Fungal diseases affect nearly a billion people worldwide, ranging from hair or nail infections to highly lethal systemic fungal diseases. A plasma membrane and a complex cell wall structure delimit fungal cells. Despite this additional barrier, these cells can secrete extracellular vesicles (Rodrigues et al., 2016). Most fungal diseases affect both immunocompromised and immunocompetent individuals, though morbidity and mortality typically markedly increase in the setting of immune dysfunction. Examples are H. capsulatum, which causes fungal pneumonia (Allen and Deepe, 2005); C. neoformans and Cryptococcus deuterogattii, which cause cryptococcosis, a meningoencephalitis (Perfect and Casadevall, 2002); Aspergillus fumigatus, which causes invasive pulmonary infection or aspergillosis (Patterson et al., 2000); and C. albicans, which can cause superficial mucosal or dermal infections, as well as disseminated candidiasis (Pfaller and Diekema, 2007; Kullberg and Arendrup, 2015). In addition, Sporothrix schenkii and Sporothrix brasiliensis cause sporotrichosis in felines and humans, leading to lesions on the skin and subcutaneous cellular tissue (Barros et al., 2011; Rodrigues et al., 2020).

Besides affecting humans, pathogenic fungi can cause disease on different plant species (Li et al., 2020). The Fusarium oxysporum f. sp. vasinfectum. is a pathogen that affects cotton leaves causing phytotoxic response (Bleackley et al., 2019); Zymoseptoria tritici affects wheat (Hill and Solomon, 2020) and Ustilago maydis affects maize (Kwon et al., 2021), both plant-pathogens causing yield reductions in crops (Hill and Solomon, 2020; Kwon et al., 2021).

RNA-binding proteins (RBPs) are essential in cellular processes involved in RNA metabolism, from transcription to decay. During transcription, RBPs associate with RNA to protect it from degradation and regulate its fate in the cell. RNA processing, including splicing, alternative splicing, and trans-splicing, relies on the proper recognition and exposure of RNA sequences to the splicing machinery (Lee and Rio, 2015). After processing, mature mRNAs are transported to the cytoplasm (Björk and Wieslander, 2017),, directed to translation or stored in RNA granules for silencing or degradation. RBPs associate with mRNAs and assemble in ribonucleoprotein complexes (mRNPs) and mediate RNA processing, transport, and localization within the cell, thereby determining mRNA fate based on the composition of each mRNP complex (Gerstberger et al., 2014; Re et al., 2014).

The arrangement and assembly of RBPs in mRNP complexes are highly dynamic (Müller-McNicoll et al., 2016; Buchan, 2014). Different sets of RBPs associate with mRNA to enable its functionality. The combinatorial arrangement of proteins onto a given mRNP complex remains elusive and is a significant challenge for researchers. In addition to the genetic and histone code, an RBP code may exist, given the number of known RBPs and proteins that function as RBPs. This code is complex and ultimately determines gene expression.

Numerous studies and databases describe the presence of RBPs in EVs of different organisms, including pathogens (Vesiclepedia [http://microvesicles.org]; (Kalra et al., 2012; Pathan et al., 2019). This analysis of fungal RBPs is based on an extensive analysis of EV proteomic data publicly available.

We systematically analyzed available proteomic datasets aiming to identify proteins involved in RNA metabolism present in fungal EVs. The species with available proteomic data were C. albicans, C. neoformans, C. deuterogattii, H. caspsulatum, P. brasiliensis, S. cerevisiae, A. fumigatus, S. brasiliensis, and S. schenckii. The most applied method to isolate fungal EVs has been through differential centrifugation followed by filtration with porosity limits varying from 0.45 μm to 1.2 μm (Albuquerque et al., 2008; Rodrigues et al., 2008; Oliveira et al., 2010; Vallejo et al., 2012; Wolf et al., 2014; Gil-Bona et al., 2015; Vargas et al., 2015; Wolf et al., 2014; Matos Baltazar et al., 2016; Ikeda et al., 2018; Zarnowski et al., 2018; Rizzo et al., 2020a; Rizzo et al., 2020b). This method allows for isolating a broader range of EVs with different sizes ( Supplemental Table 1 ).

Based on the available EV proteomic datasets (Parreira et al., 2021), we identified that around 7% of the total proteins in fungal EVs relate to RNA metabolism ( Figure 1 ; Supplemental Table 2 ). Furthermore, after removing the redundancy of the distinct strains from the same species, we identified 687 proteins associated with different steps of RNA metabolism, which we discuss in detail in this work.

The RNA life cycle is highly regulated and dynamic, with each step playing a critical role in determining the fate, functionality, and abundance of RNA molecules within a cell. In the datasets we analyzed, proteins involved in various processes that regulate the RNA molecules were identified in the EVs.

Among the proteins in EVs related to transcription, Sm-like proteins were observed. Proteins belonging to this family can form small nuclear ribonucleoprotein complexes (snRNPs) by assembling with snRNAs during nuclear mRNA splicing, the process of removing introns from pre-mRNA. Various complexes can be formed depending on the protein family bound to the snRNA molecules (Will and Luhrmann, 2011). As snRNAs and snoRNAs are significantly enriched in EVs, the presence of Sm-like proteins reinforces the important role of these snRNP complexes in the EVs ( Table 1 ).

Translation initiation factors are a group of proteins crucial for initiating protein synthesis. EVs derived from human and mouse cells also contain translation initiation factors, such as eIF4E and eIF4G, and some are used as biomarkers for certain cancer types (Dong et al., 2020). We also identified eukaryotic translation initiation factors (eIFs) in EVs from different fungal species. We observed the presence of eIF3H, eIF5A, eIF2A, and eIF6 in most of the fungal EVs species we analyzed ( Table 2 ). This finding is consistent with other proteomic studies in various cell models where many eIFs have been identified (Kalra et al., 2012; Pathan et al., 2019).

Another group of proteins related to translation identified in the EVs are elongation factors 1 (EF1) and 2 (EF2). They are vital proteins involved in the elongation phase of translation. EF2 is a very abundant protein in the cell and is also the top protein identified in EVs derived from proteomic studies in distinct species (Kalra et al., 2012; Pathan et al., 2019).

Helicases are another class of proteins involved in RNA metabolism (Linder and Jankowsky, 2011). The DEAD/DEAH protein family, found in all kingdoms, is the largest family of RNA helicases and is involved in various RNA metabolic steps such as RNA synthesis and folding, RNA-RNA interactions, RNA degradation, and localization (Gilman et al., 2017). Several RNA helicases have been described in fungal EVs, listed in Table 3 .

RNA-binding proteins are essential in most, if not all, stages of the RNA life cycle. Distinct structural features and RNA-binding mechanisms characterize several types of RNA-binding domains. The RRM domain is the most common RNA binding domain, with each domain being specific for RNA sequences and interacting with single-stranded RNA molecules. Proteins with RRM domains are involved in all steps of RNA metabolism. Many fungal species have RNA-binding proteins with RRM domains enclosed in EVs, as listed in Table 4 . One example is the polyadenylate-binding protein (PABP), a conserved protein that plays a crucial role in RNA metabolism.

Zinc finger proteins constitute a large family of proteins primarily associated with DNA binding but also capable of binding to RNA and small molecules, such as the CCCH zinc finger proteins involved in regulating RNA metabolism. Proteomics data in Table 4 lists zinc finger domain-containing proteins found in EVs derived from the pathogens reviewed in this study, with functions related to RNA binding, RNA metabolic processes, and rRNA binding. The RNA polymerase II transcription factor contains a zinc finger RNA binding domain. This protein modulates DNA-templated transcription and participates in the assembly of RNA polymerase II preinitiation complex, forming the first bonds in the RNA chain.

The hnRNP K-homology (KH) domain is one of the most prevalent RNA binding domains and is found in proteins responsible for regulating gene expression in prokaryotes and eukaryotes. In addition, this domain binds specific sequences on nucleic acids ( Table 4 ).

To determine the RNA pathways enriched in fungal EVs and to compare them among the species, we performed a gene ontology (GO) term analysis with each proteomic dataset ( Figure 2 ). Proteins related to translation were the most prominent category, followed by proteins functioning as translation initiation factors and proteins involved in translation elongation. Interestingly, the subsets showed no enrichment of proteins involved in non-coding RNA (ncRNA) biosynthesis or processing machinery. However, most RNA molecules present in the EVs belong to the ncRNA class (Peres da Silva et al., 2015).

Next, we conducted a comparative analysis of the proteins present in the EVs of various fungal species, aiming to identify orthologs. Through this analysis, we observed the formation of forty-nine protein clusters, each comprising at least three species ( Supplemental Table 3 ). A total of 22 proteins were present in at least five species studied ( Figure 3 ). Among the proteins with the highest number of orthologs, we identified: the Elongation factor 1-alpha, with a total of 10 orthologs in C. albicans, H. capsulatum, C. neoformans, C. deuterogattii, P. brasiliensis, and A. fumigatus, with at least 84% similarity among them; the ATP-dependent RNA helicase eIF4A/Helicase FAL1 with eight orthologs in H. capsulatum, C. neoformans, C. deuterogattii, C. albicans, and A. fumigatus, with at least 68% similarity; and Elongation Factor 2 with seven orthologs in C. albicans, A. fumigatus, C. neoformans, C. deuterogattii, H. capsulatum, and P. brasiliensis, with at least 75% similarity among the proteins. The top clusters are listed in Table 5 .

The relevant finding is that many proteins participating in different steps of RNA metabolism were conserved among other fungal species, such as eukaryotic translation initiation factors, helicases, elongation factors, and tRNA synthetases. Among the proteins containing known RNA binding domains, we observed the Polyadenylate-binding protein (PAB1), RRM-containing protein, KH domain RNA-binding protein, and nuclear and cytoplasmic polyadenylated RNA-binding protein (pub1).

These observations reinforce the hypothesis that RBPs, or even mRNPs, drive the selection and direction of RNA to the EVs, as many of them are conserved in phylogenetically distant fungal species. A summary of the RNA life cycle, the main proteins identified in the EVs, and the connection between the RNA metabolism and extracellular vesicles is depicted in Figure 4 .

The data analyzed in this work was collected by the database that gathered EV proteomics from nine fungal species (Parreira et al., 2021). We selected the proteins annotated with the following terms: RNA-binding protein, RNA metabolism, translation, transcription, RNA helicase, and RRM. Next, we performed an alignment for the proteins with unknown functions using BLASTp to identify proteins with RNA-related roles (Altschul et al., 1997). We removed the ribosomal proteins from the analysis to avoid bias toward translation. We used the Uniprot and GeneDB databases to analyze function prediction and other features. For the proteins annotated as hypothetical or uncharacterized, we also searched for conserved domains at Uniprot, PFAM, and Interpro that could be associated with RNA metabolism. We also performed a protein BLAST analysis using the default parameters (Altschul et al., 1990).

The GO term analysis was carried out in the database: FungiFun v. 2.2.8, available at: https://elbe.hki-jena.de/fungifun/. Using the parameters: significance test – hypergeometric distribution, test gene for category associations – over representation; adjustment method – Benjamimi-Hochberg procedure; annotation type – use Only directed annotated associations; GO advanced options – all; filter by evidence code – select all evidence codes. We then selected the terms with an FDR (Benjamimi-Hochberg) below 5%.

We used the CD-HIT v.4.8.1 software tool to cluster protein sequences from our dataset (Li and Godzik, 2006). The input file, containing all protein sequences, was formatted in FASTA, and clustering was performed with an identity threshold of 50%, a coverage cutoff of 80%, and a word size of 3. Finally, the cluster representative proteins were isolated using a custom Python script. The resulting non-redundant protein sequence set effectively represents the distinct clusters obtained from the clustering process.