Direct screening of plasma circulating microRNAs

ABSTRACT

Circulating microRNAs (miRNAs) are considered as reliable candidates for biomarker discovery. RNA-Sequencing has become the most suitable technique to accurately quantify the miRNAome. However, RNA-Sequencing relies on several technical passages before reaching the final-end.

HTG EdgeSeq technology, thanks to the abrogation of RNA extraction step, allows productivity enhancement by reducing the number of hands-on steps, the time for sample preparation and, therefore, the costs.

We found a strong correlation between qPCR and dPCR with HTG (Pearson’s coefficient of 0.93 and 0.94, respectively).

In conclusion, we showed that HTG EdgeSeq, performed on human plasma specimens without RNA extraction, is reliable, allows the simultaneous screening of more than 2,000 miRNAs, and thus, it could be applied to biomarker discovery in large cohorts.

Non-coding RNAs (ncRNA) are defined as transcripts that are not translated into functional proteins. In the last decades, several studies showed that ncRNAs have a pivotal role in the regulation of protein-coding gene transcription [1], in chromatin remodeling [2] and in several diseases [3]. In particular, a family of small ncRNAs called microRNAs (miRNAs) has been found deregulated during the development and progression of several diseases, such as cancer [4] and myocardial infarction [5]. Moreover, they have been found stably expressed in almost all human fluids, although the exact number of detectable circulating miRNAs have not been clearly established yet [6]. Hence, it has been hypothesized that miRNAs could be considered good candidates for biomarker discovery and therapeutic targets [7,8].

Recently, RNA-Sequencing (RNA-seq) has become the most suitable technique to quantify accurately the expression of transcriptome as well as miRNAome – the whole set of expressed miRNAs – overcoming previous techniques such as microarrays. In particular, RNA-Seq has a wider dynamic range than microarrays and enables the detection of all RNA species [9]. However, several preparation steps separate sample collection from the final results, thus leading to the introduction of potential technical biases in the data. In addition, some steps, such as RNA species size selection or RNA extraction, could influence the perceived levels of expression of specific miRNAs [10]. Finally, in a standard RNA-Seq workflow, the sequenced fragments (i.e. reads) must be aligned against a reference genome. Subsequently, the mapped reads are counted to estimate the expression levels. However, the choice of the reference genome, the tools for alignment, quantification, and normalization approach is crucial and dramatically affects the results. Hence, a high level of bioinformatic skills is required.

In this context, a relatively new approach, developed by HTG Molecular and called HTG EdgeSeq, allows to simplify the library and sample preparation for both RNA and miRNA targeted sequencing, by removing RNA extraction procedures [11]. In addition, HTG EdgeSeq takes advantage of probes to identify and directly quantify transcripts or miRNAs abundance, reducing consistently the bioinformatics burden. Overall, this technique allows: i) a consistent time reduction for sample preparation and for bioinformatics analysis, (results in as little as 36 hours with very little hands-on timing); ii) a simultaneous and quantitative detection of more than 2,000 human miRNA transcripts; iii) a miRNA profile identification with a low amount of samples (e.g. 15 μl for plasma); iv) a straight forward bioinformatic work-flow, leading to robust results; v) reduction of technical biases. On the other hand, specifically for whole miRNAs sequencing, the usage of probes compels the researcher to investigate a specific set of features, thus, not allowing the discovery of unknown miRNAs. Indeed, it could be difficult to validate all the results since the probe choice is led by literature evidence, thus, possibly leading to false positives. Interestingly, the HTG EdgeSeq costs are in line with the other next-generation sequencing (NGS) approaches.

In the present study, we assessed the reliability of HTG EdgeSeq approach, applied to miRNA sequencing, by comparing its performances with quantitative PCR (qPCR) and the more recent chip-based digital PCR (dPCR) [12]. The former is widely used to analyze miRNA expression and can evaluate several miRNAs at a time. The latter is the most accurate PCR technique, being highly tolerant to inhibitors and providing linear response to even the smallest change in the number of copies present in the samples. However, dPCR is implemented when a high accuracy of detection is needed (number of copies per μL) and allows the evaluation only few transcripts at a time.

HTG EdgeSeq was conducted on plasma specimens from healthy volunteers to evaluate the expression of 2,083 miRNAs. To test the suitability of the HTG technology, we performed several Pearson’s correlation analysis comparing HTG with three PCR techniques, the most common techniques used for miRNA expression evaluation, such as TaqMan Human miRNA Arrays (Card A), qPCR, and dPCR. The Pearson’s correlation coefficient (R) and the R² coefficient were adjusted for the number of observations, and p-values were computed.

Considering all the miRNAs that passed the quality thresholds in the Card A (n = 52), we showed that there is a strong correlation between HTG and Card A (r = 0.89, p < 0.0001; Figure 1(a) and Supplemental Figure S1). Of notice, only miRNAs with c_T < 28 (n = 28) seemed to correlate well with HTG (r = 0.89, p < 0.0001; Figure 1(b)), while, we did not observe a relevant relationship between HTG and Card A expression values for low expressed (28 < c_T < 38; n = 24) miRNAs (r = 0.25, p = 0.01; Figure 1(c)).

Figure 1.

Correlation between HTG EdgeSeq and Card A. (a) Scatterplot of miRNA expression values (HTG vs card A): each dot denotes a specific miRNA while the red lines represent the standard deviation (4 samples) for ‘card A’ (vertically) and ‘HTG’ (horizontally) values. The regression line and its 95% confidence interval are depicted in black and green, respectively. (b) Scatterplot between miRNAs with c_T values lower than 28. (c) Scatterplot between miRNAs with c_T values greater than 28 and lower than 38.

Concerning qPCR and dPCR, we selected and tested several candidate miRNAs (n = 13) ranging from very low (miR-1180-3p) to very high (miR-223-3p) abundance in HTG screening. A strong correlation was found between both qPCR (miRNAs n = 9) and dPCR (miRNAs n = 13) with HTG (Figure 2). The missing miRNAs, in the qPCR approach, resulted not detectable and the corresponding c_T values were ‘undetermined’, due to the differences in reaction chemistries.

Figure 2.

Correlation between HTG EdgeSeq, qPCR, and dPCR. (a) Scatterplot of miRNA expression values (HTG vs qPCR): each dot denotes a specific miRNA while the red lines represent the standard deviation (4 samples) for ‘qPCR’ (vertically) and ‘HTG’ (horizontally) values. The regression line and its 95% confidence interval are depicted in black and green, respectively. (b) Scatterplot between miRNA expression values (HTG vs digital PCR).

The Pearson’s coefficient between HTG and qPCR for nine miRNAs was 0.93 (p < 0.0001; Figure 2(a) and Supplemental Figure S2), while between HTG and dPCR (considering thirteen miRNAs) was 0.94 (p < 0.0001; Figure 2(b) and Supplemental Figure S3).

Finally, we also compared all technologies among them in order to assess the consistency of the obtained data. As shown in Figure 3, Supplemental Figure S4, and Supplemental Figure S5, all technologies showed a high correlation among them, ranging from 0.77 (qPCR vs Card A) to 0.96 (qPCR vs dPCR).

Figure 3.

Correlation matrix of the Pearson’s coefficients among the different used technologies. For each comparison, the correlation is highlighted by the Pearson’s coefficient and by a color scale: the darker the color, the higher the correlation. The cool colors represent direct correlations while warm colors represent inverse correlations.

In conclusion, we have shown for the first time that miRNA screening (over 2,000 transcripts) by HTG EdgeSeq technology, performed on human plasma specimens without RNA extraction, is consistent and reliable. Indeed, validations performed with different PCR technologies and different chemistries [12] (i.e. Card A, qPCR, and dPCR) showed comparable results to HTG EdgeSeq. Thus, this new approach could speed up the discovery process of circulating miRNAs in a several diseases.

Methods

Human specimens

The Institutional Review Board and the Ethical Committee of Centro Cardiologico Monzino IRCCS (university hospital) approved this study. The investigation conformed to the principles outlined in the Declaration of Helsinki (1964).

Peripheral blood samples were drawn from four (n = 4) healthy subjects into tubes containing EDTA (9.3 mM; Vacutainer Systems, Becton Dickinson, Franklin Lakes, NJ, USA) kept on ice and centrifuged at 3,000 g for 10 minutes at 4°C, within 30 minutes after being drawn. Plasma was separated and aliquots were stored at −80°C until analyses were performed.

HTG-sequencing

The HTG EdgeSeq system consists in a simplified targeted NGS, automating both the capture of custom targets (in this case miRNAs) and library preparations. The miRNA Whole Transcriptome Assay, performed by HTG Molecular Diagnostic, Inc. (Tucson, AZ, USA), allowed to evaluate the expression (raw read counts) of 2,083 miRNAs directly from 15 μL of plasma from each healthy subject. Sample lysis was achieved by adding 3.0 μL of Proteinase K and 12.0 μL of lysis buffer to each plasma sample (15 μL) and then incubated at 50°C for 180 minutes. Three human Brain RNA samples were used as a process control. Target capture and RNAse-protection assay were run on an HTG EdgeSeq Processor. Then, samples were individually barcoded (using a 16-cycle PCR reaction to add adapters and molecular barcodes), individually purified using AMPure XP beads, and quantified using a KAPA Library Quantification kit. The library was sequenced on an Illumina NextSeq using a High Output, 75 cycle, v2 kit with two index reads. The expression value of each miRNA is expressed as the log-counts per million (log₂CPM).

TaqMan Human miRNA card A Arrays

The TaqMan Human microRNA Card A Arrays version 2.0 (Thermo Fisher) were used for evaluating the expression of 377 miRNAs. Two specific pools of primers (Megaplex pools) were used for reverse transcription (RT) and pre-amplification steps, accordingly to the manufacturer’s protocol on 7900HT Real-Time PCR System (Thermo Fisher). In particular, RNA was extracted from plasma samples using the Total RNA Purification Plus Kit (Norgen Biotek Corp.) and it was retro-transcribed as follows: 16°C for 2 minutes, 42°C for 1 minute, and 50°C for 1 second (40 cycles), followed by an incubation at 85°C for 5 minutes. The pre-amplification was performed by a serial incubation at 95°C for 10 minutes, 55°C for 2 minutes, 72°C for 2 minutes, 12 cycles at 95°C for 15 seconds and 60°C for 4 minutes, followed by an incubation at 99.9°C for 10 minutes. Finally, the samples were diluted 1:4 using 0.1X TE pH 8.0 and then they were loaded into the run plate and incubated at 50°C for 2 minutes, 94.5°C for 10 minutes, 40 cycles at 97°C for 30 seconds and 59.7°C for 1 minute.

Analyses were performed using the Expression Suite software v1.0.3 (Thermo Fisher). MiRNA that did not pass the quality control were filtered out (Supplemental Table S1). The excluded miRNAs had one or more of the following issues, as indicated by the software: low signal in linear phase; bad passive reference signal (ROX); c_T (cycle threshold) algorithm failed or low confidence; exponential algorithm failed; and thresholding algorithm failed.

The expression value of each miRNA is expressed as – c_T and miRNA with c_T ≥ 38 are considered as not expressed.

Reverse transcription

For qPCR and dPCR, extraction of RNA was performed from plasma samples using the Total RNA Purification Plus Kit (Norgen Biotek Corp.). Two μL of RNA were used for the two-steps PCR amplification with Advanced TaqMan Reverse Transcription Reagent kit (Thermo Fisher) according to manufacturer’s protocol. In particular, we followed the described steps without modifications. 1) Poly(a) tailing addition: polyadenylation at 37°C for 45 minutes followed by an incubation at 65°C to stop the reaction. 2) 5ʹ-end Adaptor ligation: 16°C for 1 hour. 3) Reverse transcription (RT): 42°C for 15 minutes, using a universal RT primer included in the kit, followed by an incubation at 85°C for 5 minutes to stop the reaction. 4) miRNAs universal pre-amplification (using proprietary primers included in the kit): enzyme activation at 95°C for 5 minutes, denaturation at 95°C for 3 seconds and annealing/extension at 60°C for 30 seconds (14 cycles), and, finally, an incubation at 99°C for 10 minutes to stop the reaction. cDNA samples were stored at −80°C until PCR further analyses.

Real-Time PCR and chip-based digital PCR

Real Time PCR (qPCR) was performed on ABI Prism 7900 HT (Thermo Fisher), according to the manufacturer’s protocol. In particular, enzyme activation was performed at 95°C for 10 minutes, denaturation at 95°C for 30 seconds followed by annealing/extension at 59°C for 1 minute (40 cycles). The analyses were performed using the software SDS v2.4 (Thermo Fisher).

Chip-based digital PCR (dPCR) was performed on a QuantStudio 3D Digital PCR System platform composed by the QuantStudio 3D Instrument, the Dual Flat Block GeneAmp PCR System 9700 and the QuantStudio 3D Digital PCR Chip Loader (Thermo Fisher). dPCR was performed according to the manufacturer’s instructions. In particular, enzyme activation was performed at 96°C for 10 minutes, denaturation at 98°C for 30 seconds followed by annealing/extension at 56°C for 2 minute (40 cycles), then final extension at 60°C for 2 minutes. Analysis were executed with online version of the QuantStudio 3D AnalysisSuite (Thermo Fisher Cloud).

Primers purchased from Thermo Fisher were labelled with FAM dyes and used to evaluate the expression of candidate miRNAs (Supplemental Table S1).

Statistical analysis

We compared each gene expression techniques by a correlation analysis: the Pearson’s correlation coefficient (R), the R² coefficient, and the p-value were computed. We performed this kind of analysis both on the averaged expression values across all samples and for each sample independently. Each value has been previously ‘mean-centered’.

Only for the comparison ‘HTGseq versus TaqMan Human microRNA Card A Arrays’, we have also performed two additional correlation analysis, splitting genes with a C_t value greater or lower than 28 cycles. This allow evaluating the relationship between techniques for low and high expressed genes, separately.

Funding Statement

This work was supported by the Fondazione Gigi e Pupa Ferrari [FPF-14]; Ministero della Salute [RC2016-BIO34-2627243].

Acknowledgments

This work was supported by Gigi e Pupa Ferrari ONLUS and the Italian Ministry of Health [RC2016-BIO34-2627243].

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental data

Supplemental data for this article can be accessed here.