E. coli strains listed in Table 1 were cultured in LB media to ∼0.5 U (OD₆₀₀) and harvested by centrifugation. The pellet was resuspended in lysis buffer (20 mM Tris-HCl pH 7.5, 100 mM NH₄Cl, 0.5 mM EDTA, 6 mM β-mercaptoethanol). Lysozyme was added (1.5 mg/ml) to the suspension and incubated on ice for 10 min and the cell lysate centrifuged (30,000xg, 15 min, at 4°C). After incubation of the supernatant for five more min with 20 U of RQ1 DNase on ice, the suspension was re-centrifuged for 15 min at 30,000xg. The supernatant was layered on top of an equal volume of 32% sucrose solution and subject to centrifugation at 100,000xg for 18 h at 4°C to pellet the ribosomes (Mehta et al., 2012). The ribosomal pellet was resuspended in ribosome storage buffer (50 mM Tris-HCl pH 7.5, 100 mM NH₄Cl, 10 mM MgCl₂, 6 mM β-mercaptoethanol) and stored at −80 °C until use.

To obtain protein-free ribosomal RNA, ribosomes were extracted with Tri-Reagent® (Sigma-Aldrich) as per the manufacturer’s protocol. Ribosomes and ribosomal RNA were stored at −80°C.

Ribosomes (30 μg; 1A260 = 60 µg) and rRNA (7 μg; 1A260 = 40 µg) in storage buffer (0.12 μg μl⁻¹) were irradiated with UVA (370 nm peak wavelength, 360–400 nm range, 0.752 mJ cm⁻²) at room temperature (25°C) for 1 h with addition of 100 µM riboflavin (RF). Unexposed samples were kept in the dark, under the same conditions. Another set of ribosomes and rRNA samples were also incubated with 10 µM iron (II) chloride and 20 mM H₂O₂ for 1 h at 37°C. Unexposed samples were treated similarly but no H₂O₂ was added. Soon after exposure, rRNA was extracted from ribosomes using Tri-reagent. RNA was reprecipitated with ½ volume of 7.5 M ammonium acetate and three times volume of ethanol, pelleted, and dissolved in sterile water.

For in-cellular exposure, cells were grown to 0.5 OD_600 nm and exposed to 10 µM FeCl₂ and 20 mM H₂O₂ for 30 min (Willi et al., 2018) or UVA for a defined period of time, RNA extracted and ribosomal RNA fractionated by lithium chloride-based precipitation with no lag time (Nilsen, 2012).

The quality of isolated rRNA was analyzed by the Fragment Analyzer (Advanced Analytical) at Cincinnati Children core facility. The vendor-specific software (ProSize) was used to compute the signal intensity for 23S rRNA and 16S rRNA following gel electrophoresis.

E. coli strains, K12 (wild type), ΔrlmKL and ΔrlmB were grown to mid-log phase (∼0.5 OD600 nm) and diluted to ∼0.01 OD for testing growth behavior when grown with LB media. Growth was monitored through changes in OD using a microplate reader (Biotek 80 TS) at 37°C with slowest shake speed setting for 24 h.

Four µg of UVA or H₂O₂-exposed and unexposed RNA samples were digested to nucleosides following denaturation (heating at 90°C, 10 min, snap cooling on ice) and incubated with 0.3 U of nuclease P1 (Sigma-Aldrich) in 10 mM ammonium acetate at 45°C for 2 h. To this digest, ammonium bicarbonate (100 mM), 0.00125 U of snake venom phosphodiesterase and 0.1 U of bacterial alkaline phosphatase were added and incubated at 37°C for 2 h. Digests were dried and resuspended in mobile phase A (MPA) (5.3 mM ammonium acetate, pH 4.5) before subjecting it to LC-MS. A Vanquish UHPLC coupled with an Orbitrap Fusion Lumos (Thermo Scientific) or TSQ Quantiva Triple Quadrupole Mass spectrometer were used for data acquisition. Reversed phase liquid chromatography was performed using an HSS T3 column (100 Å, 1.8 μm, 2.1 × 50 mm, Waters) and the mass spectra were recorded in positive polarity. LC-MS/MS conditions were identical to those described previously (Jora et al., 2018). Nucleosides were identified based on chromatographic retention time, unique m/z values, and CID/HCD fragmentation. The data was processed using Xcalibur software (Thermo Fisher Scientific) for subsequent interpretation.

Purine nucleobases undergo oxidative damage to form oxidation products just like DNA (Scheme II). However, the prevalence of such products is not clear when RNA is bound with proteins. To obtain more insight into these events, we subjected the intact ribosomes or purified ribosomal RNA to UVA or Fenton-reaction induced oxidative stress and documented the type of oxidation products observed for each condition. The LC-MS technique employs electrospray ionization to convert the ions in the liquid phase to gaseous phase at the entrance (source) of mass spectrometer. Such a process, in general, causes oxidation of the analyte at the source. Guanosine with its low redox potential is more susceptible to such oxidation. However, such an electrospray-induced oxidation can be easily distinguished as its signal aligns identically to that of the chromatographic retention time of unoxidized nucleoside (data not shown). On the other hand, the native 8-oxo-G (present in nucleoside mix prior to electrospray) exhibited a longer retention time on the chromatographic column and generates a separate signal enabling its distinction (Supplemental Figure S1A).

We noticed the presence of 8-oxo-G (m/z 300.0944) lesions in unexposed rRNA at an average of ∼one to two for every 1,000 guanosines when cells were harvested at mid-log phase (Supplementary Figure S1B). When RNA was incubated with either FeCl₂ or RF without H₂O₂ or UVA exposure, respectively, variability at the basal levels of 8-oxo-G was observed. So, in order to account for these effects, changes in 8-oxo-G levels are expressed as fold change between the oxidant included and omitted samples but contained either FeCl₂ or RF in the respective sets. However, advanced oxidation product (further oxidation of 8-oxo-G) levels were at baseline levels in the oxidant omitted samples. In these cases, the observed peak areas of oxidation products were normalized against defined number of guanosines (10³ or 10⁵).

Guanosine oxidation products: Less than 60 min of UVA exposure of ribosomes did not cause significant increase in 8-oxo-G levels and more than 60 min exposure resulted in fragmentation of RNA (data not shown). Therefore, 60 min was chosen as the optimal exposure time to differentiate the oxidative effects due to protein presence in the ribosome. Upon UVA exposure we see higher levels of 8-oxo-G and guanidinohydantoin (Gh) (m/z 290.1097) lesions (Supplementary Figure S2A) compared to the unexposed (but riboflavin-containing) samples. However, the UVA-induced differences show contrasting features between ribosomes and rRNA even though they were exposed under identical buffer conditions. While the 8-oxo-G levels were significantly higher for ribosomes (7-fold vs 4-fold for rRNA), Gh levels were at increased levels for rRNA, 24) compared to ribosomes (4 per 10⁵ G). This suggests that association with proteins could potentially decrease the chances of further oxidation of 8-oxo-G in the ribosomes. Lack of associated proteins presumably enabled advanced oxidation to Gh (4-electron oxidation product) in protein-free rRNA. Interestingly, the observed levels for FapyG (m/z 302.1101) (Supplementary Figure S2B) were lower (7–9 per 10⁵ G) upon UVA exposure and do not show significant differences between ribosome and rRNA samples suggesting that this type of alternate oxidation process occurs at relatively low level irrespective of the protein association status (Figure 1A).

Fenton reaction-induced oxidative stress, on the other hand, generated higher levels of 8-oxo-G (11.5-fold) and FapyG (46-fold) for rRNA. On the other hand, ribosome samples did not show noticeable level of oxidation products with Fenton reaction. Gh levels were not significantly different between ribosome and rRNA. Observation of lower levels of oxidative lesions for ribosomes suggest that protein association could protect rRNA against ROS generated by the Fenton reaction. However, such protection was not observed for rRNA devoid of proteins (Figure 1B). Overall, the type of oxidation process and protein association status would determine the nature and amount of guanosine oxidation products in this experimental system.

Adenosine oxidation products: Either UVA or Fenton reaction induced oxidative stress generated 8-oxo-A and FapyA (Supplementary Figure S3) for ribosomes and rRNA. However, 8-oxo-A was significantly higher for rRNA (8-fold) compared to ribosomes for both types of oxidative stress, but FapyA did not show any significant differences whether or not RNA bound with protein (Figure 2).

To understand whether the absence of structure-stabilizing methylation modifications (Decatur and Fournier, 2002; Sergiev et al., 2011) influence the susceptibility of rRNA to ROS, we exposed the ribosomes derived from E. coli strains, K12, ΔrlmKL (exhibit loss of 7-methylguanine (m⁷G) at position 2069 and 2-methylguanine (m²G) at position 2,445) and ΔrlmB (exhibit loss of guanosine ribose methylation at position 2,251 of 23S rRNA) that are deficient in respective methyl transferases. Although these strains exhibited significantly lower levels of corresponding methylations in the ribosomal RNA (Figure 4A), the cells exhibited more or less identical growth behavior suggesting that the effects of undermethylated rRNA are minimal or insignificant (Figure 4B) under the tested culture conditions. However, upon UVA exposure, the 8-oxo-G levels increased 25-fold in the ΔrlmB ribosomes compared to K12 (12-fold) or ΔrlmKL (19-fold) suggesting that the loss of ribose methylations could make the ribosomes more susceptible to ROS damage (Figure 4C).

The wobble modification in tRNA, mnm⁵s²U, is susceptible to UVA-induced oxidative degradation and the ΔtrmU strain of E. coli deficient in this modification exhibited elevated levels of oxidized guanosine in tRNA upon exposure to UVA (Sun et al., 2018). We tested the possibility of such behavior under Fenton-reaction induced oxidative stress conditions. Exposure of this strain to the Fenton reaction caused higher levels of 8-oxo-G in tRNA consistent with previous observations, and such elevated levels were also reflected in total RNA level in comparison to the K-12 wildtype cells. (Figure 5).

Consistent with the previous reports (Liu et al., 2012), the basal levels of 8-oxo-G exhibited wide variation depending on the additives included in the samples even though all the precautions (dark incubation, minimal exposure to air) were followed. Iron chloride addition alone increased the basal levels of 8-oxo-G (Supplemental Figure S1) presumably due to spurious oxidation during sample handling and/or LC-MS analysis (Gedik and Collins, 2005). However, addition of oxidant such as hydrogen peroxide, or exposure to UVA increased the oxidation of purines up to 7-fold for 8-oxo-G, and 25 Gh lesions or 46 FapyG lesions per 10⁵ guanosines (Figure 1) compared to riboflavin or iron chloride added samples (Supplemental Figures S2, S3).

Exposure of rRNA and ribosome yielded 8-oxo-G and advanced oxidation products, Gh or FapyG. Exposure to H₂O₂ at lower concentration (1–10 mM) did not cause significant levels of advanced oxidation products (other than 8-oxo-G) (data not shown). Higher concentrations of H₂O₂ (20 mM) in combination with iron chloride (induced Fenton reaction) caused higher levels of advanced oxidation products. However, Gh formation was predominantly noticed in UVA exposure and FapyG by the Fenton reaction. These observations indicate the prevalence of specific oxidation pathways depending on photooxidation or chemical oxidation. The predominant type of ROS generated, for example, singlet oxygen by UVA (Miyamoto et al., 2003; Baier et al., 2007; Schuch et al., 2013) and hydroxyl radicals by Fenton reaction (Kehrer, 2000; Koppenol, 2001) could be one reason for operation of alternate oxidation pathways.

Ribosomal protein association with rRNA, in general, decreased the levels of advanced oxidation (further oxidation of 8-oxo-G) products for UVA exposure (Figure 1). There was general decrease in purine oxidation levels when ribosomes were exposed to the Fenton reaction conditions (Figure 2). Ribosomal proteins are located in the outer shell of ribosome, which provides an easier solvent-accessibility for ROS (Nikolay et al., 2015). They are also known for maintaining ribosomes’ structure, suggesting that their absence alters ribosome conformation and may expose rRNA nucleotides previously “sheltered” in the inner core. Aside from their localization and function, ribosomal proteins are composed of amino acids that have a lower calculated reduction potential compared to rRNA nucleotides, ranging from 0.85 V (tryptophan) to 1.67 V (lysine) (Close and Wardman, 2018). Therefore, the absence of ribosomal proteins could lead to enhanced oxidative damage to rRNA. This raises interesting questions about the role of proteins in influencing the oxidation. Do they need to be bound to the rRNA for reducing the damage or unrelated proteins or free amino acids could also influence extent of damage? Future studies involving RNA and RNA binding proteins or mixtures of RNA and proteins or amino acids could provide more insight into the specific or non-specific nature of these effects. Nevertheless, these observations clearly indicate that the chance of aggravated oxidative damage is higher if the RNA is free of proteins. Locating the oxidation sites with or without protein presence could also help answer this query. Increased fragmentation of oxidized RNA in the cellular conditions might be due to the higher affinity of oxidized RNA to enzymes such as polynucleotide phosphorylase leading to RNA degradation (Hayakawa et al., 2001; Malla and Li, 2019).

We evaluated the impact of oxidative stress on ribosomes when the cells are deficient in specific methyltransferases. These strains of E. coli (ΔrlmKL, ΔrlmB) did not show any significant differences in their growth behavior under normal cell culture conditions. The methylations catalyzed by these enzymes are located in Domain V of 23S rRNA. The ribose methylation, Gm is located at position 2,251 in the peptidyl transferase center (PTC) and is a universally conserved modification (Polikanov et al., 2015). Modifications, m⁷G (at 2069) and m²G (at 2,445) are installed by a unique dual fused methyltransferase RlmKL (Kimura et al., 2012). UVA exposure of the ribosomes isolated from these strains showed elevated 8-oxo-G levels. While the increase in 8-oxo-G levels is not significant (p value = 0.053) for ΔrlmKL, those of ΔrlmB strain exhibited significantly higher levels than K12 cells (p = 0.00064). It is interesting to note that although RlmKL installs methyl groups at two different positions (2069 and 2,445) of 23S rRNA compared to RlmB (at position 2,251) (Popova and Williamson, 2014; Sergiev et al., 2018), the latter strain seems to be more susceptible to oxidative stress (Figure 4). Further studies involving modification mapping experiments (Sun et al., 2020; Thakur et al., 2020) could identify the exact locations in the sequence that might serve as hotspots for oxidative damage. Nevertheless, these studies indicate that although some post-transcriptional methylations are not essential for normal growth, their absence could aggravate the vulnerability of rRNA to oxidative damage. This is presumably due to the instability of the folded structure, because methylations on the nitrogenous base and ribose play an important role in stabilizing the structure (Decatur and Fournier, 2002; Sergiev et al., 2011) and the interactions with other translational apparatus (Polikanov et al., 2015). Such vulnerability is different from that of the oxidative degradation of post-transcriptional modifications containing -amino, -oxy, and sulfur groups. This is because methylations, in general, were found to be less susceptible to oxidative degradation in our previous studies (Sun et al., 2018).

Similarly, the absence of thiolation in mnm⁵s²U also led to increased guanosine oxidation upon UVA exposure in the enzyme-deficient cells (ΔtrmU) (Sun et al., 2018). This observation is recapitulated in the current investigations for transfer RNA and even extended to total RNA with Fenton reaction studies, where the 8-oxo-G levels significantly increased in the mutant cells (Figure 5). These observations strongly suggest that certain nucleoside modifications are associated with oxidative stress effects. Recently, E. coli cells deficient in mnm⁵s²U are shown to be highly sensitive to hydrogen peroxide and mnm⁵s²U is required for stress-resistance responses (Jingjing et al., 2021) suggesting that the observed effects in our studies are not coincidental. Overall, these studies strongly indicate that ribosomal proteins and post-transcriptional nucleoside modifications could play multiple roles in not only maintaining optimal structure and decoding function but also protect against the adverse effects of oxidative stress to perform optimal mRNA translation in the cell.

In summary, the pattern of oxidative damage to RNA (8-oxo-G and its advanced oxidation products) is altered by the protein association and presence of post-transcriptional nucleoside modifications. RNA oxidation and degradation levels under cellular condition were higher compared to the exposure of ribosomes. Although, loss of methyl groups at two different positions did not increase the susceptibility, absence of ribose methylation at another position increased the susceptibility of rRNA to oxidative damage in ribosomes compared to that of wild type E. coli suggesting a position-specific dependence. Similarly, deficiency of ROS-susceptible nucleoside modification increased the overall oxidative damage in tRNA, and total RNA compared to wild type strain. This suggests that it is not the number, but the nature and status of modification at a defined position that determines the overall susceptibility of RNA to ROS.