Site-specific analysis of ribosomal 2′O-methylation by quantitative reverse transcription PCR under low deoxynucleotide triphosphate concentrations

2′O-methylation (Nm) of RNA (Figure 1) is a common modification in different RNA species abundantly found at multiple sites in rRNA and tRNA, some mRNAs and at the 3′-end of miRNAs and piRNAs [1]. Nm causes a substantial change in the biophysical and biochemical properties of RNA, including structural stabilization and resistance to enzymatic cleavage [2]. Recent works show that changes in the methylation status at individual Nm sites can affect ribosomal function and fidelity. Furthermore, methylation levels at specific ribosomal Nm sites demonstrate high heterogeneity in cancer tissues, which is associated with disease progression or outcomes [3–5]. These findings suggest new possibilities for the utilization of ribosomal Nm marks as biomarkers or therapeutic targets in disease and warrant the detailed profiling of individual sites in large patient cohorts and various biological model systems. A prerequisite for the successful exploitation of the biomarker and therapeutic potential of ribosomal Nm is the availability of quick, accurate and affordable detection methodologies that can be performed under standard laboratory conditions.

Several methods for Nm detection and quantification have been developed, for example, HPLC coupled with tandem mass spectrometry, primer extension preceded by endonuclease digestion and reverse transcription (RT) and, more recently, next-generation sequencing (NGS)-based methods, such as RiboMeth-Seq, 2′O-Me-Seq and Nm-seq [6,7]. However, these techniques are laborious and time-consuming, require specialized equipment or could require as much as 1000 ng of total RNA. This limits their use in the profiling of large numbers of tissue samples, for example, from biopsies, where often only a limited amount of RNA is available, and hinders routine laboratory analysis of Nm in experimental models.

RT under low deoxynucleotide triphosphate (dNTP) concentration followed by PCR (RTLP) is a methodology for Nm detection that relies on the quantitation of PCR products by agarose gel electrophoresis visualization or by quantitative real-time PCR (qPCR) and can avoid the previously mentioned limitations. qPCR analysis of rRNA Nm marks was first described by Belin and colleagues [8]. Later, Dong and colleagues modified the technique to reduce technical bias and also allow the detection of 3′-modified residues by semiquantitative PCR [9]. RTLP relies on the differential behavior of reverse transcriptase when the concentration of dNTP in the reaction mix is limiting for the enzymatic reaction. Under dNTP concentrations approximately 500-times lower than routinely used, the RT enzyme will pause and eventually abort cDNA synthesis when it encounters an Nm site in the RNA template, resulting in shorter RT products compared with those generated under standard dNTP concentrations. Differences in the amount of produced cDNA then can be detected by PCR and agarose gel electrophoresis analysis or by qPCR. Briefly, for each RNA sample, four RT reactions are performed: two under low and another two under standard (high) dNTP concentrations. For each of the RT reactions under a given concentration, a different RT primer is used: an anchored primer (RT-A) with a 3′-terminal residue complementary to the investigated Nm site in the RNA template, and an unanchored primer (RT-U) that is shifted 1 base downstream of the RT-A primer so that its 3′-terminal residue is complementary to the nucleotide at position Nm+1 (Figure 2). When cDNA synthesis is performed under low dNTP concentrations, the amount of shorter (aborted) first-strand cDNA products is proportional to the amount of RNA template with Nm at the tested site. In a subsequent PCR step, this amount can be quantified using the cDNA products as a template and a primer pair amplifying a short region upstream of the Nm-modified site. The obtained PCR product can then be visualized by agarose-gel-electrophoresis and image analysis or by qPCR. To account for RT primer bias, the reactions with RT-A and RT-U are also performed under high dNTP concentrations.

The use of RTLP for Nm quantitation has been described in several papers [10–13]. Nevertheless, its application for the detection of Nm in clinical samples has not been reported. The protocol presented here describes quantitative (q) RTLP, a qPCR protocol for quantitative analysis of ribosomal Nm optimized for use with low amounts of RNA from clinical samples, which can be performed with as little as 80 ng of total RNA. This protocol is an adaptation and further optimization of Dong et al. [9] and provides a fast, easy, reproducible and accurate method that preserves precious RNA material from tissue samples encompassed in research studies and uses standard reagents and equipment available in most laboratories.

Materials & methods

Template sequences & primer design

rRNA template sequences

Template rRNA sequences for the human 18S and 28S ribosomal subunits can be retrieved from the National Center for Biotechnology Information (NCBI) using accession numbers NR_146119.1 and NR_146118.1, respectively. Nm positions according to the old rRNA nomenclature can be extracted from the snoRNA-LBME-db database (www-snorna.biotoul.fr). Tables converting 28S-Nm positions from old to new nomenclature are provided in several publications [14,15].

RT primer design

Since the RT primers bind to specific RNA sequences, specific assessment and a new set of site-specific primers must be designed for each investigated Nm site as follows: design the anchored RT primer (RT-A) so that the 3′-terminal base anneals at the predicted Nm site and design the unanchored RT primer (RT-U) so that the 3′-terminal base anneals one position downstream of the foreseen methylated site (position Nm+1). Optimal design parameters for the RT primers are length between 20–25 nt, melting temperature between 60–65°C and a GC percentage of 50–60%. Positive (fully methylated 18S-A1031 site) and negative (nonmethylated 28S-U1991 site) control primers (Table 1) have been previously described and work well with the current protocol [16].

qPCR primer design

To ensure accurate quantification, the qPCR amplicon should be positioned in near proximity of the analyzed Nm residue so that the reverse primer (R) is located 30–50 bp upstream of the targeted Nm site. The optimal amplicon size is in the range of 150–200 bp. The forward and reverse primer of each pair should have the same or similar melting temperature (Tm) to ensure successful qPCR reactions. The recommended Tm of PCR primers is usually in the range of 55°C to 65°C and within 5°C of each other. qPCR primers can be designed and analyzed for specificity and secondary structures using Primer3 (http://primer3.ut.ee/) [17]. The concentration of storage stocks of all primers (Table 2) is 100 μM and the concentration of working solutions of each primer is 10 μM.

RNA from cell lines can be used to optimize PCR parameters and test primer specificity. Primer specificity and reaction condition optimization can be done by monitoring qRTLP products with a melting curve analysis and by agarose gel electrophoresis.

Sample preparation & RNA isolation

Before starting, ensure that all consumables, tools and work surfaces have been decontaminated with RNase-eliminating reagents. Total RNA is isolated using RNABee reagent or equivalent (Campro Scientific, GmbH, Berlin, Germany) according to the manufacturer's protocols. Good-quality RNA is preferred for the high performance of the assay. Ideally, RNA samples should have an A260/A280 OD ratio equal to or above 2.0, indicative of high purity. RNA integrity assessed by the RIN number at the 2100 Bioanalyzer or equivalent should also be confirmed.

RT under high- & low-dNTP conditions

For the detection of ribosomal Nm sites in human cells and tissue samples, qRTLP is conducted as follows:

• Thaw RNA sample, RT primers, dNTPs (10 mM), dNTPs (20 μM) and M-MLV 5X reaction buffer on ice. Keep the enzyme M-MLV-reverse transcriptase 200 U/μl (Promega, WI; M1701) in a -20°C freezer or enzyme cooler until ready to use.

• For each RNA sample, label two PCR tubes: RT-A and RT-U.

• Add 40 ng RNA template into each tube plus 1 μM RT primers to the corresponding tubes and adjust the volume to 20 μl with RNase-free water.

• Incubate at 70°C for 10 min to anneal the primers and then place on ice.

• Divide each RT reaction into two different tubes (10 μl for each), one for high dNTP and another for low dNTP concentration, so that in the end there are four RT tubes per tested sample.

• Prepare high and low RT reaction mixes as outlined in Table 3. Vortex, and spin down briefly each mix prior to use.

• Add 10 μl of the appropriate RT reaction mix into each tube, RT-A and RT-U, pipetting up and down to mix.

• Place RT reactions in a thermal cycler and carry the RT step at 42°C for 1 h (lid temperature: 45°C). Terminate RT reactions by heating at 70°C for 15 min. Hold at 4°C.

• Proceed to qPCR or freeze at -20°C until ready to use.

Quantitative real-time PCR

The high- and low-RT products are analyzed in a subsequent amplification reaction by real-time SYBR Green-based qPCR using a single primer set. Both 10-μl and 5-μl reaction volumes in 96-well or 384-well plate formats, respectively, work equally well, without loss of reaction efficiency or sensitivity. However, the use of high-sensitivity PCR plates, e.g. AmpliStar-II 96 PCR plate, (cat. nr.: WB1900WS, Westburg, Leusden, The Netherlands), MicroAmpTM 384-well reaction plate, (cat. no. 4343370, Thermo Fisher Scientific, Massachusetts, USA) or equivalent is recommended. In the current authors' lab, qPCR reactions were performed with either the 7500HT cycler (Applied Biosystems, Massachusetts, USA or the CFX384 Real-Time PCR System (Bio-Rad, California, USA). All qPCR reactions are performed in triplicate as defined in Table 4 using the SensiMix™ SYBR^® Low-ROX kit (cat. no. QT625-05, Bioline, Meridian Bioscience, Tennessee, USA) and specific qPCR primers (designed as described). Adjust the ROX concentration as recommended for the thermocycler in use if the qPCR machine is different than the one specified.

After all samples are loaded onto the qPCR plate, seal them with optical film. Centrifuge the plates at 5000 g for 1 min at room temperature and ensure that no bubbles are present in sample wells. Load the qPCR plate into the thermocycler and start the following cycle program:

Step 1: 95°C for 10 min (1x);

Step 2: 95°C for 15 s;

Step 3: 55°C for 30 s;

Step 4: 72°C for 15 s, return to step 2 (40x);

Step 5: dissociation curve (optional).

After the assay is complete, collect Ct values for quantification and analysis.

Quantification of Nm fraction

Calculate the average Ct from technical triplicates for each sample, condition and target site. The same qPCR detection threshold must be used for each condition and sample of the same target site. To calculate the relative quantity of PCR product in the low dNTP sample (RQ_Low), subtract the average Ct of the RT-U primer reaction from the average Ct of the RT-A primer reaction at low dNTP concentration. Formula:

Proceed the same way to calculate the relative quantity of PCR product in the high dNTP sample (control) but using the average Ct values from the high dNTP concentration. Formula:

The high-concentration dNTP reactions serve as controls for the amount of input RNA in the RT step, as it is expected that under high dNTP conditions there is no difference in the amount of amplified product synthesized from cDNA generated with RT-A or RT-U primers. Nevertheless, for some residues, small differences caused by the 1 nt primer shift can be observed. The control high-concentration dNTP reaction is also important to assess whether reduced RT efficiency under low-concentration dNTP is due to the presence of Nm or some other determinant (e.g., complex secondary structure) [18]. Calculate the efficiency of the qRTLP reaction (E_qRTLP) as the ratio between the experimental delta Ct at low dNTP concentration and the control delta Ct at high dNTP concentration.

The Nm fraction is obtained using the formula:

Results & discussion

qRTLP is a robust and sensitive method for site-specific Nm analysis of human rRNA using low input amounts of total RNA. The method has been successfully applied to detect previously identified Nm sites in yeast rRNAs as well as mouse piRNAs [19]. More recently, RTLP has been used to test rRNA from human cell lines [10,19]. However, to the best of our knowledge, the method has not been applied for the relative measurement of Nm of rRNA from human tissue specimens. Here we describe an adjustment and further optimization of a previously existing molecular method suitable for clinical samples.

Traditionally, only a limited amount of tissue and RNA is available from patients' samples (e.g., biopsies, archive collections). To investigate the amount of total RNA that can be used for the detection of Nm at specific positions in rRNA without affecting the sensitivity of the method, we tested RTLP with different amounts of total RNA extracted from human cell lines. For this, RTLP was performed with 200 ng, 20 ng and 5 ng of total RNA for each one of the four RT reactions (RT-A and RT-U primers at low and high dNTP conditions), and the PCR products were then analyzed by agarose gel electrophoresis. The experiments done with 5 ng of total RNA did not produce good results, as the resolution at low dNTP condition was lost. Yet, the experiments done with either 200 ng and 20 ng of total RNA gave similar results as demonstrated in Figure 3. As a positive control, we tested the rRNA methylation fraction at nucleotide position 18S-A1031 in human rRNA from normal cells, which is known to be fully methylated [15,16]. As expected, there was less RT-PCR product generated from the RT-U primer compared with the RT-A primer under low dNTP concentration, while no difference between RT-A and RT-U products was observed at high dNTP concentration. In a negative control experiment, we analyzed the presumed Nm fraction of the nucleotide at position 28S-U1991. It is known that this nucleotide is not 2′O-methylated in human ribosomes, therefore, no visible differences in the amount of RT-A and RT-U RT products were expected or observed regardless of the concentration of dNTP. Furthermore, 20 ng of total RNA per RT reaction was sufficient to clearly detect the 2′-O-methylated site in human 18S rRNA. This result demonstrated that site-specific RTLP for the detection of Nm in human rRNA can be successfully performed with as little as 80 ng of total RNA, which makes the technique suitable for the analysis of tissue material such as patient samples in medical research.

To precisely quantify the difference in amplified cDNA products for the positive and negative control residues, 18S-A1031 and 28S-U1991, respectively, we used standard 40-cycle SYBR Green qPCR (Figure 4A). For 18S-A1031, under low dNTP concentration, RT-U amplification products reached the threshold intensity 4 cycles after RT-A products, indicating at least 16-times less cDNA product generated in the RT-U condition as a result of abortive RT reaction caused by Nm at this site. Based on the obtained Ct values, the methylation fraction for this position was calculated at 95.3% (Table 5), which is in agreement with previously reported methylation levels for this residue in human tissues and cell lines [14]. Inversely, for 28S-U1991 the amplification curves of RT-A and RT-U products at low dNTP are overlapping because no methylation is present at this position in human ribosomal RNA. Subsequently, the calculated methylation fraction for this position is residual (3.8%; Table 5).

For 28S-G4593, we found that the RT-U primer yielded significantly less cDNA than the RT-A primer at low dNTP concentrations as demonstrated by the differential band intensity in PCR products (Figure 3) and by the 3-cycle difference in the amplification curve (Figure 4B, left panel). These results indicated that 28S-G4593 is indeed 2′O-methylated in a substantial fraction of all ribosomes in the cell. Remarkably, loss of Nm at 28S-G4593 due to homozygous snoRNA knockout by CRISPR/Cas9 in the same cell line led to increased read-through during RT, increasing the RT-U qPCR product, which became similar to the RT-A qPCR product (Figure 4B, right panel).

Next, we tested the applicability of qRTLP to assess Nm stoichiometry at specific rRNA position (28S-Gm4593) in fresh-frozen tissues (Figure 5). We obtained good amplification curves with 2-cycle difference between RA and RU for 28S-Gm4593, reconfirming the presence of variable, partial Nm at this ribosomal position in tissues from different organs. These results demonstrate that the described qRTLP protocol allows the detection of differences in ribomethylation levels at specific Nm sites of clinical specimens and could be used to infer heterogeneous tissue-associated ribomethylation profiles. Distinct fractional scores at specific sites may reflect the partitioning of rRNA into different pathways during folding and are also indicative that ribosome modification signatures could vary between cells in a population. Currently, it is unclear to what extent this heterogeneity has functional and/or physiological implications, but it is consistent with previous observations [20]. Furthermore, it is well known that human rRNA methylation site measurement can be influenced by 3D RNA structure, other base modification sites surrounding the methylation site and methylation sites in close proximity. This might be something to consider when using qRTLP for regions comprising clusters of methylated rRNA sites. For those cases, there are several methyl residues that may halt RT activity and, consequently, site-specific ribomethylation quantification will be hindered. The analysis of highly methylated and structured rRNA regions should be carefully conducted as the calculated methylated fraction can be misinterpreted and does not necessarily approach a zero value even in the absence of Nm.

Currently, there are other methodologies dedicated to mapping and quantification of Nm residues in rRNA. For instance, the RiboMeth-Seq analysis pipeline is used as well in tissue samples requiring as little as 5 ng of total RNA [15]. Nevertheless, this comprises a very expensive technique and not all laboratories are equipped with a sequencing facility, while the qPCR machine is a basic instrument available at every routinely based molecular biology lab. Compared with previous techniques, qRTLP is a faster and more precise method for site-specific Nm detection and quantification that eliminates the use of radiolabeled primers or chemical treatments. The optimized technique presented here has additional advantages, for example, avoiding the use of time-consuming agarose gel electrophoresis, which also requires PCR cycle number optimization and only allows for semiquantification. Additionally, qRTLP is time- and cost-effective, as it allows the simultaneous processing of several samples in as little as 5 μl in routinely used qPCR machines spending approximately 50 euros per reaction. To assess one methylated rRNA position in one sample we spend four reactions, meaning paying around 2 euros per sample. RTLP methodology is not a high-throughput technique but is rather used to interrogate a limited set of target methylated sites. Hence, the effective cost to measure up to 10 methylated rRNA positions is around 20 euros per clinical sample, which is much more reasonably priced compared with other alternatives such as sequencing technologies.

Finally, although the methodology presented here was optimized in fresh-frozen tissues, we believe the protocol is also suitable for paraffin-embedded tissues since improved RNA quality does not increase qPCR sensitivity to a large extent. In the near future, the applicability of the optimized qRTLP should also be tested on liquid biopsies, such as urine and blood, which will enhance the usefulness of the method for clinical applications and consequent disease management.

Conclusion

In conclusion, qRTLP provides a simple, sensitive and cost-efficient approach for Nm detection and quantification to investigate and characterize individual Nm positions in rRNA and makes possible the monitoring of specific Nm methylation marks in clinical specimens of limited availability and in large patient cohorts.

Future perspective

Here, we report a specific methodology to assess Nm in rRNA, the most abundant type of RNA in eukaryotic cells. Nevertheless, there are no significant methodological obstacles that prevent the adaptation of the method for the detection of Nm in other RNA species. In the near future, qRTLP can be further optimized for use with other RNA species, such as messenger and long noncoding RNAs and small RNAs. Semiquantitative methodology for the detection of 3′-terminal Nm in small RNAs has already been described [9]. Since the RT primers employed in qRTLP are target-specific, Nm assessment is always possible as long as the probable site modification is defined a priori. Nevertheless, the lower abundance of other RNA species and in particular coding and noncoding transcripts must be considered and probably will require target enrichment techniques prior to qRTLP. Higher amounts of total RNA material, rRNA depletion or mRNA and lncRNA enrichment would be logical approaches to overcome low target abundance. Further empirical optimization of primer and template concentrations and titration of appropriate low dNTP amounts are likely to be necessary. Another possible development that can enhance the sensitivity of the assay is the translation of qRTLP to digital PCR platforms and the development of a dRTLP methodology. Finally, the rapid development of long-read direct RNA sequencing technologies and the corresponding computational algorithms provide a novel platform for the detection of RNA modifications [21] including Nm and undoubtedly will become available for routine analysis in the long run.