (35), including heterogeneous nuclear ribonucleoproteins (hnRNPs). These proteins are responsible for directing pre-mRNAs in the maturation processes that culminate in transcriptional regulation, alternative splicing, transport, and localization (35). In addition, RBPs in EVs were identified in distinct models as hepatocytes, human embryonic kidney (HEK) cells, and mouse myoblast cells (35)(36)(37). Interestingly, one of the RBPs identified in EVs was SND1 (staphylococcal nuclease domain-containing protein 1), which is a main component of the RNA-induced silencing complex (RISC) that plays an important role in miRNA function (37).
Another example of a protein identified in the EVs of H. capsulatum and distinct organisms is an endonuclease of the Ago2 family. An infection model with Plasmodium falciparum demonstrated that infected red blood cells released EVs containing functional miRNA-Argonaute 2 complexes (38). Moreover, endothelial cells internalized the P. falciparum EVs, and the miRNA-Argonaute 2 complexes were transferred to the cells and acted in regulation of gene expression and in the barrier properties of the recipient cells (38). The Argonaute protein named QDE2 in H. capsulatum was identified as enriched in the EVs of the G217B strain. The small silencing RNAs include a variety of molecules, such as microRNAs (miRNAs) and various small interfering RNAs (siRNAs), including exo-siRNAs, endogenous siRNAs (endo-siRNAs), and Piwi-interacting RNAs (piRNAs) (39). Previous studies of small RNAs in fungi identified the RNAi machinery in the fission yeast species Schizosaccharomyces pombe, in the budding yeast species Saccharomyces castellii and C. albicans, and in filamentous fungi (26,27,40). One of the best-characterized models is represented by the filamentous fungus N. crassa (27,(41)(42)(43)(44)(45). The RNAi machinery in that organism functions in defense against transposons (46). A similar process has been described in C. neoformans, where RNAi is involved in the regulation of transposon activity and genome integrity during vegetative growth (47). In N. crassa, the QDE2 gene encodes an Argonaute protein that is homologous to the rde-1 gene in C. elegans, encoding a protein required for double-stranded RNA (dsRNA)-induced silencing (27). The characterization of RNAs associated with QDE2 in N. crassa led to the identification of miRNA-like RNAs (milRNAs) in this organism (48). The identification of QDE2 in H. capsulatum EVs in association with the small RNAs indicated that the QDE2-milRNA complex might be directed to the EVs and possibly delivered to recipient cells, with the potential to interfere with gene expression regulation and/or cell-cell communication.
Fungal EVs have been implicated in a number of communication processes, including transfer of virulence (49) and antifungal resistance (50). In Cryptococcus gattii, pathogen-to-pathogen communication via EVs resulted in reversion of an avirulent phenotype through mechanisms that required vesicular RNA (49). The sequences required for this process, however, remained unknown. This is an efficient illustration of the potential derived from the characterization of EV-associated RNA in fungi. In this context, our study results provide information from the H. capsulatum model that will allow the design of pathogenic experimental models aiming at characterizing the role of extracellular RNAs in fungal pathogenesis.

MATERIALS AND METHODS
Fungal strains and growth conditions. The H. capsulatum strains were subjected to long-term storage at Ϫ80°C. Aliquots were inoculated into Ham's F-12 media (Gibco; catalog no. 21700-075) supplemented with glucose (18.2 g/liter), L-cysteine (8.4 mg/liter), HEPES (6 g/liter), and glutamic acid (1 g/liter) and cultivated at 37°C with constant shaking at 150 rpm. Viability assessments were performed using Janus green 0.02%, and all aliquots used had Ͼ99% live yeast cells. EVs were then isolated from fungal culture supernatants as previously described (12).
sRNA isolation. Small RNA-enriched fractions were isolated using a miRNeasy minikit (Qiagen) and were then treated with an RNeasy MinElute cleanup kit (Qiagen), according to the manufacturer's protocol, to obtain small RNA-enriched fractions. The sRNA profile was assessed in an Agilent 2100 Bioanalyzer (Agilent Technologies).
RNA sequencing. Purified sRNA (100 ng) was used for RNA-seq analysis with two independent biological replicates. The RNA-seq analysis was performed using a SOLiD 3 Plus platform and an RNA-Seq kit (Life Sciences) according to the manufacturer's recommendations.
In silico data analysis. The sequencing data were analyzed using version 10.1 of CLC Genomics Workbench. The reads were trimmed on the basis of quality, with a threshold Phred score of 25. The reference genomes used for mapping were obtained from the NCBI database (H. capsulatum G186AR strain ABBS02 and G217B strain ABBT01). The alignment was performed using the following parameters: additional number of bases of upstream and downstream sequences, 100; minimum number of reads, 10; maximum number of mismatches, 2; nonspecific match limit, Ϫ2, minimum fraction length, 0.7 for the genome mapping or 0.8 for the RNA mapping. The minimum proportion of read similarity mapped on the reference genome was 80%. Only uniquely mapped reads were considered in the analysis. The libraries were normalized per million, and the expression values for the transcripts were recorded in RPKM (reads per kilobase per million). We also analyzed other expression values, including TPM (transcripts per million) and CPM (counts per million). The statistical test applied was the DGE (differential gene expression) test. For the ncRNA analysis, the database used was the ncRNA database from Histoplasma capsulatum (EnsemblFungi G186AR GCA_000150115 assembly ASM15011v1). The secondary structure analysis was performed using the PPFold plugin in CLC Genomics Workbench v. 10.1 and the default parameters. The entire RNA-seq database was subjected to PPFold analysis, and the putative structures were determined. Analysis of the relationship between the profile of RNA sequences detected in this study and the protein composition of H. capsulatum EVs was based on results recently obtained with strain G217B using a proteomic approach (25). The cellular RNA used in this analysis was assessed using the Sequence Read Archive (SRA) database (accession numbers SRR2015219 and SRR2015223) (28).
Data availability. The data were deposited into the SRA database under study accession number PRJNA514312.
We declare that we have no conflicts of interest.