Analysis by HPLC-ESI-MS

For the calculation of the exact analyte levels from the mass spectrometric data calibration curve equations are necessary. Their determination was executed as follows: Solutions of the labeled and the unlabeled nucleoside with defined concentration were prepared. In this way defined amounts of substance can easily be handled. Mass spectrometric data were collected for five different concentration ratios of the labeled and corresponding unlabeled nucleoside (Figure 15). For each concentration ratio an average value of three independent measurements was determined using three mixtures of the labeled nucleoside solution with one out of three solutions with different concentrations of the corresponding unlabeled nucleosides. By

plotting the weight ratios against the experimentally determined ion count area ratios the calibration curve and its corresponding calibration equation (Figure 15) were obtained.

  Figure 15: Representative example of a calibration curve and its equation.

These are necessary, because the reagents used to incorporate heavy atoms in the standard are never 100 % labeled themselves. Commercially available LiAlD4 used here for example contains 98 % deuterium. Therefore, every standard solution also comprises a small amount of unlabeled material. In addition, measurement of three different concentrations of the unlabeled nucleoside against one concentration of the labeled nucleoside averages weighing errors. The use of calibration curves also compensates for systematic differences in the processing of labeled and unlabeled nucleosides, e.g. in the mass spectrometer. This is emphasized by Figure 15, where the slope is substantially higher than one. Calculation of the modification level without normalization using the calibration curve equation would thus result in a significant error. The calibration curves and equations for all other modifications studied in this work are compiled in chapter 10.6.2.

5.1.5.2 HPLC method

Advantageously, HPLC-MS quantification utilizes two levels of separation. First, the analytes are separated due to their different retention times during HPL chromatography (Figure 16) and, second, they are differentiated by their individual masses by mass spectrometry (Figure 17 a). On the other hand, molecules with the same mass and the same retention time are not quantifiable individually. For HPLC-MS quantification of modified tRNA nucleosides all modifications with the same mass have to reach the mass spectrometer at different times. Otherwise the mass signals overlap and an individual quantification of the overlapping

tend to have similar retention times, because they are structurally related. For example methyladenosines m2A and m6A show this effect (Figure 17 b).

In addition, modified nucleosides have to be chromatographically separated from canonical nucleosides. The detector used here, a Thermo Finnigan LTQ Orbitrap XL, analyzes the ions reaching it in a two step process. First, a certain number of ions is collected. Then these accumulated ions are measured. The canonical nucleosides are present in the sample in high excess compared to the modified ones. Therefore, when a canonical nucleoside reaches the analyzer, it will be flooded by the ions of this nucleoside, which in turn leads to suppression of the ions of interest. The strength of suppression is proportional to the amount of the abundant nucleoside reaching the detector. Labeled and unlabeled nucleoside do not reach the mass spectrometer simultaneously due to a small difference of their retention times (e.g. 0.2 min for m2A and d3-m2A, Figure 17). Therefore, their amounts, which reach the detector at the same time as the canonical nucleoside, are not identical. This leads to different ion suppressions of the nucleosides of interest and frustrates their reliable quantification.

  Figure 16: Representative HPL chromatograms and gradients.

HPL chromatogram and gradient of tRNA nucleoside mixtures from a) E. coli; b) pork liver ( = 260 nm). The HPLC gradient is depicted in red as percentage of buffer D.

In order to circumvent these problems, a potent HPL chromatography method had to be established. There are numerous HPLC conditions known in literature for the separation and analysis of modified nucleosides,[59a, 106, 110] which either do not reach a sufficient level of separation or are not compatible with the equipment on site. For these reasons the HPLC method used in the Carell group for analysis of DNA digests was chosen as a starting point.[113] Additionally, the flow used for HPL chromatography (0.15 mL/min) lay within the range of the injection volume of the mass spectrometer and could therefore be adopted

without alterations and the buffer system was suitable for mass spectrometry (buffer C: 2 mM

ammonium formiate, pH 5.5; buffer D: 2 mM ammonium formiate/MeCN 20/80, pH 5.5). With the general HPLC conditions in place a screening of various gradients for maximum separation of modified nucleosides was conducted resulting in similar gradients for the various tRNA nucleoside mixtures from the cells investigated (Figure 16). In addition, coinjection studies were conducted to determine the retention times of the nucleosides in question. Detailed HPLC conditions are described in chapter 10.6.1. To our knowledge, our HPLC method for the first time allows separation of m2A and m6A making these nucleosides quantifiable (Figure 17 b).

Typical UV traces of the obtained E. coli and pork liver modification pattern are depicted in Figure 16 a and Figure 16 b, respectively. Next to the dominant signals for the canonical RNA nucleosides, the various modified nucleosides either appear as small signals or are hidden in the base line. Unambiguous assignment and quantification of the modifications is therefore only possible by MS analysis.

  Figure 17: Mass spectrometric measurements of m2A, m6A, d3-m2A, and d3-m6A.

a) Relevant high resolution mass spectrometric data for unlabeled and labeled m2A. b) Positive ion currents of the protonated nucleosides m2A and m6A and the corresponding synthetic isotope labeled derivatives.

5.1.5.3 Mass spectrometric analysis

After high resolution separation of the modified nucleosides by the optimized HPLC conditions they were quantified by mass spectrometry as described below using the example of m2A and d3-m2A. The nucleoside mixture obtained from the digest is subjected to HPLC-

MS. The nucleosides m2A and d3-m2A were separated from all other nucleosides. Due to their mass difference they appear in the mass spectra as two separate signals (Figure 17 a). After completion of the whole measurement the obtained data were analyzed with “Xcalibur”. The measurement of high resolution mass spectrometry employed here allows isolation of those ions produced by methylated adenosines. By setting a defined mass range (for methyladenosine: 282.1187 – 282.1207; for d3-methyladenosine: 285.1372 - 285.1392) the ion current of all methyladenosines and their labeled standards were extracted (Figure 16 b). Integration of the areas under the mass peaks yielded the ratio of m2A and d3-m2A ions.

Substitution of this ratio into the corresponding calibration curve (chapters 5.1.5.1 and 10.6.2) gave the molar ratio of m2A and d3-m2A present in the measured sample. This ratio was charged against the known amount of d3-m2A added to the sample prior to the measurement. This way an exact determination of the amount of m2A in the sample was achieved.

It is important to not inject the first 3 minutes of an HPLC run into the mass spectrometer. All salts from the sample do not adhere to the reverse phase material of the column and therefore elute with the void volume. Injection of these in the mass spectrometer would clog its entrance inevitably leading to adulterated results.

As a control for the reliability and reproducibility of our approach intra- and inter-assay tests were performed for nucleosides m1A, i6A, ms2i6A, and m1G. During the intra-assay tests the determined values of labeled to unlabeled nucleosides of a sample after enzymatic digestion showed excellent reproducibility for each nucleoside (N=5; 2.5 % for m1A, 0.4 % for i6A, 0.7 % for ms2i6A and 2.6 % for m1G). Inter-assay tests gave good area ratio reproducibility on six subsequent days (N=6; of 6.3 % for m1A, 1.0 % for i6A, 1.8 % for ms2i6A and 4.3 % for m1G). These tests attest the outstanding accuracy of the HPLC-MS quantification. The determined integrals for each labeled or unlabeled nucleoside not using the standards resulted in an average standard deviation of 34.5 % and showed insufficient reproducibility. Therefore, usage of the labeled nucleosides as standard is essential. No memory effect was observed during blank LC/MS experiments performed after several measurements of a sample excluding contamination by carry-over. Tuning of the spectrometer was performed with uridine. Further detailed description of the optimization of the digest, HPLC, and MS method will be discussed by D. Globisch in an upcoming Ph.D. thesis.

In summary, in this thesis work a stable isotope based method was established (Figure 10), that allows efficient, direct, and parallel quantification of theoretically all tRNA modifications from any cell or tissue with extremely high sensitivity.

5.2

Initial measurements of modified tRNA nucleosides in E. coli, HCT-116,