relative rarity of transferred RNAs it is highly recommended to
sequence on the order of 100–150 million paired-end reads that
are at least 75 bp in length. However, that can be adapted based on
the particular experimental design (see Note 12).
Another point to be considered is the length and type of
fragment (single or paired ends). Due to the high degree of homology between the human and mouse transcriptomes, using suboptimal length reads may lead to the mapping of human reads to the
mouse genome and vice versa. To determine the optimum length
and type of fragment end, we did an in silico simulation of mapping
25, 50, 75, and 100 bp long single- and paired-end reads from the
human transcriptome to the mouse transcriptome. We observed
that 25 bp, single-end reads have a very high (>85%) nonspecific
alignment with the mouse genome (Fig. 6 and Table 2). This
nonspecific alignment decreases drastically when using longer
reads. A similar analysis was done for mouse reads aligned to the
human genome. Based on this analysis, we recommend using 75 bp
paired-end reads when performing human-mouse cell co-cultures.
3.6 Identification
of Transferred RNAs
To identify human RNAs transferred from MCF7 to MBS-MEF
cells, the sequenced reads from MBS-MEF-enriched fraction of the
“co-culture” sample are aligned to the human reference genome
(e.g., hg38). Bona fide transferred human RNAs from co-cultured
cells will have more reads that align to the human reference genome
as compared to the “mix” sample, wherein the human and mouse
cells were mixed and immediately sorted. A similar analysis can be
done to detect mouse RNAs transferred from MBS-MEF to MCF7
cells by aligning the reads from MCF7-enriched fraction to a reference mouse genome (e.g., mm10). For identifying RNAs that are
enriched in the “co-culture” samples, as compared to the “mix”
samples, any statistical package designed for differential gene
expression analysis can be used. We use DESeq2 [22, 23]. Other
packages for similar analysis include edgeR, DSS, and EBSeq [24–
26]. Since we know that β-actin-MBS mRNA robustly transfers
from MBS-MEFs to MCF7 cells, it can be used as a positive control
for analysis of the sequencing data.
4 Notes
1. The protocol can be adapted for studying RNA transfer in
“in vivo” systems such as chimeric (mosaic) tissue, tumors,
and organoids. For example, chimeric tissues or tumor xenografts can be made by injecting human cells into immunosuppressed mice. After sacrificing and harvesting of the tissue, the
same protocol can be used to sort the cells and perform the
required downstream analysis. In addition, this protocol can
also be scaled up to analyze higher numbers of cells, if so
required.
Use of RNA Tagging as a Control for RNA Transferome Analysis
209
sequence on the order of 100–150 million paired-end reads that
are at least 75 bp in length. However, that can be adapted based on
the particular experimental design (see Note 12).
Another point to be considered is the length and type of
fragment (single or paired ends). Due to the high degree of homology between the human and mouse transcriptomes, using suboptimal length reads may lead to the mapping of human reads to the
mouse genome and vice versa. To determine the optimum length
and type of fragment end, we did an in silico simulation of mapping
25, 50, 75, and 100 bp long single- and paired-end reads from the
human transcriptome to the mouse transcriptome. We observed
that 25 bp, single-end reads have a very high (>85%) nonspecific
alignment with the mouse genome (Fig. 6 and Table 2). This
nonspecific alignment decreases drastically when using longer
reads. A similar analysis was done for mouse reads aligned to the
human genome. Based on this analysis, we recommend using 75 bp
paired-end reads when performing human-mouse cell co-cultures.
3.6 Identification
of Transferred RNAs
To identify human RNAs transferred from MCF7 to MBS-MEF
cells, the sequenced reads from MBS-MEF-enriched fraction of the
“co-culture” sample are aligned to the human reference genome
(e.g., hg38). Bona fide transferred human RNAs from co-cultured
cells will have more reads that align to the human reference genome
as compared to the “mix” sample, wherein the human and mouse
cells were mixed and immediately sorted. A similar analysis can be
done to detect mouse RNAs transferred from MBS-MEF to MCF7
cells by aligning the reads from MCF7-enriched fraction to a reference mouse genome (e.g., mm10). For identifying RNAs that are
enriched in the “co-culture” samples, as compared to the “mix”
samples, any statistical package designed for differential gene
expression analysis can be used. We use DESeq2 [22, 23]. Other
packages for similar analysis include edgeR, DSS, and EBSeq [24–
26]. Since we know that β-actin-MBS mRNA robustly transfers
from MBS-MEFs to MCF7 cells, it can be used as a positive control
for analysis of the sequencing data.
4 Notes
1. The protocol can be adapted for studying RNA transfer in
“in vivo” systems such as chimeric (mosaic) tissue, tumors,
and organoids. For example, chimeric tissues or tumor xenografts can be made by injecting human cells into immunosuppressed mice. After sacrificing and harvesting of the tissue, the
same protocol can be used to sort the cells and perform the
required downstream analysis. In addition, this protocol can
also be scaled up to analyze higher numbers of cells, if so
required.
Use of RNA Tagging as a Control for RNA Transferome Analysis
209