Absolute and relative quantification of RNA modifications via biosynthetic isotopomers

Abstract

In the resurging field of RNA modifications, quantification is a bottleneck blocking many exciting avenues. With currently over 150 known nucleoside alterations, detection and quantification methods must encompass multiple modifications for a comprehensive profile. LC-MS/MS approaches offer a perspective for comprehensive parallel quantification of all the various modifications found in total RNA of a given organism. By feeding (13)C-glucose as sole carbon source, we have generated a stable isotope-labeled internal standard (SIL-IS) for bacterial RNA, which facilitates relative comparison of all modifications. While conventional SIL-IS approaches require the chemical synthesis of single modifications in weighable quantities, this SIL-IS consists of a nucleoside mixture covering all detectable RNA modifications of Escherichia coli, yet in small and initially unknown quantities. For absolute in addition to relative quantification, those quantities were determined by a combination of external calibration and sample spiking of the biosynthetic SIL-IS. For each nucleoside, we thus obtained a very robust relative response factor, which permits direct conversion of the MS signal to absolute amounts of substance. The application of the validated SIL-IS allowed highly precise quantification with standard deviations<2% during a 12-week period, and a linear dynamic range that was extended by two orders of magnitude.

Figure 1.

Correlation of absolute analyte amount and signal intensity using UV and MS/MS measurements. As an example, an oligomer with equimolar amounts of each depicted nucleoside is digested and separated by column chromatography. The chromatograms above (UV) and below (MS/MS) represent the effect of the response factors on the signal intensities. The UV signal is essentially dependent on only the absorption coefficient ϵ, which is mildly dependent on solvent composition and pH. The MS/MS signal is subject to significant and varying influence by several physicochemical properties, sample parameters and instrumental parameters that cannot be assessed in one generally applicable parameter (legend: V, volt; R_t, retention time; A, ampere; eV, electron volt; Q1(2), quadrupol1(2) and CC/collision cell).

Figure 2.

Determination of response factors by usage of commercial Am and ¹³C Am from SIL-IS. (A) MS spectrum of unlabeled 2′-O-methyladenosine (Am) (left) and ¹³C-labeled Am (right). (B) Upper left: calibration measurements of commercially available, unlabeled Am. At high substance amounts, a flattening of the calibration curve due to saturation effects is highlighted in gray. Below, the signal intensity for constant amounts of ¹³C-labeled Am in the presence of increasing unlabeled Am amounts is shown. Here, the drop in signal intensity due to saturation can also be observed in the gray area. Right: By division of corresponding MS signals of unlabeled and ¹³C labeled Am, the NIF is received and a dynamic calibration curve can be plotted. The slope of the linear fit represents the relative response factor for Am = rRFN (Am).

Figure 3.

Determination of TruB turnover completeness. Saccharomyces cerevisiae tRNA^Phe transcript was incubated with TruB, digested and SIL-IS added to determine the absolute amount of pseudouridine formed. For analysis, the MS/MS traces were used to calculate the amount of injected pseudouridine in the sample. These results were compared to the amount of injected RNA, which is received by analysis of the UV chromatogram at 254 nm. Thereby, a turnover efficiency of 99.8% was found for TruB/S.cerevisiae tRNA^Phe. Numeric details and a flow chart are given in Supplementary Table S2 and Supplementary Figure S1.

Figure 4.

Relative and absolute quantifications of ribosomal RNA modifications. (A) Relative quantification displayed as fold changes for several modified nucleosides. The fold changes are the ratio of the modification level of rRNA fragments from WT1 and WT2. Fold changes <1 indicate higher levels in WT2 whereas fold changes >1 indicate higher modification levels in WT1. (B) Absolute modification of modified nucleosides from the same analysis using rRFN values. Am, Gm and Ψ are present in stoichiometrically relevant numbers whereas Cm, I, m⁶₂A and m⁶A are only present in traces.

Figure 5.

Reproducibility of rRFN values over 12 weeks. Relative standard deviations (RSDs) in % for the MS/MS signal and the isotope normalized signal of 2’-O-methyladenosine (Am) over a period of 12 weeks. (A) MS/MS abundance of constant amounts and average abundance with RSD. (B) Relative response factor (rRFN) of Am obtained by the measurements shown in (A). (C) RSD of MS/MS signal and rRFN for 10 modified ribonucleosides after 12 weeks. In black, the RSD in % of the MS abundance without ¹³C SIL-IS-based correction is shown. RSDs of the respective rRFN are in green.