Guanine-Rich Sequences Are a Dominant Feature of Exosomal microRNAs across the Mammalian Species and Cell Types

Exosome is an extracellular vesicle released from multivesicular endosomes and contains micro (mi) RNAs and functional proteins derived from the donor cells. Exosomal miRNAs act as an effector during communication with appropriate recipient cells, this can aid in the utilization of the exosomes in a drug delivery system for various disorders including malignancies. Differences in the miRNA distribution pattern between exosomes and donor cells indicate the active translocation of miRNAs into the exosome cargos in a miRNA sequence-dependent manner, although the molecular mechanism is little known. In this study, we statistically analyzed the miRNA microarray data and revealed that the guanine (G)-rich sequence is a dominant feature of exosome-dominant miRNAs, across the mammalian species-specificity and the cell types. Our results provide important information regarding the potential use of exosome cargos to develop miRNA-based drugs for the treatment of human diseases.


Introduction
Exosomes are extracellular vesicles, ranging in size from 40 to 150 nm in diameter, which are released from variety cell types by the exocytotic fusion of multivesicular bodies of the endosome with the plasma membrane [1]. Proteins and lipids are the major components of exosome membranes. Proteins on the exosome membranes are thought to function as ligands for interactions with target cells. In addition, various nucleic acids, including mRNAs and micro-RNAs (miRNAs), are found in the exosomal lumen [1,2]. Evidence suggests that miRNAs in exosome cargos are able to modulate appropriate neighboring cells or distant recipient cells [1][2][3], however the exact molecular mechanisms of endocytosis and the specific interaction between exosomes and target cells is yet to be clarified. miRNAs are small (17-24 ribonucleic acids), non-coding RNAs (located mainly within genomic introns) that regulate post-transcriptional gene silencing by binding to the 3'-Declaration of Helsinki. Written informed consent was obtained from all healthy donors participating in this study. The experimental protocols were approved by the Ethics Review Committees of the Mie University Graduate School of Medicine (Approval No.: 2879).

Isolation and purification of exosomes
Exosomes were prepared from culture supernatants by differential centrifugation and filtration. Cell cultures were subjected to centrifugation at 400 g for 10 min and 10,000 g for 20 min to remove cells and debris, and then filtration through 0.45 μm and 0.22 μm filters (Merck Millipore). Exosomes were purified from the culture supernatants by ultracentrifugation (Optima L-90K, Beckman Coulter; SW28 rotor) at 120,000 g for 70 min at 4°C, and then washed with phosphate-buffered saline (PBS). Finally, the exosome pellets were dissolved in PBS at a maximum concentration of 300 μg/ml, and stored at either 4°C or -80°C ( Fig 1A).

Protein assays
Total exosomal protein was measured by a bicinchoninic acid (BCA) assay (Micro BCA Protein Assay kit, Thermo Scientific) and analyzed using a microplate reader and SoftMax Pro Software (SpectraMax M2, Molecular Devices).

Measuring number and size of exosomes
The number and size of exosomes were determined by nanoparticle-tracking analysis (NTA) based on Brownian motion of nanovesicles using a NanoSight LM10 microscope and NTA software ver. 2.3 (NanoSight).

RNA isolation and miRNA microarray analysis
Small RNAs were extracted from the exosome or cell samples by using 3D-gene extract reagent (Toray) or miRNeasy mini kit (Qiagen), respectively, according to a manufacturer's instruction. The extracted RNAs were checked by Bioanalyzer (Agilent) and was labeled with 3D-Gene miRNA labeling kit (Toray), and hybridized onto 3D-Gene Human miRNA Oligo chips (Platform GPL18941: Toray). The annotations and oligonucleotide sequences of the probes were conformed to the miRBase (http://www.mirbase.org/). After stringent washes, fluorescent signals were scanned with the 3D-Gene Scanner and analyzed using 3D-Gene extraction software (Toray).
The raw data of each spot was normalized by substitution with a mean intensity of the background signal determined by all blank spots' signal intensities of 95% confidence intervals. Measurements of spots with the signal intensities greater than 2 standard deviations (SD) of the background signal intensity were considered to be valid. A relative expression level of a given miRNA was calculated by comparing the signal intensities of the valid spots throughout the microarray experiments. The normalized data were globally normalized per array, such that the median of the signal intensity was adjusted to 25.

Reverse transcription quantitative PCR (RT-qPCR)
Small RNA was isolated from exosomes or donor cells using miRNeasy mini kit (Qiagen) according to the manufacture's direction. Reverse transcription of RNA was performed using the Mir-X miRNA First-Strand Synthesis Kit (Clonetech). RT-qPCR was performed by StepO-nePlus Real-Time PCR system (Applied Biosystems) using SYBR Advantage qPCR Premix (Clonetech) and synthetic primers (GeneDesign). Quantity of each miRNA was measured by the comparative Ct method (the ⊿⊿ Ct method). The level of each miRNAs was normalized to that of a U6 snRNA control. Analysis of miRNA microarray data and sequences Human activated T cells or their-released exosomes were prepared from PBMCs of 3 healthy donors including 2 lots from one donor. Exosome-or donor human T cell-dominant miRNAs were selected that showed over a 2-fold change in the comparison between the two, and then visualized as a heat map of the cluster analysis. The normalized data were Z-transformed and a heat-map of selected genes was constructed by hierarchical cluster analysis using Cluster 3.0 software and the results were displayed with the TreeView program (http://rana.lbl.gov/eisen/).

Analysis of ratio and pattern of guanine in miRNA sequences
To forecast of G-rich character of exosomal miRNAs, miRNA sequences from human T cell (the averaged raw data from the miRNA microarray analysis of three donors)-, K562-DUC18 CTL-, or RAW264.7-released exosomes containing G motifs (G, GG, GGG, GGGG) as red letters and other bases (A, U, C) as white letters were lined up in order from the highest value of exosomal miRNA or ratio of exosomal/donor T cell miRNA by using Excel VBA macro programing. The percentage of G, the maximal continuity of G, the number of continuous G, and the maximal G-G interval of miRNA sequences from human T cell-, A549-, or HCT116-released exosomes were counted and analyzed using Excel software before statistical analysis. The percentage of each base in exosomal miRNAs and donor T cell miRNAs was indicated by different colors, and lined up in order from the highest value of exosomal miRNAs or donor T cell miRNAs, or the highest exosomal/donor T cell miRNA ratio. In addition, the percentage of each base in miRNAs from A549-, HCT116-, K562-, BALB/c CD8 + T-, DUC18 CTL-, or RAW264.7-released exosomes was indicated by different colors and lined up in order from the highest value of exosomal miRNAs for estimate G-rich features of exosomal miRNA across the cell types.

Statistical analysis
Pearson's correlation test was performed to confirm statistical significance of a G-rich feature of exosome-dominant miRNA sequences. The correlation coefficient (r) and p-value were calculated between the percentage of G, the maximal continuity of G, the number of continuous G, or the maximum G-G interval and the exosome/donor cell miRNA ratios, and between the percentage of each base and the amount of the miRNA (normalized raw data from the miRNA microarray) or the exosome/donor cell miRNA ratios. Positive or negative correlation was evaluated as "significant" at a correlation coefficient 0 < r ≦ 1 or -1 ≦ r < 0, respectively, and p < 0.05.

Results
Isolation and characterization of activated human T cell-, human tumor cell-, murine T cell-and murine macrophage-released exosomes hPBMCs were cultured for two weeks in medium supplemented with OKT3, RetroNectin, rIL-2 and autologous plasma, and the proportion of CD4 + or CD8 + cells was determined by flow cytometry. This protocol always resulted in a dominant expansion of the CD8 + cell population (Fig 1A and 1B). The supernatant obtained from these cultured hPBMCs was subjected to ultracentrifugation after removing cell debris and aggregated proteins for exosome purification  ( Fig 1A). The hPBMC-released exosomes exhibited an average diameter of 153 nm, as determined by NTA (Fig 1C). In addition to the dominancy of CD8 + cell population, nevertheless the 2-weeks cultured hPBMCs abundantly expressed tetraspanins (CD9, CD63, and CD81), T cell-related molecules (CD3, CD28, CD45, T cell receptor [TCR] αβ), and human leukocyte antigens (HLA class I and class II molecules), their-released exosomes exhibited the modest expression of CD4, CD8, CD9, CD45, and HLA class II molecules and the substantial expression of CD63, CD81, and HLA class I molecules by flow cytometric analysis of the mAbstained exosome/latex beads, while no expression of CD3, CD28, or TCRαβ was detected on the surface of the cultured human T cell-released exosomes (Fig 1D)

Prediction of G-rich feature of exosomal miRNAs
A total of 256 patterns of 4 nucleotide sequences in 157 exosome-and 76 donor T cell-dominant miRNA sequences were examined. Human CD8 + T cell-derived exosome-dominant miRNA sequences included G-rich sequences, such as GGGG, GGGA, AGGG and GAGG at a high frequency, which is similar with one motif from other group [16], whereas there was no such specific feature in the parent cell-dominant miRNA sequences (Fig 3A). This G-enrichment was visualized by lining each miRNA sequence up in order from the highest exosome/ donor T cell miRNA ratio or the highest amount of normalized raw data of exosomal miRNAs (Fig 3B). Statistical analysis of the percentage of G (r = 0.380, p < 0.001), the maximal continuity of G (r = 0.343, p < 0.001), the number of continuous G (r = 0.346, p < 0.001), and the Enriched G-Rich miRNAs in Exosomes maximum G-G interval (r = -0.204, p < 0.001), exhibited the concentrated G in exosomal miRNA sequences (Fig 3C). Consistent with the human T cell-released exosome-dominant miRNA sequences, the concentrated G regions also seemed to be present in the miRNA sequences of murine cytotoxic T lymphocyte (CTL)-, human K562 cell-, or murine RAW264.7 cell-derived exosomes (S4 Fig). In addition, the enrichment of G was shown statistically significance in A549-and HCT116-released exosome-dominant miRNAs (S5 Fig) similar to the miR-NAs in human T cell-released exosomes, indicating that the conserved G-rich phenomenon of exosomal miRNAs is not restricted to one species or cell type.

Statistical analysis to confirm the G abundance in human T cell-released exosome-dominant miRNA sequences
The percentage of each base was visualized using different colors, and then lined up in order of the highest value of exosomal miRNAs or donor T cell miRNAs, or the highest ratio of exosome/ donor T cell miRNAs (Fig 4A). In addition to the high G content in exosome-dominant miRNA sequences, an inverse correlation of U and C content in miRNA sequences was seen in the exosomal miRNA and the exosomal/donor T cell miRNA ratio groups, but not in a donor T cell miRNA value group. To confirm the positive correlation of G and the negative correlation of U and C, we performed a statistical analysis of each base using Pearson's correlation test. As indicated in Fig 4B, (Figs 5 and 6). On the other hand, there was a significant negative correlation of U and A in miRNA sequences from all types of murine cell-released exosomes, compared with the ambiguity of miRNA sequences form human tumor cell-derived exosomes (Figs 5B and 6B). Thus, we conclude that the G-rich sequence is a feature of mammal cell-released exosome-dominant miRNAs.

Discussion
The functional strand of the mature miRNA, called as the guide strand, interacts with the Ago proteins by A-or U-recognition in 5' region to form of the RNA-induced silencing complex Enriched G-Rich miRNAs in Exosomes with the target mRNA [18]. In addition, the seed sequence positioned at 5' end of miRNA (nucleotide two to eight) plays a pivotal role in binding with 3'-UTR of mRNA and consequent translational repression and degradation of mRNA [19]. It has been reported that the G/C content in the seed sequence of miRNA exhibits positive correlation with specific recognition of target mRNA [20], indicating the profound significance of the concentrated G in exosomal miRNAs as demonstrated in this study by statistical analysis using Pearson's correlation test. G-rich rule of exosomal miRNAs has a possibility to utilize exosome cargo in nucleotide acid medicine. For instance, transfection of the G-concentrated miRNA-expressing vectors into appropriate cells may result in the production of transfected miRNA-embedded exosomes. Dendritic cell (DC) -derived exosomes directly modulate the antigen-specific responses of CD4 + and CD8 + T cells, and participate in the activation of NK cells [1]. T cell exosomel miRNAs are transferred into DCs in an antigen-specific manner [21]. In addition, it has been reported that regulatory T cell-released exosomes act as a suppressor against pathogenic Th1 responses in a miRNA-dependent manner [22]. These findings seem to indicate the inheriting donor cell functions in the released exosomes partly via miRNAs. CD8 + T cells play as a central effector in surveillance against malignancy [23,24], suggesting the presence of cytotoxic or down-regulatory miRNAs against tumors in CD8-dominant T cell-derived exosomes used in this study. Indeed, some antitumor miRNAs reported previously were found in T cell-released exosome-dominant miRNAs by PubMed search (S1 Table) [25][26][27][28][29][30][31][32][33][34][35][36], having a possibility of T cell-derived exosomes as a tumoricidal effector, while further studies are necessary to clarify this supposition.
The difference of miRNA distribution pattern in the donor cells with the released exosomes (Fig 2, [21]), implying one possibility that RBP-mediated translocation of cellular miRNAs into exosome cargos occurs by G-recognition when multivesicular endosomes are generated in the donor cells. An in silico analysis using an RBPDB database to predict RBPs specific to 335 human T cell-released exosome-dominant miRNAs, compared to 75 miRNAs from donor T cell-released exosomes, revealed that FUS, SFRS13A and RBM4, in addition to RNPA2B1 as previously reported [16], were identified as strong candidate RBPs for the selective translocation of cellular miRNAs into exosome cargos (S2 Table). With the exception of FUS at 3.9%, these candidates were not predicted as donor T cell-dominant exosome miRNA-specific RBPs. In particular, SFRS13A could possibly be a central player in the concentration of exosomal miRNAs because of the wide recognition of G-rich domains. Further analysis by silencing the expression of these candidate RBPs will be necessary to clarify the relationship between G-rich exosome-dominant miRNAs and RBPs, including examination of RBP modifications, such as sumoylation or ubiquitination [16,37].
A low significant positive correlation of G in sequences from HCT116-released exosomedominant miRNAs was calculated by statistical analysis using Pearson's correlation test, although a strong significance was indicated in other cases. Outliers positioned at a high X-and low Y-axis area or a low X-and high Y-axis area have been shown to affect positive correlation calculations using Pearson's test [38]. In the current study, such biased miRNAs (e.g. human miR-1260b, -4286, -5100, -7975, and -7977) were scattered in an area of high exosomal miRNA value and low G percentage in HCT116-released exosome-dominant miRNAs, compared with A549-and K562-exosomal miRNAs, and therefore were determined to be the cause of the low significant positive correlation values. Indeed, HCT116 exosome-dominant miRNAs excluding 5 biased Enriched G-Rich miRNAs in Exosomes miRNAs resulted in the strong significant positive correlation of G (r = 0.123, P < 0.001). It is suggested that such outliers (miR-1260b, -4286, -5100, -7975, and -7977) are incorporated into exosome cargos from the donor cytoplasm by natural inflow rather than RBP-mediated active translocation because they exist at high levels in HCT116 cells.
Our selected GGAG sequence in exosome-dominant miRNA sequences is identical to one of four of nucleotide motifs reported previously [16]. However the other three motifs, UGAG, CCCU and UCCU, were not identified in our analysis, possibly due to changes in exosomal miRNA patterns by the activation state of the donor human T cells, such as different stimulation conditions of human T cells; the present study used a CD3-specific monoclonal antibody and RetroNectin, versus phorbol 12-myristate 13-acetate and ionomycin used in the previously published study [16]. The differences could also be due to inherent differences between the "Toray 3D-gene" and "Agilent" microarray chips for the detection of miRNA profiles. Fluctuations in miRNA patterns by the activation state of human T cells was negligible because the Grich sequences, including a GGAG motif, were generally present in miRNAs from human tumor-, murine tumor-and murine T cell-released exosomes in addition to the exosomal miRNA sequences of human activated T cells in this study. Therefore, differences in the miRNA microarray chips could be crucial to the analysis of miRNA sequences. Although both Toray 3D-gene and Agilent microarray chips are extremely accurate and reproducible compared to quantitative RT-PCR [39], it seems that the probes on the microchips from Toray or Agilent are designed for detecting total miRNAs including pri, pre and mature forms, or mature miRNAs, respectively [40]. Mature miRNAs are stabilized by 3'-end adenylation or uridylation, and 3'-end uridylated miRNAs have been reported to be concentrated in exosomes [41]. Two CCCU and UCCU motifs may be identified only when using an agilent microarray chip for mature miRNAs.
In summary, the G-rich feature of exosomal miRNAs reveal important insights into the utilization of T cell-released exosomes for the treatment of cancer, and the development of nanomedicines by artificially introducing cytotoxic short RNAs, including miRNA and siRNA, into the exosome cargo in combination with a rule of 3'-end uridylation. The invention of comprehensive G-rich sequences and RBP-bound G-rich domains of exosomal miRNAs will revolutionize the introduction of siRNAs generally designed in a restricted G/C content into exosomes. Enriched G-Rich miRNAs in Exosomes S3 Fig. Confirmation of the quantitative correlation of miRNAs between normalized raw data of microarray and RT-qPCR. Indicated 6 human T cell-released exosome-dominant or 3 donor T cell-dominant miRNAs were selected from the high E value and the high E/C value groups, or the high C value group by comparing the normalized raw data of microarray, respectively. RT-qPCR was performed by using primer-specific for the selected each miRNA. Data were expressed as the mean ± SD (duplicate) of the relative quantification of each miRNA.