Quantification of light and heavy labelled modifications during growth

In the case of Gm, Cm and m⁷G, an approximate quantification of the relative change was performed by comparison of the integral of the ion count signals of light and heavy modification with respect to either m¹G, in the case of Gm, or with respect to m⁵C, for Cm and m⁷G. m¹G and m⁵C were selected for comparison because they elute at similar retention times. These ratios were normalized with respect to the UV absorbance of the reference nucleosides and of the nucleoside A (for details, see Table 1). The resulting change in modification content is displayed in Figure 15. Approximate quantification reveals that, especially in the case of Gm, there is a large increase in modification content stating from day 2, in agreement with other modifications (Figure 14).

Figure 15 Percental change in Gm, Cm and m⁷G. Plots show the percental change of Gm (based on m¹G), Cm and m⁷G (based on m⁵C) with respect to the modification content at time point 1 h. Changes in modification content were estimated from comparison of the ion count integrals and normalized to A nucleoside content (see Table 6).

In summary, yeast tRNAs show a slight decrease in the extent of modification during the first 24 h, followed by an overall increase in modification content during entry into stationary phase, in line with previous findings.^[138] Exceptions are observed particularly for Am, which shows the opposite trend.

This modification is generally thought to stabilize poorly modified tRNA molecules. Variations in Am modification content might therefore reflect a different need for stabilization which inversely correlates with the overall extent of modification of the tRNA species.

only due to modifications introduced during growth in labelled medium, and can be used to assess turnover of tRNAs generated prior to medium exchange.

CH3/CD3 ratios were quantified by comparison of the integrals of light and heavy modifications by LC-MS analysis. Analysis of CH3/CD3 ratios was performed for the modifications m¹A, Am, m⁷G, Gm, m¹G, m²G, m²2G, m⁵C and Cm. In labelled medium, the proportion of light modification content was on average about twice as abundant as the heavy modification. Upon exchanging of the medium, the CH3/CD3 ratio increased over the first five days until it stabilized at a ratio of about 70-80 between day 5 and day 7. Most modifications followed a similar pattern, with the exception of Am which on average showed lower ratios.

Having quantified the CH3/CD3 ratios for the aforementioned modifications, the variation in ratios over time was evaluated using two different models. In the first model, expected ratios over time were estimated based purely on changes in optical density. In the second, more complex model, the variation in tRNA content per OD unit (see Figure 11) was also taken into account. In both cases, measured ratios of all modifications (except for Am) were averaged and their average ratio at time 1 h was used to construct the models. Comparison was then done between the average measured ratios over the 7 days and the expected ones. Am was analyzed separately. Both models rely on the assumption that the heavy-labelled tRNA content present at time point 1 h is not degraded over time and that it can therefore be used to calculate the expected ratios assuming that there is a linear increase in modification content with increasing OD.

In the first model, we assume that tRNA content per OD unit is constant over time, and that the CH3/CD3 modification ratio can be estimated exclusively from the increase in optical density, i.e. the increasing CH3/CD3 ratio reflects the increasing dilution in labelled tRNA modification during cell division (with tRNA content proportionally increasing with OD). Therefore, starting from an OD600 of 1.32 (time 1 h) with an average CH3/CD3 ratio of 2.24 ± 0.12, we can assume that about one third of the tRNA modifications are labelled. This number is assumed to remain constant over the 7 days, and therefore the expected CH3/CD3 ratio can simply be estimated by calculating the proportion of non-labelled tRNAs from the measured OD600 at each time point (for details, see Table 1).

Figure 16A shows the comparison of the average measured CH3/CD3 ratios (for all modifications but Am) with respect to the expected ratios calculated based on optical density. As it can be seen, expected ratios accurately predicted measured values up to time point 24 h. However, starting from day 2 the measured CH3/CD3 ratios are higher than expected, suggesting that there is a greater proportion of light (CH3)-modification than expected. This can result from two different scenarios: it can either be due to an increase in the number of modifications introduced starting from day 2, meaning that tRNA samples are more extensively modified with non-labelled SAM, or it could arise from turnover of labelled tRNA, which would lead to a gradual loss of heavy modifications and therefore to higher

CH3/CD3 ratios. Interestingly, Am follows the opposite trend: as shown in Figure 16B, starting from time point 20 h the average measured CH3/CD3 ratio for Am is lower than the ratio expected based on optical density, implying that this modification is introduced to a lesser extent than expected during growth into stationary phase. This would again support the hypothesis that Am might be introduced preferentially for low modified tRNA samples (first 24 h), later becoming less essential upon increasing extent of tRNA modification (day 2‒7).

Figure 16 Comparison of measured and expected CH3/CD3 ratios based on optical density. A) Comparison of measured and expected ratios for all modifications except Am. Measured ratios were averaged for m¹A, m¹G, m²G, m⁷G, Gm, m²₂G (CD3-, CH3-labelled only), m⁵C and Cm. Expected ratios were calculated assuming that light (CD3)-modification content increased linearly with optical density, starting from time point 0 h, with an initial CH₃/CD₃ ratio of 2.24 and OD₆₀₀ = 1.32.

B) Comparison of average measured Am ratios and expected ratios. Expected ratios were calculated as in A).(see Table 7)

In the second model, expected CH3/CD3 ratios are calculated based on varying tRNA content per OD unit as well as optical density. This model accounts for the fact that the tRNA pool per cell is not constant over time. In fact, given that the tRNA per OD unit gradually decrease after the first 24 h (see Figure 11), a model assuming a constant tRNA content per OD unit (first model) would overestimate the amount of non-labelled tRNA produced during days 2 to 7.

Once again, the model is constructed from knowledge of the initial OD600 (1.32), CH3/CD3 ratio (2.24 ± 0.12) and normalized tRNA content per OD unit (68.4 % ± 3.1 %). The labelled tRNA pool is assumed to remain constant over the 7 days (i.e. no turnover), and the expected ratios are calculated as described above (see also Table 1). Figure 17A shows the comparison of the measured and expected CH3/CD3 ratios for all modifications except Am. As in the previous model, the expected ratios closely model the measured ones up to time point 24 h, after which the expected ratios largely underestimate the real CH3/CD3 values. As in the previous case, this might arise from increased modification extent starting from day 2, or it could be due to degradation of labelled tRNA. Interestingly, the expected ratio at time point 4 h overestimates the CH3/CD3 ratio. This, knowing that at this time point we

observe the largest proportion of tRNA content per OD unit (Figure 11B), could be explained in terms of an overestimation of the extent of tRNA modification at 4 h. In fact, as discussed in Section 4.2.2, Figure 13, the extent of modification at this time point reaches a minimum, implying that the tRNA sample is less modified. The model used to calculate the CH3/CD3 ratios shown in Figure 17 assumes instead that the proportion of tRNA modification is uniform at all time points, and is therefore not surprising that it might overestimate the ratio at time point 4 h.

Once again, Am shows a different pattern, with ratios which are generally fitting with the modelled ratios, suggesting that the Am content in tRNA is not increasing over time, but that it is rather more closely correlating with the increase in tRNA during growth.

Figure 17 Comparison of measured and expected CH3/CD3 ratios based on tRNA content per OD unit. A) Comparison of measured and expected ratios for all modifications except Am. Measured ratios were averaged for m¹A, m¹G, m²G, m⁷G, Gm, m²2G (CD3-, CH3-labelled only), m⁵C and Cm. Expected ratios were calculated assuming that light (CH3)-modification content increased linearly with tRNA content, starting from time point 0 h, with an initial CH₃/CD₃ ratio of 2.24, OD600 = 1.32 and tRNA content per OD unit of 68.38 %. B) Comparison of average measured Am ratios and expected ratios.

Expected ratios were calculated as in A). (see Table 8)

In summary, evaluation of the CH3/CD3 ratios over time suggests that for all modifications but Am there is a greater proportion of light (CH3)-modification than expected starting from day 2. This might either be due to the production of more highly modified tRNAs starting from this time point, or it might result from degradation of labelled tRNA, yielding a reduced heavy (CD3)-modification content and higher CH3/CD3 ratios. Once again, Am shows a different trend, with ratios largely fitting with a linear increase in modification content following optical density and variation in tRNA content, again suggesting a different role of this modification compared to other modified nucleosides investigated in this study.