The EV cargo consists of nuclear and mitochondrial DNA, while the EV transcriptome is more diverse and includes different coding and non-coding RNA species (Malkin and Bratman, 2020; Veziroglu and Mias, 2020). The EV DNA and RNA cargo reflect heterogenic EV genomic and transcriptomic profiles of parental cells, and this feature can be used to characterize the EVs as biomarkers and future clinical diagnostic tools (Lázaro-Ibáñez et al., 2019). How the nucleotide sequences are packed into EVs is not well understood, but published studies report different mechanisms involved in nucleotide sequence EV sorting. While the EV cargo release is an energy-consuming process, the cells generally control EV release and cargo loading of specifically selected nucleotide sequence transfer to targeted cells (Malkin and Bratman, 2020; O’Brien et al., 2020). However, the EV DNA and RNA packing can also be non-directed and result in a similar transcriptomic and genomic profile as the parental cell.

EV DNA is mainly studied in apoptotic bodies, where during the apoptosis process, degraded cell components, including degraded genomic DNA, are packed in apoptotic bodies in the form of short DNA fragments of a few hundred nucleotides (Malkin and Bratman, 2020). The apoptotic body clearance is determined by the apoptotic body diameter, where smaller apoptotic bodies have a longer life span than larger ones, and can sustain in circulating body fluids longer and can transfer EV DNAs to targeted cells. Most apoptotic body EVs and their DNAs are cleared by phagocytosis by cells of the immune system or neighboring cells quickly after cell apoptosis (Malkin and Bratman, 2020). The intensive release of apoptotic bodies is present in late-stage tumors, where hypoxic environments promote the release of more extensive apoptosis- and necrosis-associated onko-EVs with larger DNA fragments (Vagner et al., 2018). These onko-EV degradomes often harbor somatic pro-oncogenic genomic aberrations such as single nucleotide mutations and copy number variations in TP53, KRAS, and PTEN genes frequently altered in cancer and metastasis (Bergsmedh et al., 2001). The transferred onko-EV DNAs can promote horizontal onko-gene transfer, associated with spreading tumor mutations, promoting a pro-oncogenic and pro-metastatic environment in other tissues and harboring novel mutations (Bergsmedh et al., 2001; Kawamura et al., 2017; Amintas et al., 2021).

The reporting of genomic DNA fragments in EVs within the diameter of exosomes and microvesicle EVs is inconsistent. Several studies report the presence of DNA in all subgroups of EVs; some of them report that DNA might not be associated with the parental cell release, but as DNA bonded outside on the surface of EVs in the extracellular space, body fluids, or during sample processing (Lázaro-Ibáñez et al., 2019; Elzanowska et al., 2021; Hur and Lee, 2021). Even if DNA is bonded on the surface of EVs, it can be biologically active and can have its biological function. The EV-transferred DNA can regulate the immune system via TLR9 receptors, which detect non-methylated CpG-containing DNA delivered by the endocytotic pathway as part of the anti-pathogen response mechanism (Hemmi et al., 2000) or potentially even modified human DNA (Ye et al., 2017). While the normal tissue cells include germline DNA, with attained somatic variants, the cells and EVs also carry different DNA epigenomic information, which is tissue-specific and contains methylated, oxidated, and other nucleotide modifications produced by cell enzymes and reactive chemical species (Mannavola et al., 2019). Cell and EV DNA epigenome with modification differences reflect environmental and tissue physiological specificity that needs to be investigated.

Cell transcriptome includes diverse coding and non-coding RNA molecules, which are specific for cell types and tissues. The transcripts are primarily generated with the DNA transcription in the nucleus, where primarily transcripts are processed and trimmed. In the following steps, the transcripts are transported to the cytoplasm, where the RNA molecules are post-transcriptionally processed and modified to the active form of coding or non-coding RNA, which can perform their functions in a cell or can be directed into EVs for EV cell release (Tosar et al., 2021). The heterogeneous cell- and tissue-dependent RNA profile is reflected in EV transcriptome heterogenicity (O’Brien et al., 2020). All RNA species were reported in the EV cargo. However, due to the thermodynamically favorable processes, only small RNA molecules are preferentially abundant in the EV cargo. RNAs are packed into EVs by selective and non-selective EV nucleotide sequence-loading mechanisms (O’Brien et al., 2020; Veziroglu and Mias, 2020). While some studies report “naked” RNA EV loading, the other results suggest RNA EV-directed loading with RNA-binding proteins, even as active regulatory complexes (O’Brien et al., 2020; Veziroglu and Mias, 2020).

Although the EV transcriptomic profiles are concentrated in short RNA sequences, EVs intact mRNA and long ncRNAs (lncRNA) were reported. Even though the mRNAs are not highly represented in EVs due to their size, EV mRNAs were reported in several studies which suggest the intracellular horizontal cell-coding transcriptome transfer (Cha et al., 2020; Bracht et al., 2021). EV lncRNAs, longer than 200 bps with no apparent coding functions, were investigated as biomarkers and mediators involved in oncogenesis as well as tissue repair and regeneration (Kogure et al., 2013; Born et al., 2020). Only a few studies report the EV presence of the larger rRNA units (18S and 28S), while it is inconclusive if these molecules are present due to non-EV RNA contaminations or parts of RNA degradome produced in cell necrosis or apoptosis. Nevertheless, smaller rRNA units, 5S and 5.8S rRNA with a length up to 150 nucleotides (Tosar et al., 2020), whose function is an enhancement of the protein synthesis and stabilization of a ribosomal structure, are commonly reported within EVs (Szymanski et al., 2002). Most of the EV transcriptomic profile belongs to small non-coding RNAs (sncRNAs), where the most abundant sncRNAs are miRNA, yRNA, siRNA, piRNA, and tRNA represented as mature sncRNAs, their precursors, or isoforms. The abundance of sncRNA types varies and is highly related to the sample type, sample processing, and the EV isolation methodology.

The EV research is mainly directed into EV miRNA (microRNA) due to the well-characterized miRNA regulatory function in the cellular transcription and regulation of the mRNA expression to coding proteins. However, miRNA function is even more diverse. miRNAs are 16–24 nucleotides long sncRNA, coded by the miRNA genes which are in the human genome often located in miRNA gene clusters. 2,654 miRNA human sequences and 1,917 miRNA precursors are identified so far and their sequences are listed in miRBase (http://www.mirbase.org/; Release 22.1) (Kozomara et al., 2019). It is estimated that miRNAs can regulate the translation of more than 60% of human mRNAs to proteins (Friedman et al., 2009). The miRNA synthesis starts in the nucleus by polymerase II transcription, where pre-miRNA transcripts are formed. In the nucleus, pre-miRNAs are cleaved with the Drosha-DGCR8 complex and a stem-loop structure, which is transported in the cytoplasm by the exportin complex, is formed. In the cytosol, pre-miRNA is diced by the Dicer protein complex to 3′ and 5′ miRNA strands, and other non-template nucleotide additions (NTA) can be added on the 3′ and 5′ ends of miRNA. One of the formed miRNA strands (5p or 3p) is degraded, while the other miRNA strand is incorporated into Ago proteins and forms the RISC complex (O’Brien et al., 2018). The RISC-miRNA complex can bind on mRNA, mostly on 3′ mRNA and can be involved in the translation regulation with protein translation silencing or mRNA degradation. Given the knowledge of miRNA and prevalence of EV miRNA studies, EV miRNAs have been investigated in most diseases and pathological conditions in the RNAs’ mature form, isoforms, or precursors isolated from different liquid biopsy–obtained samples. Blood EVs miRNAs miRNA-92a and miRNA-222 are overexpressed in metastatic colorectal cancer patients associated with a lower survival rate (de Miguel Pérez et al., 2020). miR-127-3p, miR-155-5p, miR-21-5p, miR-24-3p, and let-7a-5p are increased in plasma EVs of patients with classical Hodgkin lymphoma, convenient even for detection of small residual lesions (Drees et al., 2021), while some of these miRNAs even have a potential for therapy response monitoring (van Eijndhoven et al., 2016). Urinary EVs’ isomiRs of miR-21, miR-204, and miR-375 are highly differentially expressed in prostate cancer individuals (Koppers-Lalic et al., 2016), while urinary EV-increased expressions of miR-29a-3p and miR-200a-3p are altered in patients with nephropathy in Fabry disease (Levstek et al., 2021). The altered EV miRNA levels are also reported in autoimmune diseases such as systemic lupus erythematosus, Sjogren’s syndrome, and type 1 diabetes (Michael et al., 2010; Katsiougiannis et al., 2019; Tesovnik et al., 2020; Foers et al., 2021).

Piwi-interacting RNA (piRNA) regulation molecules are produced by similar synthesis, and have similar silencing mRNA translation and protein expression mechanisms as miRNA. This species of 26–31 nucleotides long sncRNA is highly expressed in germline cells and embryos, where pi-RNAs are involved in silencing of transposable elements, the epigenetic regulation, while in adult humans, piRNAs are essential in the process of spermatogenesis. EV piRNAs were investigated in seminal plasma EVs, where piRNAs are associated with spermatogenic ability (Chen et al., 2021), while multiple myeloma-derived EV piRNAs promote tumorigenesis through re-educating endothelial cells in the tumor environment (Li et al., 2019). Vesicle-related piRNAs are also reported as biomarkers of cardiovascular disease and diabetes and even as factors with a potential involvement in cardiovascular disease development through piRNA/LINE-1/AKT, AKT, and AMPK signal pathways (Zeng et al., 2021).

Another sncRNAs reported in EVs are yRNAs, which are regulators of the DNA replication (Kowalski and Krude, 2015) and are together with RNA-binding protein Ro60 clinical important targets of autoantibodies in systemic lupus erythematosus and Sjögren’s syndrome (Boccitto and Wolin, 2019). yRNAs are in their mature form abundant in plasma immune cell-released EVs and are investigated as biomarkers. EV-altered plasma subtype yRNA ratios are reported in inflammatory disease (Driedonks et al., 2020) and carcinogenesis (Haderk et al., 2017).

EV transcriptome also includes tRNAs and tRNA fragments (tRFs), reported as highly represented in T-cell–released EVs (Gámbaro et al., 2020). Even though the cellular function of tRNA fragments is not clear, it is reported their upregulation in cellular stress and EV tRNAs can be potentially involved in target cell signaling and immune system cell regulation (Chiou et al., 2018; Tosar et al., 2020).

Similar to miRNA, other sncRNAs can be loaded into EVs together with RNA binding with Ago proteins or Y-box binding protein 1 (O’Brien et al., 2020; Ghoshal et al., 2021), which are the mediators of EV loading. The sncRNA EVs preferential directioning is reported being related to the RNA structure and nucleotide sequence motifs. Studies also report sncRNA exporting into EVs could be 5′ and 3′ nucleotide sequence specific, where miRNA sequences with specific NTA can be preferentially loaded into EVs (Koppers-Lalic et al., 2014).

In targeted cells, EV RNA molecules modulate transcription regulation via miRNA-like silencing mechanisms, with binding on the mRNA transcription regulatory sites, other noncoding RNAs, or intracellular receptor mechanisms. If the EV RNA is released in the cytosol of the targeted cell, the released RNA can activate RNA-sensing receptors. Besides cytosol RIG1 receptors (Boelens et al., 2014), the EV RNA can activate endosomal intracellular RNA tool–like receptors (TLR), which are also part of the antimicrobial innate immune defense system. Endosomal TLR3 can detect dsRNA motifs enriched with I:C, while TLR7 and TLR8 can detect ssRNA with enriched U and GU sequence motifs (Hornung et al., 2005, 7; Forsbach et al., 2008) in phagocytes and can transmit an activation signal to the effector cells of the immune system. How the human intracellular RNA-sensing receptors distinguish between human and foreign RNA remains unclear. The answer might be in the RNA modifications. In the literature, there are reported over 100 RNA nucleotide modifications, which can suppress or enhance the RNA stability, modify the nucleotide sequence target binding activity, or can have different affinities for the immune system regulatory receptors (Boccaletto et al., 2018; Xuan et al., 2018). tRNA nucleotide modifications have been studied to some extent. It has been reported that tRNA nucleotide sequence modifications can directly affect the tRNA structure, stability, aminoacylation, protein translation accuracy, and translation rate, and these RNA modifications associate with cell stress and cancer (Endres et al., 2019). The biological function of other sncRNA nucleotide modifications is not well understood. However in vitro studies show that miRNA nucleotide modifications can increase mRNA translation with the decreased miRNA regulatory function (Lockhart et al., 2018).

There are plenty of strategies for EV characterization, and they have their unique advantages and disadvantages. Therefore, a study setup with a standardization step is essential to perform a successful EV study outcome. Compared to tissues and cell cultures, the methods for detecting human liquid biopsy–isolated EVs deal with relatively small quantities of EV material. Consequently, the EV DNA and RNA material yield is often the most significant limiting factor that researchers need to challenge (Brennan et al., 2020). EV characterization and biomarker discovery include sample collection, sample processing and storage, EV isolation and characterization, EV cargo analysis, and data interpretation. All the analytical steps can induce variability in EV analysis and can have a significant impact on the final result. Therefore, it is crucial to include homogeneous groups of subjects in the study.

Different EV isolation strategies result in considerable differences between the EV final yield and purity (Konoshenko et al., 2018) and thus, reflect in non-reproducible EV transcriptomic and genomic profiles (Figure 1). There is no such thing as a perfect method for EV isolation. The EV isolation procedure cannot be standardized for the research applications due to the diversity of the biological samples. However, EV multiple diagnostic applications show strong tendency to standardize the EV isolation. The isolation procedure should be carefully selected in the phase of study design, and it should be adjusted to the sample type, complexity of body fluid, its volume, and required EV yield and purity to provide a sufficient EV material for the downstream analysis (Brennan et al., 2020). Due to a variety of EV procedures, the International Society for Extracellular Vesicles (ISEV) organization addressed this problem and suggested the standardized reporting of the EV isolation procedure to provide reproducible and reliable EV results (Oeyen et al., 2019).

The human EV analysis starts with the human sample collection (Figure 1). In clinical practice, a least invasive sample collection is preferred. The human liquid samples of studied subjects are collected with a sterile blood draw, cerebrospinal liquid, amniotic liquid, or any other body fluids as urine or human milk (Foster et al., 2016; Bernardi and Balbi, 2020). The most commonly used starting liquid biopsy material for tracking physiological processes in a human body is the blood due to the constant environment, low-variability of human blood parameters, and relatively non-invasive collection. To preserve cells and blood plasma EV amounts in collected blood, it is advised to draw blood with anticoagulant tubes with citrate, heparin, or EDTA (Witwer et al., 2013). On the contrary, serum samples are not recommended for EV biomarkers discovery due to the platelet activation and platelet-released EVs, which can reduce the detection and sensitivity of the non–platelet derived EVs (Witwer et al., 2013; Palviainen et al., 2020). A hemolyzed and icteric sample rejection from the analysis is recommended in order to avoid sample heterogeneity, where additional protein and lipid contaminations and other cellular non-EV components can increase the amount of sample non-EV nucleotide sequences (Coumans et al., 2017). After the biological sample collection, the procedures of EV pre-isolation include multiple steps, starting with the cell removal with low speed centrifugation. In further steps, additional centrifugation at higher speeds or filtration through syringe filters is used to remove cell debris or larger protein complexes. After sample pre-processing, the processed samples are stored with deep freezing or a direct EV isolation procedure follows.

The EV isolation with an ultracentrifuge at ultra-high speeds is a gold standard for the EV isolation procedure, even though this is a time-consuming process (Momen-Heravi, 2017). Ultracentrifugation provides a low output yield, and the isolated EVs are not highly pure due to the EV co-isolation of larger protein complexes. The purity of EVs can be improved by sucrose or iodixanol gradient density ultracentrifugation, providing highly pure EV fractions with less protein contaminants, but the multiple steps of centrifugation with wash steps and EV concentration are also related to higher losses of EVs (Onódi et al., 2018; Bernardi and Balbi, 2020). Besides the impurity limitation, the obstacle of this method is often the high cost of the ultracentrifuge itself, and thus, not likely to be affordable to an average clinical research laboratory. However, there are alternatives offering more straightforward and more elegant EV isolation methods.

The precipitation reagent EV isolation is based on the addition of volume excluding polymers, where non-covalent interactions between anionic polymers and positive charge of EV membrane lipids form EV precipitates (Rider et al., 2016). The precipitated EVs are collected by centrifugation with a bench laboratory centrifuge at a lower speed up to 10.000 g. The most commonly used precipitation reagents are polyethylene glycol (PEG) polymers with different molecular masses. The PEG EV isolation yields in high EV concentrations with their wide diameter ranges in the range of exosomes, microvesicles, and small apoptotic bodies (García-Romero et al., 2019). However, with PEG, isolated EVs are co-precipitated with larger protein complexes and other non-EV particle contaminations. EVs can be isolated with commercial PEG reagent mixtures or following published protocols for PEG reagents suitable for EV isolation.

Over the past few years, there was a widespread of the EV isolation procedure with commercial size exclusion chromatography (SEC) columns, where vesicles are separated on a column by the EV diameter (Gámez-Valero et al., 2016; Lane et al., 2019). The SEC-isolated EV yields with intact EVs and with fewer contaminants than the PEG procedure and ultracentrifugation. The EV separation columns are classified for multiple uses, suitable for isolation of several samples after the wash, and column regeneration. Another similar EV isolation approach offers EV concentration or isolation with centrifugal molecular weight size cut-off spin concentrating membranes, where EV suspension concentrates on the membrane’s surface, while smaller particles and proteins are removed through the membrane (Benedikter et al., 2017; Stranska et al., 2018). Some commercial centrifugation spin membranes allow EV adsorption on the membrane surface in the presence of a binding buffer. After a wash step, the EVs are eluted or directly lysed on the membrane, and the lysate is collected for further investigation.

Tissue-specific EVs can be isolated based on specific surface proteins with immunoprecipitation, where antibodies attached on immunoprecipitation columns or magnetic beads specifically bind to EV membrane protein markers (Pedersen et al., 2017). Due to the required specific monoclonal antibodies, the isolation procedure is relatively costly and does not result in large amounts of EVs. The immunoprecipitation-isolated EVs typically do not result in high enrichment of specific EVs due to non-specific interactions with antibodies or solid-state surface interactions. There were also reports of EV isolation on liquid laminar flow microfluidic chips with the acoustic driven separation. The method has a huge potential for EV isolation with the option of EV selection by using the vesicle diameter directly from EV liquid biopsy media (Wu et al., 2017), but unfortunately, there is no commercially available isolation module on the market, yet.

All the EV isolation methods, some to a larger degree than others, also isolate non-EV impurity complexes. The most common isolated EVs from blood plasma contain EV low-density lipoproteins (LDL), high-density lipoproteins (HDL), and chylomicrones whose concentration is significantly increased in postprandial blood samples (Karimi et al., 2018). Therefore, fasting blood collection is advised for studying EVs. However, in addition to lipoprotein, even protein complexes as Ago proteins with miRNA can be co-isolated from blood plasma samples, which are not related to prandial cycles. In urine EV samples, the isolation is challenged by protein contamination with uromodulin (Tamm–Horsfall protein complex), which aggregates and forms large protein complexes (Musante et al., 2020), while synovial fluid EVs can be contaminated with glycosaminoglycan and proteoglycan complexes (Buzás et al., 2018). The extracellular non-EV RNA and DNA contamination can be degraded with RNase or DNase enzymes in pre-isolation sample treatment, while protein non-EV contaminants can be decreased by the proteinase treatment and chelating agents such as EDTA (Gutiérrez García et al., 2020; Malkin and Bratman, 2020). Before further EV processing and EV characterization nucleotide sequence extraction, it is advised to optimize and standardize the EV isolation procedure with assessment of the isolated EV average diameter and concentration, vesicle specific protein markers, and to estimate co-isolated non-EVs contaminants (Doyle and Wang, 2019). For EV reporting, the ISEV organization suggests EV detection and characterization by TEM imaging of single isolated EVs, quantification of the EV concentration, and the EV diameter using nanoparticle analysis. In addition, Western detection is appreciated to detect membrane vesicle proteins and protein contaminants (Brennan et al., 2020). The specificity of the isolated EV cargo can be characterized also by nucleotide sequence expression profiles and compare EVs with starting liquid biopsy samples, using the procedures described in the following paragraphs.

While there are different EV isolation procedures, there are also diverse DNA or RNA isolation techniques based on different chemistry adjusted to isolate particular molecules of interest. The EV cargo typically consists of small DNA and RNA fragments; consequently, the isolation must isolate preferentially small nucleotide sequence fragments. The amount of the isolated EV DNAs or RNAs is highly variable to the isolation procedure, and it is the limiting factor that determines the EV nucleotide sequence characterization. A reasonable conclusion is that the EV DNA or RNA isolation method must ensure minimum sample losses.

In order to ensure high purity and high yield of the isolated EV RNA or DNA, it is recommended to use the standardized, commercially available reagent kits for the EV nucleotide sequence extraction. The most common RNA and DNA isolation methods include the lysis reaction with the nucleotide sequence extraction, following precipitation and purification on the silicate columns and final elution (Ali et al., 2017). In the case of RNA, the most effective isolation method is based on phenol–chloroform extraction coupled with column purification. Phenol–chloroform extraction can isolate RNA without purification columns by precipitation using carriers as linear polyacrylamide or glycogen, but precipitation isolation cannot successfully remove phenols and salts from the isolated samples. There are also less harmful, less toxic, and more environment-friendly methods, but unfortunately, these methods provide a lower yield of the isolated EV nucleotide sequences. In the last decade, EV nucleotide sequences can also be isolated by solid-phase reversible immobilization (SPRI), based on lysis, nucleotide sequence desalting precipitation, and purification on the magnetic beads (Ali et al., 2017). The method simplifies the isolation procedure, but SPRI nucleotide sequence precipitation and binding require high salt concentrations, which cannot be removed within the following wash steps successfully. Furthermore, salt impurities can interfere in the downstream analysis. A few DNA/RNA isolation procedures suggest using spike-in synthetic RNA or DNA oligos to evaluate the success of the isolation procedure and allow nucleotide sequence–related RNA quantification with the assessment of fold change trueness (Everaert et al., 2019). Because spiked-in sequences reduce the available sequencing capacity, it is not advised to use the spike-in reference in the studied samples during nucleotide sequence isolation if it is planned that the EV nucleotide sequence profile will be detected by next-generation sequencing (NGS).

The EV nucleotide sequence content is still being discussed, but studies are unified in terms of a low EV nucleotide sequence quantitative amount. It was estimated that only one per 100 EVs contains miRNA (Chevillet et al., 2014; Albanese et al., 2020). However, it is hard to quantify the EV nucleotide sequence cargo due to biological sample variability, sample processing, EV isolation, and nucleotide sequence isolation losses as well as technical variability in EV nucleotide sequence detection. The amount of the isolated RNA or DNA and quality control with protein and salt contaminants can be assessed using UV-absorbance or fluorometric spectrophotometric methodology or qPCR. The isolated EV DNAs/RNAs can be additionally characterized with their size distribution and quantity, by electrophoresis on a chip. EVs’ short nucleotide sequence RNA electrophoresis profiles normally show nucleotide sequences in range up to 200 nucleotides (Figure 2). Potential 18S and 28S ribosomal rRNA peaks can be observed, but these are not specific for EVs and can indicate cell debris contamination (Tang et al., 2017). While cellular RNA and DNA quality is estimated with the RNA integrity number (RIN) and DNA integrity number (DIN), these parameters cannot evaluate the EV nucleotide sequence profile due to their short nucleotide sequence lengths.

Using quantitative PCR (qPCR), also known as real-time PCR, only the selected known sequences can be detected. EV sequences of interest can be selected based on the predetermined human organ and tissue expression data or can be predicted by known regulated pathways that are involved in the physiological and pathological conditions (Bellingham et al., 2017). The conventional RNA qPCR detection consists of the complete transcriptome reverse transcription with random hexamer oligos, following specific target amplification in DNA qPCR detection with specific oligonucleotides together with the reporting probe or reference dye in the amplification reaction. While the EV DNA and longer RNA molecules detection do not differentiate from the conventional qPCR, there are some differences in the detection of sncRNA. There are two basic qPCR approaches of miRNA detection, which are useful also for other short fragment sncRNAs detection (Figure 3). The stem-loop–mediated qPCR reaction includes a specific stem-loop reverse-transcription primer, which anneals with up to six nucleotides to the targeted miRNA and allows reverse transcription (Chen et al., 2005). The cDNA products are analyzed by PCR amplification, where specific primers align on the cDNA–miRNA complex sequence together with the reporting dye or specific probes, which complementary align on the transcribed miRNA sequence and form a detectable product and reflect the presence and quantity of miRNA. The disadvantage of the stem-loop method is the complexity of the primer design, reverse transcription variability due to stem-loop oligos, and limiting multiplexing options of different miRNA detection in the same reaction. The universal miRNA qPCR can be used to overcome the multiplexing problem and differences in the transcription efficiency. The universal miRNA and qPCR miRNA sequences are elongated by polyA polymerase, forming 3′ polyA tail, while on 5′, the universal oligo is ligated. The detection of miRNA of interest is performed by the qPCR technique, where after the reverse transcription in the PCR amplification reaction, universal oligos and specific probes are aligned on the miRNA cDNA transcript and allow its detection (Niu et al., 2015). The advantage of the universal miRNA qPCR reaction is in more uniform cDNA synthesis, which requires less sample material for multiple miRNA qPCR analyses in comparison to the loop-mediated qPCR. The qPCR EV nucleotide sequence detection allows a quick detection of the predicted differentially expressed sequences. One of the limitations of the qPCR EV nucleotide sequence detection is its data reporting, where the sequences of interest are compared to standard or reference RNAs—housekeepers, that can be selected experimentally or based on the previous results. The housekeeping RNA are independently expressed in the investigated samples regarding the variability of the studied conditions. Housekeeping RNAs allow assessing and quantifying the studied RNA with the normalized expression and relative quantification ratio. To provide convincing qPCR results, it is recommended to design an experiment with more housekeeping sequences to better verify the differential expression ratios for the studied RNAs (Gouin et al., 2017). One of the disadvantages of qPCR sncRNA detection is only mature sncRNA detection using the commercial pre-designed validated probes, while the modified sncRNA with NTA and sncRNA isoforms requires the custom probe design and PCR program standardization based on the oligo and probe melting temperature. Most qPCR methods quantify the nucleotide sequences with delta–delta–Ct and report the differential expression of the target nucleotide sequences as fold difference in the expression related to the expressed housekeeping sequence. A more precise absolute concentration of the investigated nucleotide sequence molecules can be quantitatively assessed if the qPCR measurement includes standard dilutions of known concentration by which samples’ Ct values can be translated to the nucleotide sequence concentration (Lee et al., 2020). Nevertheless, the absolute qPCR quantification approach is not used in practice. Better sensitivity and reproducibility compared to qPCR and absolute quantification can be achieved using digital qPCR (Wang et al., 2019), where sample molecules are analyzed in thousands of nanoliter-scale reactions in emulsion droplets or microchambers on a microfluidic array.

The complete EV transcriptome and EV DNA profiles can be assessed only by NGS (Amorim et al., 2017). Second-generation sequencing technology was firstly supported by Roche pyrosequencing technology and by the Applied Biosystems SOLID platform. Nowadays, the improved second generation NGS, based on sequencing by synthesis, represents the most used sequencing method, and it is offered on the market by three companies, Illumina, MGi/BGI, and IonTorrent, accessible practically to every average laboratory with different throughput sequencing machines (Hu et al., 2017). The second NGS generation allows short fragment sequencing with insert size 20–500 nucleotides at a relatively low price per read nucleotide. Many variations of the sample library preparation were developed to detect nucleotide sequences and additional sequence-related information. Both kinds of EV nucleotide sequence samples of investigation should undergo sample processing or the sample NGS library preparation procedure that includes adapter sequence ligation on 5′-phosphorylated and 3′-OH ends of sample molecules with T4 ligases. The RNA preparation procedure distinguishes from DNA library preparation in terms of the additional first step of reverse transcription, where the cDNA strand is formed. With the ligation or PCR amplification, the EV sample molecules are labeled with the index sequences (barcodes), allowing parallel multiple sample sequencing in the same sequencing run. Ligated adapter and index sequences on the sample sequences form uniform product ends, enabling uniform PCR amplification and attachment on the solid-phase surface of the flow cell in the sequencing machine. The Illumina sequencing platform as the leading sequencing technology consists of synthesis with fluorescent-labeled nucleotides, where the fluorescence of the attached nucleotides is detected by an optical system in a sequencing machine and later translated to nucleotide sequences by bioinformatics algorithms. The limitation of the second NGS generation is the limited insert size of NGS library–incorporated sequences, which is not problematic for EV sncRNA detection due to EV short sequences. An additional disadvantage presents the library processing procedure, with induced bias during the ligation step due to nucleotide sequence modifications and secondary structures in the RNA or DNA sequences (Wilusz, 2015). Additionally, the PCR step removes the information about nucleotide modifications and induce non-uniformly amplification bias of the GC-rich and homopolymeric sequences (Ross et al., 2013).

To simplify the sample sequencing preparation, several library preparation kits are on the market to prepare DNA, RNA, and sncRNA libraries. As small transcriptomic RNA molecules are abundant in EVs, it is recommended that small RNA library preparation kits are used for the NGS EV transcriptome detection (Etheridge et al., 2018; Srinivasan et al., 2019). The small RNA library preparation procedure includes the final product size selection step, where products with the incorporated miRNA or piRNA can be enriched with SPRI size selection or using gel PAGE excision of the products with specified lengths. Additionally, the size selection removes adapter dimers that are formed during the library preparation and can significantly decrease the sequencing yield (Figure 4A). Alternative small RNA library options for enriched miRNA and piRNA sequencing present tRNA/YRNA and rRNA depletion kits, which can reduce other sequences of no interest. All the NGS-prepared libraries can be enriched for particular known sequences, with the specific capture probes. Additional data improvement in the NGS detection of rare DNA or RNA molecules with a low expression can be provided by additional unique molecular identifiers (UMI), where every sequence in the sample can be marked with a specific barcoding sequence (Hong and Gresham, 2017). UMI, together with particular bioinformatics procedures, can remove PCR artefacts, decrease PCR amplification variability, and improve the detection of rear events, which are essential for an early marker detection of tumors and other diseases (Hong and Gresham, 2017; Smith et al., 2017). While different genes can be coded on the opposite DNA strands of the alleles, these alleles produce different transcripts. Strand specificity of mRNA sequencing can be achieved with RNA directional sequencing, where the degradation of non-expressed RNA PCR-formed strands produces the expression data of mRNA with overlapping genes in reverse and forward genomic sequences (Yang et al., 2017).

NGS sequencing also presents the possibility to detect epigenomic modifications, which can be tissue-specific and can be very challenging in case of EV DNA and RNA. While the DNA methylation pattern detection is supported by the commercial reagents, sample preparation requires a large amount of DNA and analysis-specialized bioinformatics pipelines (Barros-Silva et al., 2018). The DNA methylation pattern can be detected with sequencing after the bisulfite conversion of non-methylated cytosine nucleotides, which are converted to thymines (Barros-Silva et al., 2018). The alternative option of methylation nucleotide sequence detection is methylation sequence enrichment with immunoprecipitation or methylation-binding domain proteins. The detection of other nucleotide sequences’ modifications, such as oxidative stress and other reactive species–induced modifications and other amino groups, are currently still being developed and need further optimization and evaluation. The RNA modification detection is more complicated due to the lack of commercial kits and standard pipelines. While DNA commercial kits include end prep reagents for repairing 5′ and 3′ ends (Kozarewa and Turner, 2011), the commercial RNA reagent kits do not include the end repairing module. The lack of RNA repair and RNA modification causes the detection of only a fraction of the sample RNA profile (Wilusz, 2015). The non–5′-phosphoryl, remained 3′-phosphoryl groups and sequence bias can reduce the adapter ligation efficiency on RNA sample molecules. The RNA-reduced ligation efficiency can be bypassed with the application of T4 polynucleotide kinase, letter induce 3′ and 5′ end repair at sample molecules and allows further adapter molecule ligation (Sorefan et al., 2012). An additional challenge for the detection of the whole transcriptome is methylated nucleotides, which can block reverse transcription, and the RNA inserts are not included in the NGS library. The obstacle of RNA-methylated nucleotides can be neutralized by demethylation with dealkylating enzymes as α-ketoglutarate–dependent hydroxylase (AlkB) (Cozen et al., 2015). These challenges point out the importance to pay attention to nucleotide sequence modifications, which have a significant impact on the final results, even from a technical perspective.

The third NGS generation is primarily designed as long-read sequencing technology for detecting DNA or RNA with several kilobase long fragments. This technology is suitable for detecting intact transcripts and providing accurate genome assemblies with successful detection even repeat sequences throughout the genome (Xiao and Zhou, 2020). Two companies offer different technologies of the third-gen NGS technology; Oxford nanopore technology (ONT) and PacBio technology. PacBio’s SMRT sequencing technology provides more accurate long-read sequencing data than ONT, but the technology requires large amounts of starting material (Xiao and Zhou, 2020); sequencing is costly, and consequently, its usage is not reasonable for short sequences mostly present in EVs. A more promising third NGS generation suitable for the EV nucleotide sequence detection is the ONT sequencing technology. A significant advantage of it is its price accessibility and its ability to detect even short nucleotide sequences with the real-time sequencing analysis together with nucleotide sequence modifications. The library preparation for ONT sequencing is similar to the second-generation sequencing, while the adapter and index molecules are attached on the 3′-end of the sample sequence together with helicase proteins, which unwinds the double-stranded native or duplicated DNA or RNA/DNA reverse transcriptase–formed hybrid complex. When unwinded nucleotide sequences are directed through protein pores on the flow-cell membrane, sequences change permeability through pores and affect the electric current that is translated to the nucleotide sequences (Xiao and Zhou, 2020; Zhang et al., 2020). The third-generation sequencing technology is under intensive development. With the improvement of sequencing hardware, sequencing chemistry, and improved library preparation, new NGS techniques will support low input material analysis applications, suitable even for EV analysis. Along with the development of bioinformatics tools, the possibility of determining EV nucleotide modifications is also expected.

The knowledge of bioactivity of specific and differentially expressed nucleotide sequences is essential to specify the effect of the investigated EV nucleotide-targeted sequences in targeted cells and organs in vivo. Additionally, the EV nucleotide sequence role is essential for conformational in vitro studies to promote the development of the EV-related therapeutics. Bioactive EV DNA delivered to targeted cells can be transferred to cytosol or nuclei, transcribed to RNA, and later translated in cytosol from RNA to proteins (Waldenström et al., 2012; Cai et al., 2013) and can potentially change the phenotype of the cell. Although the EV DNA and mRNA function can be predicted from the coding sequence, the non-coding RNA effect prediction requires in-depth additional analysis.

The biological EV sncRNA regulatory function can be predicted by tools that find miRNA predicted and validated targets based on the sncRNA and 3′ mRNA interaction. Some of these online or standalone tools, e.g., miRNA TargetScan (Agarwal et al., 2015), miRANDA (Enright et al., 2003), miRDB (Chen and Wang, 2020), and miRWalk (Dweep et al., 2014), predict interaction targets based on different algorithms searching for 3' or central domain mRNA–targeted complemental sncRNA sequences. Bioinformatics tools can predict the targeted antisense sncRNA-mRNA/miRNA interaction based on the organism transcriptomic data (Thorsen et al., 2012; Nguyen and Chang, 2017), and with the predicted mRNA silencing enrichment pathway analysis, the phenotypic effect can be predicted. The biological effect of differentially expressed sequences can be estimated by the pathway analysis that can be performed with freely available software, such as Cytoscape, g:Profiler, GSEA, and EnrichmentMAP, with modules for the analysis of the affected gene lists (León and Calligaris, 2017; Reimand et al., 2019). Experimentally defined differentially expressed sequences are compared to a hierarchically organized gene or RNA-expressed lists and sets of standardized terms of biological processes, molecular functions in cellular components, or tissues. The pathway analysis result is an altered cellular pathways, which can be associated with physiological and phenotypic changes (Reimand et al., 2019).

The prediction tools can just estimate the EV nucleotide sequence effect on target cells, and the in vitro study on cell cultures or in vivo on animal models are essential to confirm these findings of EV nucleotide sequences in the predicted processes. Differentially expressed nucleotide sequences can be synthesized and exposed to the studied cells or organisms utilizing synthetic vesicles, and the predicted effect should be characterized by the in vitro or in vivo study, where phenotype changes can be assessed by RNA sequencing, single cell sequencing, mass spectrometry, flow cytometry, or imaging techniques. In order to estimate the immunomodulating effect of EVs delivered nucleotide sequences, the in vitro study can be conducted, with nucleotide sequences delivered by the endolysosomal pathway or released in the cytosol, activating nucleotide sequence–sensitive receptors (Cardarelli et al., 2016; Tesovnik et al., 2020) or changing translation and transcription (Figure 6).

First studies of EV therapeutics show promising results as cancer therapeutics, especially in the field of cancer treatment. With the genetics and transcriptomic background knowledge of diseases, the EVs loaded with custom nucleotide sequences present enormous potential as a next-generation drug delivery platform for developing target-directed therapeutics (Herrmann et al., 2021). The use of vesicles as communication proxies’ therapeutic agents targeting specific cells is intensively investigated. Many studies investigated vesicle-delivered siRNA and miRNA for therapeutic application already in the pre-clinical trials and some even in clinical trials (Wiklander et al., 2019; Han et al., 2021). The preclinical studies show the antitumorigenic and antimetastatic activity of EVs isolated from cultured cell lines, macrophages, dendritic cells, and EV sonification loaded with small interfering RNA (siRNA) or miRNA in case of lung, ovarian cancer, and pulmonary metastasis. Nevertheless, allogenic mesenchymal stem cell EVs (exosomes) loaded with miRNA-124 are in clinical trial for treatment of cerebrovascular disorders (NCT03384433) and conditions after stroke, while KRAS G12D siRNA–loaded mesenchymal stem cell EVs are investigated for the therapeutic possibility in case of pancreatic adenocarcinoma (NCT 03608631).

Several procedures allow the production of EVs with specific nucleotide sequence loading for potential therapeutic application. The nucleotide sequence–loaded EVs can be produced in vitro after cell plasmid transfection, where transfected cells release EVs loaded with specific nucleotide sequences. Other EV nucleotide sequence loading options are EV incubation with nucleotide sequence suspension with physical force such as shaking, electroporation, or fractioning EVs, where nucleotide sequences can overpass the lipid membrane. EVs with specific nucleotide sequences can be also produced by gentle EV transfection with the isolated EVs and cholesterol-modified nucleotide sequence incubation or nucleotide sequence incubation with the transfection reagent. However, vesicle-like particles for nucleotide sequence delivery by the endolysosomal pathway can be also produced by transfection reagents like DOTAP and lipofectamine (Han et al., 2021). The cells or tissue ex vivo isolated vesicles prepared with loading procedures produce the EVs with nucleotide sequence cargo with preserved original membrane protein components and can be used only for autogenic and allogenic in vivo transplantation (Hocine et al., 2019; Wiklander et al., 2019). While EVs prepared from ex vivo cell or tissue cultures present personalized medicine application, the preparation procedures are expensive and time-consuming and cannot be used for general use due to antigen complexes. The EV non-allogenic protein content can induce immune EV rejection or can provoke the immune system destruction of cells that engulf EVs. Due to the desired general use for the entire population, therapeutic EVs with low antigen content are under development. The future of therapeutic EVs is in synthetic vesicles, which have already been used in clinical practice. Synthetic RNA lipid vesicles and technology of the lipid nanoparticle delivery system have been approved for clinical application by FDA first in 2018. ONPATTRO^® (patisiran) includes siRNA-targeting transthyretin, used for the treatment of polyneuropathies induced by hereditary transthyretin amyloidosis, which was first-of-its kind targeted RNA-based therapy to treat a rare disease (Hu et al., 2020). The technology became useful also for human vaccines, as the vaccine against SARS-CoV2 for preventing the SARS-CoV-19 pandemic. The synthetic vesicle platform showed the possibility for quick intervention and possible therapeutic adjustment in response to new pathogenic variants. Vesicles derived RNA with the modified nucleotides is capable of avoiding the innate immune system activation by cellular RNA-sensing receptors and promoting the expression of pathogen-like antigens against whom the adaptive immune system produces antibodies and form humoral protection against the pathogen. Vesicle delivery RNA technology is already investigated to protect against many other diseases, such as seasonal flu, respiratory viral infections (An et al., 2021), malaria, and cancer.

While the synthetic vesicles are in use for the delivery of mRNA in vaccines, the vesicles can also be loaded with human RNA to substitute cellular-damaged mRNA sequences due to genomic mutations in coding genes. Futhermore, Lainšček et al. showed EV delivery of the CRISPR/Cas system with the transcriptional regulator in the in vivo mice animal model, enabling activation of gene transcription (Lainšček et al., 2018). In addition, the EV therapeutic delivery by membrane protein or lipid EVs markers will allow EVs to direct the therapeutic cargo to tissues and cells, which represents the future of vesicle-delivered nucleotide sequence precision medicine.