NEW YORK (GenomeWeb) – Investigators at Germany's Saarland University have published a study describing a new technique to sequence short RNAs from low-input sources — in this case Mitra home sample-collection devices from California-based Neoteryx.
In a report in Analytical Chemistry last week, the Saarland team showed that the method could provide efficient isolation and readout of not only microRNAs, but several other small RNA types. They also showed that their NGS results were reproducible and comparably sensitive to microarray-based detection.
Study co-author Andreas Keller, chair of clinical bioinformatics at the university, has dedicated significant research to developing technology and methods for studying small non-coding RNA molecules, which include the ubiquitously studied microRNA. As part of that work, he and his colleagues have previously published comparisons of different sequencing and array-based approaches for small RNA profiling.
"We want to reach a holistic understanding of human small-non-coding RNAs," Keller said, describing the team's work. This includes questions about which cells and which diseases involve the activity of miRNAs, and how specific molecules regulate genes and pathways in these cells.
As the group has moved forward with research on these topics, investigators have created a variety of web-based software tools (miRCarta, miRMaster, miEAA, and the miRNA Tissue Atlas) to help other researchers use or build upon their own results. Keller also founded a company in Heidelberg two years ago, called Hummingbird Diagnostics, which is focused on developing blood-based biomarkers for things like early cancer detection, neurodegenerative disorders, and autoimmune disease.
Keller stressed that the group's latest study represents early research and shouldn't be taken to represent any defined commercial plans on the company's part. That said, Hummingbird did participate in the study, performing RNA extraction from the home sampling devices., In addition, Keller said, the firm is exploring home sampling as part of its development of various strategies for screening, disease prevention, and monitoring.
Apart from demonstrating any future clinical potential, Keller said that the main goal of the effort was to nail down whether they could get RNA sequencing reads from this type of remote sampling device that correspond well to what would be possible in normal blood samples.
Analysis of Mitra-isolated RNA had previously been limited mainly to microarray studies, due to the limitations of input material. But sequencing in these samples would be beneficial for several reasons. Because it does not require a priori knowledge of specific sequence targets like microarrays do, sequencing allows for a discrete differentiation of isomiRs — sequences that vary with respect to the reference miRNA sequence. It also opens up the opportunity to analyze not only miRNAs but the total population of small RNA.
Crucial for making NGS possible in these limited dried samples, was the application of a ligation-free library preparation using a method called template switching, which combines 3’-end poly-(A) tailing and subsequent template switch-based cDNA generation.
"Library preparation procedures for NGS analysis of small RNAs are mostly built on … procedures [which] rely on the ligation of adapters causing biochemical biases for 5'-monophosphorylated sRNAs and for individual nucleotides. They are moreover heavily limiting in terms of input material," the team wrote.
According to co-author Martin Simon, also a professor at Saarland University, the approach is both much more sensitive and also less biasing against individual RNA species.
In the study, Simon, Keller, and colleagues obtained Mitra cartridge samples from four consented individuals, and isolated RNA from dried, preserved samples using modified protocols to enrich for sRNA.
Using 500 picograms of the total Mitra RNA isolates, the team performed library preparation using the described ligation-free template switching method, and then PCR amplification.
"After sequencing and trimming of polyadenylation, the read length distribution [showed] the majority of reads longer than 37 [nucleotides] indicating longer RNA fragments to be abundant in the library," the team wrote. "The mapping efficiency to the human genome assembly of 70 [percent] also indicates a good library quality and the absence of large contaminations from sampling," they added.
The authors reported that the largest proportion of the reads they produced corresponded to annotated miRNAs, but they were also able to detect piwi interacting RNAs (piRNAs as well as small nucleolar RNA (snoRNA), small nuclear RNA (snRNA, transfer RNA (tRNA), ribosomal RNA (rRNA) and long intergenic noncoding (lincRNA).
The method did not appear to offer a way to comprehensively detect the full range of isomiRs, but the team reported that dissection of these variable sequences could be improved using size selection after library preparation to limit to the miRNA fraction.
"I think the greatest advantage of NGS over microarrays is that we are not limited to miRNAs but we can analyze the total small RNA world including piRNAs, lncRNAs, tRNA-fragments," Simon wrote. And while it's too early to know whether other members from this larger spectrum of short RNA and isoforms might be useful as biomarkers, "their analysis offers great possibilities for alternative diagnostics in the future," he added.
Finally, the team evaluated the same RNA samples in parallel using sequencing and microarrays and compared both analyses. Data indicated that the NGS method could detect approximately 20 percent more miRNAs than arrays.
However, the top 10 detected miRNAs differed between the methods with only four miRNAs (miR486-5p, miR451-5p, let7b-5p, let7a-5p) shared between the two methods.
"Microarrays and NGS have advantages and disadvantages. On microarrays you have a set of miRNAs that you measure while in NGS you theoretically sequence the overall repertoire of all small non-coding RNAs. You can also detect single nucleotide variants in NGS and so-called isoforms," Keller explained.
As Keller and colleagues have found in other studies comparing NGS and arrays, a more in-depth comparison in this study showed that the array results were much more stable from sample to sample. But interestingly, one of the four human samples looked to harbor a population of small RNAs that was substantially different than the others based on the sequencing results, though this would not have been apparent via the microarray data.
Keller said that the group hopes to further validate the approach using a specific disease model that might demonstrate clinical potential, though he did not specify what that disease would be.
"For the time being, this study [is a more basic effort] demonstrating that we have the capabilities to sequence from home sampling devices that have been taken by patients and can be sent by post to laboratories," he wrote.
"Whether there is indeed a clinical relevance for such applications has to be explored," he added. But "potentially, such tests could work in rural areas where diagnosis is required but no high coverage of specialized doctors exists [or] … for direct to consumer tests." For miRNA tests focused on known markers, NGS would mostly likely be less applicable than PCR.
But Simon added that because isolating RNA with good integrity is more difficult than DNA, future studies that use sequencing to discover or validate RNA biomarkers may also require techniques like this, which make it possible to perform remote, large cohort sampling.