Processes 2018, 6,42
by either repressing translation of the mRNA or triggering its degradation [21]. MiRNAs have been
found to play a role in numerous cellular processes, from cell cycle control and apoptosis regulation
to hormone production and immune response [22]. Importantly, misexpression of miRNAs has been
implicated in a number of different disease states ranging from cardiovascular [23] and neurological
disorders [24] to many types of cancer [25]. Having been associated with such a wide array of processes
and pathologies, these molecules have garnered increased attention recently as investigators begin to
evaluate their potential utility as biomarkers and therapies.
While these two mechanisms are fairly well understood independently, only recently have reports
profiled the functional connection between RNA editing and miRNAs. For example, it has been shown
that ADAR1 forms a complex with Dicer through direct protein interactions and enhances global
miRNA processing [26]. Further, ADAR deamination of pri-miRNA transcripts can cause alterations to
their structural conformations and subsequent maturation and processing by Drosha and Dicer [27,28].
In addition, while any editing of miRNA transcripts can have functional implications, arguably the
most critical changes are to the seed regions of the miRNAs as this can drastically alter the set of
genes able to be regulated [29]. This is especially true in cancer where altered miRNA regulation of
oncogenes and tumor suppressors can lead to tumor formation [17]. Importantly, it should be noted
that, in addition to editing the miRNA transcript, ADAR can also edit the 3 ′ UTR of target mRNAs.
This modification dramatically increases the interplay between miRNAs and their targets by allowing
a different set of miRNAs to regulate a given mRNA depending on if the transcript has been edited
or not. Unfortunately, while the effects of editing the miRNA transcripts themselves have been well
documented, this opposite effect of editing mRNAs in regions complementary to miRNA seeds is less
understood. However, a number of reports suggesting this plays a significantly underappreciated role
in miRNA targeting have surfaced within the last year [30–33]. To examine this, our study identified
edit sites within two breast cancer cell lines (MCF-7 and MDA-MB-231) and analyzed the effect these
edits had on subsequent regulation by miRNAs.
2. Materials and Methods
2.1. NGS Sequencing of MCF-7 and MDA-MB-231
Two breast cancer cell lines (MCF-7 and MDA-MB-231) grown under standard procedures were
obtained from colleagues at the Mitchell Cancer Institute (Mobile, AL, USA). RNA was isolated
and suspended in Trizol per standard manufacture protocol before being shipped to Otogenetics
(Otogenetics Corporation, Atlanta, GA, USA) for commercial next-generation sequencing on an
Illumina HiSeq2000 sequencer. Two RNA-Seq protocols were requested: (1) a total polyA selected
RNA-Seq to provide mRNA transcripts, and (2) a small size selected RNA-Seq to provide small RNAs
ranging from 17 to 35 nt in length. Raw paired-end reads were received totaling around 6 billion base
pairs per cell line. Reads were uploaded to the NCBI Sequence Read Archive (SRA) and assigned the
project number SRP101635.
2.2. Identification of A-to-G Edits in Breast Cancer Cells
Reads from the polyA selected RNA-Seq were filtered for low-quality reads and adapter
contamination using Trimmomatic [34] and then aligned to the GRCh38 human reference genome
using TopHat [35] (one mismatch allowed per alignment, only unique mappings reported). Edit sites
were identified using the ‘mpileup’ command of SAMtools [36] which generates a VCF file
containing location information for observed variations between the reads and the reference.
All identified variations other than A-to-G and T-to-C were removed, and the remaining locations were
cross-referenced with the dbSNP database to exclude variations that are known SNPs. To be considered
a probable edit, at least 10% of transcriptome reads were required to differ from the reference genome
at the edit position (with a minimum of 30 total reads).
132
by either repressing translation of the mRNA or triggering its degradation [21]. MiRNAs have been
found to play a role in numerous cellular processes, from cell cycle control and apoptosis regulation
to hormone production and immune response [22]. Importantly, misexpression of miRNAs has been
implicated in a number of different disease states ranging from cardiovascular [23] and neurological
disorders [24] to many types of cancer [25]. Having been associated with such a wide array of processes
and pathologies, these molecules have garnered increased attention recently as investigators begin to
evaluate their potential utility as biomarkers and therapies.
While these two mechanisms are fairly well understood independently, only recently have reports
profiled the functional connection between RNA editing and miRNAs. For example, it has been shown
that ADAR1 forms a complex with Dicer through direct protein interactions and enhances global
miRNA processing [26]. Further, ADAR deamination of pri-miRNA transcripts can cause alterations to
their structural conformations and subsequent maturation and processing by Drosha and Dicer [27,28].
In addition, while any editing of miRNA transcripts can have functional implications, arguably the
most critical changes are to the seed regions of the miRNAs as this can drastically alter the set of
genes able to be regulated [29]. This is especially true in cancer where altered miRNA regulation of
oncogenes and tumor suppressors can lead to tumor formation [17]. Importantly, it should be noted
that, in addition to editing the miRNA transcript, ADAR can also edit the 3 ′ UTR of target mRNAs.
This modification dramatically increases the interplay between miRNAs and their targets by allowing
a different set of miRNAs to regulate a given mRNA depending on if the transcript has been edited
or not. Unfortunately, while the effects of editing the miRNA transcripts themselves have been well
documented, this opposite effect of editing mRNAs in regions complementary to miRNA seeds is less
understood. However, a number of reports suggesting this plays a significantly underappreciated role
in miRNA targeting have surfaced within the last year [30–33]. To examine this, our study identified
edit sites within two breast cancer cell lines (MCF-7 and MDA-MB-231) and analyzed the effect these
edits had on subsequent regulation by miRNAs.
2. Materials and Methods
2.1. NGS Sequencing of MCF-7 and MDA-MB-231
Two breast cancer cell lines (MCF-7 and MDA-MB-231) grown under standard procedures were
obtained from colleagues at the Mitchell Cancer Institute (Mobile, AL, USA). RNA was isolated
and suspended in Trizol per standard manufacture protocol before being shipped to Otogenetics
(Otogenetics Corporation, Atlanta, GA, USA) for commercial next-generation sequencing on an
Illumina HiSeq2000 sequencer. Two RNA-Seq protocols were requested: (1) a total polyA selected
RNA-Seq to provide mRNA transcripts, and (2) a small size selected RNA-Seq to provide small RNAs
ranging from 17 to 35 nt in length. Raw paired-end reads were received totaling around 6 billion base
pairs per cell line. Reads were uploaded to the NCBI Sequence Read Archive (SRA) and assigned the
project number SRP101635.
2.2. Identification of A-to-G Edits in Breast Cancer Cells
Reads from the polyA selected RNA-Seq were filtered for low-quality reads and adapter
contamination using Trimmomatic [34] and then aligned to the GRCh38 human reference genome
using TopHat [35] (one mismatch allowed per alignment, only unique mappings reported). Edit sites
were identified using the ‘mpileup’ command of SAMtools [36] which generates a VCF file
containing location information for observed variations between the reads and the reference.
All identified variations other than A-to-G and T-to-C were removed, and the remaining locations were
cross-referenced with the dbSNP database to exclude variations that are known SNPs. To be considered
a probable edit, at least 10% of transcriptome reads were required to differ from the reference genome
at the edit position (with a minimum of 30 total reads).
132