The human TRAM1 locus expresses circular RNAs

Numerous indirect and in silico produced evidences suggest circular RNAs (circRNA) in mammals while thorough experimental proofs of their existence have rarely been reported. Biological studies of circRNA, however, should be based on experimentally verified circRNAs. Here, we describe the identification of two circRNAs originating from the gene locus of the translocation associated membrane protein 1 (TRAM1). Linear and potentially circular TRAM1-specific transcripts were identified in a transcriptome analysis of urine RNA of bladder cancer (BCa) patients versus healthy donors. Thus, we first focused on the topology of TRAM1-specific transcripts. We describe conclusive experimental evidence for the existence of TRAM1-specific circRNAs in the human BCa cell lines ECV-304 and RT-4. PCR-based methodology followed by cloning and sequencing strongly indicated the circular topology of two TRAM1 RNAs. Further, studies with exon fusion sequence-specific antisense oligonucleotides (asON) and RNase H as well as studies in the use of RNase R contribute to conclusive set of experiments supporting the circular topology of TRAM1 transcripts. On the biological side, TRAM1-specific circRNAs showed low expression levels and minor differences in BCa cell lines while linear TRAM1 transcripts displayed down-regulated expression in the higher cancer stage model ECV-304 versus more differentiated RT-4 cells.


Supplementary Methods
RNA Isolation from urine samples: Spontaneously voided urine of donors was collected and stabilized with one volume of a lysis buffer (6 mol/l guanidinium isothiocyanate, 0.05 mol/l sodium acetate, and 0.5% N-lauroylsarcosine) as previously described 38 . Stabilized urine samples were frozen in liquid nitrogen and stored at -80°C. To isolate RNA from human urine, stabilized samples were thawed slowly and adjusted to pH 7.0 by adding 1 M HEPES buffer. RNeasy Midi Kit (QIAGEN, Hilden, Germany) was used for RNA preparation with minor modifications: Instead of using RLT buffer, recommended volumes of 70% ethanol and mercaptoethanol were added directly to the samples. Subsequent steps were performed following the manufacturer's instructions. RNA samples were eluted in 320 µl of RNase-free water, lyophylized, and re-suspended in 16 µl of RNase-free water. Urine RNA was stored at -80 °C.
RNA quantification and quality assessment: RNA samples were quantified using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, U.S.A.). Agilent 2100 Bioanalyzer in combination with the Agilent RNA 6000 Pico Kit (Agilent Technologies, Santa Clara, CA, U.S.A.) were used to measure concentration and integrity of pooled urinary RNA.

Synthesis, purification and quantification of double-stranded cDNA libraries:
Seven urine samples were pooled and used to compose the final patient pools C and HR (Supplementary Table 1). Both urine RNA pools were used to produce cDNA libraries which were synthesized using the SMARTer Stranded Total RNA-Seq Kit -Pico Input Mammalian (TaKaRa Bio Inc., Kusatsu, prefecture Shiga, Japan). According to manufacturer's instructions, 10 ng of urine RNA was fragmented and used for first-strand synthesis. For sequencing on Illumina platforms, Illumina adapters were added and ribosomal cDNA were depleted. The final double-stranded cDNA library was amplified via PCR according to manufacturer's instructions and purified using the Agencourt AMPure XP PCR purification system (Beckman Coulter, Brea, CA, U.S.A.). The first purification step was performed according to the manufacturer´s instructions. The second purification step contained minor modifications: The amount of beads was increased to 100 µl volume and the cDNA was eluted in 20 µl elution buffer. Quantification of double-stranded cDNA libraries was performed using the Qubit dsDNA HS Assay Kit and the Qubit Fluorometer (Thermo Fisher Scientific, Waltham, MA, U.S.A.).
RNA sequencing: Double-stranded cDNA libraries of patient pools C and HR were sequenced by GATC Biotech AG (GATC Biotech AG, Konstanz, Germany) using HiSeq4000 (Illumina, San Diego, CA, U.S.A). The Illumina platform was used with paired-end (PE) mode and read lengths of 125 nucleotides.
Data quality control: Quality of reads provided by GATC Biotech AG was checked using the software FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) by Simon Andrews at Babraham Bioinformatics. Data analysis: Codes for analyses of transcriptome data were written in the UNIX command line. First, the cDNA and ncRNA databases provided by Ensembl (version e87) and the circRNA database provided by circBase (version updated December 2015) were downloaded as reference data sets. Next, the number of raw reads for each sequencing file was counted.
Furthermore, this step provides information if sequencing depth of files was comparable. Databases were indexed to accelerate the mapping process and read mapping was performed using the program BWA. This algorithm does not perform global alignments but searches for base pair matches long enough to identify a single transcript. Resulting mapped reads were recorded and transcript names, lengths and read counts were reported. Finally, reads were normalized to TPMs (transcripts per million) to eliminate biases introduced by transcript length and sequencing depth. Gene and transcript lengths provided by the database versions of Ensembl and circBase were used for calculation of TPM values. This step allows a direct comparison of transcript quantities between the two different patient groups C and HR. .$*-#)* 0"'(. ? [=6>] (log(fold change) > 0 for transcripts with a higher expression in patient group H, log(fold change) = 0 without differences in gene expression in both patient groups, and log(fold change) < 0 for transcripts with a lower expression in patient group H)