We collected peripheral blood (about 2 ml) from 16 individuals, then the blood samples were centrifugated at 3000 rpm for 10 min at 4˚C. The serum samples were collected and 500 μl serum of every individual were mixed, then the mixed serum samples were divided into 500 μl, 1.5 ml, 2.5 ml, 3.5 ml, stored at −80˚C until use [12]. All subjects gave their informed consent for inclusion before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of Jiangsu Province Hospital.

The total RNA from serum (extracellular RNA) was extracted by using miRNeasy Serum/Plasma Kit according to the manufacturer’s instructions with modifications. Simply, every 500 μl serum add 1 ml TRIzol reagent, mixed by vortex and incubated at room temperature for 5 min, followed 250 μl chloroform were added to each tube, vortexed and incubated at room temperature for 15 min [13]. After centrifuge the sample at 12,000 g for 15 minutes at 4˚C, the aqueous phase was placed into a new tube. Then we added equal volume of 100% isopropanol to the aqueous phase and incubated at room temperature for 40 min. After that, the aqueous phase transferred to one spin column according to the instructions. Finally, exRNA was dissolved 25 μl RNase-free water. The quality of total exRNA was determined by NanoDrop ND-1000.

DNA contaminations from extracted total RNA were removed by Recombination DNase I according to the manufacturer’s instructions. After incubated at 37˚C for 25 min, RNA was purified by VAHTS RNA clean beads following the protocol. Because the concentration of extracted RNA is too low, we adopted the exRNA library preparation without rRNA depletion. The exRNA-seq library constructed following manufacturer instructions using the NEBNext® Ultra™ Directional RNA Library Prep Kit for Illumina®. And exRNAlibraries were amplified by PCR cycles using Illumina compatible index primers. The amplified libraries were resolved on a 2% agarose gel and recollected the 250 - 500 bp cDNA fragments by using TIANgel Midi Purification Kit. Finally, 4 sequencing libraries were pooled into a single sequencing lane and sequenced by Hiseq-PE150 (Illumina).

Raw data of fastq format were firstly processed through in-house perl scripts. In this step, reads containing ploy-N, adapter and low-quality reads were filtered. Q20, Q30 and GC content the clean data were calculated. All the downstream analyses were based on the clean data with high quality. HISAT2 (version 2.1.0) was used to align exRNA-seq paired-end clean reads to the reference genome (hg38). The number RNA transcripts were counted by feature Counts [14] (version 1.6.3) using the aligned reads and the gene annotation file (gencode.v30.annotation.gtf) as the input file. Finally, using CIRI (CircRNA Identifier) software [15] to predict the “junction read” of the sequence, and identify circRNA from these “junction read”, and use circbase to annotate and count circRNA.

We prepared 4 different volumes (500 μl, 1.5 ml, 2.5 ml, 3.5 ml) of serum samples, then the exRNAs were extracted and the RNA-seq library were prepared. Using Qubit2.0 to measure the concentration of the extracted RNA, but we found that the RNA concentration was too low to be detected. And using agarose gel electrophoresis (Figure 1(A)) and Agilent 2100 (Figure 1(B)) to show the quality of cDNA library, and we found the library fragment size was 250 - 500 bp, the average concentration of the cDNA library is 1.84 nmol/l.

We use fastQC software to analyze the quality of sequencing data. In the exRNA sequencing data, most of the reads reached the Q30 standard (the error rate of each base is 0.1%) (Figure 2(A)). The average GC content of each read also tends to the standard level (Figure 2(B)).

We evaluated the results of serum exRNA sequencing of different volumes (500 μl, 1.5 ml, 2.5 ml, 3.5 ml) and the circRNA prediction. We found that the

number of genes, the reads of gene, circRNA number, and circRNA counts obtained from the analysis of different volumes of serumexRNA sequencing did not change with the volume (Table 1).

To get a better understanding about different volumes of serum’s exRNA biotypes distribution, we analyzed the total aligned reads of each biotype between different volumes of serum (Figure 3(A)) and visualized this using a donut graph where each layer represents the reads and numbers of RNA biotype for each volume (Figure 3(B)). The results show the reads and numbers of the main biotypes of different volumes serum did not change with the volume.

We select the transcripts that have over 20 reads in each sample and analyze their expression level in different volumes of serum. The results show that there

is no obvious connection between gene expression and serum volume, and most genes are averagely distributed among the 4 samples (Figure 4).

We used CIRI software to predict the circRNA of RNA sequencing data from 4 samples. In total, we found that the total reading of 48 circRNA in serum was greater than 3 (Figure 5(A)). In general, most circRNAs have very low abundance, and only a few circRNAs are detected with higher counts (Figure 5(B)). However, the highly expressed circRNA expressed the similar counts in the 4 samples.