Diversity and signature of small RNA in different bodily fluids using next generation sequencing

Diversity and signature of small RNA in different bodily fluids using next generation sequencing Background: Small RNAs are critical components in regulating various cellular pathways. These molecules may be tissue-associated or circulating in bodily fluids and have been shown to associate with different tumors. Next generation sequencing (NGS) on small RNAs is a powerful tool for profiling and discovery of microRNAs (miRNAs). Results: In this study, we isolated total RNA from various bodily fluids: blood, leukocytes, serum, plasma, saliva, cell-free saliva, urine and cell-free urine. Next, we used Illumina’s NGS platform and intensive bioinformatics analysis to investigate the distribution and signature of small RNAs in the various fluids. Successful NGS was accomplished despite the variations in RNA concentrations among the different fluids. Among the fluids studied, blood and plasma were found to be the most promising fluids for small RNA profiling as well as novel miRNA prediction. Saliva and urine yielded lower numbers of identifiable molecules and therefore were less reliable in small RNA profiling and less useful in predicting novel molecules. In addition, all fluids shared many molecules, including 139 miRNAs, the most abundant tRNAs, and the most abundant piwi-interacting RNAs (piRNAs). Fluids of similar origin (blood, urine or saliva) displayed closer clustering, while each fluid still retains its own characteristic signature based on its unique molecules and its levels of the common molecules. Donor urine samples showed sex-dependent differential clustering, which may prove useful for future studies. Conclusions: This study shows the successful clustering and unique signatures of bodily fluids based on their miRNA, tRNA and piRNA content. With this information, cohorts may be differentiated based on multiple molecules from each small RNA class by a multidimensional assessment of the overall molecular signature. Keywords: miRNA, tRNA, piRNA, Next generation sequencing, Blood, Plasma, Serum, Saliva, Urine Background bind to and downregulate messenger RNAs (mRNAs) Small RNAs are a class of mainly non-coding RNAs [5]. They down regulate gene expression, playing a major (ncRNAs) characterized by their small nucleotide length role in essential biological pathways, such as differenti- of less than 200 nt [1]. Within this class there are key ation, proliferation, metastasis and apoptosis. [6–11]. RNA types with a size range of 14–35 nt that are highly MicroRNAs represent an entire layer of gene expression important for diagnostic biomarker discovery and the regulation, regulating more than 50% of protein coding development of therapeutic agents [2–4]. These include mRNAs in mammalian cells [12]. To date, 2588 human microRNAs (miRNAs), transfer RNA-derived RNAs mature miRNAs have been identified and are currently (tDRs) and Piwi-interacting RNAs (piRNAs). Micro- included in miRBase 21 [13]. Aside from being found in RNAs are non-coding molecules of about 19–23 nt that tissues and cells, miRNAs are found in bodily fluids in extracellular vesicles or in complexes with argonaute or lipoproteins [14–17]. They have been reported in bodily * Correspondence: melmogy@norgenbiotek.com; melmogy@hotmail.com fluids such as blood, plasma, serum, urine, tears, saliva, Norgen Biotek Corp, Thorold, ON L2V 4Y6, Canada breast milk, amniotic fluid, seminal fluid and colostrum Molecular Biology Department, National Research Centre, Dokki, Giza, Egypt Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. El-Mogy et al. BMC Genomics (2018) 19:408 Page 2 of 24 [18–23]. MicroRNAs have been linked to many diseases Methods and are highly promising molecular biomarkers [24–27]. Sample collection, preparation and RNA purification Mature tRNAs and nascent pre-tRNA transcripts are Blood, saliva and urine were collected from 4 healthy processed enzymatically to produce well defined tDRs in donors, 2 females and 2 males between the ages of 20 to a regulated process, suggesting that they are not random 30. The various bodily fluids were collected from each degradation products [28, 29]. Sizes of tDRs range from individual within a 2-h period. Collection and sample 30 to 35 nt for tRNA halves and 14 to 26 nt for the de-identification was performed under an IRB approved shorter fragments [4]. Various studies suggest that tDRs protocol (16198–16:02:416–11-2017). Three 10 mL are involved in different functions including stress re- blood samples were collected from each donor. Two of sponses in human diseases, where they act as inhibitors these samples were collected in Vacutainer® plastic of global translation and transcription regulation [30– EDTA tubes (BD, USA) and used for RNA isolation from 36]. Like miRNAs, they can conduct specific gene silen- whole blood, leukocytes and plasma. RNA was isolated cing and have potential as cancer biomarkers [7, 37–47]. directly from 0.2 mL of blood using the Total RNA Puri- Finally, piRNAs have a nucleotide size range between 26 fication Plus Kit (Norgen Biotek Corp., Canada). Leuko- to 31. They modulate different gene expression pathways cytes were prepared from 0.5 mL of blood by using the by interacting with Piwi proteins [48, 49]. Currently Leukocyte RNA Purification Kit (Norgen Biotek Corp., there are 23,439 piRNA molecules in the piRNABank Canada). One entire tube was used to prepare plasma [50]. They are abundant in gonads and mediate trans- and was centrifuged at 200 RCF for 10 min at room poson repression to conserve genome integrity [51–53]. temperature. Plasma was collected and stored at − 70 °C The use of next generation sequencing (NGS) technol- until isolation. The last blood sample was collected in a ogy in small RNA detection has advanced research in the Vacutainer® glass serum tube with silicon coated interior field at unprecedented speed. NGS shines light on the key (BD, USA) and used for RNA isolation from serum. The role of ncRNAs in transcriptome regulation in healthy and tube was left to stand at room temperature for 45 min, disease conditions and accelerates the profiling and dis- and then it was centrifuged at 1300 RCF for 15 min. covery of molecules [54]. The technology enables the ana- Serum was collected and stored at − 70 °C until isola- lysis of multiple samples in parallel and provides precise tion. Both plasma and serum RNA were isolated from quantification of each molecule, making it superior to pre- 0.2 mL using the Plasma/Serum RNA Isolation Mini Kit vious genomic technologies. The demonstrated capacities (Norgen Biotek Corp., Canada). All kits were used of NGS have led to advances in biological and medical according to the manufacturer’s instructions. genomics and transcriptomics [55–57]. Two milliliters of saliva was collected from each donor Solid tumor miRNAs are well represented in bodily into Falcon 50 mL centrifuge tubes (BD, USA) and fluids, indicating their importance as cancer biomarkers 0.3 mL was used directly for saliva RNA isolation using [58–60]. Almost all bodily fluids from healthy individ- the Total RNA Purification Kit (Norgen Biotek Corp., uals contain miRNAs. Therefore, bodily fluids represent Canada), according to the manufacturer’s instructions. an excellent candidate for non-invasive detection of Another 0.3 mL of saliva was transferred into a 1.5 mL miRNAs and have been used in applications such as tube (Eppendorf, Germany) and spun down at 200 RCF cancer biomarker discovery [8]. Transfer RNA-derived for 10 min to remove cells, and then the supernatant small RNAs are thought to have a dual function, as they was used for cell-free saliva RNA purification by the act as suppressors and oncogenes [42]. In addition, al- same kit. A similar approach was used for urine sample tered piRNAs levels were found to be associated with preparation, where 100 mL of urine was collected from lung, breast, gastric and colon cancers [61–64]. How- each donor into disposable cups (Sarstedt, Germany). ever, no comprehensive study has been reported on tDRs RNA was isolated from 30 mL of urine using the Urine or piRNAs in different bodily fluids. Cell-Free Circulating RNA Purification Maxi Kit (Nor- The field of small RNA is expanding with the profiling gen Biotek Corp., Canada). The kit’s procedure was and discovery of molecules in various disease conditions modified by skipping the initial centrifugation steps to and treatments. Therefore, it is important to explore the purify RNA from total urine. Another urine sample was small RNA content in normal individuals to better processed by the same kit without any modifications to understand the small RNA profile in each fluid as well the manufacturer’s procedures to isolate RNA from as their relative distribution among the different fluids. 30 mL of cell-free urine. Purified RNA from all samples To gain insights into the distribution and signature of were tested for positive amplification by miR-21 small RNA in bodily fluids, we carried out a comparative stem-loop RT-PCR [65]. RNA concentration was then study on RNA from different fluids collected from the estimated by the Agilent 2100 Bioanalyzer System (Agi- same donors and used NGS to explore and describe lent Technologies, USA) using the RNA 6000 Nano their small RNA content. Total RNA chip. El-Mogy et al. BMC Genomics (2018) 19:408 Page 3 of 24 Small RNA library construction and high-throughput trimming, read quality was assessed by FASTQC to filter sequencing out reads with a quality score lower than 30 on the The small RNA libraries were prepared from the RNA PHRED scale. Reads were first mapped to the UniVec isolated from each sample using the Small RNA Library and human ribosomal RNA (rRNA) sequences to Prep Kit for Illumina (Norgen Biotek Corp., Canada) ac- exclude them before mapping to databases of miRBase cording to the manufacturer’s instructions. Briefly, 6 μL version 21, gtRNAdb and piRNABank to assign reads to of purified RNA was mixed with the 3′ adapter and miRNAs, tRNAs and piRNAs, respectively. Identified incubated at 70 °C for 2 min before being used in a tRNAs are tRNA-derived RNA fragments due to the fact ligation step by adding T4 RNA ligase 2 (truncated), buf- that the library insert size is below 50 nt. Remaining fer and RNase inhibitor. The reaction was incubated at sequences were then annotated to gencode version 24 28 °C for 1 h then heat inactivated at 70 °C for 10 min. (hg38) which includes protein coding transcripts (pro- The excess 3′ adapters were removed by the addition of tein_coding), mitochondrial rRNA (Mt_rRNA), mito- the reverse primer and incubating the reaction at 75 °C chondrial tRNA (Mt_tRNA), small nuclear RNA for 5 min, 37 °C for 15 min and 25 °C for 5 min. The 5′ (snRNA), small nucleolar RNA (snoRNA), long inter- adapter was denatured at 70 °C for 2 min and then genic noncoding RNA (lincRNA) and miscellaneous added together with 10 mM ATP and T4 RNA ligase 1 RNA (misc_RNA). to the reaction and incubated at 28 °C for 1 h followed by heat inactivation at 70 °C for 10 min. The two Data analysis adapters were diluted 1:1 before being added to the reac- Raw read counts obtained from the Genboree Work- tions and all the incubation steps were performed in a bench’s exceRpt small RNA-seq pipeline were further thermocycler with cooling on ice between the different analyzed using R (version 3.4.0) and R studio (version steps. Reverse transcription was performed on the 1.0.143). The following R packages were used in the ana- ligation reaction product by adding a mixture containing lysis: RnaSeqGeneEdgeRQL (version 1.0.0) for counts 10 mM dNTPs, first strand buffer and TruScript reverse per million (CPM) filtration and normalization by using transcriptase, and incubating the reaction at 50 °C for trimmed mean of M-values (TMM) [67], ggfortify (ver- 1 h before heat inactivation at 70 °C for 15 min. Finally, sion 0.4.1) and ComplexHeatmap (version 1.14.0) for the reverse transcription reaction product was amplified principal component analysis (PCA) plot and heatmaps and indexed in a 15 cycle PCR reaction by adding the based on the filtered and normalized data, respectively. NGS PCR master mix, PCR reverse primer and the VennDiagram (version 1.6.17) was used to illustrate unique index primer for each sample. Venn diagrams. miRDeep2 (version 2.0.0.8) was used to The PCR reaction product was cleaned and separated predict novel miRNA candidates and tDRmapper was on a 6% Novex® TBE PAGE gel (Life Technologies, used to identify tDRs. USA). The gel was stained with SYBR® Gold Nucleic Acid Gel Stain (Life Technologies, USA) and a library Results size range from 125 bp to 170 bp was excised from the Small RNA profiles in the various bodily fluids used in gel and placed in a Gel Breaker Tube (IST Engineering, this study provide an atlas of miRNAs, tRNAs and piR- USA), then centrifuged at 14000 RCF for 2 min. The NAs relative distribution. They also provide in depth prepared libraries were then eluted overnight in molecular analysis and a guide for NGS-based small nuclease-free water (Ambion, USA) and cleaned. The RNA expression studies that employ one or more of library was quantified by the High Sensitivity DNA Ana- these bodily fluids as a source of biological data. It is im- lysis Kit on the Agilent 2100 Bioanalyzer System (Agilent portant to look at the normal characteristics of small Technologies, USA). Libraries were diluted to 4 nM, RNA molecules in each fluid in terms of abundance and pooled, and sequenced on the Illumina HiSeq 4000 at representation. The origin and nature of these fluids can The McGill University and Génome Québec Innovation pose a significant effect on their use in certain studies Centre (Montreal, Canada), using the HiSeq 3000/4000 that might require specific handling during preparation SBS Kit (50 cycles). and sequencing to ensure the validity of results. Read mapping and small RNA annotation RNA concentration variations in the different bodily fluids The sequence raw data from the Illumina HiSeq 4000 Concentration of RNA from each bodily fluid tested was were converted to fastq format. Files were then used in measured using an Agilent 2100 Bioanalyzer. The aver- the Genboree Workbench’s exceRpt small RNA-seq age range of RNA content in 1 L of bodily fluids was as pipeline (version 4.6.2) for read mapping to the hg38 hu- low as 0.01 mg in urine to as high as 11.2 mg in saliva. man genome version [66]. This allowed for a single mis- Bodily fluids can be categorized based on their RNA matched base down to 18 nucleotides. After adapter content; significantly higher amounts of RNA can be El-Mogy et al. BMC Genomics (2018) 19:408 Page 4 of 24 recovered from saliva, blood and cell-free saliva (4.2 to respectively. This is indirectly proportional to reads used 11.2 mg/L). Leukocytes, serum and plasma had moder- for alignment to the human genome. More than 50% of ate yields of 0.8 to 1.8 mg/L, while urine and cell-free reads used for alignment were mapped to the human urine had significantly lower RNA content of 0.01 mg/L. genome in blood, plasma, serum, urine and cell-free Blood, leukocytes, saliva and cell-free saliva had lower urine. The percentage was lower in the leukocytes as standard deviations in their RNA content (< 50% of aver- well as total and cell-free samples of saliva. age), whereas plasma, serum, urine and cell-free urine In saliva and cell-free saliva, the average percentage of had higher deviations between their samples (70–85% of unmapped reads was about 50% of reads used for align- average). The average concentration of the isolated RNA ment (48.9 ± 19.7% and 50.3 ± 10.1% of input reads, from all bodily fluids ranged from 67.2 ng/μL to 3.4 ng/ respectively). Conversely, urine and cell-free urine had μL. They can be classified into high (> 20 ng/μL) from an average percentage of unmapped reads relative to blood, leukocytes, saliva and cell-free saliva and low (< reads used in alignment of 21.2 ± 17.8% and 25.5 ± 10 ng/μL) from plasma, serum, urine and cell-free urine. 23.9%, respectively. The percentage of input reads align- The RNA integrity number (RIN) was more than 7 for ment from each bodily fluid can be found in Table 2. RNA from leukocytes and lower for RNA from blood, saliva and cell-free saliva (about 2–3). RNA from serum, Small RNA biotype mapping urine and cell-free urine had a low RIN of 1 or less, Reads that were mapped to the human genome were while RNA from plasma had no measurable RIN from then mapped and classified to the various small RNA any sample (Table 1). biotypes. The average total reads mapped to small RNA biotypes within each bodily fluid ranged from 1.3 to 12.8 Input read alignment million reads. Blood, plasma, cell-free urine and urine Reads obtained from sequencing were used for align- had more than 8 million reads mapped to biotypes (12.8, ment and mapping to the human genome after adapter 9.7, 8.7 and 8.0 million reads, respectively). Serum, leu- clipping and quality filtering. The range of average input kocytes, cell-free saliva and saliva had 5.4 million reads reads from the various bodily fluids was between 9.5 or less (5.4, 2.6, 1.3 and 1.3 million reads, respectively). million reads (serum) to 15.7 million reads (blood). The The distribution of biotypes within each bodily fluid average input reads from all the bodily fluid samples showed distinct patterns. Plasma had a high percentage tested was 12.57 ± 3.54 million reads. The descending of miscellaneous RNA (misc_RNA; 58.0 ± 39.4), while order of samples based on their number of input reads urine and cell-free urine had high amounts of tRNAs was: blood, cell-free urine, urine, leukocyte, plasma, (91.3 ± 77.5% and 91.3 ± 90.3%, respectively). The other cell-free saliva, saliva and serum. The percentage of suc- bodily fluids had a more diverse pattern with no single cessfully clipped reads was more than 60% from all sam- biotype exceeding 50% of the content. MicroRNAs rep- ple types, with a minimum percentage of reads failing resented more than 85% of blood biotypes, 25% of leu- quality filters. Reads were mapped to human rRNA to kocytes, and 15–25% of plasma, serum and cell-free exclude rRNA sequences before mapping to human gen- saliva. Saliva, cell-free urine and urine contained the ome. The average percentage of reads aligned to human lowest miRNA content (5.3–12.0%). Transfer RNA was rRNA was less than 12% in all bodily fluids except saliva, the predominant biotype in urine and cell-free urine (> cell-free saliva and leukocytes, which had average per- 90%), while serum, saliva and cell-free saliva contained centages of 16.6 ± 7.5, 13.0 ± 6.2 and 36.0 ± 1.5, moderate tRNA content (20–50%). Leukocytes had 18.4 Table 1 Variations in RNA concentration, RIN value, and yield among the different bodily fluids Bodily fluid Concentration (ng/uL) RIN Value RNA amount in 1 L of fluid (mg) Ave STDEV Ave STDEV Ave STDEV Blood 21.200 4.764 2.740 2.243 5.300 1.191 Leukocytes 18.200 5.848 7.600 0.474 1.820 0.585 Plasma 3.400 2.881 N/A N/A 0.850 0.720 Serum 6.600 5.320 1.000 N/A 1.650 1.330 Saliva 67.200 33.056 2.360 0.602 11.200 5.509 Cell-Free Saliva 25.333 10.970 2.133 1.002 4.222 1.828 Urine 5.750 4.113 1.000 N/A 0.010 0.007 Cell-Free Urine 6.600 5.683 1.000 N/A 0.011 0.009 RNA concentration and RIN value were determined by the Agilent 2100 Bioanalyzer System El-Mogy et al. BMC Genomics (2018) 19:408 Page 5 of 24 Table 2 Percentage of input reads alignment from each bodily fluid Bodily Fluid Blood Leukocytes Plasma Serum Saliva Cell-Free Urine Cell-Free Saliva Urine Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Input (million reads) 15.7 2.3 12.5 1.0 12.4 6.7 9.5 1.5 10.4 1.1 10.9 1.4 14.1 2.6 15.1 4.2 Successfully clipped 94.4 0.7 63.4 4.6 78.7 20.7 80.7 5.3 82.1 12.2 78.3 2.1 83.2 9.4 82.1 8.8 Failed quality filter 0.2 0.0 0.1 0.0 0.1 0.0 0.1 0.0 0.2 0.0 0.2 0.0 0.1 0.0 0.1 0.0 Failed homopolymer filter 0.0 0.0 0.2 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.1 0.0 0.1 0.0 0.0 0.0 UniVec contaminants 0.4 0.0 0.7 0.3 0.8 0.9 0.5 0.2 0.6 0.4 0.6 0.2 0.5 0.3 0.6 0.4 rRNA 9.1 1.4 36.0 1.5 2.0 0.9 7.0 3.4 16.6 7.5 13.0 6.2 5.3 4.4 2.1 1.3 Reads used for alignment 84.7 2.2 26.4 2.8 75.8 21.0 73.1 7.6 64.6 19.5 64.5 7.2 77.2 14.0 79.3 9.8 Genome 82.2 1.9 22.7 2.3 69.0 26.7 58.0 9.5 15.7 5.3 14.2 3.3 56.1 31.7 53.7 33.6 miRNA sense 70.5 3.1 6.1 0.9 11.6 7.7 10.2 3.4 1.4 0.9 2.7 0.7 3.2 1.9 4.4 2.8 miRNAprecursor sense 0.2 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 tRNA sense 1.7 0.3 3.8 1.0 8.9 11.1 22.0 6.1 5.4 4.1 4.6 1.3 47.4 37.6 44.5 40.0 piRNA sense 1.5 0.5 1.2 0.2 4.9 3.3 1.1 0.4 0.1 0.0 0.1 0.0 0.1 0.1 0.1 0.1 Gencode sense 7.9 0.7 9.6 1.1 42.7 28.1 23.0 3.7 5.1 2.7 4.1 2.5 1.9 1.5 1.1 1.0 Not mapped to genome or transcriptome 2.5 0.2 3.8 0.5 6.7 5.8 15.1 5.3 48.9 19.7 50.3 10.1 21.2 17.8 25.6 23.9 Number of million input reads obtained from sequencing of each sample and used in the alignment ± 4.0 tRNA while plasma and blood contained the lowest miRNA tRNA fractions of 5.8 ± 2.5 and 2.1 ± 0.7, respectively. Fi- The most variable 50 miRNAs were calculated based on nally, piRNAs represented less than 2% of the reads in TMM-normalized miRNA counts at a CPM correspond- blood, serum, saliva, cell-free saliva, urine and cell-free ing to a minimum of 5 counts in a library to achieve a urine, while more than 5% of the reads were piRNAs high confidence level. These miRNAs were then used for in leukocytes and plasma. The percentages of the PCA which showed the separation of various bodily various biotypes in each bodily fluid are listed in fluids based on their miRNA expression (Fig. 2). The Table 3 and illustrated in Fig. 1.The biotype distribu- analysis revealed closeness between saliva and cell-free tion in each donor is illustrated in an additional file (Add- saliva samples and between urine and cell-free urine itional file 1: Figure S2) and shows a relatively similar samples. Close clustering was seen between blood and pattern for each bodily fluid between the donors. How- leukocytes, and between plasma and serum. The data ever, in urine samples there was a difference in patterns disclosed consistent biological origin and miRNA between male and female donors. expression-based separation of bodily fluid profiles. The Table 3 Percentage of biotype counts in the various bodily fluids Biotype Blood Leukocytes Plasma Serum Saliva Cell-Free Saliva Urine Cell-Free Urine Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Million reads mapped to biotypes 12.8 1.9 2.6 0.1 9.7 6.2 5.4 1.6 1.3 0.8 1.3 0.5 8.0 5.9 8.7 7.6 miRNA 86.6 12.3 29.9 3.3 18.8 14.1 18.1 6.4 12.0 8.3 23.6 7.6 5.3 2.2 6.8 2.9 tRNA 2.1 0.7 18.4 4.0 5.8 2.5 39.7 16.9 46.0 39.2 39.0 11.6 91.3 77.5 91.3 90.3 piRNA 1.9 0.8 5.8 0.9 8.0 5.8 1.9 0.5 0.8 0.4 0.5 0.3 0.3 0.2 0.2 0.2 misc_RNA 7.7 1.5 12.7 2.6 58.0 39.4 35.1 10.3 4.6 2.6 7.1 2.0 1.7 1.7 1.1 0.8 protein_coding 0.4 0.1 5.8 0.6 0.5 0.4 0.6 0.2 6.6 3.9 9.2 7.0 0.4 0.3 0.2 0.1 Mt_rRNA 0.0 0.0 1.9 0.4 0.9 0.7 0.9 0.4 6.8 4.3 2.1 1.8 0.3 0.1 0.0 0.0 Mt_tRNA 0.1 0.0 0.7 0.1 7.7 5.1 3.1 1.3 5.5 2.4 1.2 1.1 0.2 0.1 0.1 0.0 snRNA 0.2 0.0 4.9 0.4 0.0 0.0 0.1 0.1 5.2 5.4 11.9 11.0 0.1 0.1 0.0 0.0 snoRNA 0.3 0.0 8.7 0.8 0.0 0.0 0.1 0.1 4.1 2.6 1.3 1.4 0.1 0.1 0.0 0.0 lincRNA 0.0 0.0 0.7 0.1 0.0 0.0 0.0 0.0 0.7 0.4 0.5 0.3 0.0 0.0 0.0 0.0 Others 0.6 0.1 10.5 1.0 0.2 0.1 0.3 0.1 7.6 4.8 3.7 2.5 0.3 0.2 0.1 0.1 Number of million reads mapped to biotypes El-Mogy et al. BMC Genomics (2018) 19:408 Page 6 of 24 Fig. 1 Relative biotype distribution among the various bodily fluids. The graph was generated using average percentage of biotype counts of each fluid. MicroRNAs are the largest biotype of blood, while tRNAs are the major biotype of urine. Saliva has the highest biotype diversity among all fluids Z scores of the most variable 50 miRNAs were used to leukocytes and plasma. Saliva had 98.9% of its miRNA generate a heatmap that illustrated the pattern of ex- identical to cell-free saliva, while the latter had only pression as well as the relationship between samples 75.5% of its miRNAs overlapped with saliva. About 85– (Fig. 3). It clustered bodily fluids based on their biology. 91% of urine and cell-free urine miRNAs were over- The dendrogram showed that invasive bodily fluids lapped with blood, leukocytes, plasma and serum. (blood, leukocytes, plasma and serum) branched apart Cell-free urine had 91.3% similarity with urine, while from non-invasive fluids (saliva, cell-free saliva, urine urine had 92.7% similarity to cell-free urine. Saliva and and cell-free urine). Furthermore, it showed that urine shared more than 77% of their miRNAs. In cell-depletion from saliva or urine did not have a major addition, Venn diagrams were used to demonstrate the effect on their clustering. A large set of miRNAs ap- overlap between invasive and non-invasive bodily fluids peared to be highly abundant or severely depleted in the (Fig. 5). The invasive fluids had 230 common miRNAs various bodily fluids. Particularly, urine and cell-free (Additional file 2: Table S1). Blood had 98 unique miR- urine as well as plasma and serum had different sets of NAs, which was 2 to 3-fold higher than plasma and leu- upregulated miRNAs. kocytes. In contrast, non-invasive fluids had 148 Bodily fluids can be classified into two groups based common miRNAs (Additional file 3: Table S2), and had on their collection procedures: invasive (blood, leuko- lower numbers of unique miRNAs. The non-invasive cytes, plasma and serum) and non-invasive (saliva and fluids shared 144 common miRNAs with blood (Add- urine). Variations can be seen in the detectable number itional file 4: Table S3), while the latter had 209 unique of miRNAs, at a minimum of 5 counts in 3 or more in- miRNAs that were absent from all non-invasive fluids dividuals, between the two groups. The range of de- (Additional file 5: Figure S1). Common miRNAs be- tected miRNAs from invasive fluids was 307–440 while tween all fluids were 139 (Additional file 6: Table S4). the range from the non-invasive fluids was 178–233. We ran a detailed analysis on the miRNA composition Blood had the largest number of detected miRNAs and of each bodily fluid. We looked at the 20 most abundant saliva had the lowest. Plasma contained more miRNAs miRNAs and calculated their fractions of the total compared to serum. Similarly, more miRNAs were de- miRNA content of each sample (Table 5). These 20 most tected from the cell-free preparation of saliva than saliva. abundant miRNAs covered about 74 to 94% of all Almost no differences could be seen between urine and miRNA counts. They represented 94 and 91% of blood cell-free urine (Table 4). The overlap between detected and cell-free urine, respectively. They were only 75% of miRNAs in the different bodily fluids was illustrated in all miRNA counts in saliva and cell-free saliva, while Venn diagrams (Fig. 4). About 97% of serum miRNAs representing 80–88% of the rest of the other bodily were shared with plasma. More than 90% of miRNAs in fluids. The deep analysis of miRNAs revealed that 2 spe- saliva and cell-free saliva were shared with blood, cific miRNAs were dominating the counts of 4 bodily El-Mogy et al. BMC Genomics (2018) 19:408 Page 7 of 24 Fig. 2 Principal component analysis of the most variable miRNAs in each bodily fluid. Analysis of the most variable 50 miRNAs was calculated based on TMM-normalized miRNA counts. Four pairs of fluids show close clustering: blood/leukocyte, plasma/serum, saliva/cell-free saliva and urine/cell-free urine fluids. Hsa-miR-486-5p made up 53.6 ± 1.9% and 43.2 ± include: hsa-let-7a-5p, hsa-let-7f-5p, hsa-miR-191-5p, 13.2% of miRNA counts of blood and serum, respect- hsa-miR-26a-5p and hsa-miR-486-5p. These five miR- ively, while hsa-miR-10b-5p represented 38.4 ± 8.3% and NAs represent more than 50% of blood and serum 45.6 ± 5.6% of miRNA counts of urine and cell-free miRNA counts, 25 to 45% of plasma and saliva (total urine, respectively. Other miRNAs that represented 10– and cell-free) and less than 11% of urine and cell-free 20% of total miRNA counts included let-7f-5p (11.11 ± urine miRNA counts (Fig. 7). 1.12%) in blood, miR-146b-5p (11.23 ± 1.85) and Analysis of unique and novel miRNAs are valuable in let-7f-5p (10.58 ± 1.33) in leukocytes, miR-486-5p (15.16 evaluating the usefulness of using a specific specimen as ± 1.63) and miR-191-5p (10.84 ± 0.25) in plasma, a source of information. We ran the analysis of unique miR-143-3p (10.65 ± 4.16) in saliva, miR-143-3p (14.92 ± miRNAs in our study using 3 comparison groups: all 4.47) and miR-191-5p (11.59 ± 1.64) in cell-free saliva, fluids, invasive fluids and non-invasive fluids. The num- and miR-10a-5p in urine (11.56 ± 2.67) and cell-free ber of unique miRNAs detected in each comparison urine (13.6 ± 0.97). The proportions of the top 10 most group are listed in Table 6. Blood, leukocytes and plasma abundant miRNAs in each fluid are illustrated in Fig. 6. had significantly higher numbers of unique miRNAs Among the top 20 most abundant miRNAs of each fluid, compared to the rest of fluids. Blood had the highest 5 miRNAs were found common to all fluids. These number of unique molecules (94 miRNAs), while plasma El-Mogy et al. BMC Genomics (2018) 19:408 Page 8 of 24 Fig. 3 Heatmap clustering of the most variable miRNAs in each of the bodily fluids. The sex of the sample donor is indicated as (F) for female donors or (M) for male donors. The analysis was generated using Z scores of the most variable 50 miRNAs. The dendrogram shows distinct clustering of invasive fluids (blood, leukocyte, plasma and serum) and non-invasive fluids (saliva, cell-free saliva, urine and cell-free urine) El-Mogy et al. BMC Genomics (2018) 19:408 Page 9 of 24 Table 4 Detected number of miRNAs in the various bodily fluids Bodily fluid Blood Leukocytes Plasma Serum Saliva Cell-Free Saliva Urine Cell-Free Urine Number of miRNAs 440 352 403 307 178 233 205 208 Only miRNAs that are present at a minimum of 5 counts in 3 or more individuals were considered detectable and leukocytes had 42 and 30 unique miRNAs, re- All bodily fluids were analyzed for their novel miRNA spectively. The comparison within the invasive group candidates using miRDeep2. Signal-to-noise ratio of showed a similar trend. The comparison within the more than 10 was used to select for miRDeep2 score non-invasive group showed that cell-free saliva had cutoff [68]. In bodily fluids where signal-to-noise ratio the most unique miRNAs (32 miRNAs) followed by was less than 10 (saliva, urine and cell-free urine), we se- cell-free urine (11 miRNAs). Saliva and urine had lected the score cutoff that corresponds to the highest minimal numbers of unique miRNAs compared to signal-to-noise ratio (Table 7). Invasive bodily fluids had their cell-free preparations. The list of unique miR- higher numbers of novel miRNA candidates than NAs from the three comparison groups are listed in non-invasive fluids. The highest number of novel three additional files (Additional file 7:Table S5,Add- miRNA candidates was observed in plasma and blood, itional file 8: Table S6, Additional file 9:Table S7). with 50 and 48 candidates, respectively. Serum had 20 Fig. 4 Overlap of miRNA content between various bodily fluids. MicroRNAs of each fluid were filtered to keep molecules that have a minimum of 5 counts in 3 or more individuals. The highest overlap is seen between fluids within the same category: invasive fluids (blood, leukocyte, plasma and serum) or non-invasive fluids (saliva, cell-free saliva, urine and cell-free urine) El-Mogy et al. BMC Genomics (2018) 19:408 Page 10 of 24 Fig. 5 Venn diagram showing the overlap between (a) invasive and (b) non-invasive bodily fluids. Only miRNAs that are present at a minimum of 5 counts in 3 or more individuals were included in the comparison. Invasive fluids have higher number of shared and unique miRNAs compared to non-invasive fluids novel candidates while leukocytes had 16 candidates. Read counts of tRNAs were normalized using TMM at All the non-invasive fluids had 7 or less novel candi- a CPM corresponding to a minimum of 5 counts in a dates. Sequences of novel miRNA candidates were library. The normalized reads were then used to gener- matched to the miRCarta database (v1.0) of newly ate principal component analysis which showed the sep- predicted human miRNAs [69]. More than 50% of aration of various bodily fluids based on their tRNA blood and leukocytes novel miRNA candidates were levels (Fig. 8). The analysis revealed closeness between present in miRCarta (66 and 56%, respectively). saliva and cell-free saliva as well as closeness between Plasma and serum had less miRNA candidates match- the invasive fluids. However, urine and cell-free urine ing miRCarta database (30 and 45%, respectively). No were dispersed between both saliva and serum. The Z miRCarta matches were found for the novel miRNA scores of these tRNAs were used to generate a heatmap candidates of the non-invasive fluids. An additional that indicates the levels of various tRNAs in the different file containing the list of novel miRNA candidates in samples (Fig. 9). Serum and the male urine/cell-free each bodily fluid and their sequences, as well as the urine samples showed distant clustering from the rest of matching results to the miRCarta database is provided samples. The female urine/cell-free urine clustered with (Additional file 10:Table S8). saliva/cell-free saliva. Blood, leukocytes and plasma showed similar clustering. The data shows clustering tRNA patterns based on sample biology and no difference Mapped tRNAs represented tDRs down to 18 nucleo- between cell-depleted and non-depleted conditions. tides. The predominant tRNA fragments in all the bodily Gly fluids was tRNA . This tRNA composed 86.5 and piRNA 87.6% of the total tRNA content in urine and cell-free All bodily fluids had piR-016658 at different levels. The urine, respectively. For the remaining bodily fluids, it highest levels were seen in blood and serum (92.3 ± 1.8% made up 72.0 to 84.1% of the total tRNA content. and 94.0 ± 2.7%, respectively), followed by plasma and Glu The second most abundant tRNA was tRNA , with leukocytes (81.8 ± 33.1% and 73.5 ± 3.8%, respectively). It a range of 6.7 to 21.4% of the tRNA content. Further was the highest piRNA in cell-free saliva (40.3 ± 17.0%). analysis of tRNAs by tDRmapper to look at the exact It had lower concentrations, yet more than 10%, in sal- Gly-GCC tDRs composition showed that tRNA and iva, urine and urine-cell free (14.9 ± 1.7%, 15.6 ± 12.9% Glu-CTC tRNA were the predominant fragments in all and 14.5 ± 12.5, respectively). Urine and urine-cell free fluids. All samples, regardless of the fluid origin, had piR-019825 as the highest piRNA (46.0 ± 40.4% and shared similar tDRs composition. The quantification 58.7 ± 32.1%, respectively). Interestingly, piR-019825 was and coverage of the top 50 tDRs in each fluid are the second highest piRNA in plasma where it repre- presented in Additional files 11 and 12.Noneofthe sented 15.2 ± 30.4% of the piRNA content (Table 9). An remaining tRNAs in any of the fluids exceeded 3.1% additional file contains the list of piRNAs at an average of the tRNA content (Table 8). For blood, plasma, of 1% or more of the entire piRNAs counts of each bod- saliva and cell-free saliva, there was a higher diversity ily fluid (Additional file 13: Table S9). of tRNAs that represented 1% or more of the total Read counts of piRNAs were TMM-normalized at a tRNA content of the sample (5 tRNAs or more). CPM corresponding to a minimum of 5 counts in a li- However, for leukocytes, serum, urine and cell-free brary. The normalized counts were used to generate a urine, there was a lower diversity (3 to 4 tRNAs). PCA plot (Fig. 10). Close clustering was obtained El-Mogy et al. BMC Genomics (2018) 19:408 Page 11 of 24 Table 5 Twenty most abundant miRNAs detected in each bodily fluid # Blood Leukocytes Plasma Serum Saliva Cell-Free Urine Cell-Free Urine Saliva 1 hsa-miR-486-5p hsa-miR- hsa-miR-486-5p hsa-miR-486-5p hsa-miR-143- hsa-miR-143- hsa-miR-10b- hsa-miR-10b-5p (53.64 ± 1.89) 146b-5p (15.16 ± 1.63) (43.22 ± 13.23) 3p (10.65 ± 3p (14.92 ± 5p (38.4 ± (45.58 ± 5.62) (11.23 ± 1.85) 4.16) 4.47) 8.33) 2 hsa-let-7f-5p (11.11 hsa-let-7f-5p hsa-miR-191-5p hsa-miR-92a-3p hsa-miR- hsa-miR-191- hsa-miR-10a- hsa-miR-10a-5p ± 1.12) (10.58 ± 1.33) (10.84 ± 0.25) (4.85 ± 0.48) 203a-3p 5p (11.59 ± 5p (11.56 ± (13.6 ± 0.97) (8.36 ± 5.89) 1.64) 2.67) 3 hsa-miR-451a (4.79 hsa-miR-26a- hsa-miR-26a-5p hsa-miR-191-5p hsa-miR-191- hsa-miR-26a- hsa-miR-30a- hsa-miR-30a-5p ± 1.08) 5p (8.32 ± (8.21 ± 0.77) (4.51 ± 1.77) 5p (7.66 ± 5p (8.06 ± 5p (6.58 ± (6.59 ± 1.95) 0.33) 4.8) 0.69) 1.5) 4 hsa-miR-92a-3p hsa-let-7 g-5p hsa-let-7f-5p (6.99 hsa-let-7f-5p (4.5 ± hsa-miR-26a- hsa-miR- hsa-miR-192- hsa-miR-192-5p (4.09 ± 0.55) (6.68 ± 0.97) ± 0.7) 1.5) 5p (5.94 ± 148a-3p 5p (4.5 ± (4.53 ± 2.42) 4.11) (4.01 ± 0.63) 2.12) 5 hsa-miR-191-5p hsa-miR-150- hsa-miR-92a-3p hsa-miR-26a-5p hsa-let-7f-5p hsa-miR-375 hsa-let-7f-5p hsa-let-7f-5p (2.47 (3.74 ± 0.64) 5p (6.48 ± (5.41 ± 0.09) (3.45 ± 1.74) (5.32 ± 2) (3.56 ± 1.92) (3.32 ± 0.77) ± 0.51) 2.23) 6 hsa-let-7a-5p (3.12 hsa-miR-191- hsa-miR-146a-5p hsa-let-7a-5p (3.21 hsa-miR-486- hsa-miR-27b- hsa-miR-100- hsa-miR-27b-3p ± 0.86) 5p (5.3 ± 0.68) (5.17 ± 0.13) ± 0.87) 5p (4.77 ± 3p (3.55 ± 5p (2.61 ± (2.41 ± 0.76) 3.31) 1.66) 0.64) 7 hsa-let-7i-5p (2.6 ± hsa-miR-342- hsa-miR-30d-5p hsa-miR-146a-5p hsa-miR- hsa-let-7f-5p hsa-miR-27b- hsa-miR-100-5p 0.09) 3p (4.84 ± (3.58 ± 0.43) (3.11 ± 1.21) 378a-3p (3.49 ± 0.57) 3p (2.52 ± (2.36 ± 0.57) 2.52) (4.75 ± 7.51) 1.13) 8 hsa-let-7 g-5p (2.42 hsa-miR-486- hsa-miR-151a- hsa-miR-423-5p hsa-miR-27b- hsa-miR- hsa-miR-26a- hsa-miR-26a-5p ± 0.12) 5p (3.27 ± 5p|hsa-miR-151b (2.27 ± 0.45) 3p (3.59 ± 203a-3p 5p (2.38 ± (1.94 ± 0.34) 1.37) (2.67 ± 0.09) 1.13) (3.39 ± 1.96) 0.41) 9 hsa-miR-182-5p hsa-miR-21- hsa-miR-146b-5p hsa-miR-30d-5p hsa-let-7 g- hsa-let-7a-5p hsa-let-7a-5p hsa-miR-99a-5p (1.25 ± 0.25) 5p (3.21 ± (2.54 ± 0.16) (2.01 ± 0.51) 5p (3.11 ± (2.94 ± 0.25) (2.06 ± 0.91) (1.43 ± 0.36) 0.52) 1.36) 10 hsa-let-7b-5p (1.07 hsa-miR-92a- hsa-miR-21-5p hsa-let-7b-5p (1.69 hsa-miR-24- hsa-miR-1246 hsa-miR- hsa-let-7a-5p (1.27 ± 0.26) 3p (3.19 ± (2.18 ± 0.17) ± 0.25) 3p (2.97 ± (2.74 ± 2.33) 200b-3p ± 0.16) 0.23) 1.86) (1.75 ± 0.77) 11 hsa-miR-185-5p hsa-miR-146a- hsa-let-7a-5p (2.12 hsa-miR-122-5p hsa-let-7a-5p hsa-miR-92a- hsa-miR-486- hsa-miR-486-5p (1.01 ± 0.09) 5p (3.13 ± ± 0.26) (1.32 ± 1.46) (2.93 ± 1.31) 3p (2.45 ± 5p (1.71 ± (1.26 ± 0.95) 0.77) 0.28) 1.05) 12 hsa-miR-16-5p hsa-miR-143- hsa-miR-423-5p hsa-miR-151a- hsa-miR-375 hsa-miR-423- hsa-miR-99a- hsa-let-7b-5p (1.21 (0.95 ± 0.21) 3p (2.49 ± (2.11 ± 0.19) 5p|hsa-miR-151b (2.31 ± 1.17) 5p (1.74 ± 5p (1.65 ± ± 0.34) 1.42) (1.29 ± 0.38) 1.49) 0.43) 13 hsa-miR-26a-5p hsa-let-7i-5p hsa-miR-423-3p hsa-miR-451a (1.23 hsa-miR- hsa-let-7b-5p hsa-miR- hsa-miR-200b-3p (0.86 ± 0.11) (1.86 ± 0.21) (2.03 ± 0.17) ± 0.6) 148a-3p (2 ± (1.72 ± 0.79) 148a-3p (1.13 ± 0.25) 0.86) (1.48 ± 0.6) 14 hsa-miR-25-3p hsa-miR-30d- hsa-miR-99b-5p hsa-miR-146b-5p hsa-miR-21- hsa-miR-30e- hsa-miR- hsa-miR-148a-3p (0.64 ± 0.08) 5p (1.78 ± 0.3) (1.98 ± 0.38) (1.09 ± 0.38) 5p (1.85 ± 5p (1.71 ± 203a-3p (1.07 ± 0.36) 1.27) 0.26) (1.36 ± 1.7) 15 hsa-miR-183-5p hsa-miR-451a hsa-miR-151a-3p hsa-miR-151a-3p hsa-miR-205- hsa-miR-99a- hsa-miR-21- hsa-miR-30d-5p (0.62 ± 0.05) (1.4 ± 0.59) (1.94 ± 0.33) (1.05 ± 0.23) 5p (1.62 ± 5p (1.71 ± 5p (1.17 ± (0.93 ± 0.32) 0.92) 0.82) 0.92) 16 hsa-miR-181a-5p hsa-let-7a-5p hsa-let-7i-5p (1.88 hsa-miR-320a (0.92 hsa-miR-320a hsa-miR-24- hsa-let-7b-5p hsa-miR-99b-5p (0.58 ± 0.15) (1.38 ± 0.4) ± 0.2) ± 0.36) (1.44 ± 1.94) 3p (1.64 ± (1.17 ± 0.23) (0.86 ± 0.44) 0.11) 17 hsa-miR-151a- hsa-miR-181a- hsa-miR-127-3p hsa-miR-10a-5p hsa-miR-99a- hsa-miR-486- hsa-miR-30d- hsa-miR-423-5p 5p|hsa-miR-151b 5p (1.36 ± (1.36 ± 0.21) (0.89 ± 0.42) 5p (1.4 ± 5p (1.63 ± 5p (1.04 ± (0.7 ± 0.17) (0.53 ± 0.05) 0.27) 0.16) 0.53) 0.28) 18 hsa-miR-101-3p hsa-miR-223- hsa-let-7 g-5p (1.34 hsa-let-7i-5p (0.87 hsa-let-7i-5p hsa-miR-23a- hsa-miR-99b- hsa-miR-151a- (0.42 ± 0.1) 3p (1.23 ± ± 0.19) ± 0.14) (1.4 ± 0.33) 3p (1.52 ± 5p (0.88 ± 5p|hsa-miR-151b 0.71) 0.38) 0.34) (0.55 ± 0.07) 19 hsa-miR-30d-5p hsa-miR-30e- hsa-miR-222-3p hsa-miR-99b-5p hsa-miR-92a- hsa-miR- hsa-miR-191- hsa-miR-30a-3p (0.4 ± 0.09) 5p (1.16 ± (1.2 ± 0.32) (0.8 ± 0.3) 3p (1.39 ± 200b-3p 5p (0.8 ± (0.53 ± 0.07) 0.09) 0.41) (1.26 ± 0.69) 0.41) El-Mogy et al. BMC Genomics (2018) 19:408 Page 12 of 24 Table 5 Twenty most abundant miRNAs detected in each bodily fluid (Continued) # Blood Leukocytes Plasma Serum Saliva Cell-Free Urine Cell-Free Urine Saliva 20 hsa-miR-30e-5p hsa-miR-101- hsa-miR-10a-5p hsa-miR-22-3p (0.8 hsa-let-7b-5p hsa-miR-223- hsa-miR-423- hsa-miR-191-5p (0.36 ± 0.07) 3p (1.07 ± (1.18 ± 0.23) ± 0.19) (1.22 ± 0.33) 3p (1.23 ± 5p (0.65 ± (0.46 ± 0.09) 0.16) 0.19) 0.22) MicroRNAs are arranged in a descending order from highest to lowest. Percentage of miRNA to all miRNAs in the bodily fluid is shown between brackets. None of the unique miRNAs of each bodily fluid were found among its top 20 miRNAs between blood and leukocytes, saliva and cell-free saliva, For better representation of the actual library prepar- and urine and cell-free urine. Plasma and serum showed ation workflow, we used a standard NGS preparation relatively distant clustering from each other and from protocol based on equal input volumes from each bodily the other fluids. A similar clustering pattern was ob- fluid preparation. An average of 12.57 ± 3.54 million tained from the heatmap. However, urine samples and reads were obtained from all samples, despite the large most of the cell-free urine samples showed inter-fluid and intra-fluid variations in RNA yield and sex-dependent clustering (Fig. 11). Female urine samples integrity. This indicates that regardless of sample type, a clustered close to saliva samples, while male urine sam- clean purification and robust library preparation can ples clustered close to plasma and serum samples. While yield similar sequencing read outcomes. The critical par- clustering reflects the close biology of the samples, it has ameter that would then define suitability of a sample to a distinct trend when compared to that of miRNA as it be used in small RNA profiling and discovery would be showed more overlap between invasive and non-invasive its actual biotype content. Read alignment to human fluids. genome varied between the different fluids based on their molecular composition. The lower percentage of Discussion leukocytes reads successfully aligned to the human gen- In this study, we investigated small RNA profiles in vari- ome is a result of their higher rRNA content. The lower ous bodily fluids using NGS in order to understand the percentage of saliva and cell-free saliva reads successfully distribution of the various biotypes between fluids as aligned to the human genome is due to the high per- well as the molecular signature of each fluid. Purified centage of unmapped reads. We conducted exogenous total RNA from each fluid showed large variations in the mapping analysis on the saliva and cell-free saliva sam- RNA content and integrity. Saliva, cell-free saliva and ples and 85–90% of the unmapped reads were mapped blood have the highest RNA content. These elevated to bacteria (Additional file 14: Figure S3 and Additional levels in both blood and saliva are due to their high file 15: Figure S4). This agrees with results of a recent number of cells. However, the high RNA levels in the study by Yeri et al. [70]. This in turn reduces the amount cell-free saliva preparation were most likely due to the of valuable human RNA molecules that can be used in high bacterial content, as the cell-free preparation steps profiling or discovery. Efficient removal of salivary bac- utilized in this study were aimed at removing mamma- teria can be achieved by centrifugation [71]. However, re- lian but not bacterial cells. The lowest RNA yields were moval of bacteria in this manner would reduce the found in both total and cell-free urine preparations. This number of reads directed towards bacterial sequences, indicated very low cell content as well as minimal thereby hindering the study of bacterial communities and/ cell-free RNA content of urine samples collected from or pathogens that might be contained within these fluids. healthy individuals. Intra-fluid RNA yields are more con- While the profile of leukocytes showed a fairly even sistent from fluids that have high cell content (blood, distribution of biotypes, all the other bodily fluids saliva and leukocytes). RNA integrity as measured by showed predominant reads from one or more biotypes. RIN value was also dependent on the cellular content of This can be of a significant value if these predominant the fluid. This severely affected the RNA integrity of biotypes are of known biological importance such as urine, cell-free urine, plasma and serum where they al- miRNAs, tRNAs or piRNAs. Blood has the highest levels most have no measurable integrity. In fact, while many of miRNAs (86.6 ± 12.3), which were about 3-fold the small RNA-Seq library preparation pipelines require levels of leukocytes miRNAs (29.9 ± 3.3). A large portion RNA with a minimum RIN value, this study showed of these blood miRNAs are lost from plasma and serum, that many bodily fluids of low cell content may not as the separation and coagulation processes might be of meet such requirements. However, in this study li- the factors that affect miRNA distribution and recovery. braries were successfully constructed from RNA sam- Variations between plasma and serum miRNA content ples with low or no RIN value, suggesting that using results from the stress during coagulation [18]. RIN value as a sole determination of RNA quality Non-invasive fluids had lower miRNA fractions, keeping may not be universally applicable. in mind that they had a large percentage of unmapped El-Mogy et al. BMC Genomics (2018) 19:408 Page 13 of 24 Fig. 6 Top 10 most abundant miRNAs relative to all miRNA counts in each fluid. The 10 miRNAs that have the highest read counts in each fluid were illustrated relative to the total miRNA read counts of the fluid. Counts of the remaining miRNAs were summed up and illustrated as “other” El-Mogy et al. BMC Genomics (2018) 19:408 Page 14 of 24 Fig. 7 Relative abundance of the top 5 common miRNAs between the different fluids. Counts of the 5 common miRNAs are presented relative to the total miRNA counts of each fluid. The five common miRNAs represent a large fraction of the invasive fluids with the highest percentage in blood. They represent lower fractions in the non-invasive fluids reads (about 50% of saliva and 20% of urine reads). The The different samples are well clustered based on miR- relatively high bacterial content of saliva as well as the NAs according to sample type and their biology. In filtered and diluted nature of urine were key factors in addition, invasive and non-invasive fluids have distinct this result. Recent analysis of urine, saliva and plasma profiles and less variations between the fluids within the miRNAs from NGS data showed lower miRNA counts same group. PCA plots and heatmaps generated for from urine and saliva [70]. Profiling of miRNA in bodily tRNAs and piRNAs show a biology-related clustering, fluids by RT-PCR in an earlier study showed similar low with overlap between invasive and non-invasive fluids. urine RNA concentrations and low numbers of detected An interesting observation was the differential clustering miRNAs, while saliva had the highest number of miR- of tRNAs and piRNAs from urine and cell-free urine NAs among the studied fluids [72]. The information ob- samples based on the sex of the donor, indicating tained in our results could be used to guide methods for sex-related expression of these molecules. In addition, targeting specific biotypes in bodily fluids (via enrich- urine showed close clustering to serum, indicating that ment, separation, or depletion, for example). the latter might be the true liquid part of blood. The The various bodily fluids have unique miRNA, tRNA Table 7 Number of novel miRNA candidates in the different and piRNA profiles that characterize the type and origin bodily fluids of the fluid as seen from the PCA plots and heatmaps. Bodily Signal- miRDeep2 Number of Number of fluid to- score novel miRNA candidates present in noise candidates miRCarta database Table 6 Number of unique miRNAs in each bodily fluid at a minimum of 5 counts Blood 15.9 5 48 32 Bodily fluid All Invasive Non-invasive Leukocytes 11.5 5 16 9 Blood 94 96 Not compared Plasma 17 5 50 15 Leukocytes 30 35 Not compared Serum 17.1 5 20 9 a a Plasma 42 43 Not compared Saliva 4.8 54 0 Serum 1 1 Not compared Cell-Free 10.5 5 7 0 Saliva Saliva 1 Not compared 1 a a Urine 7.6 55 0 Cell-Free Saliva 3 Not compared 32 a a Cell-Free 7.7 55 0 Urine 1 Not compared 4 Urine Cell-Free Urine 3 Not compared 11 Signal-to-noise ratio is below the minimum accepted cutoff (10). The highest Three comparison groups were used: “All” indicating unique miRNA among all signal-to-noise ratio was used. However, predicted novel miRNAs at this value fluids, “invasive” for comparison within the invasive fluids (blood, leukocytes, might not be real. Predicted novel miRNAs with a non-significant p-value of serum and plasma), and “non-invasive” for comparison within the non-invasive the RNA minimum free energy of folding randomization test (Randfold) have fluids (saliva, cell-free saliva, urine and cell-free urine) not been counted El-Mogy et al. BMC Genomics (2018) 19:408 Page 15 of 24 Table 8 List of tRNAs that represents ≥1% of all tRNA counts in each bodily fluid Blood Leukocyte Plasma Serum Saliva Cell-Free Saliva Urine Cell-Free Urine Gly (84.1 ± 1.8) Gly (77.8 ± 1.9) Gly (79.9 ± 12.7) Gly (73.5 ± 8.2) Gly (83.7 ± 6.6) Gly (72.0 ± 7.4) Gly (86.5 ± 6.4) Gly (87.6 ± 6.0) Glu (6.7 ± 0.5) Glu (11.8 ± 2.6) Glu (9.5 ± 6.0) Glu (21.4 ± 7.5) Glu (7.9 ± 4.5) Glu (15.4 ± 2.6) Glu (8.0 ± 4.2) Glu (8.4 ± 3.9) Lys (2.1 ± 0.4) Lys (3.1 ± 0.7) SeC (1.8 ± 1.5) Val (1.5 ± 0.2) Val (2.1 ± 0.5) Lys (2.8 ± 0.9) Val (1.8 ± 1.9) Lys (1.5 ± 0.5) SeC (1.0 ± 0.3) Val (2.8 ± 0.4) His (1.5 ± 1.2) Lys (1.3 ± 0.5) Lys (1.6 ± 0.5) Ala (2.6 ± 2.7) Lys (1.6 ± 0.7) His (1.2 ± 0.1) Arg (1.2 ± 1.0) His (1.0 ± 0.5) Val (1.6 ± 0.2) Val (1.2 ± 0.2) Gln (1.2 ± 0.8) Asp (1.5 ± 0.7) Lys (1.1 ± 0.6) Pro (1.0 ± 0.9) Average percentage and standard deviation for each tRNA relative to the total tRNA content of the bodily fluid is included between the brackets Fig. 8 Principal component analysis of tRNAs in each bodily fluid. The plot was generated based on TMM-normalized tRNA counts. Samples of the same origin clustered closer to each other. However, urine samples are more dispersed from each other El-Mogy et al. BMC Genomics (2018) 19:408 Page 16 of 24 Fig. 9 Heatmap clustering of tRNAs in the various bodily fluids. The sex of sample donor is indicated as (F) for female donors or (M) for male donors. The analysis was generated using Z scores of TMM-normalized tRNA counts. The dendrogram shows distant clustering of urine samples (total and cell-free) based on sex. Male urine samples (total and cell-free) clustered close to serum while female urine samples (total and cell-free) are clustered with saliva differentiation of the three biotypes (miRNA, tRNA and where some molecules are enriched, while others are de- piRNA) in the different bodily fluids might be a result of pleted. None of the unique miRNAs were found among fluid origin and biological functions. The higher impact the top 20 most abundant molecules of each fluid. This of origin of fluid on miRNA distribution may refer to indicated the need for higher read depth to detect more specialized functions of miRNA in comparison to miRNAs that might have specific functions. Five piRNAs and tRNAs, which might be involved in more miRNAs: hsa-let-7a-5p, hsa-let-7f-5p, hsa-miR-191-5p, general biological roles. hsa-miR-26a-5p and hsa-miR-486-5p were common Non-invasive fluids had almost half the number of among the top 20 most abundant molecules in all fluids, identified miRNAs. For urine, this may result from the indicating shared origin or function. These five abundant filtering process by the kidneys. For saliva, the lower cir- common miRNAs represented a large portion of the culating nucleic acid content, relative to blood-related miRNA counts of invasive fluids (more than 40%), while fluids may be the cause [73–76]. The larger miRNA frac- they were relatively lower in non-invasive fluids (less tion and lower numbers of unmapped reads for the inva- than 30%). There was a set of 139 core miRNAs that are sive fluids explained their higher number of identified common among the different fluids and a set of 144 miRNAs. Almost every fluid had unique miRNAs that miRNAs that were shared between non-invasive fluids are specific only to that fluid, providing a specific signa- and blood. While the levels of these molecules may vary ture for each fluid. The number of common and unique between fluid types, they might be promising biomarker miRNAs between two fluids varied depending on bio- candidates that can be detected from multiple sources, logical relatedness. The invasive fluids collected in this including non-invasive fluids. study were more similar to the other invasive fluids, and An interesting observation was the variation between the non-invasive fluids were more similar to the other the most expressed miRNAs in the different fluids. In non-invasive fluids collected. This may be due to the fact blood, plasma and serum, hsa-miR-486-5p was the most that the invasive fluids collected were all blood derived. expressed, while hsa-miR-143-3p was the most expressed The fluid-specific unique miRNAs can result from differ- in saliva and cell-free saliva and hsa-miR-10b-5p was the ent cells secreting the different fluids. They may also re- predominant miRNA in urine and cell-free urine. We sult from the natural filtration process of some fluids, searched these 3 miRNAs on the human miRNA tissue El-Mogy et al. BMC Genomics (2018) 19:408 Page 17 of 24 Table 9 Predominant piRNAs in each bodily fluid Body fluid piRNA GenBank Accession number Chromosomal position Percentage Blood hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (92.3 + 1.8) Leukocyte hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (73.5 + 3.8) hsa_piR_000552 DQ570687 Homo_sapiens:22:38045003:38045030:Minus (5.6 + 0.6) Plasma hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (81.8 + 33.1) hsa_piR_019825 DQ597218 Homo_sapiens:1:227740227:227740256:Plus (15.2 + 30.4) Serum hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (94.0 + 2.7) Saliva hsa_piR_014620 DQ590013 Homo_sapiens:5:93930930:93930956:Minus (16.3 + 14.1) hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (14.9 + 1.7) hsa_piR_019521 DQ596805 Homo_sapiens:11:10487516:10487542:Minus (10.5 + 2.7) hsa_piR_000552 DQ570687 Homo_sapiens:22:38045003:38045030:Minus (7.6 + 2.7) hsa_piR_018780 DQ595807 Homo_sapiens:17:72068837:72068864:Plus (5.8 + 3.3) hsa_piR_000805 DQ571003 Homo_sapiens:1:212438966:212438997:Plus (5.5 + 1.5) Cell-Free Saliva hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (40.3 + 17.0) hsa_piR_016659 DQ592932 Homo_sapiens:14:22388242:22388267:Plus (7.4 + 6.7) hsa_piR_019521 DQ596805 Homo_sapiens:11:10487516:10487542:Minus (5.2 + 2.0) hsa_piR_000552 DQ570687 Homo_sapiens:22:38045003:38045030:Minus (5.1 + 3.3) hsa_piR_020450 DQ598104 Homo_sapiens:9:133350930:133350959:Plus (5.1 + 0.9) Urine hsa_piR_019825 DQ597218 Homo_sapiens:1:227740227:227740256:Plus (46.0 + 40.4) hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (15.6 + 12.9) hsa_piR_014620 DQ590013 Homo_sapiens:5:93930930:93930956:Minus (5.6 + 5.0) Cell-Free Urine hsa_piR_019825 DQ597218 Homo_sapiens:1:227740227:227740256:Plus (58.7 + 32.1) hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (14.5 + 12.5) hsa_piR_004153 DQ575660 Homo_sapiens:3:156861576:156861607:Plus (6.0 + 3.3) Molecules that represent an average of 5% or more of the entire piRNAs count of each bodily fluid are listed atlas [77] to identify the tissue of origin. High quantile nor- serum. Plasma did not suffer from the presence of a pre- malized expression levels of hsa-miR-486-5p were found in dominant molecule as did blood and serum. This vein and muscle specimens, while hsa-miR-143-3p was resulted in a high number of predicted novel miRNA highly elevated in esophagus and relatively high in colon, candidates compared to all the other bodily fluids, mak- bladder and prostate specimens. The expression levels of ing plasma a good alternative to blood. However, deple- hsa-miR-10b-5p were very high in the epididymis and ele- tion of hsa-miR-486-5p from blood and serum could be vated in kidney, colon and muscle specimens which ex- a useful tool to direct a greater proportion of reads to plains the relatively higher expression levels of this miRNA other miRNA sequences. in male urine samples. Recent studies indicate the import- Both saliva and urine did not offer the same advantage ance of hsa-miR-486-5P as a cancer biomarker in as the invasive fluids. They had lower miRNA content non-small cell lung cancer [78], gastric cancer [79]and oral and this affected their molecular diversity. The most tongue squamous cell carcinoma [80]. It may act as a expressed miRNA in saliva and cell-free saliva was tumorsuppressormiRNA [81]and mayalso beusedto hsa-miR-143-3p (10–15%). It is also the second most predict the efficacy of cancer vaccine treatment for colorec- expressed miRNA in saliva exosomes [83]. It is differen- tal cancer [82]. However, in many other cancer studies, this tially expressed in senescence [84] and as a tumor miRNA was not deregulated. suppressor in gliomas [85]. MicroRNA hsa-miR-10b-5p The large breadth of unique miRNAs found in blood, represented about 38–45% of urine and cell-free urine combined with an abundance of predicted novel miRNA miRNAs. It has been recently reported to be the most candidates demonstrates the superiority of blood for expressed miRNA in urine samples [86]. hsa-miR-10b-5p miRNA profiling and discovery. However, blood has high plays a role in carcinoma metastasis and is overexpressed levels of hsa-miR-486-5p, representing over 50% of its in colorectal cancer [87–90]. Due to its high expression, a miRNA content. Other bodily fluids that had a relatively lower proportion of reads will map to other miRNA se- high miRNA content are plasma and, to some extent, quences and its depletion should be considered as a El-Mogy et al. BMC Genomics (2018) 19:408 Page 18 of 24 Fig. 10 Principal component analysis of piRNAs in each bodily fluid. Analysis was generated based on TMM-normalized piRNA counts. Samples are clustered based on biology and fluids that share similar origin have close clustering. Close clustering is seen between the following fluid pairs: blood/leukocytes, saliva/cell-free saliva and urine/cell-free urine. Serum and plasma show distant clustering from the other fluids priority for improving the diversity of miRNAs within miRNAs and the predicted novel miRNA candidates of urine specimens. the invasive fluids (30 to 66%), while no matches were Novel miRNA prediction was not as efficient when found between miRCarta and the non-invasive fluids. dealing with non-invasive fluids. Their low This might be due to the large number of miRNA stud- signal-to-noise ratio made it hard to obtain accurate pre- ies from the invasive fluids as well as the higher counts diction. The only exception was the cell-free saliva, and diversity of miRNAs from these fluids compared to where a fair signal-to-noise ratio was achieved, and 7 the non-invasive fluids. This further indicates the higher novel miRNA candidates had been identified. It also had potential of the invasive fluids in novel miRNA predic- double the number of unique miRNAs compared to the tion. Although our findings are based on prediction of other saliva and urine samples. Due to the removal of candidate miRNA and have not been validated by an- mammalian cells by centrifugation, cell-free saliva usu- other technique, they showed the potential of these vari- ally captures more circulating miRNAs than total saliva ous fluids in novel miRNA discovery. or the cellular fraction of saliva [20, 91]. This made the While tRNA fragments are a minor portion of blood cell free saliva sample superior in terms of discovering and plasma small RNAs, they were well represented in unique and novel miRNAs candidates compared to the serum and saliva preparations (39–46%) and were the other non-invasive fluids. Matches were found between major small RNA species of urine and cell-free urine the miRCarta database of newly predicted human (> 90%). The main component of these tRNAs in all El-Mogy et al. BMC Genomics (2018) 19:408 Page 19 of 24 Fig. 11 Heatmap clustering of piRNAs in each bodily fluid. The sex of sample donor is indicated as (F) for female donors or (M) for male donors. Heatmap was generated using Z scores of TMM-normalized tRNA counts. The dendrogram shows distant clustering of urine samples (total and cell-free) based on sex. Male urine and cell-free urine samples clustered close to serum and plasma while female urine and most of cell-free urine samples are clustered with saliva El-Mogy et al. BMC Genomics (2018) 19:408 Page 20 of 24 Gly the fluids was tRNA (72.0 to 87.6%), followed by tRNA- plasma, serum, saliva, cell-free saliva, urine and cell-free Glu (6.7 to 21.4%). Urine samples, unlike the other fluids, urine), even with the high variations in the volumes used had high sample-to-sample variations, with tRNAs ran- for RNA purification as well as the high variations in ging from 47 to 98% of small RNAs. Similar variations concentrations of the isolated RNA. Despite the ease of have been reported in a recent study on urine from ovar- collection and handling of non-invasive fluids, they did ian cancer patients [86]. However, these variations may be not provide the same small RNA diversity and sample correlated to the sex of the individual, where male urine consistency as invasive fluids. However, this study has over 90% and female urine has about 70% or less. A showed that these samples can still be routinely profiled. larger study is needed to validate these findings. It is also Furthermore, the signatures of these non-invasive fluids interesting that the specific tRNA molecular composition are very likely linked to their origin. For example, urine of these tRNA fractions is consistent. Despite the overall may be a good candidate for studying diseases related to fluctuations in urine tRNA fractions, changes in the mo- organs such as kidney and bladder, although careful re- lecular signature of tRNA molecules might still be valid sult interpretation should be considered when investigat- for potential biomarker discovery. However, it may be lim- ing male and female urine, as their biotypes may be ited by the lower urine tRNA molecular diversity, com- sex-dependent. This observation is limited by the sample pared to blood, plasma and saliva. These variations in size of our study and is yet to be investigated on a large diversity were also observed between plasma, saliva and sample size study. An organ and fluid small RNA index urine in a recent study [70]. However, the percentage might be needed to track and correlate origins and func- abundance of molecules was different. tions of the various molecules. Processing of larger vol- Plasma and leukocytes contain relatively high amounts umes of urine, and bacterial removal from saliva of piRNA (8 and 5.8%, respectively). All the other bodily preparations might improve their NGS mapping to human fluids contain less than 2%, which is related to the small targets. In addition, depletion of specific molecules or se- fraction of piRNAs that are consistently being expressed lection/enrichment of target molecules from almost every in normal and cancer cells [92]. It is interesting that a sin- bodily fluid may significantly increase flow cell capacity gle piRNA molecule, hsa-piR-016658, was the most for target molecules and in turn provide a meaningful read expressed in all bodily fluids except in saliva, urine and depth. Successful clustering of bodily fluids based on their cell-free urine, where it was the second most abundant. miRNA distribution can be expanded to cohorts that can This molecule is associated with patients with prostate be differentiated according to their miRNA, and possibly cancer [93]. The most abundant piRNA molecule in urine in combination with other small RNAs. Therefore, a bio- and cell-free urine was hsa-piR-019825, which is deregu- marker within these fluids would be the overall biotype lated in colorectal cancer patients [93]. Given the high distribution and the molecular signature within these bio- number of human piRNAs, they may play a role as an im- types, rather than a single molecule. portant small RNA species with functional targets that are yet to be elucidated and correlated with various disease Additional files conditions. Recent studies have identified differentially expressed piRNA molecules as potential biomarkers of Additional file 1: Figure S2. Relative biotype distribution among the various cancers [94–97]. The relatively high levels of these various bodily fluids of each donor. (TIF 566 kb) molecules in plasma might prove important as potential Additional file 2: Table S1. Common miRNAs between invasive fluids. (DOCX 14 kb) biomarkers. The low piRNAs levels in the other fluids can Additional file 3: Table S2. Common miRNAs between non-invasive be overcome by size selection methods, albeit not easily as fluids. (DOCX 13 kb) they overlap with other small RNA species. Additional file 4: Table S3. Common miRNAs between non-invasive Plasma and serum had a large fraction of reads map- fluids and blood. (DOCX 13 kb) ping to miscellaneous RNA (misc_RNA) (58 and 35%, Additional file 5: Figure S1. Venn diagram showing the overlap respectively). Only 4 YRNA-derived small RNAs between blood and the non-invasive bodily fluids. (TIF 536 kb) (s-RNYs) sequences were elevated within the misc_RNA Additional file 6: Table S4. Common miRNAs between all fluids. (DOCX 13 kb) fractions of these two fluids: RNY4, RNY4P10, RNY4P7 Additional file 7: Table S5. Unique miRNAs detected in each bodily and YRNA.295. It has been previously reported that fluid. (DOCX 14 kb) s-RNYs are abundant in human serum and plasma [98]. Additional file 8: Table S6. Unique miRNAs detected in the invasive They are potential cancer biomarkers and regulators of bodily fluids. (DOCX 14 kb) inflammation and cell death [99, 100]. Additional file 9: Table S7. Unique miRNAs detected in the non- invasive bodily fluids. (DOCX 13 kb) Conclusions Additional file 10: Table S8. Candidate novel miRNAs detected by miRDeep2 in the different bodily fluids and their matching result to the Our study showed that it is possible to successfully ac- miRCarta database. (XLSX 43 kb) complish NGS of the different bodily fluids (blood, El-Mogy et al. BMC Genomics (2018) 19:408 Page 21 of 24 therapeutic strategies in oncology (review). Int. J. Oncol. [internet]. 2016;49:5–32. Additional file 11: Top 50 tDRs in each fluid with their quantification Available from: http://www.ncbi.nlm.nih.gov/pubmed/27175518 and coverage percentage. (XLSX 4259 kb) 3. Keller A, Meese E. Can circulating miRNAs live up to the promise of being Additional file 12: Profiles of the top 50 mature tDR in each fluid. minimal invasive biomarkers in clinical settings? Wiley Interdiscip. Rev. RNA (PDF 4737 kb) [internet]. 2016;7:148–156. Available from: http://www.ncbi.nlm.nih.gov/ pubmed/26670867 Additional file 13: Table S9. piRNAs that represent an average of 1% or more of the entire piRNA counts of each bodily fluid. (DOCX 18 kb) 4. Cristodero M, Polacek N. The multifaceted regulatory potential of tRNA- derived fragments. Non-coding RNA Investig. [Internet]. 2017;7–7. Available Additional file 14: Figure S3. Exogenous mapping of unmapped saliva from: http://ncri.amegroups.com/article/view/3820/4459 reads. (PDF 24 kb) 5. Guay C, Regazzi R. Circulating microRNAs as novel biomarkers for diabetes Additional file 15: Figure S4. Exogenous mapping of unmapped mellitus. Nat Rev Endocrinol [Internet] 2013;9:513–521. Available from: cell-free saliva reads. (PDF 24 kb) http://www.nature.com/doifinder/10.1038/nrendo.2013.86 6. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell [internet]. 2004;116:281–297. Available from: http://www.ncbi.nlm.nih.gov/ Abbreviations pubmed/14744438 CPM: Counts per million; lincRNA: Long intergenic noncoding RNA; 7. Kim VN, Han J, Siomi MC. Biogenesis of small RNAs in animals. Nat Rev Mol miRNA: MicroRNA; misc_RNA: Miscellaneous RNA; mRNA: Messenger RNA; Cell Biol [Internet] 2009;10:126–139. Available from: http://www.nature.com/ Mt_rRNA: Mitochondrial rRNA; Mt_tRNA: Mitochondrial tRNA; ncRNA: Non- doifinder/10.1038/nrm2632 coding RNA; NGS: Next generation sequencing; PCA: Principal component 8. Ortiz-Quintero B. Cell-free microRNAs in blood and other body fluids, as analysis; piRNA: Piwi-interacting RNA; RIN: RNA integrity number; cancer biomarkers. Cell Prolif. [internet]. 2016;49:281–303. Available from: rRNA: Ribosomal RNA; snoRNA: Small nucleolar RNA; snRNA: Small nuclear http://www.ncbi.nlm.nih.gov/pubmed/27218664 RNA; s-RNYs: Y RNA-derived small RNAs; TMM: Trimmed mean of M-values 9. Brennecke J, Hipfner DR, Stark A, Russell RB, Cohen SM. Bantam encodes a developmentally regulated microRNA that controls cell proliferation and Availability of data and materials regulates the proapoptotic gene hid in Drosophila. Cell [internet]. 2003;113: The raw sequencing files of the study are available on the NCBI Sequence 25–36. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12679032 Read Archive (SRA) through the following link: https://www.ncbi.nlm.nih.gov/ 10. Scapoli L, Palmieri A, Lo Muzio L, Pezzetti F, Rubini C, Girardi A, et al. sra/SRP136264, SRA accession: SRP136264. MicroRNA expression profiling of oral carcinoma identifies new markers of tumor progression. Int. J. Immunopathol. Pharmacol. 2010;23:1229–34. Authors’ contributions Available from: http://www.ncbi.nlm.nih.gov/pubmed/21244772 ME, BL, NR and YHA conceived, designed and initiated the project. ME, BL, 11. Liang Z, Bian X, Shim H. Downregulation of microRNA-206 promotes SM performed experiments. THA and DY contributed to the analysis tools. invasion and angiogenesis of triple negative breast cancer. Biochem. THA, LN and ME carried out data analysis. ME and PR drafted the manuscript. Biophys. Res. Commun. 2016;477:461–6. Available from: http://www.ncbi. BL, THA, NR and YHA helped to revise the manuscript. LN re-edited the lan- nlm.nih.gov/pubmed/27318091 guage of the manuscript. All authors read and approved the final 12. Friedman RC, Farh KKH, Burge CB, Bartel DP. Most mammalian mRNAs are manuscript. conserved targets of microRNAs. Genome Res. 2009;19:92–105. Available from: http://www.ncbi.nlm.nih.gov/pubmed/18955434 Ethics approval and consent to participate 13. miRBase: the microRNA database [Internet]. [cited 2018 Mar 7]. Available The study protocol and consent were reviewed and approved by Veritas IRB from: http://www.mirbase.org/cgi-bin/browse.pl?org=hsa Ethics Review Board (Veritas IRB, Montreal, Canada. IRB tracking number: 14. Arroyo JD, Chevillet JR, Kroh EM, Ruf IK, Pritchard CC, Gibson DF, et al. 16198–16:02:416–11-2017). Healthy volunteer donors were recruited by Argonaute2 complexes carry a population of circulating microRNAs advertising in local communities and all participants gave their written independent of vesicles in human plasma. [cited 2018 Mar 7]; Available informed consent. from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3064324/pdf/pnas. 201019055.pdf 15. Hunter MP, Ismail N, Zhang X, Aguda BD, Lee EJ, Yu L, et al. Detection of Competing interests microRNA expression in human peripheral blood microvesicles. PLoS One. The following authors are employees at Norgen Biotek Corp.: ME, BL, THA, 2008 [cited 2018 Mar 7];3. Available from: https://www.ncbi.nlm.nih.gov/ DY, LN, PR and NR. YHA is the President and CEO of Norgen Biotek Corp. SM pmc/articles/PMC2577891/pdf/pone.0003694.pdf received an Industrial Undergraduate Student Research Award from the 16. Valadi H, Ekström K, Bossios A, Sjöstrand M, Lee JJ, Lötvall JO. Exosome- Natural Sciences and Engineering Research Council of Canada (NSERC) at mediated transfer of mRNAs and microRNAs is a novel mechanism of Norgen Biotek Corp. Some of Norgen Biotek’s products have been used in genetic exchange between cells. Nat. Cell Biol; 2007 [cited 2018 Mar 7]; the study, however the study is for basic scientific exploratory purposes and 9:654–659. Available from: http://www.nature.com/articles/ncb1596. is not intended to promote or test any of Norgen Biotek’s products. Nature Publishing Group 17. Vickers KC, Remaley AT. Lipid-based carriers of microRNAs and intercellular Publisher’sNote communication. [cited 2018 Mar 7]; Available from: https://www.ncbi.nlm. Springer Nature remains neutral with regard to jurisdictional claims in nih.gov/pmc/articles/PMC5570485/pdf/nihms892052.pdf published maps and institutional affiliations. 18. Wang K, Zhang S, Weber J, Baxter D, Galas DJ. Export of microRNAs and microRNA-protective protein by mammalian cells. Nucleic Acids Res. 2010; Author details 38:7248–59. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20615901 1 2 Norgen Biotek Corp, Thorold, ON L2V 4Y6, Canada. Molecular Biology 19. DumacheR,Ciocan V, Muresan C, Rogobete AF,Enache A.CirculatingmicroRNAs Department, National Research Centre, Dokki, Giza, Egypt. Department of as promising biomarkers in forensic body fluids identification. Clin. Lab. 2015;61: Biological Sciences, Brock University, St. Catharines, ON L2S 3A1, Canada. 1129–35. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26554231 20. Park NJ, Zhou H, Elashoff D, Henson BS, Kastratovic DA, Abemayor E, et al. Received: 22 November 2017 Accepted: 11 May 2018 Salivary microRNA: Discovery, characterization, and clinical utility for oral cancer detection. Clin. Cancer Res. 2009;15:5473–7. Available from: http:// www.ncbi.nlm.nih.gov/pubmed/19706812 References 21. Tölle A, Jung M, Rabenhorst S, Kilic E, Jung K, Weikert S. Identification of 1. Storz G. An expanding universe of noncoding RNAs. Science. 2002;296: microRNAs in blood and urine as tumour markers for the detection of 1260–1263. Available from: http://www.sciencemag.org/cgi/doi/10.1126/ urinary bladder cancer. Oncol. Rep. 2013;30:1949–56. Available from: http:// science.1072249 www.ncbi.nlm.nih.gov/pubmed/23877086 2. Gambari R, Brognara E, Spandidos DA, Fabbri E. Targeting oncomiRNAs and 22. Suryawanshi S, Vlad AM, Lin HM, Mantia-Smaldone G, Laskey R, Lee M, et al. mimicking tumor suppressor miRNAs: Ew trends in the development of miRNA Plasma MicroRNAs as novel biomarkers for endometriosis and El-Mogy et al. BMC Genomics (2018) 19:408 Page 22 of 24 endometriosis-associated ovarian cancer. Clin. Cancer Res. 2013;19:1213–24. 42. Green D, Fraser WD, Dalmay T. Transfer RNA-derived small RNAs in the Available from: http://www.ncbi.nlm.nih.gov/pubmed/23362326 cancer transcriptome. Pflugers Arch. Eur. J. Physiol. 2016;468:1041–7. 23. Hu Z, Chen X, Zhao Y, Tian T, Jin G, Shu Y, et al. Serum microRNA signatures Available from: http://www.ncbi.nlm.nih.gov/pubmed/27095039 identified in a genome-wide serum microRNA expression profiling predict 43. Garcia-Silva MR, Cabrera-Cabrera F, Güida MC, Cayota A. Hints of tRNA-derived survival of non-small-cell lung cancer. J. Clin. Oncol. 2010;28:1721–6. small RNAs role in RNA silencing mechanisms. Genes (Basel), Available from. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20194856 2012;3:603–14. http://www.ncbi.nlm.nih.gov/pubmed/24705078 24. Gilad S, Meiri E, Yogev Y, Benjamin S, Lebanony D, Yerushalmi N, et al. 44. Goodarzi H, Liu X, Nguyen HCB, Zhang S, Fish L, Tavazoie SF. Endogenous Serum microRNAs are promising novel biomarkers. PLoS One. 2008;3:e3148. tRNA-derived fragments suppress breast cancer progression via YBX1 Available from: http://www.ncbi.nlm.nih.gov/pubmed/18773077 displacement. Cell. 2015;161:790–802. Available from: http://www.ncbi.nlm. 25. Xie Z, Chen G, Zhang X, Li D, Huang J, Yang C, et al. Salivary MicroRNAs as nih.gov/pubmed/25957686 Promising Biomarkers for Detection of Esophageal Cancer. Lo AWI, editor. 45. Maute RL, Schneider C, SumazinP,HolmesA,CalifanoA,BassoK, et al. tRNA- PLoS One. 2013 [cited 2016 Aug 10];8:e57502. Available from: http://dx.plos. derived microRNA modulates proliferation and the DNA damage response and org/10.1371/journal.pone.0057502. Public Library of Science is down-regulated in B cell lymphoma. Proc. Natl. Acad. Sci. 2013;110:1404–1409. 26. Kosaka N, Iguchi H, Ochiya T. Circulating microRNA in body fluid: a new Available from: http://www.pnas.org/lookup/doi/10.1073/pnas.1206761110 potential biomarker for cancer diagnosis and prognosis. Cancer Sci. 2010;101: 46. Atala A. Re: sex hormone-dependent tRNA halves enhance cell proliferation 2087–92. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20624164 in breast and prostate cancers. J. Urol. 2016;195:1168–9. Available from: 27. Zhao N, Jin L, Fei G, Zheng Z, Zhong C. Serum microRNA-133b is http://www.ncbi.nlm.nih.gov/pubmed/26124144 associated with low ceruloplasmin levels in Parkinson’sdisease. Park. 47. Balatti V, Pekarsky Y, Croce CM. Role of the tRNA-derived small RNAs in Cancer: new potential biomarkers and target for therapy [internet]. 1st ed. Elsevier Inc.; Relat. Disord, Available from. 2014;20:1177–80. http://www.ncbi.nlm.nih. gov/pubmed/25218846 2017. Available from: https://doi.org/10.1016/bs.acr.2017.06.007.Adv.CancerRes 28. Kirchner S, Ignatova Z. Emerging roles of tRNA in adaptive translation, 48. Hatfield SD, Shcherbata HR, Fischer KA, Nakahara K, Carthew RW, signalling dynamics and disease. Nat. Rev. Genet. 2015;16:98–112. Available Ruohola-Baker H. Stem cell division is regulated by the microRNA from: http://www.nature.com/doifinder/10.1038/nrg3861 pathway. Nature. 2005;435:974–978. Available from: http://www.nature. 29. Selitsky SR, Baran-Gale J, Honda M, Yamane D, Masaki T, Fannin EE, et al. com/doifinder/10.1038/nature03816 Small tRNA-derived RNAs are increased and more abundant than 49. Rouget C, Papin C, Boureux A, Meunier AC, Franco B, Robine N, et al. microRNAs in chronic hepatitis B and C. Sci. Rep. 2015;5:7675. Available Maternal mRNA deadenylation and decay by the piRNA pathway in the from: http://www.nature.com/articles/srep07675 early Drosophila embryo. Nature. 2010;467:1128–1132. Available from: http:// 30. Ivanov P, Emara MM, Villen J, Gygi SP, Anderson P. Angiogenin-induced www.nature.com/doifinder/10.1038/nature09465 tRNA fragments inhibit translation initiation. Mol. Cell. 2011;43:613–23. 50. piRNABank: : a web resource on classified and clustered Piwi-interacting RNAs Available from: http://www.ncbi.nlm.nih.gov/pubmed/21855800 [Internet]. [cited 2018 Mar 7]. Available from: http://pirnabank.ibab.ac.in/stats.html 31. Yamasaki S, Ivanov P, Hu GF, Anderson P. Angiogenin cleaves tRNA and 51. Aravin AA, Sachidanandam R, Bourc’his D, Schaefer C, Pezic D, Toth KF, et al. promotes stress-induced translational repression. J. Cell Biol. 2009;185:35–42. A piRNA Pathway Primed by Individual Transposons Is Linked to De Novo Available from: http://www.ncbi.nlm.nih.gov/pubmed/19332886 DNA Methylation in Mice. Mol. Cell. 2008;31:785–99. Available from: http:// 32. Saikia M, Jobava R, Parisien M, Putnam A, Krokowski D, Gao X-H, et al. www.ncbi.nlm.nih.gov/pubmed/18922463 Angiogenin-Cleaved tRNA Halves Interact with Cytochrome c, Protecting 52. Aravin AA, Bourc’his D. Small RNA guides for de novo DNA methylation in Cells from Apoptosis during Osmotic Stress. Mol. Cell. Biol. 2014;34:2450– mammalian germ cells. Genes Dev. 2008;22:970–5. Available from: http:// 2463. Available from: http://mcb.asm.org/cgi/doi/10.1128/MCB.00136-14 www.ncbi.nlm.nih.gov/pubmed/18413711 33. Schaffer AE, Eggens VRC, Caglayan AO, Reuter MS, Scott E, Coufal NG, et al. 53. Hirakata S, Siomi MC. piRNA biogenesis in the germline: From transcription CLP1 founder mutation links tRNA splicing and maturation to cerebellar of piRNA genomic sources to piRNA maturation. Biochim. Biophys. Acta - development and neurodegeneration. Cell. 2014;157:651–63. Available from: Gene Regul. Mech. 2016;1859:82–92. Available from: https://doi.org/10.1016/ http://www.ncbi.nlm.nih.gov/pubmed/24766810 j.bbagrm.2015.09.002. Elsevier B.V 54. Esteller M. Non-coding RNAs in human disease. Nat. Rev. Genet. [Internet]. 2011; 34. Gebetsberger J, Zywicki M, Künzi A, Polacek N. TRNA-derived fragments 12:861–874. Available from: http://www.nature.com/doifinder/10.1038/nrg3074 target the ribosome and function as regulatory non-coding RNA in Haloferax volcanii. Archaea. 2012;2012:260909. Available from: http://www. 55. Zhang J, Chiodini R, Badr A, Zhang G. The impact of next-generation ncbi.nlm.nih.gov/pubmed/23326205 sequencing on genomics [Internet]. J. Genet. Genomics. 2011 [cited 2017 35. Gebetsberger J, Wyss L, Mleczko AM, Reuther J, Polacek N. A tRNA-derived Jul 16]. p. 95–109. Available from: https://www.ncbi.nlm.nih.gov/pmc/ fragment competes with mRNA for ribosome binding and regulates articles/PMC3076108/pdf/nihms-282401.pdf translation during stress. RNA Biol. 2017;14:1364–73. Available from: http:// 56. Byron SA, Van Keuren-Jensen KR, Engelthaler DM, Carpten JD, Craig DW. www.ncbi.nlm.nih.gov/pubmed/27892771 Translating RNA sequencing into clinical diagnostics: Opportunities and 36. Sharma U, Conine CC, Shea JM, Boskovic A, Derr AG, Bing XY, et al. challenges. Nat. Rev. Genet. 2016 [cited 2017 Jul 16];17:257–271. Available Biogenesis and function of tRNA fragments during sperm maturation and from: http://www.nature.com/doifinder/10.1038/nrg.2016.10 fertilization in mammals. Science. 2016;351:391–396. Available from: http:// 57. Shore S, Henderson JM, Lebedev A, Salcedo MP, Zon G, McCaffrey AP, et al. www.sciencemag.org/cgi/doi/10.1126/science.aad6780 Small RNA library preparation method for next-generation sequencing using chemical modifications to prevent adapter dimer formation. PLoS 37. Venkatesh T, Suresh PS, Tsutsumi R. TRFs: miRNAs in disguise. Gene. 2016; 579:133–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26743126 One. 2016 [cited 2017 Jul 16];11. Available from: http://journals.plos.org/ plosone/article/file?id=10.1371/journal.pone.0167009&type=printable 38. Haussecker D, Huang Y, Lau A, Parameswaran P, Fire AZ, Kay MA. Human tRNA-derived small RNAs in the global regulation of RNA 58. Chen X, Ba Y, Ma L, Cai X, Yin Y, Wang K, et al. Characterization of silencing. Rna. 2010;16:673–695. Available from: http://rnajournal.cshlp. microRNAs in serum: A novel class of biomarkers for diagnosis of cancer org/cgi/doi/10.1261/rna.2000810 and other diseases. Cell Res. 2008;18:997–1006. Available from: http://www. 39. Elbarbary RA, Takaku H, Uchiumi N, Tamiya H, Abe M, Takahashi M, et al. nature.com/doifinder/10.1038/cr.2008.282 Modulation of gene expression by human cytosolic tRNase Z(L) through 5′- 59. Mitchell PS, Parkin RK, Kroh EM, Fritz BR, Wyman SK, Pogosova-Agadjanyan half-tRNA. PLoS One. 2009;4:e5908. Available from: http://www.ncbi.nlm.nih. EL, et al. Circulating microRNAs as stable blood-based markers for cancer gov/pubmed/19526060 detection. Proc. Natl. Acad. Sci. 2008;105:10513–10518. Available from: 40. Ghildiyal M, Zamore PD. Small silencing RNAs: an expanding universe. Nat http://www.pnas.org/cgi/doi/10.1073/pnas.0804549105 Rev Genet [Internet] 2009;10:94–108. Available from: http://www.nature. 60. Wang J, Zhang KY, Liu SM, Sen S. Tumor-associated circulating micrornas as com/doifinder/10.1038/nrg2504 biomarkers of cancer. Molecules. 2014;19:1912–38. Available from: http:// 41. Kumar A, Karmarkar AM, Tan A, Graham JE, Arcari CM, Ottenbacher KJ, et al. www.ncbi.nlm.nih.gov/pubmed/24518808 The effect of obesity on incidence of disability and mortality in Mexicans 61. Cheng J, Guo JM, Xiao BX, Miao Y, Jiang Z, Zhou H, et al. PiRNA, the new aged 50 years and older. Salud Publica Mex. 2015 [cited 2017 Jul 12];57: non-coding RNA, is aberrantly expressed in human cancer cells. Clin. Chim. S31–S38. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/ Acta. 2011;412:1621–5. Available from: http://www.ncbi.nlm.nih.gov/ PMC4572697/pdf/nihms673180.pdf pubmed/21616063 El-Mogy et al. BMC Genomics (2018) 19:408 Page 23 of 24 62. Hashim A, Rizzo F, Marchese G, Ravo M, Tarallo R, Nassa G, et al. RNA 80. Chen Z, Yu T, Cabay RJ, Jin Y, Mahjabeen I, Luan X, et al. miR-486-3p, miR- sequencing identifies specific PIWI-interacting small non-coding RNA 139-5p, and miR-21 as Biomarkers for the Detection of Oral Tongue expression patterns in breast cancer. Oncotarget. 2014;5:9901–10. Available Squamous Cell Carcinoma. Biomark. Cancer [Internet]. 2017;9:1–8. Available from: http://www.oncotarget.com/fulltext/2476 from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5224348/ 63. Li Y, Wu X. Piwi-Interacting RNAs (piRNAs) Are Dysregulated in Renal Cell 81. Ye H, Yu X, Xia J, Tang X, Tang L, Chen F. MiR-486-3p targeting ECM1 Carcinoma and Associated with Tumor Metastasis and Cancer-Specific represses cell proliferation and metastasis in cervical cancer. Biomed. Survival. Mol. Med. 2015;21:1. Available from: http://www.molmed.org/ Pharmacother. [Internet], Available from. 2016;80:109–14. http://www.ncbi. content/pdfstore/14_203_Li.pdf nlm.nih.gov/pubmed/27133046 64. Reeves ME, Firek M, Jliedi A, Amaar YG. Identification and characterization of 82. Shindo Y, Hazama S, Nakamura Y, Inoue Y, Kanekiyo S, Suzuki N, et al. miR- RASSF1C piRNA target genes in lung cancer cells. Oncotarget. 2017; 196b, miR-378a and miR-486 are predictive biomarkers for the efficacy of Available from: http://www.oncotarget.com/abstract/15965 vaccine treatment in colorectal cancer. Oncol. Lett. [Internet]. 2017;14:1355– 65. Chen C, Ridzon DA, Broomer AJ, Zhou Z, Lee DH, Nguyen JT, et al. Real- 62. Available from: http://www.ncbi.nlm.nih.gov/pubmed/28789351 time quantification of microRNAs by stem-loop RT-PCR. Nucleic Acids Res. 83. Ogawa Y, Taketomi Y, Murakami M, Tsujimoto M, Yanoshita R. Small 2005 [cited 2017 Jan 1];33:e179. Available from: https://academic.oup.com/ RNA transcriptomes of two types of exosomes in human whole saliva nar/article-lookup/doi/10.1093/nar/gni178 determined by next generation sequencing. Biol Pharm Bull [Internet]. 66. Subramanian SL, Kitchen RR, Alexander R, Carter BS, Cheung KH, 2013;36:66–75. Available from: http://www.ncbi.nlm.nih.gov/entrez/ Laurent LC, et al. Integration of extracellular RNA profiling data using query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids= metadata, biomedical ontologies and Linked Data technologies. J. 23302638 Extracell. Vesicles. 2015;4:27497. Available from: http://www.ncbi.nlm.nih. 84. Bonifacio LN, Jarstfer MB. MiRNA profile associated with replicative gov/pubmed/26320941 senescence, extended cell culture, and ectopic telomerase expression in 67. Robinson MD, Oshlack A. A scaling normalization method for human foreskin fibroblasts. PLoS One [Internet]. 2010;5:1–8. Available from: differential expression analysis of RNA-seq data. Genome Biol. 2010;11: http://www.ncbi.nlm.nih.gov/pubmed/20824140 85. Wang L, Shi Z, Jiang C, Liu X, Chen Q, Qian X, et al. MiR-143 acts as a tumor R25. Available from: http://genomebiology.biomedcentral.com/articles/10. 1186/gb-2010-11-3-r25 suppressor by targeting N-RAS and enhances temozolomide-induced apoptosis in glioma. Oncotarget [Internet]. 2014;5:5416–27. Available from: 68. Dhahbi JM, Atamna H, Boffelli D, Magis W, Spindler SR, Martin DIK. Deep http://www.oncotarget.com/fulltext/2116 sequencing reveals novel micrornas and regulation of microRNA expression during cell senescence. PLoS One. 2011;6:e20509. Available from: http:// 86. Zhou K, Spillman MA, Behbakht K, Komatsu JM, Abrahante JE, Hicks D, et al. www.ncbi.nlm.nih.gov/pubmed/21637828 A method for extracting and characterizing RNA from urine: For 69. Backes C, Fehlmann T, Kern F, Kehl T, Lenhof HP, Meese E, et al. MiRCarta: A downstream PCR and RNAseq analysis. Anal. Biochem. [Internet]. 2017;536: central repository for collecting miRNA candidates. Nucleic Acids Res. 2018 8–15. Available from: https://doi.org/10.1016/j.ab.2017.08.003. Elsevier Inc [cited 2018 Apr 12];46:D160–D167. Available from: https://www.ncbi.nlm.nih. 87. Zhang L, Sun J, Wang B, Ren JC, Su W, Zhang T. MicroRNA-10b triggers gov/pmc/articles/PMC5753177/pdf/gkx851.pdf the epithelial–mesenchymal transition (EMT) of laryngeal carcinoma 70. Yeri A, Courtright A, Reiman R, Carlson E, Beecroft T, Janss A, et al. Total Hep-2 cells by directly targeting the E-cadherin. Appl. Biochem. extracellular small RNA profiles from plasma, saliva, and urine of healthy Biotechnol. [internet]. 2015;176:33–44. Available from: http://www.ncbi. subjects. Sci. Rep. 2017 [cited 2017 Jul 11];7:44061. Available from: http:// nlm.nih.gov/pubmed/25875782 www.nature.com/articles/srep44061 88. Xiao H, Li H, Yu G, Xiao W, Hu J, Tang K, et al. MicroRNA-10b promotes migration and invasion through KLF4 and HOXD10 in human bladder 71. Park NJ, Zhou X, Yu T, Brinkman BMN, Zimmermann BG, Palanisamy V, et al. cancer. Oncol. Rep. [Internet]. 2014;31:1832–8. Available from: http://www. Characterization of salivary RNA by cDNA library analysis. Arch. Oral Biol. ncbi.nlm.nih.gov/pubmed/26311318 2007 [cited 2017 Oct 11];52:30–35. Available from: https://www.ncbi.nlm.nih. gov/pmc/articles/PMC2743855/pdf/nihms15843.pdf 89. Ma Z, Chen Y, Min L, Li L, Huang H, Li J, et al. Augmented miR-10b expression 72. Weber JA, Baxter DH, Zhang S, Huang DY, Huang KH, Lee MJ, et al. The associated with depressed expression of its target gene KLF4 involved in microRNA spectrum in 12 body fluids. Clin Chem. 2010;56:1733–41. gastric carcinoma. Int. J. Clin. Exp. Pathol. [Internet]. 2015;8:5071–9. Available 73. Gallo A, Tandon M, Alevizos I, Illei GG. The majority of microRNAs detectable from: http://www.ncbi.nlm.nih.gov/pubmed/26191201 in serum and saliva is concentrated in exosomes. Afarinkia K, editor. PLoS 90. Abdelmaksoud-Dammak R, Chamtouri N, Triki M, Saadallah-Kallel A, Ayadi One. 2012 [cited 2016 Aug 10];7:e30679. Available from: http://dx.plos.org/ W, Charfi S, et al. Overexpression of miR-10b in colorectal cancer patients: 10.1371/journal.pone.0030679. Public Library of Science Correlation with TWIST-1 and E-cadherin expression. Tumor Biol. [Internet]. 2017 [cited 2017 Oct 20];39:101042831769591. Available from: http:// 74. Majem B, Rigau M, Reventós J, Wong DT. Non-coding RNAs in saliva: journals.sagepub.com/doi/10.1177/1010428317695916. SAGE emerging biomarkers for molecular diagnostics. Int. J. Mol. Sci. 2015;16: PublicationsSage UK: London, England 8676–98. Available from: http://www.ncbi.nlm.nih.gov/pubmed/25898412 75. Spielmann N, Ilsley D, Gu J, Lea K, Brockman J, Heater S, et al. The human 91. Lin X, Lo H-C, Wong DTW, Xiao X. Noncoding RNAs in human saliva as salivary RNA transcriptome revealed by massively parallel sequencing. Clin. potential disease biomarkers. Front Genet [Internet] 2015;6:1–6. Available Chem. 2012;58:1314–21. Available from: http://www.ncbi.nlm.nih.gov/ from: http://www.frontiersin.org/RNA/10.3389/fgene.2015.00175/full pubmed/22773539 92. Martinez VD, Vucic EA, Thu KL, Hubaux R, Enfield KSS, Pikor LA, et al. Unique 76. Li M, Zeringer E, Barta T, Schageman J, Cheng A, Vlassov A V. Analysis of the somatic and malignant expression patterns implicate PIWI-interacting RNAs RNA content of the exosomes derived from blood serum and urine and its in cancer-type specific biology. Sci. Rep. [Internet]. 2015 [cited 2017 Oct 10]; potential as biomarkers. Philos. Trans. R. Soc. B Biol. Sci. 2014;369:20130502. 5. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4444957/ Available from: http://rstb.royalsocietypublishing.org/cgi/doi/10.1098/rstb. pdf/srep10423.pdf 2013.0502 93. Yuan T, Huang X, Woodcock M, Du M, Dittmar R, Wang Y, et al. Plasma 77. Ludwig N, Leidinger P, Becker K, Backes C, Fehlmann T, Pallasch C, et al. extracellular RNA profiles in healthy and cancer patients. Sci. Rep. [Internet]. Distribution of miRNA expression across human tissues. Nucleic Acids Res. 2016;6:19413. Available from: http://www.nature.com/articles/srep19413 2016 [cited 2018 Apr 13];44:3865–3877. Available from: https://www.ncbi. 94. Lim SL, Ricciardelli C, Oehler MK, De Arao Tan IMD, Russell D, Grützner F. nlm.nih.gov/pmc/articles/PMC4856985/pdf/gkw116.pdf Overexpression of piRNA pathway genes in epithelial ovarian cancer. PLoS 78. Sromek M, Glogowski M, Chechlinska M, Kulinczak M, Szafron L, Zakrzewska One [Internet]. 2014 [cited 2017 Oct 10];9. Available from: http://journals. K, et al. Changes in plasma miR-9, miR-16, miR-205 and miR-486 levels after plos.org/plosone/article/file?id=10.1371/journal.pone.0099687&type= non-small cell lung cancer resection. Cell. Oncol. 2017;40:529–36. Available printable from: http://www.ncbi.nlm.nih.gov/pubmed/28634901 95. Müller S, Raulefs S, Bruns P, Afonso-Grunz F, Plötner A, Thermann R, et al. 79. Sierzega M, Kaczor M, Kolodziejczyk P, Kulig J, Sanak M, Richter P. Evaluation Next-generation sequencing reveals novel differentially regulated mRNAs, of serum microRNA biomarkers for gastric cancer based on blood and lncRNAs, miRNAs, sdRNAs and a piRNA in pancreatic cancer. Mol. Cancer tissue pools profiling: The importance of MIR-21 and MIR-331. Br. J. Cancer. [Internet]. 2015;14:94. Available from: http://molecular-cancer.biomedcentral. 2017;117:266–73. Available from: http://www.nature.com/doifinder/10.1038/ com/articles/10.1186/s12943-015-0358-5 bjc.2017.190 El-Mogy et al. BMC Genomics (2018) 19:408 Page 24 of 24 96. Martinez VD, Enfield KSS, Rowbotham DA, Lam WL. An atlas of gastric PIWI- interacting RNA transcriptomes and their utility for identifying signatures of gastric cancer recurrence. Gastric Cancer [internet]. 2016;19:660–5. Available from: http://www.ncbi.nlm.nih.gov/pubmed/25779424 97. Koduru S V, Tiwari AK, Hazard SW, Mahajan M, Ravnic DJ. Exploration of small RNA-seq data for small non-coding RNAs in Human Colorectal Cancer. J. Genomics [Internet]. 2017 [cited 2017 Oct 10];5:16–31. Available from: http:// www.jgenomics.com/v05p0016.htm 98. Dhahbi JM, Spindler SR, Atamna H, Boffelli D, Mote P, Martin DIK. 5’-YRNA fragments derived by processing of transcripts from specific YRNA genes and pseudogenes are abundant in human serum and plasma. Physiol. Genomics [Internet]. 2013;45:990–8. Available from: http://physiolgenomics. physiology.org/cgi/doi/10.1152/physiolgenomics.00129.2013 99. Dhahbi, Spinder S, Atamna H, Boffelli D, Martin D. Deep Sequencing of Serum Small RNAs Identifies Patterns of 5&amp;#39; tRNA Half and YRNA Fragment Expression Associated with Breast Cancer. Biomark. Cancer [Internet]. 2014 [cited 2016 Apr 11];6:37. Available from: http://www.la-press. com/deep-sequencing-of-serum-small-rnas-identifies-patterns-of-5-trna-half- article-a4553 100. Hizir Z, Bottini S, Grandjean V, Trabucchi M, Repetto E. RNY (YRNA)-derived small RNAs regulate cell death and inflammation in monocytes/ macrophages. Cell Death Dis. [Internet]. 2017 [cited 2017 Oct 11];8:e2530. Available from: http://www.nature.com/doifinder/10.1038/cddis.2016.429 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Genomics Springer Journals

Diversity and signature of small RNA in different bodily fluids using next generation sequencing

Free
24 pages

Loading next page...
 
/lp/springer_journal/diversity-and-signature-of-small-rna-in-different-bodily-fluids-using-mzE4KuSeoN
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s).
Subject
Life Sciences; Life Sciences, general; Microarrays; Proteomics; Animal Genetics and Genomics; Microbial Genetics and Genomics; Plant Genetics and Genomics
eISSN
1471-2164
D.O.I.
10.1186/s12864-018-4785-8
Publisher site
See Article on Publisher Site

Abstract

Background: Small RNAs are critical components in regulating various cellular pathways. These molecules may be tissue-associated or circulating in bodily fluids and have been shown to associate with different tumors. Next generation sequencing (NGS) on small RNAs is a powerful tool for profiling and discovery of microRNAs (miRNAs). Results: In this study, we isolated total RNA from various bodily fluids: blood, leukocytes, serum, plasma, saliva, cell-free saliva, urine and cell-free urine. Next, we used Illumina’s NGS platform and intensive bioinformatics analysis to investigate the distribution and signature of small RNAs in the various fluids. Successful NGS was accomplished despite the variations in RNA concentrations among the different fluids. Among the fluids studied, blood and plasma were found to be the most promising fluids for small RNA profiling as well as novel miRNA prediction. Saliva and urine yielded lower numbers of identifiable molecules and therefore were less reliable in small RNA profiling and less useful in predicting novel molecules. In addition, all fluids shared many molecules, including 139 miRNAs, the most abundant tRNAs, and the most abundant piwi-interacting RNAs (piRNAs). Fluids of similar origin (blood, urine or saliva) displayed closer clustering, while each fluid still retains its own characteristic signature based on its unique molecules and its levels of the common molecules. Donor urine samples showed sex-dependent differential clustering, which may prove useful for future studies. Conclusions: This study shows the successful clustering and unique signatures of bodily fluids based on their miRNA, tRNA and piRNA content. With this information, cohorts may be differentiated based on multiple molecules from each small RNA class by a multidimensional assessment of the overall molecular signature. Keywords: miRNA, tRNA, piRNA, Next generation sequencing, Blood, Plasma, Serum, Saliva, Urine Background bind to and downregulate messenger RNAs (mRNAs) Small RNAs are a class of mainly non-coding RNAs [5]. They down regulate gene expression, playing a major (ncRNAs) characterized by their small nucleotide length role in essential biological pathways, such as differenti- of less than 200 nt [1]. Within this class there are key ation, proliferation, metastasis and apoptosis. [6–11]. RNA types with a size range of 14–35 nt that are highly MicroRNAs represent an entire layer of gene expression important for diagnostic biomarker discovery and the regulation, regulating more than 50% of protein coding development of therapeutic agents [2–4]. These include mRNAs in mammalian cells [12]. To date, 2588 human microRNAs (miRNAs), transfer RNA-derived RNAs mature miRNAs have been identified and are currently (tDRs) and Piwi-interacting RNAs (piRNAs). Micro- included in miRBase 21 [13]. Aside from being found in RNAs are non-coding molecules of about 19–23 nt that tissues and cells, miRNAs are found in bodily fluids in extracellular vesicles or in complexes with argonaute or lipoproteins [14–17]. They have been reported in bodily * Correspondence: melmogy@norgenbiotek.com; melmogy@hotmail.com fluids such as blood, plasma, serum, urine, tears, saliva, Norgen Biotek Corp, Thorold, ON L2V 4Y6, Canada breast milk, amniotic fluid, seminal fluid and colostrum Molecular Biology Department, National Research Centre, Dokki, Giza, Egypt Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. El-Mogy et al. BMC Genomics (2018) 19:408 Page 2 of 24 [18–23]. MicroRNAs have been linked to many diseases Methods and are highly promising molecular biomarkers [24–27]. Sample collection, preparation and RNA purification Mature tRNAs and nascent pre-tRNA transcripts are Blood, saliva and urine were collected from 4 healthy processed enzymatically to produce well defined tDRs in donors, 2 females and 2 males between the ages of 20 to a regulated process, suggesting that they are not random 30. The various bodily fluids were collected from each degradation products [28, 29]. Sizes of tDRs range from individual within a 2-h period. Collection and sample 30 to 35 nt for tRNA halves and 14 to 26 nt for the de-identification was performed under an IRB approved shorter fragments [4]. Various studies suggest that tDRs protocol (16198–16:02:416–11-2017). Three 10 mL are involved in different functions including stress re- blood samples were collected from each donor. Two of sponses in human diseases, where they act as inhibitors these samples were collected in Vacutainer® plastic of global translation and transcription regulation [30– EDTA tubes (BD, USA) and used for RNA isolation from 36]. Like miRNAs, they can conduct specific gene silen- whole blood, leukocytes and plasma. RNA was isolated cing and have potential as cancer biomarkers [7, 37–47]. directly from 0.2 mL of blood using the Total RNA Puri- Finally, piRNAs have a nucleotide size range between 26 fication Plus Kit (Norgen Biotek Corp., Canada). Leuko- to 31. They modulate different gene expression pathways cytes were prepared from 0.5 mL of blood by using the by interacting with Piwi proteins [48, 49]. Currently Leukocyte RNA Purification Kit (Norgen Biotek Corp., there are 23,439 piRNA molecules in the piRNABank Canada). One entire tube was used to prepare plasma [50]. They are abundant in gonads and mediate trans- and was centrifuged at 200 RCF for 10 min at room poson repression to conserve genome integrity [51–53]. temperature. Plasma was collected and stored at − 70 °C The use of next generation sequencing (NGS) technol- until isolation. The last blood sample was collected in a ogy in small RNA detection has advanced research in the Vacutainer® glass serum tube with silicon coated interior field at unprecedented speed. NGS shines light on the key (BD, USA) and used for RNA isolation from serum. The role of ncRNAs in transcriptome regulation in healthy and tube was left to stand at room temperature for 45 min, disease conditions and accelerates the profiling and dis- and then it was centrifuged at 1300 RCF for 15 min. covery of molecules [54]. The technology enables the ana- Serum was collected and stored at − 70 °C until isola- lysis of multiple samples in parallel and provides precise tion. Both plasma and serum RNA were isolated from quantification of each molecule, making it superior to pre- 0.2 mL using the Plasma/Serum RNA Isolation Mini Kit vious genomic technologies. The demonstrated capacities (Norgen Biotek Corp., Canada). All kits were used of NGS have led to advances in biological and medical according to the manufacturer’s instructions. genomics and transcriptomics [55–57]. Two milliliters of saliva was collected from each donor Solid tumor miRNAs are well represented in bodily into Falcon 50 mL centrifuge tubes (BD, USA) and fluids, indicating their importance as cancer biomarkers 0.3 mL was used directly for saliva RNA isolation using [58–60]. Almost all bodily fluids from healthy individ- the Total RNA Purification Kit (Norgen Biotek Corp., uals contain miRNAs. Therefore, bodily fluids represent Canada), according to the manufacturer’s instructions. an excellent candidate for non-invasive detection of Another 0.3 mL of saliva was transferred into a 1.5 mL miRNAs and have been used in applications such as tube (Eppendorf, Germany) and spun down at 200 RCF cancer biomarker discovery [8]. Transfer RNA-derived for 10 min to remove cells, and then the supernatant small RNAs are thought to have a dual function, as they was used for cell-free saliva RNA purification by the act as suppressors and oncogenes [42]. In addition, al- same kit. A similar approach was used for urine sample tered piRNAs levels were found to be associated with preparation, where 100 mL of urine was collected from lung, breast, gastric and colon cancers [61–64]. How- each donor into disposable cups (Sarstedt, Germany). ever, no comprehensive study has been reported on tDRs RNA was isolated from 30 mL of urine using the Urine or piRNAs in different bodily fluids. Cell-Free Circulating RNA Purification Maxi Kit (Nor- The field of small RNA is expanding with the profiling gen Biotek Corp., Canada). The kit’s procedure was and discovery of molecules in various disease conditions modified by skipping the initial centrifugation steps to and treatments. Therefore, it is important to explore the purify RNA from total urine. Another urine sample was small RNA content in normal individuals to better processed by the same kit without any modifications to understand the small RNA profile in each fluid as well the manufacturer’s procedures to isolate RNA from as their relative distribution among the different fluids. 30 mL of cell-free urine. Purified RNA from all samples To gain insights into the distribution and signature of were tested for positive amplification by miR-21 small RNA in bodily fluids, we carried out a comparative stem-loop RT-PCR [65]. RNA concentration was then study on RNA from different fluids collected from the estimated by the Agilent 2100 Bioanalyzer System (Agi- same donors and used NGS to explore and describe lent Technologies, USA) using the RNA 6000 Nano their small RNA content. Total RNA chip. El-Mogy et al. BMC Genomics (2018) 19:408 Page 3 of 24 Small RNA library construction and high-throughput trimming, read quality was assessed by FASTQC to filter sequencing out reads with a quality score lower than 30 on the The small RNA libraries were prepared from the RNA PHRED scale. Reads were first mapped to the UniVec isolated from each sample using the Small RNA Library and human ribosomal RNA (rRNA) sequences to Prep Kit for Illumina (Norgen Biotek Corp., Canada) ac- exclude them before mapping to databases of miRBase cording to the manufacturer’s instructions. Briefly, 6 μL version 21, gtRNAdb and piRNABank to assign reads to of purified RNA was mixed with the 3′ adapter and miRNAs, tRNAs and piRNAs, respectively. Identified incubated at 70 °C for 2 min before being used in a tRNAs are tRNA-derived RNA fragments due to the fact ligation step by adding T4 RNA ligase 2 (truncated), buf- that the library insert size is below 50 nt. Remaining fer and RNase inhibitor. The reaction was incubated at sequences were then annotated to gencode version 24 28 °C for 1 h then heat inactivated at 70 °C for 10 min. (hg38) which includes protein coding transcripts (pro- The excess 3′ adapters were removed by the addition of tein_coding), mitochondrial rRNA (Mt_rRNA), mito- the reverse primer and incubating the reaction at 75 °C chondrial tRNA (Mt_tRNA), small nuclear RNA for 5 min, 37 °C for 15 min and 25 °C for 5 min. The 5′ (snRNA), small nucleolar RNA (snoRNA), long inter- adapter was denatured at 70 °C for 2 min and then genic noncoding RNA (lincRNA) and miscellaneous added together with 10 mM ATP and T4 RNA ligase 1 RNA (misc_RNA). to the reaction and incubated at 28 °C for 1 h followed by heat inactivation at 70 °C for 10 min. The two Data analysis adapters were diluted 1:1 before being added to the reac- Raw read counts obtained from the Genboree Work- tions and all the incubation steps were performed in a bench’s exceRpt small RNA-seq pipeline were further thermocycler with cooling on ice between the different analyzed using R (version 3.4.0) and R studio (version steps. Reverse transcription was performed on the 1.0.143). The following R packages were used in the ana- ligation reaction product by adding a mixture containing lysis: RnaSeqGeneEdgeRQL (version 1.0.0) for counts 10 mM dNTPs, first strand buffer and TruScript reverse per million (CPM) filtration and normalization by using transcriptase, and incubating the reaction at 50 °C for trimmed mean of M-values (TMM) [67], ggfortify (ver- 1 h before heat inactivation at 70 °C for 15 min. Finally, sion 0.4.1) and ComplexHeatmap (version 1.14.0) for the reverse transcription reaction product was amplified principal component analysis (PCA) plot and heatmaps and indexed in a 15 cycle PCR reaction by adding the based on the filtered and normalized data, respectively. NGS PCR master mix, PCR reverse primer and the VennDiagram (version 1.6.17) was used to illustrate unique index primer for each sample. Venn diagrams. miRDeep2 (version 2.0.0.8) was used to The PCR reaction product was cleaned and separated predict novel miRNA candidates and tDRmapper was on a 6% Novex® TBE PAGE gel (Life Technologies, used to identify tDRs. USA). The gel was stained with SYBR® Gold Nucleic Acid Gel Stain (Life Technologies, USA) and a library Results size range from 125 bp to 170 bp was excised from the Small RNA profiles in the various bodily fluids used in gel and placed in a Gel Breaker Tube (IST Engineering, this study provide an atlas of miRNAs, tRNAs and piR- USA), then centrifuged at 14000 RCF for 2 min. The NAs relative distribution. They also provide in depth prepared libraries were then eluted overnight in molecular analysis and a guide for NGS-based small nuclease-free water (Ambion, USA) and cleaned. The RNA expression studies that employ one or more of library was quantified by the High Sensitivity DNA Ana- these bodily fluids as a source of biological data. It is im- lysis Kit on the Agilent 2100 Bioanalyzer System (Agilent portant to look at the normal characteristics of small Technologies, USA). Libraries were diluted to 4 nM, RNA molecules in each fluid in terms of abundance and pooled, and sequenced on the Illumina HiSeq 4000 at representation. The origin and nature of these fluids can The McGill University and Génome Québec Innovation pose a significant effect on their use in certain studies Centre (Montreal, Canada), using the HiSeq 3000/4000 that might require specific handling during preparation SBS Kit (50 cycles). and sequencing to ensure the validity of results. Read mapping and small RNA annotation RNA concentration variations in the different bodily fluids The sequence raw data from the Illumina HiSeq 4000 Concentration of RNA from each bodily fluid tested was were converted to fastq format. Files were then used in measured using an Agilent 2100 Bioanalyzer. The aver- the Genboree Workbench’s exceRpt small RNA-seq age range of RNA content in 1 L of bodily fluids was as pipeline (version 4.6.2) for read mapping to the hg38 hu- low as 0.01 mg in urine to as high as 11.2 mg in saliva. man genome version [66]. This allowed for a single mis- Bodily fluids can be categorized based on their RNA matched base down to 18 nucleotides. After adapter content; significantly higher amounts of RNA can be El-Mogy et al. BMC Genomics (2018) 19:408 Page 4 of 24 recovered from saliva, blood and cell-free saliva (4.2 to respectively. This is indirectly proportional to reads used 11.2 mg/L). Leukocytes, serum and plasma had moder- for alignment to the human genome. More than 50% of ate yields of 0.8 to 1.8 mg/L, while urine and cell-free reads used for alignment were mapped to the human urine had significantly lower RNA content of 0.01 mg/L. genome in blood, plasma, serum, urine and cell-free Blood, leukocytes, saliva and cell-free saliva had lower urine. The percentage was lower in the leukocytes as standard deviations in their RNA content (< 50% of aver- well as total and cell-free samples of saliva. age), whereas plasma, serum, urine and cell-free urine In saliva and cell-free saliva, the average percentage of had higher deviations between their samples (70–85% of unmapped reads was about 50% of reads used for align- average). The average concentration of the isolated RNA ment (48.9 ± 19.7% and 50.3 ± 10.1% of input reads, from all bodily fluids ranged from 67.2 ng/μL to 3.4 ng/ respectively). Conversely, urine and cell-free urine had μL. They can be classified into high (> 20 ng/μL) from an average percentage of unmapped reads relative to blood, leukocytes, saliva and cell-free saliva and low (< reads used in alignment of 21.2 ± 17.8% and 25.5 ± 10 ng/μL) from plasma, serum, urine and cell-free urine. 23.9%, respectively. The percentage of input reads align- The RNA integrity number (RIN) was more than 7 for ment from each bodily fluid can be found in Table 2. RNA from leukocytes and lower for RNA from blood, saliva and cell-free saliva (about 2–3). RNA from serum, Small RNA biotype mapping urine and cell-free urine had a low RIN of 1 or less, Reads that were mapped to the human genome were while RNA from plasma had no measurable RIN from then mapped and classified to the various small RNA any sample (Table 1). biotypes. The average total reads mapped to small RNA biotypes within each bodily fluid ranged from 1.3 to 12.8 Input read alignment million reads. Blood, plasma, cell-free urine and urine Reads obtained from sequencing were used for align- had more than 8 million reads mapped to biotypes (12.8, ment and mapping to the human genome after adapter 9.7, 8.7 and 8.0 million reads, respectively). Serum, leu- clipping and quality filtering. The range of average input kocytes, cell-free saliva and saliva had 5.4 million reads reads from the various bodily fluids was between 9.5 or less (5.4, 2.6, 1.3 and 1.3 million reads, respectively). million reads (serum) to 15.7 million reads (blood). The The distribution of biotypes within each bodily fluid average input reads from all the bodily fluid samples showed distinct patterns. Plasma had a high percentage tested was 12.57 ± 3.54 million reads. The descending of miscellaneous RNA (misc_RNA; 58.0 ± 39.4), while order of samples based on their number of input reads urine and cell-free urine had high amounts of tRNAs was: blood, cell-free urine, urine, leukocyte, plasma, (91.3 ± 77.5% and 91.3 ± 90.3%, respectively). The other cell-free saliva, saliva and serum. The percentage of suc- bodily fluids had a more diverse pattern with no single cessfully clipped reads was more than 60% from all sam- biotype exceeding 50% of the content. MicroRNAs rep- ple types, with a minimum percentage of reads failing resented more than 85% of blood biotypes, 25% of leu- quality filters. Reads were mapped to human rRNA to kocytes, and 15–25% of plasma, serum and cell-free exclude rRNA sequences before mapping to human gen- saliva. Saliva, cell-free urine and urine contained the ome. The average percentage of reads aligned to human lowest miRNA content (5.3–12.0%). Transfer RNA was rRNA was less than 12% in all bodily fluids except saliva, the predominant biotype in urine and cell-free urine (> cell-free saliva and leukocytes, which had average per- 90%), while serum, saliva and cell-free saliva contained centages of 16.6 ± 7.5, 13.0 ± 6.2 and 36.0 ± 1.5, moderate tRNA content (20–50%). Leukocytes had 18.4 Table 1 Variations in RNA concentration, RIN value, and yield among the different bodily fluids Bodily fluid Concentration (ng/uL) RIN Value RNA amount in 1 L of fluid (mg) Ave STDEV Ave STDEV Ave STDEV Blood 21.200 4.764 2.740 2.243 5.300 1.191 Leukocytes 18.200 5.848 7.600 0.474 1.820 0.585 Plasma 3.400 2.881 N/A N/A 0.850 0.720 Serum 6.600 5.320 1.000 N/A 1.650 1.330 Saliva 67.200 33.056 2.360 0.602 11.200 5.509 Cell-Free Saliva 25.333 10.970 2.133 1.002 4.222 1.828 Urine 5.750 4.113 1.000 N/A 0.010 0.007 Cell-Free Urine 6.600 5.683 1.000 N/A 0.011 0.009 RNA concentration and RIN value were determined by the Agilent 2100 Bioanalyzer System El-Mogy et al. BMC Genomics (2018) 19:408 Page 5 of 24 Table 2 Percentage of input reads alignment from each bodily fluid Bodily Fluid Blood Leukocytes Plasma Serum Saliva Cell-Free Urine Cell-Free Saliva Urine Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Input (million reads) 15.7 2.3 12.5 1.0 12.4 6.7 9.5 1.5 10.4 1.1 10.9 1.4 14.1 2.6 15.1 4.2 Successfully clipped 94.4 0.7 63.4 4.6 78.7 20.7 80.7 5.3 82.1 12.2 78.3 2.1 83.2 9.4 82.1 8.8 Failed quality filter 0.2 0.0 0.1 0.0 0.1 0.0 0.1 0.0 0.2 0.0 0.2 0.0 0.1 0.0 0.1 0.0 Failed homopolymer filter 0.0 0.0 0.2 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.1 0.0 0.1 0.0 0.0 0.0 UniVec contaminants 0.4 0.0 0.7 0.3 0.8 0.9 0.5 0.2 0.6 0.4 0.6 0.2 0.5 0.3 0.6 0.4 rRNA 9.1 1.4 36.0 1.5 2.0 0.9 7.0 3.4 16.6 7.5 13.0 6.2 5.3 4.4 2.1 1.3 Reads used for alignment 84.7 2.2 26.4 2.8 75.8 21.0 73.1 7.6 64.6 19.5 64.5 7.2 77.2 14.0 79.3 9.8 Genome 82.2 1.9 22.7 2.3 69.0 26.7 58.0 9.5 15.7 5.3 14.2 3.3 56.1 31.7 53.7 33.6 miRNA sense 70.5 3.1 6.1 0.9 11.6 7.7 10.2 3.4 1.4 0.9 2.7 0.7 3.2 1.9 4.4 2.8 miRNAprecursor sense 0.2 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 tRNA sense 1.7 0.3 3.8 1.0 8.9 11.1 22.0 6.1 5.4 4.1 4.6 1.3 47.4 37.6 44.5 40.0 piRNA sense 1.5 0.5 1.2 0.2 4.9 3.3 1.1 0.4 0.1 0.0 0.1 0.0 0.1 0.1 0.1 0.1 Gencode sense 7.9 0.7 9.6 1.1 42.7 28.1 23.0 3.7 5.1 2.7 4.1 2.5 1.9 1.5 1.1 1.0 Not mapped to genome or transcriptome 2.5 0.2 3.8 0.5 6.7 5.8 15.1 5.3 48.9 19.7 50.3 10.1 21.2 17.8 25.6 23.9 Number of million input reads obtained from sequencing of each sample and used in the alignment ± 4.0 tRNA while plasma and blood contained the lowest miRNA tRNA fractions of 5.8 ± 2.5 and 2.1 ± 0.7, respectively. Fi- The most variable 50 miRNAs were calculated based on nally, piRNAs represented less than 2% of the reads in TMM-normalized miRNA counts at a CPM correspond- blood, serum, saliva, cell-free saliva, urine and cell-free ing to a minimum of 5 counts in a library to achieve a urine, while more than 5% of the reads were piRNAs high confidence level. These miRNAs were then used for in leukocytes and plasma. The percentages of the PCA which showed the separation of various bodily various biotypes in each bodily fluid are listed in fluids based on their miRNA expression (Fig. 2). The Table 3 and illustrated in Fig. 1.The biotype distribu- analysis revealed closeness between saliva and cell-free tion in each donor is illustrated in an additional file (Add- saliva samples and between urine and cell-free urine itional file 1: Figure S2) and shows a relatively similar samples. Close clustering was seen between blood and pattern for each bodily fluid between the donors. How- leukocytes, and between plasma and serum. The data ever, in urine samples there was a difference in patterns disclosed consistent biological origin and miRNA between male and female donors. expression-based separation of bodily fluid profiles. The Table 3 Percentage of biotype counts in the various bodily fluids Biotype Blood Leukocytes Plasma Serum Saliva Cell-Free Saliva Urine Cell-Free Urine Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Ave SDEV Million reads mapped to biotypes 12.8 1.9 2.6 0.1 9.7 6.2 5.4 1.6 1.3 0.8 1.3 0.5 8.0 5.9 8.7 7.6 miRNA 86.6 12.3 29.9 3.3 18.8 14.1 18.1 6.4 12.0 8.3 23.6 7.6 5.3 2.2 6.8 2.9 tRNA 2.1 0.7 18.4 4.0 5.8 2.5 39.7 16.9 46.0 39.2 39.0 11.6 91.3 77.5 91.3 90.3 piRNA 1.9 0.8 5.8 0.9 8.0 5.8 1.9 0.5 0.8 0.4 0.5 0.3 0.3 0.2 0.2 0.2 misc_RNA 7.7 1.5 12.7 2.6 58.0 39.4 35.1 10.3 4.6 2.6 7.1 2.0 1.7 1.7 1.1 0.8 protein_coding 0.4 0.1 5.8 0.6 0.5 0.4 0.6 0.2 6.6 3.9 9.2 7.0 0.4 0.3 0.2 0.1 Mt_rRNA 0.0 0.0 1.9 0.4 0.9 0.7 0.9 0.4 6.8 4.3 2.1 1.8 0.3 0.1 0.0 0.0 Mt_tRNA 0.1 0.0 0.7 0.1 7.7 5.1 3.1 1.3 5.5 2.4 1.2 1.1 0.2 0.1 0.1 0.0 snRNA 0.2 0.0 4.9 0.4 0.0 0.0 0.1 0.1 5.2 5.4 11.9 11.0 0.1 0.1 0.0 0.0 snoRNA 0.3 0.0 8.7 0.8 0.0 0.0 0.1 0.1 4.1 2.6 1.3 1.4 0.1 0.1 0.0 0.0 lincRNA 0.0 0.0 0.7 0.1 0.0 0.0 0.0 0.0 0.7 0.4 0.5 0.3 0.0 0.0 0.0 0.0 Others 0.6 0.1 10.5 1.0 0.2 0.1 0.3 0.1 7.6 4.8 3.7 2.5 0.3 0.2 0.1 0.1 Number of million reads mapped to biotypes El-Mogy et al. BMC Genomics (2018) 19:408 Page 6 of 24 Fig. 1 Relative biotype distribution among the various bodily fluids. The graph was generated using average percentage of biotype counts of each fluid. MicroRNAs are the largest biotype of blood, while tRNAs are the major biotype of urine. Saliva has the highest biotype diversity among all fluids Z scores of the most variable 50 miRNAs were used to leukocytes and plasma. Saliva had 98.9% of its miRNA generate a heatmap that illustrated the pattern of ex- identical to cell-free saliva, while the latter had only pression as well as the relationship between samples 75.5% of its miRNAs overlapped with saliva. About 85– (Fig. 3). It clustered bodily fluids based on their biology. 91% of urine and cell-free urine miRNAs were over- The dendrogram showed that invasive bodily fluids lapped with blood, leukocytes, plasma and serum. (blood, leukocytes, plasma and serum) branched apart Cell-free urine had 91.3% similarity with urine, while from non-invasive fluids (saliva, cell-free saliva, urine urine had 92.7% similarity to cell-free urine. Saliva and and cell-free urine). Furthermore, it showed that urine shared more than 77% of their miRNAs. In cell-depletion from saliva or urine did not have a major addition, Venn diagrams were used to demonstrate the effect on their clustering. A large set of miRNAs ap- overlap between invasive and non-invasive bodily fluids peared to be highly abundant or severely depleted in the (Fig. 5). The invasive fluids had 230 common miRNAs various bodily fluids. Particularly, urine and cell-free (Additional file 2: Table S1). Blood had 98 unique miR- urine as well as plasma and serum had different sets of NAs, which was 2 to 3-fold higher than plasma and leu- upregulated miRNAs. kocytes. In contrast, non-invasive fluids had 148 Bodily fluids can be classified into two groups based common miRNAs (Additional file 3: Table S2), and had on their collection procedures: invasive (blood, leuko- lower numbers of unique miRNAs. The non-invasive cytes, plasma and serum) and non-invasive (saliva and fluids shared 144 common miRNAs with blood (Add- urine). Variations can be seen in the detectable number itional file 4: Table S3), while the latter had 209 unique of miRNAs, at a minimum of 5 counts in 3 or more in- miRNAs that were absent from all non-invasive fluids dividuals, between the two groups. The range of de- (Additional file 5: Figure S1). Common miRNAs be- tected miRNAs from invasive fluids was 307–440 while tween all fluids were 139 (Additional file 6: Table S4). the range from the non-invasive fluids was 178–233. We ran a detailed analysis on the miRNA composition Blood had the largest number of detected miRNAs and of each bodily fluid. We looked at the 20 most abundant saliva had the lowest. Plasma contained more miRNAs miRNAs and calculated their fractions of the total compared to serum. Similarly, more miRNAs were de- miRNA content of each sample (Table 5). These 20 most tected from the cell-free preparation of saliva than saliva. abundant miRNAs covered about 74 to 94% of all Almost no differences could be seen between urine and miRNA counts. They represented 94 and 91% of blood cell-free urine (Table 4). The overlap between detected and cell-free urine, respectively. They were only 75% of miRNAs in the different bodily fluids was illustrated in all miRNA counts in saliva and cell-free saliva, while Venn diagrams (Fig. 4). About 97% of serum miRNAs representing 80–88% of the rest of the other bodily were shared with plasma. More than 90% of miRNAs in fluids. The deep analysis of miRNAs revealed that 2 spe- saliva and cell-free saliva were shared with blood, cific miRNAs were dominating the counts of 4 bodily El-Mogy et al. BMC Genomics (2018) 19:408 Page 7 of 24 Fig. 2 Principal component analysis of the most variable miRNAs in each bodily fluid. Analysis of the most variable 50 miRNAs was calculated based on TMM-normalized miRNA counts. Four pairs of fluids show close clustering: blood/leukocyte, plasma/serum, saliva/cell-free saliva and urine/cell-free urine fluids. Hsa-miR-486-5p made up 53.6 ± 1.9% and 43.2 ± include: hsa-let-7a-5p, hsa-let-7f-5p, hsa-miR-191-5p, 13.2% of miRNA counts of blood and serum, respect- hsa-miR-26a-5p and hsa-miR-486-5p. These five miR- ively, while hsa-miR-10b-5p represented 38.4 ± 8.3% and NAs represent more than 50% of blood and serum 45.6 ± 5.6% of miRNA counts of urine and cell-free miRNA counts, 25 to 45% of plasma and saliva (total urine, respectively. Other miRNAs that represented 10– and cell-free) and less than 11% of urine and cell-free 20% of total miRNA counts included let-7f-5p (11.11 ± urine miRNA counts (Fig. 7). 1.12%) in blood, miR-146b-5p (11.23 ± 1.85) and Analysis of unique and novel miRNAs are valuable in let-7f-5p (10.58 ± 1.33) in leukocytes, miR-486-5p (15.16 evaluating the usefulness of using a specific specimen as ± 1.63) and miR-191-5p (10.84 ± 0.25) in plasma, a source of information. We ran the analysis of unique miR-143-3p (10.65 ± 4.16) in saliva, miR-143-3p (14.92 ± miRNAs in our study using 3 comparison groups: all 4.47) and miR-191-5p (11.59 ± 1.64) in cell-free saliva, fluids, invasive fluids and non-invasive fluids. The num- and miR-10a-5p in urine (11.56 ± 2.67) and cell-free ber of unique miRNAs detected in each comparison urine (13.6 ± 0.97). The proportions of the top 10 most group are listed in Table 6. Blood, leukocytes and plasma abundant miRNAs in each fluid are illustrated in Fig. 6. had significantly higher numbers of unique miRNAs Among the top 20 most abundant miRNAs of each fluid, compared to the rest of fluids. Blood had the highest 5 miRNAs were found common to all fluids. These number of unique molecules (94 miRNAs), while plasma El-Mogy et al. BMC Genomics (2018) 19:408 Page 8 of 24 Fig. 3 Heatmap clustering of the most variable miRNAs in each of the bodily fluids. The sex of the sample donor is indicated as (F) for female donors or (M) for male donors. The analysis was generated using Z scores of the most variable 50 miRNAs. The dendrogram shows distinct clustering of invasive fluids (blood, leukocyte, plasma and serum) and non-invasive fluids (saliva, cell-free saliva, urine and cell-free urine) El-Mogy et al. BMC Genomics (2018) 19:408 Page 9 of 24 Table 4 Detected number of miRNAs in the various bodily fluids Bodily fluid Blood Leukocytes Plasma Serum Saliva Cell-Free Saliva Urine Cell-Free Urine Number of miRNAs 440 352 403 307 178 233 205 208 Only miRNAs that are present at a minimum of 5 counts in 3 or more individuals were considered detectable and leukocytes had 42 and 30 unique miRNAs, re- All bodily fluids were analyzed for their novel miRNA spectively. The comparison within the invasive group candidates using miRDeep2. Signal-to-noise ratio of showed a similar trend. The comparison within the more than 10 was used to select for miRDeep2 score non-invasive group showed that cell-free saliva had cutoff [68]. In bodily fluids where signal-to-noise ratio the most unique miRNAs (32 miRNAs) followed by was less than 10 (saliva, urine and cell-free urine), we se- cell-free urine (11 miRNAs). Saliva and urine had lected the score cutoff that corresponds to the highest minimal numbers of unique miRNAs compared to signal-to-noise ratio (Table 7). Invasive bodily fluids had their cell-free preparations. The list of unique miR- higher numbers of novel miRNA candidates than NAs from the three comparison groups are listed in non-invasive fluids. The highest number of novel three additional files (Additional file 7:Table S5,Add- miRNA candidates was observed in plasma and blood, itional file 8: Table S6, Additional file 9:Table S7). with 50 and 48 candidates, respectively. Serum had 20 Fig. 4 Overlap of miRNA content between various bodily fluids. MicroRNAs of each fluid were filtered to keep molecules that have a minimum of 5 counts in 3 or more individuals. The highest overlap is seen between fluids within the same category: invasive fluids (blood, leukocyte, plasma and serum) or non-invasive fluids (saliva, cell-free saliva, urine and cell-free urine) El-Mogy et al. BMC Genomics (2018) 19:408 Page 10 of 24 Fig. 5 Venn diagram showing the overlap between (a) invasive and (b) non-invasive bodily fluids. Only miRNAs that are present at a minimum of 5 counts in 3 or more individuals were included in the comparison. Invasive fluids have higher number of shared and unique miRNAs compared to non-invasive fluids novel candidates while leukocytes had 16 candidates. Read counts of tRNAs were normalized using TMM at All the non-invasive fluids had 7 or less novel candi- a CPM corresponding to a minimum of 5 counts in a dates. Sequences of novel miRNA candidates were library. The normalized reads were then used to gener- matched to the miRCarta database (v1.0) of newly ate principal component analysis which showed the sep- predicted human miRNAs [69]. More than 50% of aration of various bodily fluids based on their tRNA blood and leukocytes novel miRNA candidates were levels (Fig. 8). The analysis revealed closeness between present in miRCarta (66 and 56%, respectively). saliva and cell-free saliva as well as closeness between Plasma and serum had less miRNA candidates match- the invasive fluids. However, urine and cell-free urine ing miRCarta database (30 and 45%, respectively). No were dispersed between both saliva and serum. The Z miRCarta matches were found for the novel miRNA scores of these tRNAs were used to generate a heatmap candidates of the non-invasive fluids. An additional that indicates the levels of various tRNAs in the different file containing the list of novel miRNA candidates in samples (Fig. 9). Serum and the male urine/cell-free each bodily fluid and their sequences, as well as the urine samples showed distant clustering from the rest of matching results to the miRCarta database is provided samples. The female urine/cell-free urine clustered with (Additional file 10:Table S8). saliva/cell-free saliva. Blood, leukocytes and plasma showed similar clustering. The data shows clustering tRNA patterns based on sample biology and no difference Mapped tRNAs represented tDRs down to 18 nucleo- between cell-depleted and non-depleted conditions. tides. The predominant tRNA fragments in all the bodily Gly fluids was tRNA . This tRNA composed 86.5 and piRNA 87.6% of the total tRNA content in urine and cell-free All bodily fluids had piR-016658 at different levels. The urine, respectively. For the remaining bodily fluids, it highest levels were seen in blood and serum (92.3 ± 1.8% made up 72.0 to 84.1% of the total tRNA content. and 94.0 ± 2.7%, respectively), followed by plasma and Glu The second most abundant tRNA was tRNA , with leukocytes (81.8 ± 33.1% and 73.5 ± 3.8%, respectively). It a range of 6.7 to 21.4% of the tRNA content. Further was the highest piRNA in cell-free saliva (40.3 ± 17.0%). analysis of tRNAs by tDRmapper to look at the exact It had lower concentrations, yet more than 10%, in sal- Gly-GCC tDRs composition showed that tRNA and iva, urine and urine-cell free (14.9 ± 1.7%, 15.6 ± 12.9% Glu-CTC tRNA were the predominant fragments in all and 14.5 ± 12.5, respectively). Urine and urine-cell free fluids. All samples, regardless of the fluid origin, had piR-019825 as the highest piRNA (46.0 ± 40.4% and shared similar tDRs composition. The quantification 58.7 ± 32.1%, respectively). Interestingly, piR-019825 was and coverage of the top 50 tDRs in each fluid are the second highest piRNA in plasma where it repre- presented in Additional files 11 and 12.Noneofthe sented 15.2 ± 30.4% of the piRNA content (Table 9). An remaining tRNAs in any of the fluids exceeded 3.1% additional file contains the list of piRNAs at an average of the tRNA content (Table 8). For blood, plasma, of 1% or more of the entire piRNAs counts of each bod- saliva and cell-free saliva, there was a higher diversity ily fluid (Additional file 13: Table S9). of tRNAs that represented 1% or more of the total Read counts of piRNAs were TMM-normalized at a tRNA content of the sample (5 tRNAs or more). CPM corresponding to a minimum of 5 counts in a li- However, for leukocytes, serum, urine and cell-free brary. The normalized counts were used to generate a urine, there was a lower diversity (3 to 4 tRNAs). PCA plot (Fig. 10). Close clustering was obtained El-Mogy et al. BMC Genomics (2018) 19:408 Page 11 of 24 Table 5 Twenty most abundant miRNAs detected in each bodily fluid # Blood Leukocytes Plasma Serum Saliva Cell-Free Urine Cell-Free Urine Saliva 1 hsa-miR-486-5p hsa-miR- hsa-miR-486-5p hsa-miR-486-5p hsa-miR-143- hsa-miR-143- hsa-miR-10b- hsa-miR-10b-5p (53.64 ± 1.89) 146b-5p (15.16 ± 1.63) (43.22 ± 13.23) 3p (10.65 ± 3p (14.92 ± 5p (38.4 ± (45.58 ± 5.62) (11.23 ± 1.85) 4.16) 4.47) 8.33) 2 hsa-let-7f-5p (11.11 hsa-let-7f-5p hsa-miR-191-5p hsa-miR-92a-3p hsa-miR- hsa-miR-191- hsa-miR-10a- hsa-miR-10a-5p ± 1.12) (10.58 ± 1.33) (10.84 ± 0.25) (4.85 ± 0.48) 203a-3p 5p (11.59 ± 5p (11.56 ± (13.6 ± 0.97) (8.36 ± 5.89) 1.64) 2.67) 3 hsa-miR-451a (4.79 hsa-miR-26a- hsa-miR-26a-5p hsa-miR-191-5p hsa-miR-191- hsa-miR-26a- hsa-miR-30a- hsa-miR-30a-5p ± 1.08) 5p (8.32 ± (8.21 ± 0.77) (4.51 ± 1.77) 5p (7.66 ± 5p (8.06 ± 5p (6.58 ± (6.59 ± 1.95) 0.33) 4.8) 0.69) 1.5) 4 hsa-miR-92a-3p hsa-let-7 g-5p hsa-let-7f-5p (6.99 hsa-let-7f-5p (4.5 ± hsa-miR-26a- hsa-miR- hsa-miR-192- hsa-miR-192-5p (4.09 ± 0.55) (6.68 ± 0.97) ± 0.7) 1.5) 5p (5.94 ± 148a-3p 5p (4.5 ± (4.53 ± 2.42) 4.11) (4.01 ± 0.63) 2.12) 5 hsa-miR-191-5p hsa-miR-150- hsa-miR-92a-3p hsa-miR-26a-5p hsa-let-7f-5p hsa-miR-375 hsa-let-7f-5p hsa-let-7f-5p (2.47 (3.74 ± 0.64) 5p (6.48 ± (5.41 ± 0.09) (3.45 ± 1.74) (5.32 ± 2) (3.56 ± 1.92) (3.32 ± 0.77) ± 0.51) 2.23) 6 hsa-let-7a-5p (3.12 hsa-miR-191- hsa-miR-146a-5p hsa-let-7a-5p (3.21 hsa-miR-486- hsa-miR-27b- hsa-miR-100- hsa-miR-27b-3p ± 0.86) 5p (5.3 ± 0.68) (5.17 ± 0.13) ± 0.87) 5p (4.77 ± 3p (3.55 ± 5p (2.61 ± (2.41 ± 0.76) 3.31) 1.66) 0.64) 7 hsa-let-7i-5p (2.6 ± hsa-miR-342- hsa-miR-30d-5p hsa-miR-146a-5p hsa-miR- hsa-let-7f-5p hsa-miR-27b- hsa-miR-100-5p 0.09) 3p (4.84 ± (3.58 ± 0.43) (3.11 ± 1.21) 378a-3p (3.49 ± 0.57) 3p (2.52 ± (2.36 ± 0.57) 2.52) (4.75 ± 7.51) 1.13) 8 hsa-let-7 g-5p (2.42 hsa-miR-486- hsa-miR-151a- hsa-miR-423-5p hsa-miR-27b- hsa-miR- hsa-miR-26a- hsa-miR-26a-5p ± 0.12) 5p (3.27 ± 5p|hsa-miR-151b (2.27 ± 0.45) 3p (3.59 ± 203a-3p 5p (2.38 ± (1.94 ± 0.34) 1.37) (2.67 ± 0.09) 1.13) (3.39 ± 1.96) 0.41) 9 hsa-miR-182-5p hsa-miR-21- hsa-miR-146b-5p hsa-miR-30d-5p hsa-let-7 g- hsa-let-7a-5p hsa-let-7a-5p hsa-miR-99a-5p (1.25 ± 0.25) 5p (3.21 ± (2.54 ± 0.16) (2.01 ± 0.51) 5p (3.11 ± (2.94 ± 0.25) (2.06 ± 0.91) (1.43 ± 0.36) 0.52) 1.36) 10 hsa-let-7b-5p (1.07 hsa-miR-92a- hsa-miR-21-5p hsa-let-7b-5p (1.69 hsa-miR-24- hsa-miR-1246 hsa-miR- hsa-let-7a-5p (1.27 ± 0.26) 3p (3.19 ± (2.18 ± 0.17) ± 0.25) 3p (2.97 ± (2.74 ± 2.33) 200b-3p ± 0.16) 0.23) 1.86) (1.75 ± 0.77) 11 hsa-miR-185-5p hsa-miR-146a- hsa-let-7a-5p (2.12 hsa-miR-122-5p hsa-let-7a-5p hsa-miR-92a- hsa-miR-486- hsa-miR-486-5p (1.01 ± 0.09) 5p (3.13 ± ± 0.26) (1.32 ± 1.46) (2.93 ± 1.31) 3p (2.45 ± 5p (1.71 ± (1.26 ± 0.95) 0.77) 0.28) 1.05) 12 hsa-miR-16-5p hsa-miR-143- hsa-miR-423-5p hsa-miR-151a- hsa-miR-375 hsa-miR-423- hsa-miR-99a- hsa-let-7b-5p (1.21 (0.95 ± 0.21) 3p (2.49 ± (2.11 ± 0.19) 5p|hsa-miR-151b (2.31 ± 1.17) 5p (1.74 ± 5p (1.65 ± ± 0.34) 1.42) (1.29 ± 0.38) 1.49) 0.43) 13 hsa-miR-26a-5p hsa-let-7i-5p hsa-miR-423-3p hsa-miR-451a (1.23 hsa-miR- hsa-let-7b-5p hsa-miR- hsa-miR-200b-3p (0.86 ± 0.11) (1.86 ± 0.21) (2.03 ± 0.17) ± 0.6) 148a-3p (2 ± (1.72 ± 0.79) 148a-3p (1.13 ± 0.25) 0.86) (1.48 ± 0.6) 14 hsa-miR-25-3p hsa-miR-30d- hsa-miR-99b-5p hsa-miR-146b-5p hsa-miR-21- hsa-miR-30e- hsa-miR- hsa-miR-148a-3p (0.64 ± 0.08) 5p (1.78 ± 0.3) (1.98 ± 0.38) (1.09 ± 0.38) 5p (1.85 ± 5p (1.71 ± 203a-3p (1.07 ± 0.36) 1.27) 0.26) (1.36 ± 1.7) 15 hsa-miR-183-5p hsa-miR-451a hsa-miR-151a-3p hsa-miR-151a-3p hsa-miR-205- hsa-miR-99a- hsa-miR-21- hsa-miR-30d-5p (0.62 ± 0.05) (1.4 ± 0.59) (1.94 ± 0.33) (1.05 ± 0.23) 5p (1.62 ± 5p (1.71 ± 5p (1.17 ± (0.93 ± 0.32) 0.92) 0.82) 0.92) 16 hsa-miR-181a-5p hsa-let-7a-5p hsa-let-7i-5p (1.88 hsa-miR-320a (0.92 hsa-miR-320a hsa-miR-24- hsa-let-7b-5p hsa-miR-99b-5p (0.58 ± 0.15) (1.38 ± 0.4) ± 0.2) ± 0.36) (1.44 ± 1.94) 3p (1.64 ± (1.17 ± 0.23) (0.86 ± 0.44) 0.11) 17 hsa-miR-151a- hsa-miR-181a- hsa-miR-127-3p hsa-miR-10a-5p hsa-miR-99a- hsa-miR-486- hsa-miR-30d- hsa-miR-423-5p 5p|hsa-miR-151b 5p (1.36 ± (1.36 ± 0.21) (0.89 ± 0.42) 5p (1.4 ± 5p (1.63 ± 5p (1.04 ± (0.7 ± 0.17) (0.53 ± 0.05) 0.27) 0.16) 0.53) 0.28) 18 hsa-miR-101-3p hsa-miR-223- hsa-let-7 g-5p (1.34 hsa-let-7i-5p (0.87 hsa-let-7i-5p hsa-miR-23a- hsa-miR-99b- hsa-miR-151a- (0.42 ± 0.1) 3p (1.23 ± ± 0.19) ± 0.14) (1.4 ± 0.33) 3p (1.52 ± 5p (0.88 ± 5p|hsa-miR-151b 0.71) 0.38) 0.34) (0.55 ± 0.07) 19 hsa-miR-30d-5p hsa-miR-30e- hsa-miR-222-3p hsa-miR-99b-5p hsa-miR-92a- hsa-miR- hsa-miR-191- hsa-miR-30a-3p (0.4 ± 0.09) 5p (1.16 ± (1.2 ± 0.32) (0.8 ± 0.3) 3p (1.39 ± 200b-3p 5p (0.8 ± (0.53 ± 0.07) 0.09) 0.41) (1.26 ± 0.69) 0.41) El-Mogy et al. BMC Genomics (2018) 19:408 Page 12 of 24 Table 5 Twenty most abundant miRNAs detected in each bodily fluid (Continued) # Blood Leukocytes Plasma Serum Saliva Cell-Free Urine Cell-Free Urine Saliva 20 hsa-miR-30e-5p hsa-miR-101- hsa-miR-10a-5p hsa-miR-22-3p (0.8 hsa-let-7b-5p hsa-miR-223- hsa-miR-423- hsa-miR-191-5p (0.36 ± 0.07) 3p (1.07 ± (1.18 ± 0.23) ± 0.19) (1.22 ± 0.33) 3p (1.23 ± 5p (0.65 ± (0.46 ± 0.09) 0.16) 0.19) 0.22) MicroRNAs are arranged in a descending order from highest to lowest. Percentage of miRNA to all miRNAs in the bodily fluid is shown between brackets. None of the unique miRNAs of each bodily fluid were found among its top 20 miRNAs between blood and leukocytes, saliva and cell-free saliva, For better representation of the actual library prepar- and urine and cell-free urine. Plasma and serum showed ation workflow, we used a standard NGS preparation relatively distant clustering from each other and from protocol based on equal input volumes from each bodily the other fluids. A similar clustering pattern was ob- fluid preparation. An average of 12.57 ± 3.54 million tained from the heatmap. However, urine samples and reads were obtained from all samples, despite the large most of the cell-free urine samples showed inter-fluid and intra-fluid variations in RNA yield and sex-dependent clustering (Fig. 11). Female urine samples integrity. This indicates that regardless of sample type, a clustered close to saliva samples, while male urine sam- clean purification and robust library preparation can ples clustered close to plasma and serum samples. While yield similar sequencing read outcomes. The critical par- clustering reflects the close biology of the samples, it has ameter that would then define suitability of a sample to a distinct trend when compared to that of miRNA as it be used in small RNA profiling and discovery would be showed more overlap between invasive and non-invasive its actual biotype content. Read alignment to human fluids. genome varied between the different fluids based on their molecular composition. The lower percentage of Discussion leukocytes reads successfully aligned to the human gen- In this study, we investigated small RNA profiles in vari- ome is a result of their higher rRNA content. The lower ous bodily fluids using NGS in order to understand the percentage of saliva and cell-free saliva reads successfully distribution of the various biotypes between fluids as aligned to the human genome is due to the high per- well as the molecular signature of each fluid. Purified centage of unmapped reads. We conducted exogenous total RNA from each fluid showed large variations in the mapping analysis on the saliva and cell-free saliva sam- RNA content and integrity. Saliva, cell-free saliva and ples and 85–90% of the unmapped reads were mapped blood have the highest RNA content. These elevated to bacteria (Additional file 14: Figure S3 and Additional levels in both blood and saliva are due to their high file 15: Figure S4). This agrees with results of a recent number of cells. However, the high RNA levels in the study by Yeri et al. [70]. This in turn reduces the amount cell-free saliva preparation were most likely due to the of valuable human RNA molecules that can be used in high bacterial content, as the cell-free preparation steps profiling or discovery. Efficient removal of salivary bac- utilized in this study were aimed at removing mamma- teria can be achieved by centrifugation [71]. However, re- lian but not bacterial cells. The lowest RNA yields were moval of bacteria in this manner would reduce the found in both total and cell-free urine preparations. This number of reads directed towards bacterial sequences, indicated very low cell content as well as minimal thereby hindering the study of bacterial communities and/ cell-free RNA content of urine samples collected from or pathogens that might be contained within these fluids. healthy individuals. Intra-fluid RNA yields are more con- While the profile of leukocytes showed a fairly even sistent from fluids that have high cell content (blood, distribution of biotypes, all the other bodily fluids saliva and leukocytes). RNA integrity as measured by showed predominant reads from one or more biotypes. RIN value was also dependent on the cellular content of This can be of a significant value if these predominant the fluid. This severely affected the RNA integrity of biotypes are of known biological importance such as urine, cell-free urine, plasma and serum where they al- miRNAs, tRNAs or piRNAs. Blood has the highest levels most have no measurable integrity. In fact, while many of miRNAs (86.6 ± 12.3), which were about 3-fold the small RNA-Seq library preparation pipelines require levels of leukocytes miRNAs (29.9 ± 3.3). A large portion RNA with a minimum RIN value, this study showed of these blood miRNAs are lost from plasma and serum, that many bodily fluids of low cell content may not as the separation and coagulation processes might be of meet such requirements. However, in this study li- the factors that affect miRNA distribution and recovery. braries were successfully constructed from RNA sam- Variations between plasma and serum miRNA content ples with low or no RIN value, suggesting that using results from the stress during coagulation [18]. RIN value as a sole determination of RNA quality Non-invasive fluids had lower miRNA fractions, keeping may not be universally applicable. in mind that they had a large percentage of unmapped El-Mogy et al. BMC Genomics (2018) 19:408 Page 13 of 24 Fig. 6 Top 10 most abundant miRNAs relative to all miRNA counts in each fluid. The 10 miRNAs that have the highest read counts in each fluid were illustrated relative to the total miRNA read counts of the fluid. Counts of the remaining miRNAs were summed up and illustrated as “other” El-Mogy et al. BMC Genomics (2018) 19:408 Page 14 of 24 Fig. 7 Relative abundance of the top 5 common miRNAs between the different fluids. Counts of the 5 common miRNAs are presented relative to the total miRNA counts of each fluid. The five common miRNAs represent a large fraction of the invasive fluids with the highest percentage in blood. They represent lower fractions in the non-invasive fluids reads (about 50% of saliva and 20% of urine reads). The The different samples are well clustered based on miR- relatively high bacterial content of saliva as well as the NAs according to sample type and their biology. In filtered and diluted nature of urine were key factors in addition, invasive and non-invasive fluids have distinct this result. Recent analysis of urine, saliva and plasma profiles and less variations between the fluids within the miRNAs from NGS data showed lower miRNA counts same group. PCA plots and heatmaps generated for from urine and saliva [70]. Profiling of miRNA in bodily tRNAs and piRNAs show a biology-related clustering, fluids by RT-PCR in an earlier study showed similar low with overlap between invasive and non-invasive fluids. urine RNA concentrations and low numbers of detected An interesting observation was the differential clustering miRNAs, while saliva had the highest number of miR- of tRNAs and piRNAs from urine and cell-free urine NAs among the studied fluids [72]. The information ob- samples based on the sex of the donor, indicating tained in our results could be used to guide methods for sex-related expression of these molecules. In addition, targeting specific biotypes in bodily fluids (via enrich- urine showed close clustering to serum, indicating that ment, separation, or depletion, for example). the latter might be the true liquid part of blood. The The various bodily fluids have unique miRNA, tRNA Table 7 Number of novel miRNA candidates in the different and piRNA profiles that characterize the type and origin bodily fluids of the fluid as seen from the PCA plots and heatmaps. Bodily Signal- miRDeep2 Number of Number of fluid to- score novel miRNA candidates present in noise candidates miRCarta database Table 6 Number of unique miRNAs in each bodily fluid at a minimum of 5 counts Blood 15.9 5 48 32 Bodily fluid All Invasive Non-invasive Leukocytes 11.5 5 16 9 Blood 94 96 Not compared Plasma 17 5 50 15 Leukocytes 30 35 Not compared Serum 17.1 5 20 9 a a Plasma 42 43 Not compared Saliva 4.8 54 0 Serum 1 1 Not compared Cell-Free 10.5 5 7 0 Saliva Saliva 1 Not compared 1 a a Urine 7.6 55 0 Cell-Free Saliva 3 Not compared 32 a a Cell-Free 7.7 55 0 Urine 1 Not compared 4 Urine Cell-Free Urine 3 Not compared 11 Signal-to-noise ratio is below the minimum accepted cutoff (10). The highest Three comparison groups were used: “All” indicating unique miRNA among all signal-to-noise ratio was used. However, predicted novel miRNAs at this value fluids, “invasive” for comparison within the invasive fluids (blood, leukocytes, might not be real. Predicted novel miRNAs with a non-significant p-value of serum and plasma), and “non-invasive” for comparison within the non-invasive the RNA minimum free energy of folding randomization test (Randfold) have fluids (saliva, cell-free saliva, urine and cell-free urine) not been counted El-Mogy et al. BMC Genomics (2018) 19:408 Page 15 of 24 Table 8 List of tRNAs that represents ≥1% of all tRNA counts in each bodily fluid Blood Leukocyte Plasma Serum Saliva Cell-Free Saliva Urine Cell-Free Urine Gly (84.1 ± 1.8) Gly (77.8 ± 1.9) Gly (79.9 ± 12.7) Gly (73.5 ± 8.2) Gly (83.7 ± 6.6) Gly (72.0 ± 7.4) Gly (86.5 ± 6.4) Gly (87.6 ± 6.0) Glu (6.7 ± 0.5) Glu (11.8 ± 2.6) Glu (9.5 ± 6.0) Glu (21.4 ± 7.5) Glu (7.9 ± 4.5) Glu (15.4 ± 2.6) Glu (8.0 ± 4.2) Glu (8.4 ± 3.9) Lys (2.1 ± 0.4) Lys (3.1 ± 0.7) SeC (1.8 ± 1.5) Val (1.5 ± 0.2) Val (2.1 ± 0.5) Lys (2.8 ± 0.9) Val (1.8 ± 1.9) Lys (1.5 ± 0.5) SeC (1.0 ± 0.3) Val (2.8 ± 0.4) His (1.5 ± 1.2) Lys (1.3 ± 0.5) Lys (1.6 ± 0.5) Ala (2.6 ± 2.7) Lys (1.6 ± 0.7) His (1.2 ± 0.1) Arg (1.2 ± 1.0) His (1.0 ± 0.5) Val (1.6 ± 0.2) Val (1.2 ± 0.2) Gln (1.2 ± 0.8) Asp (1.5 ± 0.7) Lys (1.1 ± 0.6) Pro (1.0 ± 0.9) Average percentage and standard deviation for each tRNA relative to the total tRNA content of the bodily fluid is included between the brackets Fig. 8 Principal component analysis of tRNAs in each bodily fluid. The plot was generated based on TMM-normalized tRNA counts. Samples of the same origin clustered closer to each other. However, urine samples are more dispersed from each other El-Mogy et al. BMC Genomics (2018) 19:408 Page 16 of 24 Fig. 9 Heatmap clustering of tRNAs in the various bodily fluids. The sex of sample donor is indicated as (F) for female donors or (M) for male donors. The analysis was generated using Z scores of TMM-normalized tRNA counts. The dendrogram shows distant clustering of urine samples (total and cell-free) based on sex. Male urine samples (total and cell-free) clustered close to serum while female urine samples (total and cell-free) are clustered with saliva differentiation of the three biotypes (miRNA, tRNA and where some molecules are enriched, while others are de- piRNA) in the different bodily fluids might be a result of pleted. None of the unique miRNAs were found among fluid origin and biological functions. The higher impact the top 20 most abundant molecules of each fluid. This of origin of fluid on miRNA distribution may refer to indicated the need for higher read depth to detect more specialized functions of miRNA in comparison to miRNAs that might have specific functions. Five piRNAs and tRNAs, which might be involved in more miRNAs: hsa-let-7a-5p, hsa-let-7f-5p, hsa-miR-191-5p, general biological roles. hsa-miR-26a-5p and hsa-miR-486-5p were common Non-invasive fluids had almost half the number of among the top 20 most abundant molecules in all fluids, identified miRNAs. For urine, this may result from the indicating shared origin or function. These five abundant filtering process by the kidneys. For saliva, the lower cir- common miRNAs represented a large portion of the culating nucleic acid content, relative to blood-related miRNA counts of invasive fluids (more than 40%), while fluids may be the cause [73–76]. The larger miRNA frac- they were relatively lower in non-invasive fluids (less tion and lower numbers of unmapped reads for the inva- than 30%). There was a set of 139 core miRNAs that are sive fluids explained their higher number of identified common among the different fluids and a set of 144 miRNAs. Almost every fluid had unique miRNAs that miRNAs that were shared between non-invasive fluids are specific only to that fluid, providing a specific signa- and blood. While the levels of these molecules may vary ture for each fluid. The number of common and unique between fluid types, they might be promising biomarker miRNAs between two fluids varied depending on bio- candidates that can be detected from multiple sources, logical relatedness. The invasive fluids collected in this including non-invasive fluids. study were more similar to the other invasive fluids, and An interesting observation was the variation between the non-invasive fluids were more similar to the other the most expressed miRNAs in the different fluids. In non-invasive fluids collected. This may be due to the fact blood, plasma and serum, hsa-miR-486-5p was the most that the invasive fluids collected were all blood derived. expressed, while hsa-miR-143-3p was the most expressed The fluid-specific unique miRNAs can result from differ- in saliva and cell-free saliva and hsa-miR-10b-5p was the ent cells secreting the different fluids. They may also re- predominant miRNA in urine and cell-free urine. We sult from the natural filtration process of some fluids, searched these 3 miRNAs on the human miRNA tissue El-Mogy et al. BMC Genomics (2018) 19:408 Page 17 of 24 Table 9 Predominant piRNAs in each bodily fluid Body fluid piRNA GenBank Accession number Chromosomal position Percentage Blood hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (92.3 + 1.8) Leukocyte hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (73.5 + 3.8) hsa_piR_000552 DQ570687 Homo_sapiens:22:38045003:38045030:Minus (5.6 + 0.6) Plasma hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (81.8 + 33.1) hsa_piR_019825 DQ597218 Homo_sapiens:1:227740227:227740256:Plus (15.2 + 30.4) Serum hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (94.0 + 2.7) Saliva hsa_piR_014620 DQ590013 Homo_sapiens:5:93930930:93930956:Minus (16.3 + 14.1) hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (14.9 + 1.7) hsa_piR_019521 DQ596805 Homo_sapiens:11:10487516:10487542:Minus (10.5 + 2.7) hsa_piR_000552 DQ570687 Homo_sapiens:22:38045003:38045030:Minus (7.6 + 2.7) hsa_piR_018780 DQ595807 Homo_sapiens:17:72068837:72068864:Plus (5.8 + 3.3) hsa_piR_000805 DQ571003 Homo_sapiens:1:212438966:212438997:Plus (5.5 + 1.5) Cell-Free Saliva hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (40.3 + 17.0) hsa_piR_016659 DQ592932 Homo_sapiens:14:22388242:22388267:Plus (7.4 + 6.7) hsa_piR_019521 DQ596805 Homo_sapiens:11:10487516:10487542:Minus (5.2 + 2.0) hsa_piR_000552 DQ570687 Homo_sapiens:22:38045003:38045030:Minus (5.1 + 3.3) hsa_piR_020450 DQ598104 Homo_sapiens:9:133350930:133350959:Plus (5.1 + 0.9) Urine hsa_piR_019825 DQ597218 Homo_sapiens:1:227740227:227740256:Plus (46.0 + 40.4) hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (15.6 + 12.9) hsa_piR_014620 DQ590013 Homo_sapiens:5:93930930:93930956:Minus (5.6 + 5.0) Cell-Free Urine hsa_piR_019825 DQ597218 Homo_sapiens:1:227740227:227740256:Plus (58.7 + 32.1) hsa_piR_016658 DQ592931 Homo_sapiens:6:80508363:80508389:Plus (14.5 + 12.5) hsa_piR_004153 DQ575660 Homo_sapiens:3:156861576:156861607:Plus (6.0 + 3.3) Molecules that represent an average of 5% or more of the entire piRNAs count of each bodily fluid are listed atlas [77] to identify the tissue of origin. High quantile nor- serum. Plasma did not suffer from the presence of a pre- malized expression levels of hsa-miR-486-5p were found in dominant molecule as did blood and serum. This vein and muscle specimens, while hsa-miR-143-3p was resulted in a high number of predicted novel miRNA highly elevated in esophagus and relatively high in colon, candidates compared to all the other bodily fluids, mak- bladder and prostate specimens. The expression levels of ing plasma a good alternative to blood. However, deple- hsa-miR-10b-5p were very high in the epididymis and ele- tion of hsa-miR-486-5p from blood and serum could be vated in kidney, colon and muscle specimens which ex- a useful tool to direct a greater proportion of reads to plains the relatively higher expression levels of this miRNA other miRNA sequences. in male urine samples. Recent studies indicate the import- Both saliva and urine did not offer the same advantage ance of hsa-miR-486-5P as a cancer biomarker in as the invasive fluids. They had lower miRNA content non-small cell lung cancer [78], gastric cancer [79]and oral and this affected their molecular diversity. The most tongue squamous cell carcinoma [80]. It may act as a expressed miRNA in saliva and cell-free saliva was tumorsuppressormiRNA [81]and mayalso beusedto hsa-miR-143-3p (10–15%). It is also the second most predict the efficacy of cancer vaccine treatment for colorec- expressed miRNA in saliva exosomes [83]. It is differen- tal cancer [82]. However, in many other cancer studies, this tially expressed in senescence [84] and as a tumor miRNA was not deregulated. suppressor in gliomas [85]. MicroRNA hsa-miR-10b-5p The large breadth of unique miRNAs found in blood, represented about 38–45% of urine and cell-free urine combined with an abundance of predicted novel miRNA miRNAs. It has been recently reported to be the most candidates demonstrates the superiority of blood for expressed miRNA in urine samples [86]. hsa-miR-10b-5p miRNA profiling and discovery. However, blood has high plays a role in carcinoma metastasis and is overexpressed levels of hsa-miR-486-5p, representing over 50% of its in colorectal cancer [87–90]. Due to its high expression, a miRNA content. Other bodily fluids that had a relatively lower proportion of reads will map to other miRNA se- high miRNA content are plasma and, to some extent, quences and its depletion should be considered as a El-Mogy et al. BMC Genomics (2018) 19:408 Page 18 of 24 Fig. 10 Principal component analysis of piRNAs in each bodily fluid. Analysis was generated based on TMM-normalized piRNA counts. Samples are clustered based on biology and fluids that share similar origin have close clustering. Close clustering is seen between the following fluid pairs: blood/leukocytes, saliva/cell-free saliva and urine/cell-free urine. Serum and plasma show distant clustering from the other fluids priority for improving the diversity of miRNAs within miRNAs and the predicted novel miRNA candidates of urine specimens. the invasive fluids (30 to 66%), while no matches were Novel miRNA prediction was not as efficient when found between miRCarta and the non-invasive fluids. dealing with non-invasive fluids. Their low This might be due to the large number of miRNA stud- signal-to-noise ratio made it hard to obtain accurate pre- ies from the invasive fluids as well as the higher counts diction. The only exception was the cell-free saliva, and diversity of miRNAs from these fluids compared to where a fair signal-to-noise ratio was achieved, and 7 the non-invasive fluids. This further indicates the higher novel miRNA candidates had been identified. It also had potential of the invasive fluids in novel miRNA predic- double the number of unique miRNAs compared to the tion. Although our findings are based on prediction of other saliva and urine samples. Due to the removal of candidate miRNA and have not been validated by an- mammalian cells by centrifugation, cell-free saliva usu- other technique, they showed the potential of these vari- ally captures more circulating miRNAs than total saliva ous fluids in novel miRNA discovery. or the cellular fraction of saliva [20, 91]. This made the While tRNA fragments are a minor portion of blood cell free saliva sample superior in terms of discovering and plasma small RNAs, they were well represented in unique and novel miRNAs candidates compared to the serum and saliva preparations (39–46%) and were the other non-invasive fluids. Matches were found between major small RNA species of urine and cell-free urine the miRCarta database of newly predicted human (> 90%). The main component of these tRNAs in all El-Mogy et al. BMC Genomics (2018) 19:408 Page 19 of 24 Fig. 11 Heatmap clustering of piRNAs in each bodily fluid. The sex of sample donor is indicated as (F) for female donors or (M) for male donors. Heatmap was generated using Z scores of TMM-normalized tRNA counts. The dendrogram shows distant clustering of urine samples (total and cell-free) based on sex. Male urine and cell-free urine samples clustered close to serum and plasma while female urine and most of cell-free urine samples are clustered with saliva El-Mogy et al. BMC Genomics (2018) 19:408 Page 20 of 24 Gly the fluids was tRNA (72.0 to 87.6%), followed by tRNA- plasma, serum, saliva, cell-free saliva, urine and cell-free Glu (6.7 to 21.4%). Urine samples, unlike the other fluids, urine), even with the high variations in the volumes used had high sample-to-sample variations, with tRNAs ran- for RNA purification as well as the high variations in ging from 47 to 98% of small RNAs. Similar variations concentrations of the isolated RNA. Despite the ease of have been reported in a recent study on urine from ovar- collection and handling of non-invasive fluids, they did ian cancer patients [86]. However, these variations may be not provide the same small RNA diversity and sample correlated to the sex of the individual, where male urine consistency as invasive fluids. However, this study has over 90% and female urine has about 70% or less. A showed that these samples can still be routinely profiled. larger study is needed to validate these findings. It is also Furthermore, the signatures of these non-invasive fluids interesting that the specific tRNA molecular composition are very likely linked to their origin. For example, urine of these tRNA fractions is consistent. Despite the overall may be a good candidate for studying diseases related to fluctuations in urine tRNA fractions, changes in the mo- organs such as kidney and bladder, although careful re- lecular signature of tRNA molecules might still be valid sult interpretation should be considered when investigat- for potential biomarker discovery. However, it may be lim- ing male and female urine, as their biotypes may be ited by the lower urine tRNA molecular diversity, com- sex-dependent. This observation is limited by the sample pared to blood, plasma and saliva. These variations in size of our study and is yet to be investigated on a large diversity were also observed between plasma, saliva and sample size study. An organ and fluid small RNA index urine in a recent study [70]. However, the percentage might be needed to track and correlate origins and func- abundance of molecules was different. tions of the various molecules. Processing of larger vol- Plasma and leukocytes contain relatively high amounts umes of urine, and bacterial removal from saliva of piRNA (8 and 5.8%, respectively). All the other bodily preparations might improve their NGS mapping to human fluids contain less than 2%, which is related to the small targets. In addition, depletion of specific molecules or se- fraction of piRNAs that are consistently being expressed lection/enrichment of target molecules from almost every in normal and cancer cells [92]. It is interesting that a sin- bodily fluid may significantly increase flow cell capacity gle piRNA molecule, hsa-piR-016658, was the most for target molecules and in turn provide a meaningful read expressed in all bodily fluids except in saliva, urine and depth. Successful clustering of bodily fluids based on their cell-free urine, where it was the second most abundant. miRNA distribution can be expanded to cohorts that can This molecule is associated with patients with prostate be differentiated according to their miRNA, and possibly cancer [93]. The most abundant piRNA molecule in urine in combination with other small RNAs. Therefore, a bio- and cell-free urine was hsa-piR-019825, which is deregu- marker within these fluids would be the overall biotype lated in colorectal cancer patients [93]. Given the high distribution and the molecular signature within these bio- number of human piRNAs, they may play a role as an im- types, rather than a single molecule. portant small RNA species with functional targets that are yet to be elucidated and correlated with various disease Additional files conditions. Recent studies have identified differentially expressed piRNA molecules as potential biomarkers of Additional file 1: Figure S2. Relative biotype distribution among the various cancers [94–97]. The relatively high levels of these various bodily fluids of each donor. (TIF 566 kb) molecules in plasma might prove important as potential Additional file 2: Table S1. Common miRNAs between invasive fluids. (DOCX 14 kb) biomarkers. The low piRNAs levels in the other fluids can Additional file 3: Table S2. Common miRNAs between non-invasive be overcome by size selection methods, albeit not easily as fluids. (DOCX 13 kb) they overlap with other small RNA species. Additional file 4: Table S3. Common miRNAs between non-invasive Plasma and serum had a large fraction of reads map- fluids and blood. (DOCX 13 kb) ping to miscellaneous RNA (misc_RNA) (58 and 35%, Additional file 5: Figure S1. Venn diagram showing the overlap respectively). Only 4 YRNA-derived small RNAs between blood and the non-invasive bodily fluids. (TIF 536 kb) (s-RNYs) sequences were elevated within the misc_RNA Additional file 6: Table S4. Common miRNAs between all fluids. (DOCX 13 kb) fractions of these two fluids: RNY4, RNY4P10, RNY4P7 Additional file 7: Table S5. Unique miRNAs detected in each bodily and YRNA.295. It has been previously reported that fluid. (DOCX 14 kb) s-RNYs are abundant in human serum and plasma [98]. Additional file 8: Table S6. Unique miRNAs detected in the invasive They are potential cancer biomarkers and regulators of bodily fluids. (DOCX 14 kb) inflammation and cell death [99, 100]. Additional file 9: Table S7. Unique miRNAs detected in the non- invasive bodily fluids. (DOCX 13 kb) Conclusions Additional file 10: Table S8. Candidate novel miRNAs detected by miRDeep2 in the different bodily fluids and their matching result to the Our study showed that it is possible to successfully ac- miRCarta database. (XLSX 43 kb) complish NGS of the different bodily fluids (blood, El-Mogy et al. BMC Genomics (2018) 19:408 Page 21 of 24 therapeutic strategies in oncology (review). Int. J. Oncol. [internet]. 2016;49:5–32. Additional file 11: Top 50 tDRs in each fluid with their quantification Available from: http://www.ncbi.nlm.nih.gov/pubmed/27175518 and coverage percentage. (XLSX 4259 kb) 3. Keller A, Meese E. Can circulating miRNAs live up to the promise of being Additional file 12: Profiles of the top 50 mature tDR in each fluid. minimal invasive biomarkers in clinical settings? Wiley Interdiscip. Rev. RNA (PDF 4737 kb) [internet]. 2016;7:148–156. Available from: http://www.ncbi.nlm.nih.gov/ pubmed/26670867 Additional file 13: Table S9. piRNAs that represent an average of 1% or more of the entire piRNA counts of each bodily fluid. (DOCX 18 kb) 4. Cristodero M, Polacek N. The multifaceted regulatory potential of tRNA- derived fragments. Non-coding RNA Investig. [Internet]. 2017;7–7. Available Additional file 14: Figure S3. Exogenous mapping of unmapped saliva from: http://ncri.amegroups.com/article/view/3820/4459 reads. (PDF 24 kb) 5. Guay C, Regazzi R. Circulating microRNAs as novel biomarkers for diabetes Additional file 15: Figure S4. Exogenous mapping of unmapped mellitus. Nat Rev Endocrinol [Internet] 2013;9:513–521. Available from: cell-free saliva reads. (PDF 24 kb) http://www.nature.com/doifinder/10.1038/nrendo.2013.86 6. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell [internet]. 2004;116:281–297. Available from: http://www.ncbi.nlm.nih.gov/ Abbreviations pubmed/14744438 CPM: Counts per million; lincRNA: Long intergenic noncoding RNA; 7. Kim VN, Han J, Siomi MC. Biogenesis of small RNAs in animals. Nat Rev Mol miRNA: MicroRNA; misc_RNA: Miscellaneous RNA; mRNA: Messenger RNA; Cell Biol [Internet] 2009;10:126–139. Available from: http://www.nature.com/ Mt_rRNA: Mitochondrial rRNA; Mt_tRNA: Mitochondrial tRNA; ncRNA: Non- doifinder/10.1038/nrm2632 coding RNA; NGS: Next generation sequencing; PCA: Principal component 8. Ortiz-Quintero B. Cell-free microRNAs in blood and other body fluids, as analysis; piRNA: Piwi-interacting RNA; RIN: RNA integrity number; cancer biomarkers. Cell Prolif. [internet]. 2016;49:281–303. Available from: rRNA: Ribosomal RNA; snoRNA: Small nucleolar RNA; snRNA: Small nuclear http://www.ncbi.nlm.nih.gov/pubmed/27218664 RNA; s-RNYs: Y RNA-derived small RNAs; TMM: Trimmed mean of M-values 9. Brennecke J, Hipfner DR, Stark A, Russell RB, Cohen SM. Bantam encodes a developmentally regulated microRNA that controls cell proliferation and Availability of data and materials regulates the proapoptotic gene hid in Drosophila. Cell [internet]. 2003;113: The raw sequencing files of the study are available on the NCBI Sequence 25–36. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12679032 Read Archive (SRA) through the following link: https://www.ncbi.nlm.nih.gov/ 10. Scapoli L, Palmieri A, Lo Muzio L, Pezzetti F, Rubini C, Girardi A, et al. sra/SRP136264, SRA accession: SRP136264. MicroRNA expression profiling of oral carcinoma identifies new markers of tumor progression. Int. J. Immunopathol. Pharmacol. 2010;23:1229–34. Authors’ contributions Available from: http://www.ncbi.nlm.nih.gov/pubmed/21244772 ME, BL, NR and YHA conceived, designed and initiated the project. ME, BL, 11. Liang Z, Bian X, Shim H. Downregulation of microRNA-206 promotes SM performed experiments. THA and DY contributed to the analysis tools. invasion and angiogenesis of triple negative breast cancer. Biochem. THA, LN and ME carried out data analysis. ME and PR drafted the manuscript. Biophys. Res. Commun. 2016;477:461–6. Available from: http://www.ncbi. BL, THA, NR and YHA helped to revise the manuscript. LN re-edited the lan- nlm.nih.gov/pubmed/27318091 guage of the manuscript. All authors read and approved the final 12. Friedman RC, Farh KKH, Burge CB, Bartel DP. Most mammalian mRNAs are manuscript. conserved targets of microRNAs. Genome Res. 2009;19:92–105. Available from: http://www.ncbi.nlm.nih.gov/pubmed/18955434 Ethics approval and consent to participate 13. miRBase: the microRNA database [Internet]. [cited 2018 Mar 7]. Available The study protocol and consent were reviewed and approved by Veritas IRB from: http://www.mirbase.org/cgi-bin/browse.pl?org=hsa Ethics Review Board (Veritas IRB, Montreal, Canada. IRB tracking number: 14. Arroyo JD, Chevillet JR, Kroh EM, Ruf IK, Pritchard CC, Gibson DF, et al. 16198–16:02:416–11-2017). Healthy volunteer donors were recruited by Argonaute2 complexes carry a population of circulating microRNAs advertising in local communities and all participants gave their written independent of vesicles in human plasma. [cited 2018 Mar 7]; Available informed consent. from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3064324/pdf/pnas. 201019055.pdf 15. Hunter MP, Ismail N, Zhang X, Aguda BD, Lee EJ, Yu L, et al. Detection of Competing interests microRNA expression in human peripheral blood microvesicles. PLoS One. The following authors are employees at Norgen Biotek Corp.: ME, BL, THA, 2008 [cited 2018 Mar 7];3. Available from: https://www.ncbi.nlm.nih.gov/ DY, LN, PR and NR. YHA is the President and CEO of Norgen Biotek Corp. SM pmc/articles/PMC2577891/pdf/pone.0003694.pdf received an Industrial Undergraduate Student Research Award from the 16. Valadi H, Ekström K, Bossios A, Sjöstrand M, Lee JJ, Lötvall JO. Exosome- Natural Sciences and Engineering Research Council of Canada (NSERC) at mediated transfer of mRNAs and microRNAs is a novel mechanism of Norgen Biotek Corp. Some of Norgen Biotek’s products have been used in genetic exchange between cells. Nat. Cell Biol; 2007 [cited 2018 Mar 7]; the study, however the study is for basic scientific exploratory purposes and 9:654–659. Available from: http://www.nature.com/articles/ncb1596. is not intended to promote or test any of Norgen Biotek’s products. Nature Publishing Group 17. Vickers KC, Remaley AT. Lipid-based carriers of microRNAs and intercellular Publisher’sNote communication. [cited 2018 Mar 7]; Available from: https://www.ncbi.nlm. Springer Nature remains neutral with regard to jurisdictional claims in nih.gov/pmc/articles/PMC5570485/pdf/nihms892052.pdf published maps and institutional affiliations. 18. Wang K, Zhang S, Weber J, Baxter D, Galas DJ. Export of microRNAs and microRNA-protective protein by mammalian cells. Nucleic Acids Res. 2010; Author details 38:7248–59. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20615901 1 2 Norgen Biotek Corp, Thorold, ON L2V 4Y6, Canada. Molecular Biology 19. DumacheR,Ciocan V, Muresan C, Rogobete AF,Enache A.CirculatingmicroRNAs Department, National Research Centre, Dokki, Giza, Egypt. Department of as promising biomarkers in forensic body fluids identification. Clin. Lab. 2015;61: Biological Sciences, Brock University, St. Catharines, ON L2S 3A1, Canada. 1129–35. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26554231 20. Park NJ, Zhou H, Elashoff D, Henson BS, Kastratovic DA, Abemayor E, et al. Received: 22 November 2017 Accepted: 11 May 2018 Salivary microRNA: Discovery, characterization, and clinical utility for oral cancer detection. Clin. Cancer Res. 2009;15:5473–7. Available from: http:// www.ncbi.nlm.nih.gov/pubmed/19706812 References 21. Tölle A, Jung M, Rabenhorst S, Kilic E, Jung K, Weikert S. Identification of 1. Storz G. An expanding universe of noncoding RNAs. Science. 2002;296: microRNAs in blood and urine as tumour markers for the detection of 1260–1263. Available from: http://www.sciencemag.org/cgi/doi/10.1126/ urinary bladder cancer. Oncol. Rep. 2013;30:1949–56. Available from: http:// science.1072249 www.ncbi.nlm.nih.gov/pubmed/23877086 2. Gambari R, Brognara E, Spandidos DA, Fabbri E. Targeting oncomiRNAs and 22. Suryawanshi S, Vlad AM, Lin HM, Mantia-Smaldone G, Laskey R, Lee M, et al. mimicking tumor suppressor miRNAs: Ew trends in the development of miRNA Plasma MicroRNAs as novel biomarkers for endometriosis and El-Mogy et al. BMC Genomics (2018) 19:408 Page 22 of 24 endometriosis-associated ovarian cancer. Clin. Cancer Res. 2013;19:1213–24. 42. Green D, Fraser WD, Dalmay T. Transfer RNA-derived small RNAs in the Available from: http://www.ncbi.nlm.nih.gov/pubmed/23362326 cancer transcriptome. Pflugers Arch. Eur. J. Physiol. 2016;468:1041–7. 23. Hu Z, Chen X, Zhao Y, Tian T, Jin G, Shu Y, et al. Serum microRNA signatures Available from: http://www.ncbi.nlm.nih.gov/pubmed/27095039 identified in a genome-wide serum microRNA expression profiling predict 43. Garcia-Silva MR, Cabrera-Cabrera F, Güida MC, Cayota A. Hints of tRNA-derived survival of non-small-cell lung cancer. J. Clin. Oncol. 2010;28:1721–6. small RNAs role in RNA silencing mechanisms. Genes (Basel), Available from. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20194856 2012;3:603–14. http://www.ncbi.nlm.nih.gov/pubmed/24705078 24. Gilad S, Meiri E, Yogev Y, Benjamin S, Lebanony D, Yerushalmi N, et al. 44. Goodarzi H, Liu X, Nguyen HCB, Zhang S, Fish L, Tavazoie SF. Endogenous Serum microRNAs are promising novel biomarkers. PLoS One. 2008;3:e3148. tRNA-derived fragments suppress breast cancer progression via YBX1 Available from: http://www.ncbi.nlm.nih.gov/pubmed/18773077 displacement. Cell. 2015;161:790–802. Available from: http://www.ncbi.nlm. 25. Xie Z, Chen G, Zhang X, Li D, Huang J, Yang C, et al. Salivary MicroRNAs as nih.gov/pubmed/25957686 Promising Biomarkers for Detection of Esophageal Cancer. Lo AWI, editor. 45. Maute RL, Schneider C, SumazinP,HolmesA,CalifanoA,BassoK, et al. tRNA- PLoS One. 2013 [cited 2016 Aug 10];8:e57502. Available from: http://dx.plos. derived microRNA modulates proliferation and the DNA damage response and org/10.1371/journal.pone.0057502. Public Library of Science is down-regulated in B cell lymphoma. Proc. Natl. Acad. Sci. 2013;110:1404–1409. 26. Kosaka N, Iguchi H, Ochiya T. Circulating microRNA in body fluid: a new Available from: http://www.pnas.org/lookup/doi/10.1073/pnas.1206761110 potential biomarker for cancer diagnosis and prognosis. Cancer Sci. 2010;101: 46. Atala A. Re: sex hormone-dependent tRNA halves enhance cell proliferation 2087–92. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20624164 in breast and prostate cancers. J. Urol. 2016;195:1168–9. Available from: 27. Zhao N, Jin L, Fei G, Zheng Z, Zhong C. Serum microRNA-133b is http://www.ncbi.nlm.nih.gov/pubmed/26124144 associated with low ceruloplasmin levels in Parkinson’sdisease. Park. 47. Balatti V, Pekarsky Y, Croce CM. Role of the tRNA-derived small RNAs in Cancer: new potential biomarkers and target for therapy [internet]. 1st ed. Elsevier Inc.; Relat. Disord, Available from. 2014;20:1177–80. http://www.ncbi.nlm.nih. gov/pubmed/25218846 2017. Available from: https://doi.org/10.1016/bs.acr.2017.06.007.Adv.CancerRes 28. Kirchner S, Ignatova Z. Emerging roles of tRNA in adaptive translation, 48. Hatfield SD, Shcherbata HR, Fischer KA, Nakahara K, Carthew RW, signalling dynamics and disease. Nat. Rev. Genet. 2015;16:98–112. Available Ruohola-Baker H. Stem cell division is regulated by the microRNA from: http://www.nature.com/doifinder/10.1038/nrg3861 pathway. Nature. 2005;435:974–978. Available from: http://www.nature. 29. Selitsky SR, Baran-Gale J, Honda M, Yamane D, Masaki T, Fannin EE, et al. com/doifinder/10.1038/nature03816 Small tRNA-derived RNAs are increased and more abundant than 49. Rouget C, Papin C, Boureux A, Meunier AC, Franco B, Robine N, et al. microRNAs in chronic hepatitis B and C. Sci. Rep. 2015;5:7675. Available Maternal mRNA deadenylation and decay by the piRNA pathway in the from: http://www.nature.com/articles/srep07675 early Drosophila embryo. Nature. 2010;467:1128–1132. Available from: http:// 30. Ivanov P, Emara MM, Villen J, Gygi SP, Anderson P. Angiogenin-induced www.nature.com/doifinder/10.1038/nature09465 tRNA fragments inhibit translation initiation. Mol. Cell. 2011;43:613–23. 50. piRNABank: : a web resource on classified and clustered Piwi-interacting RNAs Available from: http://www.ncbi.nlm.nih.gov/pubmed/21855800 [Internet]. [cited 2018 Mar 7]. Available from: http://pirnabank.ibab.ac.in/stats.html 31. Yamasaki S, Ivanov P, Hu GF, Anderson P. Angiogenin cleaves tRNA and 51. Aravin AA, Sachidanandam R, Bourc’his D, Schaefer C, Pezic D, Toth KF, et al. promotes stress-induced translational repression. J. Cell Biol. 2009;185:35–42. A piRNA Pathway Primed by Individual Transposons Is Linked to De Novo Available from: http://www.ncbi.nlm.nih.gov/pubmed/19332886 DNA Methylation in Mice. Mol. Cell. 2008;31:785–99. Available from: http:// 32. Saikia M, Jobava R, Parisien M, Putnam A, Krokowski D, Gao X-H, et al. www.ncbi.nlm.nih.gov/pubmed/18922463 Angiogenin-Cleaved tRNA Halves Interact with Cytochrome c, Protecting 52. Aravin AA, Bourc’his D. Small RNA guides for de novo DNA methylation in Cells from Apoptosis during Osmotic Stress. Mol. Cell. Biol. 2014;34:2450– mammalian germ cells. Genes Dev. 2008;22:970–5. Available from: http:// 2463. Available from: http://mcb.asm.org/cgi/doi/10.1128/MCB.00136-14 www.ncbi.nlm.nih.gov/pubmed/18413711 33. Schaffer AE, Eggens VRC, Caglayan AO, Reuter MS, Scott E, Coufal NG, et al. 53. Hirakata S, Siomi MC. piRNA biogenesis in the germline: From transcription CLP1 founder mutation links tRNA splicing and maturation to cerebellar of piRNA genomic sources to piRNA maturation. Biochim. Biophys. Acta - development and neurodegeneration. Cell. 2014;157:651–63. Available from: Gene Regul. Mech. 2016;1859:82–92. Available from: https://doi.org/10.1016/ http://www.ncbi.nlm.nih.gov/pubmed/24766810 j.bbagrm.2015.09.002. Elsevier B.V 54. Esteller M. Non-coding RNAs in human disease. Nat. Rev. Genet. [Internet]. 2011; 34. Gebetsberger J, Zywicki M, Künzi A, Polacek N. TRNA-derived fragments 12:861–874. Available from: http://www.nature.com/doifinder/10.1038/nrg3074 target the ribosome and function as regulatory non-coding RNA in Haloferax volcanii. Archaea. 2012;2012:260909. Available from: http://www. 55. Zhang J, Chiodini R, Badr A, Zhang G. The impact of next-generation ncbi.nlm.nih.gov/pubmed/23326205 sequencing on genomics [Internet]. J. Genet. Genomics. 2011 [cited 2017 35. Gebetsberger J, Wyss L, Mleczko AM, Reuther J, Polacek N. A tRNA-derived Jul 16]. p. 95–109. Available from: https://www.ncbi.nlm.nih.gov/pmc/ fragment competes with mRNA for ribosome binding and regulates articles/PMC3076108/pdf/nihms-282401.pdf translation during stress. RNA Biol. 2017;14:1364–73. Available from: http:// 56. Byron SA, Van Keuren-Jensen KR, Engelthaler DM, Carpten JD, Craig DW. www.ncbi.nlm.nih.gov/pubmed/27892771 Translating RNA sequencing into clinical diagnostics: Opportunities and 36. Sharma U, Conine CC, Shea JM, Boskovic A, Derr AG, Bing XY, et al. challenges. Nat. Rev. Genet. 2016 [cited 2017 Jul 16];17:257–271. Available Biogenesis and function of tRNA fragments during sperm maturation and from: http://www.nature.com/doifinder/10.1038/nrg.2016.10 fertilization in mammals. Science. 2016;351:391–396. Available from: http:// 57. Shore S, Henderson JM, Lebedev A, Salcedo MP, Zon G, McCaffrey AP, et al. www.sciencemag.org/cgi/doi/10.1126/science.aad6780 Small RNA library preparation method for next-generation sequencing using chemical modifications to prevent adapter dimer formation. PLoS 37. Venkatesh T, Suresh PS, Tsutsumi R. TRFs: miRNAs in disguise. Gene. 2016; 579:133–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26743126 One. 2016 [cited 2017 Jul 16];11. Available from: http://journals.plos.org/ plosone/article/file?id=10.1371/journal.pone.0167009&type=printable 38. Haussecker D, Huang Y, Lau A, Parameswaran P, Fire AZ, Kay MA. Human tRNA-derived small RNAs in the global regulation of RNA 58. Chen X, Ba Y, Ma L, Cai X, Yin Y, Wang K, et al. Characterization of silencing. Rna. 2010;16:673–695. Available from: http://rnajournal.cshlp. microRNAs in serum: A novel class of biomarkers for diagnosis of cancer org/cgi/doi/10.1261/rna.2000810 and other diseases. Cell Res. 2008;18:997–1006. Available from: http://www. 39. Elbarbary RA, Takaku H, Uchiumi N, Tamiya H, Abe M, Takahashi M, et al. nature.com/doifinder/10.1038/cr.2008.282 Modulation of gene expression by human cytosolic tRNase Z(L) through 5′- 59. Mitchell PS, Parkin RK, Kroh EM, Fritz BR, Wyman SK, Pogosova-Agadjanyan half-tRNA. PLoS One. 2009;4:e5908. Available from: http://www.ncbi.nlm.nih. EL, et al. Circulating microRNAs as stable blood-based markers for cancer gov/pubmed/19526060 detection. Proc. Natl. Acad. Sci. 2008;105:10513–10518. Available from: 40. Ghildiyal M, Zamore PD. Small silencing RNAs: an expanding universe. Nat http://www.pnas.org/cgi/doi/10.1073/pnas.0804549105 Rev Genet [Internet] 2009;10:94–108. Available from: http://www.nature. 60. Wang J, Zhang KY, Liu SM, Sen S. Tumor-associated circulating micrornas as com/doifinder/10.1038/nrg2504 biomarkers of cancer. Molecules. 2014;19:1912–38. Available from: http:// 41. Kumar A, Karmarkar AM, Tan A, Graham JE, Arcari CM, Ottenbacher KJ, et al. www.ncbi.nlm.nih.gov/pubmed/24518808 The effect of obesity on incidence of disability and mortality in Mexicans 61. Cheng J, Guo JM, Xiao BX, Miao Y, Jiang Z, Zhou H, et al. PiRNA, the new aged 50 years and older. Salud Publica Mex. 2015 [cited 2017 Jul 12];57: non-coding RNA, is aberrantly expressed in human cancer cells. Clin. Chim. S31–S38. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/ Acta. 2011;412:1621–5. Available from: http://www.ncbi.nlm.nih.gov/ PMC4572697/pdf/nihms673180.pdf pubmed/21616063 El-Mogy et al. BMC Genomics (2018) 19:408 Page 23 of 24 62. Hashim A, Rizzo F, Marchese G, Ravo M, Tarallo R, Nassa G, et al. RNA 80. Chen Z, Yu T, Cabay RJ, Jin Y, Mahjabeen I, Luan X, et al. miR-486-3p, miR- sequencing identifies specific PIWI-interacting small non-coding RNA 139-5p, and miR-21 as Biomarkers for the Detection of Oral Tongue expression patterns in breast cancer. Oncotarget. 2014;5:9901–10. Available Squamous Cell Carcinoma. Biomark. Cancer [Internet]. 2017;9:1–8. Available from: http://www.oncotarget.com/fulltext/2476 from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5224348/ 63. Li Y, Wu X. Piwi-Interacting RNAs (piRNAs) Are Dysregulated in Renal Cell 81. Ye H, Yu X, Xia J, Tang X, Tang L, Chen F. MiR-486-3p targeting ECM1 Carcinoma and Associated with Tumor Metastasis and Cancer-Specific represses cell proliferation and metastasis in cervical cancer. Biomed. Survival. Mol. Med. 2015;21:1. Available from: http://www.molmed.org/ Pharmacother. [Internet], Available from. 2016;80:109–14. http://www.ncbi. content/pdfstore/14_203_Li.pdf nlm.nih.gov/pubmed/27133046 64. Reeves ME, Firek M, Jliedi A, Amaar YG. Identification and characterization of 82. Shindo Y, Hazama S, Nakamura Y, Inoue Y, Kanekiyo S, Suzuki N, et al. miR- RASSF1C piRNA target genes in lung cancer cells. Oncotarget. 2017; 196b, miR-378a and miR-486 are predictive biomarkers for the efficacy of Available from: http://www.oncotarget.com/abstract/15965 vaccine treatment in colorectal cancer. Oncol. Lett. [Internet]. 2017;14:1355– 65. Chen C, Ridzon DA, Broomer AJ, Zhou Z, Lee DH, Nguyen JT, et al. Real- 62. Available from: http://www.ncbi.nlm.nih.gov/pubmed/28789351 time quantification of microRNAs by stem-loop RT-PCR. Nucleic Acids Res. 83. Ogawa Y, Taketomi Y, Murakami M, Tsujimoto M, Yanoshita R. Small 2005 [cited 2017 Jan 1];33:e179. Available from: https://academic.oup.com/ RNA transcriptomes of two types of exosomes in human whole saliva nar/article-lookup/doi/10.1093/nar/gni178 determined by next generation sequencing. Biol Pharm Bull [Internet]. 66. Subramanian SL, Kitchen RR, Alexander R, Carter BS, Cheung KH, 2013;36:66–75. Available from: http://www.ncbi.nlm.nih.gov/entrez/ Laurent LC, et al. Integration of extracellular RNA profiling data using query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids= metadata, biomedical ontologies and Linked Data technologies. J. 23302638 Extracell. Vesicles. 2015;4:27497. Available from: http://www.ncbi.nlm.nih. 84. Bonifacio LN, Jarstfer MB. MiRNA profile associated with replicative gov/pubmed/26320941 senescence, extended cell culture, and ectopic telomerase expression in 67. Robinson MD, Oshlack A. A scaling normalization method for human foreskin fibroblasts. PLoS One [Internet]. 2010;5:1–8. Available from: differential expression analysis of RNA-seq data. Genome Biol. 2010;11: http://www.ncbi.nlm.nih.gov/pubmed/20824140 85. Wang L, Shi Z, Jiang C, Liu X, Chen Q, Qian X, et al. MiR-143 acts as a tumor R25. Available from: http://genomebiology.biomedcentral.com/articles/10. 1186/gb-2010-11-3-r25 suppressor by targeting N-RAS and enhances temozolomide-induced apoptosis in glioma. Oncotarget [Internet]. 2014;5:5416–27. Available from: 68. Dhahbi JM, Atamna H, Boffelli D, Magis W, Spindler SR, Martin DIK. Deep http://www.oncotarget.com/fulltext/2116 sequencing reveals novel micrornas and regulation of microRNA expression during cell senescence. PLoS One. 2011;6:e20509. Available from: http:// 86. Zhou K, Spillman MA, Behbakht K, Komatsu JM, Abrahante JE, Hicks D, et al. www.ncbi.nlm.nih.gov/pubmed/21637828 A method for extracting and characterizing RNA from urine: For 69. Backes C, Fehlmann T, Kern F, Kehl T, Lenhof HP, Meese E, et al. MiRCarta: A downstream PCR and RNAseq analysis. Anal. Biochem. [Internet]. 2017;536: central repository for collecting miRNA candidates. Nucleic Acids Res. 2018 8–15. Available from: https://doi.org/10.1016/j.ab.2017.08.003. Elsevier Inc [cited 2018 Apr 12];46:D160–D167. Available from: https://www.ncbi.nlm.nih. 87. Zhang L, Sun J, Wang B, Ren JC, Su W, Zhang T. MicroRNA-10b triggers gov/pmc/articles/PMC5753177/pdf/gkx851.pdf the epithelial–mesenchymal transition (EMT) of laryngeal carcinoma 70. Yeri A, Courtright A, Reiman R, Carlson E, Beecroft T, Janss A, et al. Total Hep-2 cells by directly targeting the E-cadherin. Appl. Biochem. extracellular small RNA profiles from plasma, saliva, and urine of healthy Biotechnol. [internet]. 2015;176:33–44. Available from: http://www.ncbi. subjects. Sci. Rep. 2017 [cited 2017 Jul 11];7:44061. Available from: http:// nlm.nih.gov/pubmed/25875782 www.nature.com/articles/srep44061 88. Xiao H, Li H, Yu G, Xiao W, Hu J, Tang K, et al. MicroRNA-10b promotes migration and invasion through KLF4 and HOXD10 in human bladder 71. Park NJ, Zhou X, Yu T, Brinkman BMN, Zimmermann BG, Palanisamy V, et al. cancer. Oncol. Rep. [Internet]. 2014;31:1832–8. Available from: http://www. Characterization of salivary RNA by cDNA library analysis. Arch. Oral Biol. ncbi.nlm.nih.gov/pubmed/26311318 2007 [cited 2017 Oct 11];52:30–35. Available from: https://www.ncbi.nlm.nih. gov/pmc/articles/PMC2743855/pdf/nihms15843.pdf 89. Ma Z, Chen Y, Min L, Li L, Huang H, Li J, et al. Augmented miR-10b expression 72. Weber JA, Baxter DH, Zhang S, Huang DY, Huang KH, Lee MJ, et al. The associated with depressed expression of its target gene KLF4 involved in microRNA spectrum in 12 body fluids. Clin Chem. 2010;56:1733–41. gastric carcinoma. Int. J. Clin. Exp. Pathol. [Internet]. 2015;8:5071–9. Available 73. Gallo A, Tandon M, Alevizos I, Illei GG. The majority of microRNAs detectable from: http://www.ncbi.nlm.nih.gov/pubmed/26191201 in serum and saliva is concentrated in exosomes. Afarinkia K, editor. PLoS 90. Abdelmaksoud-Dammak R, Chamtouri N, Triki M, Saadallah-Kallel A, Ayadi One. 2012 [cited 2016 Aug 10];7:e30679. Available from: http://dx.plos.org/ W, Charfi S, et al. Overexpression of miR-10b in colorectal cancer patients: 10.1371/journal.pone.0030679. Public Library of Science Correlation with TWIST-1 and E-cadherin expression. Tumor Biol. [Internet]. 2017 [cited 2017 Oct 20];39:101042831769591. Available from: http:// 74. Majem B, Rigau M, Reventós J, Wong DT. Non-coding RNAs in saliva: journals.sagepub.com/doi/10.1177/1010428317695916. SAGE emerging biomarkers for molecular diagnostics. Int. J. Mol. Sci. 2015;16: PublicationsSage UK: London, England 8676–98. Available from: http://www.ncbi.nlm.nih.gov/pubmed/25898412 75. Spielmann N, Ilsley D, Gu J, Lea K, Brockman J, Heater S, et al. The human 91. Lin X, Lo H-C, Wong DTW, Xiao X. Noncoding RNAs in human saliva as salivary RNA transcriptome revealed by massively parallel sequencing. Clin. potential disease biomarkers. Front Genet [Internet] 2015;6:1–6. Available Chem. 2012;58:1314–21. Available from: http://www.ncbi.nlm.nih.gov/ from: http://www.frontiersin.org/RNA/10.3389/fgene.2015.00175/full pubmed/22773539 92. Martinez VD, Vucic EA, Thu KL, Hubaux R, Enfield KSS, Pikor LA, et al. Unique 76. Li M, Zeringer E, Barta T, Schageman J, Cheng A, Vlassov A V. Analysis of the somatic and malignant expression patterns implicate PIWI-interacting RNAs RNA content of the exosomes derived from blood serum and urine and its in cancer-type specific biology. Sci. Rep. [Internet]. 2015 [cited 2017 Oct 10]; potential as biomarkers. Philos. Trans. R. Soc. B Biol. Sci. 2014;369:20130502. 5. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4444957/ Available from: http://rstb.royalsocietypublishing.org/cgi/doi/10.1098/rstb. pdf/srep10423.pdf 2013.0502 93. Yuan T, Huang X, Woodcock M, Du M, Dittmar R, Wang Y, et al. Plasma 77. Ludwig N, Leidinger P, Becker K, Backes C, Fehlmann T, Pallasch C, et al. extracellular RNA profiles in healthy and cancer patients. Sci. Rep. [Internet]. Distribution of miRNA expression across human tissues. Nucleic Acids Res. 2016;6:19413. Available from: http://www.nature.com/articles/srep19413 2016 [cited 2018 Apr 13];44:3865–3877. Available from: https://www.ncbi. 94. Lim SL, Ricciardelli C, Oehler MK, De Arao Tan IMD, Russell D, Grützner F. nlm.nih.gov/pmc/articles/PMC4856985/pdf/gkw116.pdf Overexpression of piRNA pathway genes in epithelial ovarian cancer. PLoS 78. Sromek M, Glogowski M, Chechlinska M, Kulinczak M, Szafron L, Zakrzewska One [Internet]. 2014 [cited 2017 Oct 10];9. Available from: http://journals. K, et al. Changes in plasma miR-9, miR-16, miR-205 and miR-486 levels after plos.org/plosone/article/file?id=10.1371/journal.pone.0099687&type= non-small cell lung cancer resection. Cell. Oncol. 2017;40:529–36. Available printable from: http://www.ncbi.nlm.nih.gov/pubmed/28634901 95. Müller S, Raulefs S, Bruns P, Afonso-Grunz F, Plötner A, Thermann R, et al. 79. Sierzega M, Kaczor M, Kolodziejczyk P, Kulig J, Sanak M, Richter P. Evaluation Next-generation sequencing reveals novel differentially regulated mRNAs, of serum microRNA biomarkers for gastric cancer based on blood and lncRNAs, miRNAs, sdRNAs and a piRNA in pancreatic cancer. Mol. Cancer tissue pools profiling: The importance of MIR-21 and MIR-331. Br. J. Cancer. [Internet]. 2015;14:94. Available from: http://molecular-cancer.biomedcentral. 2017;117:266–73. Available from: http://www.nature.com/doifinder/10.1038/ com/articles/10.1186/s12943-015-0358-5 bjc.2017.190 El-Mogy et al. BMC Genomics (2018) 19:408 Page 24 of 24 96. Martinez VD, Enfield KSS, Rowbotham DA, Lam WL. An atlas of gastric PIWI- interacting RNA transcriptomes and their utility for identifying signatures of gastric cancer recurrence. Gastric Cancer [internet]. 2016;19:660–5. Available from: http://www.ncbi.nlm.nih.gov/pubmed/25779424 97. Koduru S V, Tiwari AK, Hazard SW, Mahajan M, Ravnic DJ. Exploration of small RNA-seq data for small non-coding RNAs in Human Colorectal Cancer. J. Genomics [Internet]. 2017 [cited 2017 Oct 10];5:16–31. Available from: http:// www.jgenomics.com/v05p0016.htm 98. Dhahbi JM, Spindler SR, Atamna H, Boffelli D, Mote P, Martin DIK. 5’-YRNA fragments derived by processing of transcripts from specific YRNA genes and pseudogenes are abundant in human serum and plasma. Physiol. Genomics [Internet]. 2013;45:990–8. Available from: http://physiolgenomics. physiology.org/cgi/doi/10.1152/physiolgenomics.00129.2013 99. Dhahbi, Spinder S, Atamna H, Boffelli D, Martin D. Deep Sequencing of Serum Small RNAs Identifies Patterns of 5&amp;#39; tRNA Half and YRNA Fragment Expression Associated with Breast Cancer. Biomark. Cancer [Internet]. 2014 [cited 2016 Apr 11];6:37. Available from: http://www.la-press. com/deep-sequencing-of-serum-small-rnas-identifies-patterns-of-5-trna-half- article-a4553 100. Hizir Z, Bottini S, Grandjean V, Trabucchi M, Repetto E. RNY (YRNA)-derived small RNAs regulate cell death and inflammation in monocytes/ macrophages. Cell Death Dis. [Internet]. 2017 [cited 2017 Oct 11];8:e2530. Available from: http://www.nature.com/doifinder/10.1038/cddis.2016.429

Journal

BMC GenomicsSpringer Journals

Published: May 29, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off