Antisense pairing and SNORD13 structure guide RNA cytidine acetylation

Abstract

N4-acetylcytidine (ac⁴C) is an RNA nucleobase found in all domains of life. The establishment of ac⁴C in helix 45 (h45) of human 18S ribosomal RNA (rRNA) requires the combined activity of the acetyltransferase NAT10 and the box C/D snoRNA SNORD13. However, the molecular mechanisms governing RNA-guided nucleobase acetylation in humans remain unexplored. After applying comparative sequence analysis and site-directed mutagenesis to provide evidence that SNORD13 folds into three main RNA helices, we report two assays that enable the study of SNORD13-dependent RNA acetylation in human cells. First, we demonstrate that ectopic expression of SNORD13 rescues h45 in a SNORD13 knockout cell line. Next, we show that mutant snoRNAs can be used in combination with nucleotide resolution ac⁴C sequencing to define structure and sequence elements critical for SNORD13 function. Finally, we develop a second method that reports on the substrate specificity of endogenous NAT10–SNORD13 via mutational analysis of an ectopically expressed pre-rRNA substrate. By combining mutational analysis of these reconstituted systems with nucleotide resolution ac⁴C sequencing, our studies reveal plasticity in the molecular determinants underlying RNA-guided cytidine acetylation that is distinct from deposition of other well-studied rRNA modifications (e.g., pseudouridine). Overall, our studies provide a new approach to reconstitute RNA-guided cytidine acetylation in human cells as well as nucleotide resolution insights into the mechanisms governing this process.

INTRODUCTION

More than 100 modified RNA nucleotides are known across all domains of life. One of the most highly conserved of these is N4-acetylcytidine (ac⁴C) (Dunin-Horkawicz et al. 2006). In humans, ac⁴C has been mapped to two dominant sites in serine and leucine tRNA, as well as two dominant sites in small subunit (SSU) 18S rRNA occurring at helix 34 and 45 (Sharma et al. 2015; Sas-Chen et al. 2020). Acetylation of all of these RNA targets is catalyzed by NAT10, the only RNA acetyltransferase known in the human genome. NAT10's selectivity toward tRNA or rRNA substrates is governed by the protein adapter THUMPD1 and the short nucleolar RNA (snoRNA) adapter SNORD13, respectively (Sharma et al. 2015).

Recently, we used genetic ablation to demonstrate that SNORD13 is required for acetylation of a single cytidine residue in helix 45 (h45) of small subunit 18S ribosomal RNA (SSU-ac⁴C1842; Fig. 1A; Bortolin-Cavaille et al. 2022). In addition to its biological function, another question is how SNORD13 helps NAT10 recognize its targets in human cells. SNORD13 is a box C/D snoRNA, so termed for the presence of C (5′-RUGAUGA-3′) and D (5′-CUGA-3′) motifs in its primary sequence. However, several features of SNORD13 are atypical of this RNA class. First, while many box C/D snoRNAs play an important role in rRNA 2′-O-methylation by nucleating the assembly of short nucleolar ribonucleoprotein (snoRNP) complexes and guiding the 2′-O-methyltransferase Fibrillarin (FBL) to its target sites via antisense pairing, SNORD13 is the only human box C/D snoRNA known to be involved in cytidine acetylation (Bratkovic et al. 2020). Second, SNORD13 is one of only three snoRNAs known to be independently transcribed by RNA Polymerase II (Pol II) in human cells (Tyc and Steitz 1989). Third, SNORD13 contains two rRNA antisense elements, one located at the 5′ end and the other positioned more internally along the SNORD sequence (Fig. 1B). It is also currently unclear to what extent the mechanisms box C/D snoRNAs use to guide cytidine acetylation are conserved across evolution. For example, the yeast SNORD13 homolog snR45 directly associates with the RNA acetyltransferase Kre33, and may belong to a larger C/D box snoRNP containing the FBL homolog Nop1 (Sharma et al. 2017). However, vertebrate SNORD13 shows little similarity with snR45 (vide infra) and has a longer antisense region and shorter intervening structural loop. Another contrast with the yeast system comes from the fact that coimmunoprecipitation studies have not revealed a direct interaction of NAT10 and SNORD13 in human cells, despite many attempts (Sharma et al. 2015; Thomas et al. 2018). Finally, homologs of the snoRNA that guides helix 34 cytidine acetylation in yeast (snR4) have yet to be identified in humans, and whether such a guide mechanism exists remains unknown.

FIGURE 1.

(A) NAT10 and SNORD13 are required for deposition of ac⁴C in helix 45 of 18S rRNA. (B) Schematic of complementarity between human SNORD13 (purple) and 18S rRNA (blue). The site of cytidine acetylation (C1842) is specified.

A major challenge to understanding box C/D snoRNP function has been the difficulty of their in vitro reconstitution, which has heretofore been limited to archaeal enzymes (Omer et al. 2002; Tran et al. 2003; Graziadei et al. 2020). In contrast, the study of eukaryotic systems has been enabled by the development of cellular reconstitution assays (Kiss-Laszlo et al. 1996; Liang and Fournier 1997; Liang et al. 2009; Deryusheva and Gall 2013). In this strategy, mutants of an endogenous snoRNA are expressed ectopically in cells and tested for their ability to guide the modification of a native (or mutated nonnative) substrate. A limitation of this approach is that failure of a snoRNA to guide its cognate modification can have multiple causes, including disruption of activity, stability, structure, or RNP interactions (discussed further below). However, applied to study the box C/D snoRNAs involved in ribose methylation, these methods have enabled the identification of motifs critical for snoRNA stability, rules for substrate targeting, and guidance of ribose methylation to novel rRNA sites in yeast as well as mammalian systems (Cavaille et al. 1996b; Kiss-Laszlo et al. 1996; Cavaille and Bachellerie 1998). To date, a single study has used this approach to study snoRNA-guided ac⁴C in yeast (Sharma et al. 2017). Similar systems to study vertebrate cytidine acetylation have yet to be developed or applied.

Here, we report two experimental approaches to probe SNORD13-dependent RNA acetyltransferase activity in human cells. Our first strategy hinges on the use of a SNORD13 knockout cell line in which SSU-ac⁴C1842 is lost (Bortolin-Cavaille et al. 2022). Ectopic expression of SNORD13 rescues h45 acetylation and can be quantitatively assessed using nucleotide resolution ac⁴C sequencing. Using this assay enabled us to combine comparative phylogenetic analyses and site-directed mutagenesis of SNORD13 and identify structural and antisense elements critical for activity. To study endogenous SNORD13's substrate specificity we developed a second approach. This assay uses a RNA Polymerase I (Pol I) transcribed minigene to express a nucleolar pre-rRNA fragment which is efficiently acetylated by endogenous NAT10/SNORD13. Systematic mutation of this substrate is used to validate a consensus sequence necessary for ac⁴C deposition and highlights the potential reprogrammability of this system. Our studies demonstrate the power of combining mutational analysis with nucleotide resolution ac⁴C sequencing, and provide the basis for understanding and exploiting RNA-guided RNA modification catalysts in biology, biotechnology, and disease.

RESULTS

Conserved base-pairing interactions of vertebrate SNORD13

Toward the goal of defining functional elements in SNORD13, we first analyzed this snoRNA for evidence of conserved base-pairing interactions (Fig. 2A). All vertebrate SNORD13 genes contain canonical C- and D-boxes (yellow) as well as antisense segments with complementarity to 18S rRNA; however, little sequence similarity is evident outside these regions (Bortolin-Cavaille et al. 2022). Of note, the regions of complementarity of human SNORD13 contain bulges and unpaired bases and architecturally are not identical to SNORD13-like RNAs found in A. thaliana (snoR105, snoR108, and snoR146) (Kim et al. 2010) and S. cerevisiae (snR45) (Sharma et al. 2017). For clarity, we refer to SNORD13's antisense elements as 18S-A and 18S-B (located at the 5′-end) and 18S-C (positioned more internally; Figs. 1B, 2A).

FIGURE 2.

(A) Sequence alignment of vertebrate pre-SNORD13 homologs. Antisense regions 18S-A, 18S-B, and 18S-C are highlighted in red, C and D-box motifs involved in putative kink-turn formation in yellow, and putative stem regions in gray. The small verticle black arrow indicates the 3′ end of SNORD13 as experimentally determined by RACE analysis. (B) Proposed SNORD13 base-pairing interactions with colors coded as above. Bases written in uppercase are part of stable, fully processed SNORD13 while those written in lowercase correspond to the 3′-extension of the transient pre-SNORD13 intermediate. (C) Proposed maturation of pre-SNORD13, containing a 3′-extension which masks its 5′-antisense region, to mature SNORD13 capable of interacting with 18S rRNA.

Taking into account the formation of the predicted kink-turn motif, previously found to be crucial for box C/D snoRNP assembly (Watkins et al. 2007), in silico structural analysis predicted three short, conserved stems in SNORD13 which we designate as stems I, II, and III, respectively (Fig. 2B). To further validate these putative intramolecular helical structures, a comparative approach was conducted in Vertebrates as well as more phylogenetically distant Metazoans (multicellular animals) and Viridiplantae (“green plants”). As illustrated in Supplemental Figures S1 and S2, the formation of Stem I and Stem II is evolutionarily supported by many mutations that do not disrupt base pairings, either as fully compensatory base changes or G:U wobble base pair instead of the regular Watson–Crick G:C pairing. With the notable exception of stem I, which does not appear to be well conserved in Diptera, these two stems were detected in all organisms we examined, including in early diverging nonbilaterian phyla such as Porifera (sponges) or Cnidaria (corals, sea anemones) as well as in the Characeae which are thought to be ancestors of land plants. Stem III is also present in many, but not all, SNORD13 in metazoan while it is more rarely detected in plants (Supplemental Fig. S3). We further observed that the 18S-A and 18S-B antisense elements have the potential to base-pair with an ∼15 nt region downstream from the 3′-end of mature SNORD13, defining two additional helices which we refer to as stem IV and stem V, respectively (Fig. 2B). Although these two stems appear less conserved, particularly in noneutherian species, the ability to form an extended, irregular stem structure remains nevertheless a recurrent feature (Supplemental Fig. S3). The formation of stem IV and stem V requires that transcription of SNORD13 gene goes beyond its mature 3′-end, which has been previously observed (Ohno et al. 2000; Boulon et al. 2004; Badrock et al. 2022). These external stems may be important to formation of the kink-turn motif by bringing together the C- and D-boxes (Darzacq and Kiss 2000), or serve as signal that delineates the mature 3′-end. Of note, a process analogous to the latter has been recently found to play a role in regulating expression of SNORD118 (also known as U8) (Jenkinson et al. 2016; Badrock et al. 2020). Overall, our analyses suggest fully processed SNORD13 folds into three evolutionarily conserved stems which scaffold three distinct 18S rRNA antisense elements, two of which are unmasked from a 3′-extended precursor to enable facile base-pairing with rRNA (Fig. 2C).

Base-pairing elements in stem I, III, and IV dictate SNORD13 expression

Base-pairing interactions play a critical role in snoRNA processing, protein interactions, and cellular accumulation (Huang et al. 1992; Peculis and Steitz 1994). To begin to build sequence-function relationships for SNORD13, we first examined how mutating the base-pairing elements predicted above affected its stability in cells (Fig. 3A). To achieve this objective, we transfected a 1200 nt-long genomic human DNA fragment predicted to contain all cis-acting elements required for transcription and 3′-end processing of SNORD13. Our experimental strategy used 11 human SNORD13 constructs comprising four groups: (i) the wild-type SNORD13 sequence, (ii) two mutants designed to disrupt the C and D boxes, (iii) four mutants designed to disrupt stems I, II, III, and IV, and (iv) four mutants designed to rescue complementarity in stem-disrupting mutations (Fig. 3B). In order to distinguish ectopically expressed human SNORD13 from its endogenously expressed mouse counterpart, each construct was individually transfected into mouse L929 cells. SNORD13 expression was then assessed 48 h post-transfection by ribonuclease (RNase A/T1) protection assay using ³²P-labeled antisense riboprobes specific for each mutated version of SNORD13 (Fig. 3A). Accordingly, protected RNA fragments in the top panel of Figure 3C, which are only detected in transfected samples, represent a full length transfected version of SNORD13. Those in the bottom panel, which are detected in both transfected and nontransfected samples, originate from partial protection of endogenous background RNAs (including mouse SNORD13, due to sequence divergence between human and mouse), providing a rough measure of loading.

FIGURE 3.

(A) Schematic for analysis of SNORD13 stability by RNase A/T1 assay. (B) Sequence of human SNORD13s analyzed. Antisense regions 18S-A, 18S-B, and 18S-C are highlighted in red, C and D-box motifs involved in k-turn formation in yellow, and putative stem regions in gray. In addition to wild-type (“WT,” black) sequence, disruptive mutations (“mut,” red), and rescue mutations (“comp,” red) were explored for stems I–IV. C and D box mutations were also analyzed (sequences provided in Supplemental Information). (C) Results of RNase A/T1 stability assay. Top box indicates protected RNA fragment corresponding to full length, transfected human versions of SNORD13. Bottom box indicates protected RNA fragments corresponding to endogenously expressed mouse SNORD13 background.

Mutation of either the C or D box abolished detection of SNORD13 (Fig. 3C). This is consistent with SNORD13 functioning within a larger snoRNP complex, as previous analyses have shown the C/D box is critical for RNP formation and snoRNA accumulation (Huang et al. 1992; Peculis and Steitz 1994). Mutations causing disruption of stems I, III, and IV also led to a complete loss of observed SNORD13 expression. In the cases of stems I and IV, SNORD13 levels could be restored by introducing complementary mutations that rescue the anticipated helical structures (Fig. 3B,C). However, stem III showed minimal rescue by this approach, suggesting a critical role for its sequence in SNORD13 folding or function (Fig. 3C). Previous studies have proposed the formation of a pseudoknot between nucleotides in stem III and 18S-C in the yeast SNORD13 homolog may be required to bring the antisense regions into close proximity (Sharma et al. 2017), and it is possible complementary mutations fail to rescue this structure in SNORD13. Alternatively, the specific base pairs of native stem III may help nucleate the box C/D interactions. Mutations that disrupt stem II and flanking the 18S-C antisense element were well-tolerated and did not appear to affect SNORD13 abundance (Fig. 3C). This may imply stem II does not form or is dispensable for the processing and stability of SNORD13. These studies provide experimental evidence for the existence of a C/D box and base-pairing interactions in human SNORD13 that are critical to its cellular accumulation. Furthermore, we observe that functional features of SNORD13's predicted structure, such as the occurrence of antisense elements in single-stranded regions, are shared with yeast. This implies that despite low sequence similarity, the structure of human SNORD13 may resemble that of yeast snR45.

Rescue of cytidine acetylation depends on specific base-pairing elements in SNORD13

Recently (Bortolin-Cavaille et al. 2022), we determined that inactivation of SNORD13 in the human chronic myelogenous leukemia-derived HAP1 cell line causes loss of SSU-ac⁴C1842 in 18S rRNA. To understand the molecular determinants of this process, we next evaluated the ability of several ectopically expressed SNORD13 mutants to rescue SSU-ac⁴C1842 in this cell line (Fig. 4A). Briefly, SNORD13-deficient HAP1 cells were cotransfected with two plasmids expressing SNORD13 and eGFP, respectively. After 48 h, flow cytometry was used to isolate cells expressing eGFP (presumably also enriched in SNORD13 expression), and RNA from these cells was analyzed by nucleotide resolution ac⁴C sequencing. This method uses a chemical reaction that modifies the structure of ac⁴C residues, leading to misincorporations (C-to-T) when this RNA is reverse transcribed (Thomas et al. 2018; Sas-Chen et al. 2020). PCR amplification and sequencing are then used to provide a quantitative readout of ac⁴C's presence in RNA at a position of interest. This enables relative comparisons of SNORD13 rescue and detects differences in the rescue of ∼25% with statistical significance (vide infra).

FIGURE 4.

(A) Schematic for analysis of SNORD13 function using quantitative ac⁴C sequencing. (B) Sequence of human SNORD13s analyzed. In addition to wild-type (“WT”) sequence, disruptive mutations (18S-A, 18S-B*, 18S-C mutants), and a mutant with increased complementarity (“18S-A/18S-B full comp”) were explored for rescue of ac⁴C in SNORD13 KO cells. Sequences and verification of expression are provided in Supplemental Figure S4. Note the finding that the 18S-A mutant is expressed when its stem V is disrupted, implies stem V is not strictly required for SNORD13 synthesis. (C) Stem IV is required for accumulation of mutant SNORD13s. Mutation of SNORD13 in mutant 18S-B disrupts stem IV. Reintroduction of complementarity in construct 18-B* (bottom) allows accumulation and testing of function. Expression of 18S-B* is verified by RNase A/T1 mapping in Supplemental Figure S4c. (D) Rescue of SSU-ac⁴C1842 by SNORD13 antisense mutants. Mutants rescue values are normalized relative to the WT SNORD13, which was set to equal 100%. Background misincorporation rates in SNORD13 KO cells were 0%–4%. Values represent n = three biological replicates, analyzed by two-tailed Welch's t-test (ns = not significant, [*] P < 0.05, [**] P < 0.01, and [***] P < 0.001). Exemplary sequencing traces are provided in Supplemental Figure S4d. (E) Rescue of SSU-ac⁴C1842 by SNORD13 stem comp mutants. Structures of comp mutants are provided in Figure 3B. Values represent n = three biological replicates, analyzed by two-tailed Welch's t-test (ns = not significant, [*] P < 0.05, [**] P < 0.01, and [***] P < 0.001).

To explore whether SNORD13's antisense guides are essential for acetylation, we compared ac⁴C rescue by wild-type SNORD13 to three mutants in which 18S-A, 18S-B, and 18S-C were disrupted (18S-A, 18S-B, and 18S-C mutants; Fig. 4B, Supplemental Fig. S4a). Since 18S-B's antisense element is part of stem IV, whose disruption destabilizes SNORD13 (Fig. 3C), we used a second-generation mutant of 18S-B (termed 18S-B*) which restores the formation of stem IV (Fig. 4C; Supplemental Fig. S4b) and whose expression level is stronger than that observed with the WT SNORD13 construct (Supplemental Fig. S4c,e). As expected, rescue of SSU-ac⁴C1842 by ectopically expressed wild-type SNORD13 was readily observed using ac⁴C sequencing (Fig. 4D; Supplemental Fig. S4d). Disruption of antisense complementarity diminished rescue by 60%–92%, with the deleterious effect of mutations following the order 18S-A >> 18S-C > 18S-B (Fig. 4D; Supplemental Fig. S4d). The finding that 18S-A, which covers nt 1–6 of SNORD13 and base-pairs to nts 1854–1859 of 18S rRNA, had the most profound effect on SNORD13 activity is consistent with observations in yeast (Sharma et al. 2017) but contrasts with our recent analysis of SNORD13 homologs in Drosophila (Bortolin-Cavaille et al. 2022), where 5′ antisense elements do not appear to be found. To better understand this finding, we used RT-qPCR to quantify the relative amounts of each SNORD13 mutant studied. In contrast to mouse cells (Supplemental Fig. S4c), RT-qPCR analysis of transfected human cells revealed poor expression of the 18S-A mutant (Supplemental Fig. S4e). Thus, the limited rescue of the 18S-A mutant may also reflect its decreased transcription or stability rather than a strict requirement for its antisense element.

To more directly probe the impact of antisense complementarity on ac⁴C rescue, we designed a SNORD13 mutant in which we corrected five mismatched or unpaired bases found in the 18S-A and B regions to make it fully complementary (Fig. 4B; “18S-A/18S-B full comp”). Analysis by RT-qPCR revealed this construct was produced at high levels in HAP1 cells (Supplemental Fig. S4e). However, compared to wild-type SNORD13, this mutant also exhibited reduced rescue of SSU-ac⁴C1842. Previous studies of C/D box snoRNAs that target FBL have found methylation levels do not necessarily correlate with the stability of the substrate-guide duplex or snoRNA levels, and may be subtly tuned by a complex interaction network depending on the sequence of its RNA substrate (Krogh et al. 2016; Graziadei et al. 2020). A potential rationale is that stronger binding of SNORD13 to 18S rRNA in this mutant may limit its recycling and availability to NAT10. Finally, we applied this assay to examine the effect of stem I, II, and III on rescue, exclusively focusing on mutations that maintain stem complementarity (Fig. 3B; “comp”). Replacement of stems I and III with stems of equivalent length but altered sequence did not rescue ac⁴C, suggesting a functional role for these elements (Fig. 4E), with the caveat that the stem I and III comp mutants may be poorly expressed in sorted GFP HAP1 cells as compared to mouse cells (Fig. 3C; Supplemental Fig. S4e). In contrast, a mutant SNORD13 with altered sequences in stem II is well-expressed and exhibits good rescue (Fig. 4E; Supplemental Fig. S4e). Thus, the specific sequence forming stem II appears dispensable for both SNORD13 stability and NAT10 substrate targeting. Overall, our studies reveal features of SNORD13 critical to guiding ac⁴C deposition and emphasize the ability of antisense elements to impact the function of ectopically expressed snoRNAs by modulating expression level and guide activity.

Sequence requirements of SNORD13 substrates

The presence of multiple copies of genes encoding ribosomal RNA presents a challenge to effective mutagenesis and hinders analysis of the substrate specificity of snoRNA-guided RNA modification in human cells. Beyond this technical barrier, prior attempts to mutagenize SSU-ac⁴C substrates in yeast found every mutation introduced in helix 45 of 18S rRNA to be lethal (Sharma et al. 2017). Seeking to circumvent these limitations, we were inspired by the utility of ribosomal minigenes expressing the 3′ terminus of 18S rRNA fused to internal transcribed spacer 1 (referred to here as h45-ITS1) in investigating the role of SNORD13 in pre-rRNA processing (Hadjiolova et al. 1994; Cavaille et al. 1996a; Cavaille and Bachellerie 1998). The expected recognition of the h45-ITS1 minigene by the pre-rRNA processing machinery led us to hypothesize it may also be able to interact with endogenous NAT10/SNORD13, enabling its facile mutagenesis for substrate specificity studies. To test this, we expressed an h45-ITS1 minigene construct in HEK-293T cells, isolated RNA, and analyzed it using the aforementioned ac⁴C sequencing workflow (Fig. 5A). Importantly, these experiments used primers that specifically amplify minigene h45-ITS1, rather than the endogenous 18S pre-RNA (Supplemental Fig. S5a,b). Ribosomal minigenes expressed from a RNA Polymerase I (Pol I) (Ideue et al. 2009) but not an RNA Polymerase II (Pol II) promoter were efficiently acetylated, consistent with the predominant nucleolar localization of NAT10 and SNORD13 (Supplemental Fig. S5c). Modification occurred at the expected position corresponding to SSU-C1842 of 18S rRNA. Of note, detection of ac⁴C in this ectopically expressed pre-rRNA fragment requires the use of the nucleotide resolution sequencing assays, which enables PCR amplification of the ac⁴C signal and represents an advance over previous analytical methods.

FIGURE 5.

(A) Schematic for analysis of endogenous SNORD13 substrate specificity using a Pol I-transcribed pre-rRNA h45-ITS1 minigene substrate and quantitative ac⁴C sequencing. (B) Structure of h45-ITS1 substrates with mutations lying proximal to natural ac⁴C site. Values in red represent ac⁴C-dependent misincorporation rates normalized relative to the WT h45-ITS1 sequence, which was set to equal 100%. (C) Summary of percent misincorporation observed upon mutation of 5′-CCG-3′ consensus sequence in h45-ITS1 substrates. (D) Bar graph of mutants specified in Figure 5B and E. Values represent n = three biological replicates, analyzed by two-tailed Welch's t-test in comparison to WT (ns = not significant, [*] P < 0.05, [**] P < 0.01, [***] P < 0.001, and [****] P < 0.0001) (E) Structure of “triple mutant” h45-ITS1 substrate engineered to have increased complementarity to SNORD13. (F,G). Structure of h45-ITS1 substrates with bases inserted 5′ and/or 3′ relative to the 5′-CCG-3′ consensus sequence. (H) Structure of h45-ITS1 substrates in which the 5′-CCG-3′ consensus sequence is shifted 1 bp (+1 bp) or 3 bp (+3 bp). (I) Bar graph of mutants specified in Figure 5F–H. Values represent n = three biological replicates, analyzed by two-tailed Welch's t-test in comparison to WT (ns = not significant, [*] P < 0.05, [**] P < 0.01, [***] P < 0.001, and [****] P < 0.0001). Exemplary sequencing traces provided in Supplemental Figure S6 and Supplemental Figure S8b.

To begin to understand the substrate specificity of SNORD13-dependent RNA acetylation, we analyzed h45-ITS1 constructs in which individual residues flanking SSU-C1842 were mutated, but SNORD13 complementarity was preserved (Fig. 5B,C; Supplemental Fig. S6). Previous studies have shown eukaryotic NAT10 demonstrates a strong preference for modification of 5′-CCG-3′ sequences (Ito et al. 2014a,b; Sas-Chen et al. 2020). Whether SNORD13-dependent acetylation is able to tolerate mutation of this consensus remains unknown. Maintaining the central acetylation site and focusing on the 5′ and 3′ residues (Fig. 5C), we unexpectedly observed significant acetylation (∼44% misincorporation) of a noncanonical 5′-ACG-3′ sequence. Reanalysis of a previously reported degenerate substrate library furnished further evidence that human NAT10 can modify 5′-ACG-3′ as well as 5′-UCG-3′ sequences when massively overexpressed in HEK-293T cells (Supplemental Fig. S7), albeit with greatly reduced efficiency relative to 5′-CCG-3′. In contrast, further changes–including all substitutions of the 3′ guanosine–reduced modification of the h45-ITS1 pre-rRNA to negligible levels (<3% misincorporation). Evaluating constructs with mutations lying outside of the consensus, we found that 5′ mutations (m1) were seemingly more well-tolerated than those lying 3′ (m4/m5). Simultaneous mutation of all six residues completely eliminated modification (m7), as expected (Fig. 5B–D). Mutations in h45 designed to improve snoRNA base-pairing (m8) paradoxically lessened modification of the h45-ITS1 substrate (Fig. 5D,E). This is consistent with our analysis of SNORD13 above (Fig. 4B–D; “18S-A/18S-B full comp”), and further emphasizes that ac⁴C deposition does not strictly correlate with the apparent strength of guide-substrate complementarity.

Many snoRNA-dependent RNA modification systems exhibit strict spatial requirements for catalysis. For example, box C/D snoRNAs direct ribose methylation to the nucleotide base-paired to the fifth residue upstream of the D or D′ boxes, and insertion of an intervening base can shift what nucleotide is methylated (Cavaille et al. 1996b; Kiss-Laszlo et al. 1996). H/ACA box snoRNAs contain two antisense elements that form a three-way junction with substrates, leaving the targeted uridine and a flanking residue unpaired and presented to the pseudouridine synthase in the center of a pseudouridinylation pocket which occurs a defined distance (14–16 nt) from the H box or ACA box (Ganot et al. 1997; Bortolin et al. 1999; Czekay and Kothe 2021). These strict spatial requirements reflect the precision with which guide snoRNAs must properly position their substrates in the active site of modification enzymes to enable catalysis. To probe whether similar requirements exist for SNORD13-dependent RNA acetylation, we inserted nucleotides in the putatively unpaired “acetylation pocket” of the h45-ITS1 pre-RNA substrate (Fig. 5F–I). Addition of one or two residues 5′ to the modification site (ins1, ins2; Fig. 5F) had little affect on ac⁴C deposition (Fig. 5I). Inserting a single nucleotide 3′ to the modification site (ins3) was also tolerated. However, inserting two nucleotides 3′ to the modification site of h45 (ins4) led to a >50% loss of acetylation, and when combined with an additional upstream cytidine (ins6) abolished modification entirely (Fig. 5F–I). The finding that the ins4 construct contains an additional 5′-CCG-3′ that is not modified implies the relative location of the 5′-CCG-3′ consensus sequence matters. To test this, we attempted to shift the position of the 5′-CCG-3′ consensus without altering the overall length of the loops (+1 and +3 bp mutants; Fig. 5H). These mutant substrates were not modified at the novel 5′-CCG-3′ sequence; however, in the +3 bp mutant, a degree of modification (65%) was observed at the site corresponding to SSU-C1842 (Fig. 5H; Supplemental Fig. S8). Interestingly, our studies found disruption of 18S-B complementarity enables a degree of SNORD13-dependent ac⁴C deposition (Fig. 4D). This would seemingly preclude a rule in which a defined number of intervening bases must lie between the antisense-paired rRNA substrate and nucleotide to be modified, although additional possibilities exist. Overall, our studies indicate unusual plasticity in the NAT10 active site which accommodates diverse presentation of 5′-HCG-3′ (H = A/C/U) pre-rRNA substrates at the endogenous C1842 position.

Previous studies have exploited designer snoRNAs to direct modifications to novel RNA targets (Cavaille et al. 1996b). This has been used as an alternative to mutagenesis to explore ribosome function (Liu and Fournier 2004; Liu et al. 2008), and to influence modification-sensitive processes such as nonsense suppression (Karijolich and Yu 2011). As a simple proof-of-concept, we tested whether a mutant snoRNA could increase acetylation of an h45-ITS1 substrate that is poorly recognized by endogenous SNORD13 (Fig. 6A). Mutation of h45-ITS1 to disrupt its interaction with 18S-A (nts 1–6) largely abrogated the ability of endogenous SNORD13 to guide its modification, leading to an ∼90% decrease in ac⁴C (Fig. 6B). Ectopic expression of a SNORD13 containing an 18S-A region designed to base-pair with the mutant substrate partially rescued its modification (Fig. 6B; Supplemental Fig. S9). Although additional applications of this system lie beyond the scope of our current study, this experiment demonstrates the potential of our assays to explore how cytidine acetylation may be directed to nonnative substrates using synthetic SNORD13 analogs.

FIGURE 6.

(A) Schematic for analysis of orthogonal SNORD13-substrate pairs. Exemplary sequencing traces are provided in Supplemental Figure S9. (B) Bar graph of ac⁴C levels at site corresponding to SSU-1842 in orthogonal SNORD13-substrate pairs. Values represent ac⁴C-dependent misincorporation rates normalized relative to the WT h45-ITS1 and WT SNORD13 pair, which was set to equal 100%. Values represent n = three biological replicates, analyzed by two-tailed Welch's t-test in comparison to WT (ns = not significant, [*] P < 0.05, [**] P < 0.01, [***] P < 0.001, and [****] P < 0.0001). Sequences of SNORD13 18S-A mutant and h45-ITS1 18S-A mutant are provided in the Supplemental Information. (C) Overview of SNORD13 structure-function relationships probed in this study.

DISCUSSION

Here, we have reported the development of parallel experimental strategies to study SNORD13-dependent RNA acetylation in human cells. We demonstrate that SNORD13 knockout cells provide an optimal genetic background to examine rescue of SSU-ac⁴C1842 by ectopically expressed snoRNAs, while h45-ITS1 ribosomal minigenes can be used to profile the substrate selectivity of endogenous SNORD13 (Fig. 6C). Quantitative analysis of each of these systems is enabled by a signal amplified ac⁴C sequencing method. Applying these assays highlighted a critical role for stem I, stem III, and the 18S-A antisense regions in SNORD13 activity and validated the preference for substrate acetylation to occur in a 5′-CCG-3′ consensus sequence, with 5′-ACG-3′ also tolerated. These systems also allowed us to probe spatial constraints in the putative pocket formed by the putative snoRNP responsible for acetylation, which appears to tolerate insertion of unpaired nucleotides upstream, but not downstream, of the SSU-C1842 modification site.

A technical limitation of our studies is the precision of Sanger sequencing-based ac⁴C detection, which as used here is most conducive to relative comparisons of SNORD13 activity (detecting differences of ∼25%). Integrating ac⁴C detection chemistry with next-generation sequencing (NGS) can provide enhanced signal-to-noise relative to Sanger sequencing (Sas-Chen et al. 2020; Thalalla Gamage et al. 2021) and we recommend NGS for more sensitive quantification of small differences. Future studies may also benefit from the use of alternative methods to knockdown SNORD13, for example, using modified antisense oligonucleotides (ASOs) (Ideue et al. 2009), which could enable the functionality of SNORD13 analogs to be tested in new cell lines and model organisms not amenable to genetic manipulation.

The composition and mechanism of the snoRNP which catalyzes SSU-ac⁴C1842 remain unknown. While both NAT10 and SNORD13 are clearly required for h45 acetylation, we and others (Sharma et al. 2015) have not yet found evidence that the acetyltransferase and snoRNA directly interact. Indeed, the most well-characterized protein binding partner of SNORD13 is FBL (Tyc and Steitz 1989; Baserga et al. 1991), suggesting its likely integration into a C/D snoRNP comprised of 15.5K, NOP56, and NOP58 (Watkins et al. 2002, 2007). Yeast NAT10 has been proposed to co-occupy snR45 together with this organism's C/D snoRNP protein homologs (Sharma et al. 2017), and interactions between NAT10, FBL, and NOP56 identified in recent proteome-wide studies raise the possibility of a similar scenario in human cells (Go et al. 2021). One model for this to occur would be if a core snoRNP comprising SNORD13 bound to 18S rRNA, recruited NAT10 via a transient protein–protein interaction, and presented C1842 to its active site for acetylation. The in vitro characterization of NAT10-containing snoRNPs may benefit from the development of simplified archaeal systems, as has been useful in studies of snoRNA-dependent methylation and pseudouridinylation (Tran et al. 2003; Sas-Chen et al. 2020; Czekay and Kothe 2021). For the moment, the assays reported here should provide a powerful tool with which to determine the contribution of different snoRNP components to SSU-C1842 acetylation in human cells.

An additional motivation for studying snoRNAs lies in their potential to be engineered to direct RNA modifications to novel targets (Cavaille et al. 1996b; Liu and Fournier 2004). As an initial step in this direction, we demonstrated that a SNORD13 mutant can be used to drive acetylation of an ectopic pre-rRNA substrate that is not efficiently modified in endogenous cells. It is important to specify that this represents an extremely simple model system, and the extent of SNORD13's reprogrammability, as well as whether analogs may direct NAT10 to novel substrates, remains unknown. Considering our collective observations, the following features of SNORD13 appear necessary for ac⁴C deposition: (i) antisense complementarity, with apparent importance following 18S-A > 18S-C, 18S-B, (ii) stem I and III sequences, (iii) a stem IV engineered to complement 18S-B for stability, (iv) a 5′-MCG-3′ substrate sequence (M = A or C), and (v) a nonhybridized substrate loop which tolerates flexibility but not shifting of the cytidine modification site within it (Fig. 6C). The tolerance for a 5′ C → A mutation in the CCG consensus is in line with our recent report of acetylation in a 5′-UCG-3′ sequence in the model organism P. polycephalum (Bortolin-Cavaille et al. 2022). In contrast, substrates harboring mutations 3′ to the acetylation site (e.g., 5′-CCC-3′) were not modified. This suggests human NAT10 strongly prefers modification of 5′-HCG-3′ sequences, a consideration that should be taken into account when evaluating novel substrates (Arango et al. 2018; Sas-Chen et al. 2020). It has been previously noted that SNORD13 resembles a H/ACA snoRNA in terms of its bipartite substrate complementarity (Bachellerie and Cavaille 1997; Sharma et al. 2017). Our mutational analyses of the antisense regions of SNORD13—as well as our recent discovery of an atypical D. melanogaster SNORD13 homolog (Bortolin-Cavaille et al. 2022)—suggest that unlike H/ACA snoRNAs, a single antisense strand can suffice for ac⁴C deposition while the second may increase efficiency. Additional studies will be required to characterize this model and establish the scope of substrates NAT10 may be directed to via mutated SNORD13 analogs. Such research will be greatly aided by high-throughput methods to characterize RNA–RNA interactions (Dudnakova et al. 2018) as well as the transcriptome-wide distribution of ac⁴C (Sas-Chen et al. 2020). Such studies are in progress and will be reported in due course.

MATERIALS AND METHODS

Cell culture, transfection and RNA extraction

HEK-293T and L929 cells were grown in Dulbecco's modified Eagle medium (DMEM; 4.5 g/L glucose). HAP1 cells were grown in Iscove's modified Dulbecco's medium (IMDM; 4.5 g/L glucose). All cell culture media were supplemented with 10% fetal bovine serum (PAN-Biotech), 1 mM sodium pyruvate (Gibco) and 1% penicillin/streptomycin (Sigma-Aldrich). Cells were incubated at 37 °C with 5% CO₂. L929 and HEK-293T cells (70–80% confluency in six-well plates) were transiently transfected with plasmids using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Briefly, 5 μL of Lipofectamine 2000 was diluted in 250 μL Opti-MEM I medium (solution A) and incubated for 5 min at room temperature. A solution B was prepared with 4 μg of plasmid diluted in 250 μL Opti-MEM I. Solutions A and B were mixed gently (drop-by-drop) and incubated for 20 min at room temperature. Plasmid-Lipofectamine complexes (500 μL) were then added to seeded L929 or HEK-293T cells. Using GenePulser Xcell (Bio-Rad; 270V, 950μF, 4 mm cuvette), half a T175 culture flask containing SNORD13-KO HAP1 cells at 70–80% confluency was co-electroporated with 8 μg of SNORD13-expressing plasmids and 1 μg of GFP-expressing plasmid (pGFPmax). Forty-eight hours post-transfection, GFP positive cells (∼1 million) were sorted using fluorescence-activated cell sorting (BD Influx System [USB]), and total RNA was prepared using Tri-Reagent (Euromedex) according to the manufacturer's instructions before treatment with RQ1 DNase-RNase free (1 u/μl; Promega) and proteinase K (25 mg/mL; Sigma).

Preparation of SNORD13 constructs, SNORD13 mutagenesis and RNase A/T1 protection

A 1.2 kb-long genomic DNA fragment overlapping human SNORD13 gene (hg38: Chr8: 33,512,975-33,514,174) was PCR amplified and cloned into the EcoRI and XbaI sites of the pKS plasmid. SNORD13 mutants were then generated by multi-step PCR (SNORD13 mutant sequences and primer sequences are provided below and in Supplemental Table S1). DNA sequencing was used to confirm the introduced mutations. Plasmids were transfected in L929 cells and total RNA was analyzed by RNase A/T1 mapping. Briefly, 10 μg of total RNA was hybridized overnight at 42°C with ∼50,000 cpm of gel-purified [α32P]-labeled riboprobes in 10 μL of S1 solution (40 mM PIPES, 400 mM NaCl, 1 mM EDTA, 80% formamide). Samples were treated for 60 min at room temperature with 100 μL of RNase A/T1 buffer (10 mM Tris-Cl pH 7.5, 200 mM NaCl, 5 mM EDTA, 100 mM LiCl, 20-40μg/mL RNase A [Sigma], RNase T1 100 u/mL [Ambion]) followed by a 15-min treatment at 37°C with proteinase K (25 mg/mL, Sigma). After extraction with phenol chloroform-isoamyl alcohol, RNA was precipitated and fractionated by electrophoresis on a 6% acrylamide, 7 M urea denaturing gel. SNORD13-expressing plasmids were used to produce PCR templates for producing an antisense riboprobe via T7 in vitro transcription.

Ac4C-sequencing of endogenous SSU-C1842

DNase treated total RNA samples (1 μg) were treated with sodium cyanoborohydride (100 mM in H₂O) or vehicle (H₂O) in a final reaction volume of 100 μL. Reactions were initiated by the addition of 1 M HCl to a final concentration of 100 mM and incubated for 20 min at room temperature. Reactions were stopped by the addition of 30 μL 1 M tris−HCl pH 8.0. The quenched reactions were adjusted to 200 μL with H₂O and purified via ethanol precipitation. The pelleted RNA was dried using a Speedvac, resuspended in ddH₂O, and quantified using a Nanodrop 2000 spectrophotometer. For the reverse transcription reaction with the SuperScript III enzyme, first ∼500 pg of treated RNA were incubated with 4.0 pmole of the h45 reverse primer (5'-TAATGATCCTTCCGCAGGTTCACCTAC-3') in 1X SuperScript III buffer at 65 °C for 5 min and transferred to ice for 1 min to facilitate annealing. After annealing, reverse transcriptions were performed by adding 200 units of the SuperScript III enzyme, 5 mM DTT, 25 units of RNasin, 500 μM dNTPs (5 mM GTP, 10 mM CTP, ATP, and TTP) and incubating at 55°C for 60 min. Reactions were quenched by increasing the temperature to 70°C for 15 min. The cDNA products from the reactions and controls were directly used in PCR. PCR reactions were set up with 2 μL cDNA in 50 μL PCR reaction with Phusion Hot Start Flex (New England Biolabs). Reaction conditions: 1X supplied HF buffer, 200 μM each dNTP, 2.5 pmol each forward (5'-CGTCGCTACTACCGATTGGATGG-3') and reverse (5'-TAATGATCCTTCCGCAGGTTCACCTAC-3') primers, 2 units of Phusion Hot Start Enzyme, 2 μL template (thermocycling conditions: 67°C annealing, 34 cycles). PCR products were run on a 2% agarose gel, stained with SYBR safe, and visualized on a UV transilluminator at 302 nm. Bands of the desired size were excised from the gel. DNA was extracted using a QIAquick Gel Extraction Kit from Zymo and submitted for Sanger sequencing (GeneWiz) using 5 μM of the forward PCR primer. Processed sequencing traces were viewed using 4Peaks software. Peak height for each base was measured, and the percent misincorporation was determined using the equation: “percent misincorporation = (peak intensity of T)/ (sum of C and T base peaks) X 100%”. Misincorporation values were determined by subtracting the average background water control misincorporation levels from that of the corresponding reactions. Then the misincorporation values of each construct were normalized to empty vector value followed by the wild type value. Values represent n =3 biological replicates, analyzed by two-tailed Welch's t-test (ns = not significant, [*] P < 0.05, [**] P < 0.01, [***] P < 0.001, and [****] P < 0.0001).

Preparation of h45-ITS1 minigene constructs

A PCR amplified fragment corresponding to the human 18S (57 nt)-ITS1 (13 nt) junction was cloned into the Hind III and BamHI sites of the human ribosomal RNA (Pol-I) minigene phPol1EX (Idueu et al. 2009) or pcDNA 3.1, which drives expression from a RNA Pol-II (CMV) promoter. Mutated versions were generated by multistep PCR, and DNA sequencing was used to confirm mutations (h45-ITS1 and primer sequences are provided below and in Supplemental Table S1).

Ac4C-sequencing of h45-ITS1 constructs

DNase treated total RNA samples (∼5 μg) were treated with sodium cyanoborohydride (100 mM in H₂O) or vehicle (H₂O) in a final reaction volume of 100 μL. Reactions were initiated by the addition of 1 M HCl to a final concentration of 100 mM and incubated for 20 min at room temperature. Reactions were stopped by the addition of 30 μL 1 M tris−HCl pH 8.0. The quenched reactions were adjusted to 200 μL with H₂O and purified via ethanol precipitation. The pelleted RNA was dried using a Speedvac, resuspended in ddH₂O, and quantified using a Nanodrop 2000 spectrophotometer. For the reverse transcription reaction with the SuperScript III enzyme, first ∼2 μg of treated RNA were incubated with 4.0 pmole of the T7 reverse primer (5'-TAATACGACTCACTATAG-3') in 1X Superscript III buffer at 65°C for 5 min and transferred to ice for 1 min to facilitate annealing. After annealing, reverse transcriptions were performed by adding 200 units of the SuperScript III enzyme, 5 mM DTT, 25 units of RNasin, 500 μM dNTPs (5 mM GTP, 10 mM CTP, ATP, and TTP) and incubating at 55 °C for 60 min. Reactions were quenched by increasing the temperature to 70°C for 15 min. The cDNA products from the reactions and controls were directly used in PCR. PCR reactions were set up with 2 μL cDNA in 50 μL PCR reaction with Phusion Hot Start Flex (New England Biolabs). Reaction conditions: 1X supplied HF buffer, 2.5 pmol each SP6 forward (5'-ATTTAGGTGACACTATAGAA-3') and T7 reverse (5'-TAATACGACTCACTATAG-3') primers, 200 μM each dNTP, 2 units of Phusion Hot Start Enzyme, 2 μL template (thermocycling conditions: 52°C annealing, 33 cycles). PCR products were run on a 2% agarose gel, stained with SYBR safe, and visualized on a UV transilluminator at 302 nm. Bands of the desired size were excised from the gel. DNA was extracted using a QIAquick Gel Extraction Kit from Zymo and submitted for Sanger sequencing (GeneWiz) using 5 μM of the SP6 forward PCR primer. Processed sequencing traces were viewed using 4Peaks software. Peak height for each base was measured, and the percent misincorporation was determined using the equation: “percent misincorporation = (peak intensity of T)/ (sum of C and T base peaks) X 100%”. Misincorporation values were determined by subtracting the average background water control misincorporation levels from that of the corresponding reactions. Then the misincorporation values of each construct were normalized to the wild type value. Values represent n =3 biological replicates, analyzed by two-tailed Welch's t-test (ns = not significant, [*] P < 0.05, [**] P < 0.01, [***] P < 0.001, and [****] P < 0.0001).

RT-qPCR analysis of SNORD13 mutant expression

DNase treated total RNA (100 ng) from GFP-sorted cells were used for the RT-qPCR analysis. Specific primers were designed for the RT-qPCR of each SNORD13-18S and stem mutant RNA (see Table 1 above) and used with the Luna Universal One-Step RT-qPCR kit (cat. no. E3005, NEB) according to the manufacturer's protocol. Briefly, 20 μL reactions were set up with 10 μL of the 2x Luna Universal One-Step Reaction Mix, 1 μL of the 20X Luna WarmStart RT enzyme mix, 0.8 μL of 10 μM forward and reverse primers, 100 ng of template RNA, and water. U6 snoRNA was used as the housekeeping gene. Assay mix was set up in triplicate for each reaction and control. They were loaded onto a 96-well plate, and the plate was centrifuged at 1500 rpm for 1 min to remove bubbles and collect liquid. RT-qPCR assay was performed on a Roche LightCycler 480 II instrument under the SYBR Green/ROX2 detection format setting. Assay conditions: reverse transcription at 55°C for 10 min, initial denaturation 95°C for 1 min, denaturation 95°C for 10 sec, extension 60°C for 30 sec–45 cycles, and melt curve at 60–95°C. The average of three Ct values per sample was calculated, and ΔCt values were determined using the equation “ΔCt=Ct (average of SNORD13 mutant) – Ct (average of U6)”. Then the 2^–(ΔCt) values of each construct were normalized to the wild type value using the equation “Normalization = 2^–(ΔCt) of SNORD mutant/ 2^–(ΔCt) of WT SNORD13”. Values represent n = 3 biological replicates, analyzed by two-tailed Welch's t-test (ns = not significant, [*] P < 0.05, [**] P < 0.01, [***] P < 0.001, and [****] P < 0.0001).

TABLE 1.

RT-qPCR primer sequences

Multiple sequence alignments of SNORD13 in representative organisms

Multiple sequence alignments of SNORD13 in representative organisms of Metazoan and Viridiplantae kingdoms (Bortolin-Cavaille et al. 2022) were generated with Molecular Evolutionary Genetic Analysis (MEGAX) (Kumar et al. 2018) and ClustalW (Eddy 1995). If necessary, the alignment of sequences were curated manually to correctly align well-established RNA motifs (e.g., box C, box D, 18S rRNA antisense elements). The human SNORD13 sequence and its 3'-flanking region (∼15 nt-long) were manually curated for putative intramolecular base-pairing interactions. Within this framework, we inferred that the C- and D-boxes are brought together to form a kink-turn motif, as previously observed for box C/D snoRNAs. This allowed us to predict five stem structures denoted as stem I, II, III, IV and V. Stem I was already proposed by Tyc and Steitz (1989). Note that, at that time, the authors were not aware of the existence of a k-turn in box C/D snoRNAs, nor of the ability of SNORD13 to base-pair with 18S rRNA. The relevance of these five putative RNA helical structures was then established by a phylogenetic comparative approach that consisted in identifying covarying nucleotides that maintain base-pairing interactions in the stems, either Watson-Crick base-pair compensatory mutations or GU Wobble pairs. The schematic representation of the stems were manually drawn using Adobe Illustrator.

SUPPLEMENTAL MATERIAL

Supplemental material is available for this article.

MEET THE FIRST AUTHOR

Supuni Thalalla Gamage.

Meet the First Author(s) is a new editorial feature within RNA, in which the first author(s) of research-based papers in each issue have the opportunity to introduce themselves and their work to readers of RNA and the RNA research community. Supuni Thalalla Gamage is co-first author of this paper, “Antisense pairing and SNORD13 structure guide RNA cytidine acetylation,” along with Marie-Line Bortolin-Cavaillé. Supuni is a visiting postdoctoral fellow in Dr. Jordan Meier's laboratory at the National Cancer Institute. Her research focuses on developing and applying new methods to study post-transcriptional modifications of RNA.

What are the major results described in your paper and how do they impact this branch of the field?

In this paper, we use a novel cell assay to determine the sequence requirements for snoRNA-guided deposition of the highly conserved RNA modification N4-acetylcytidine (ac⁴C) by the human acetyltransferase enzyme Nat10. This involved first introducing different variants of SNORD13 (a snoRNA that guides ac⁴C modification in rRNA) into a knockout cell line and used a nucleotide resolution ac⁴C sequencing method to quantify the level of ac⁴C rescue at an rRNA site. To complement this approach, we developed a parallel method that mutates the Nat10/snoRNA substrate in helix 45 of rRNA. Together, these two approaches defined a strong preference for acetylation in a 5′-CCG-3′ sequence context and also provided evidence that putative stem and antisense regions in SNORD13 play an important role in directing acetylation to RNA substrates. Overall, our work provides insights into the molecular mechanisms underlying RNA-guided cytidine acetylation.

What led you to study RNA or this aspect of RNA science?

I earned my PhD in Professor Christine Chow's laboratory at Wayne State where I became fascinated by how chemical modifications and small molecules can alter nucleic acid structure and function. Since joining the Meier laboratory at the NCI, I've been able to develop this passion by contributing to sequencing methods and biochemical and biological studies of modified nucleobases in RNA. I find it amazing and exciting that even for modifications such as ac⁴C—first discovered in the 1960s—there remains so much to be learned.

During the course of these experiments, were there any surprising results or particular difficulties that altered your thinking and subsequent focus?

One unexpected result stemming from the assay system was that human NAT10 can modify a noncanonical 5′-ACG-3′ sequence. This was surprising, as previous work from our laboratory and others had shown that nearly all metazoan RNA acetyltransferases act exclusively on 5′-CCG-3′ sequences. This result demonstrates that NAT10 can have some plasticity for modification of noncanonical sequences in overexpression contexts. Another unexpected turn was when analysis of the SNORD13 structure by our collaborators in the Cavaillé laboratory led us to study the conservation of rRNA acetylation across evolution. This led to the discovery of new SNORD13s and Metazoans that have dispensed with rRNA acetylation. It also required extraction and nucleotide resolution detection of ac⁴C in organisms ranging from worms to spiders to mushrooms. Certainly not what I had expected when I started my postdoc!

What are some of the landmark moments that provoked your interest in science or your development as a scientist?

A memorable landmark moment for myself was learning from my science teacher in middle school. She made the lessons so enjoyable and digestible. It made an impact and sparked the idea that I may want to pursue a career as a scientist.

If you were able to give one piece of advice to your younger self, what would that be?

Have confidence in yourself! And as this project shows, be open to new things.

Are there specific individuals or groups who have influenced your philosophy or approach to science?

My PhD and postdoctoral mentors, Professor Christine Chow and Dr. Jordan Meier, have influenced my approach to science. As mentioned above, I was first introduced into the world of RNA in Professor Chow's laboratory at Wayne State University in Michigan. Professor Chow always encouraged me to understand the basic science behind experiments and to critically think about my research. My postdoc advisor, Dr. Meier, has also had a big influence on my approach to science. His enthusiasm for research always motivates me and encourages me to enjoy the process of problem solving.

What are your subsequent near- or long-term career plans?

My immediate plans are to continue to grow my skills to understand the biology and biochemistry of RNA modifications. I find this area fascinating and impactful given the use of modifications in the vaccines used to combat the ongoing pandemic. After completing my training at NCI, I am planning to continue my scientific career by contributing to therapeutic development in the biotech sector.

What were the strongest aspects of your collaboration as co-first authors?

We benefited enormously from the hard work, expertise, and enthusiasm of our collaborators in this project. Drs. Jerome Cavaillé and Marie-Line Bortolin-Cavaillé have been pioneers in defining the mechanism and biology of small RNAs, with an emphasis (relevant to this project) on box C/D snoRNAs. Following the discovery of SNORD13 by the Steitz group in 1989, Jerome's PhD advisor Dr. Jean-Pierre Bachellerie noted its potential for 18S rRNA interaction. This was actually an area Jerome explored in his thesis! All this is to say that this was a question the Cavaillé laboratory had been considering for years. This project succeeded by combining this strong intellectual foundation with my co-first author Marie-Line Bortolin-Cavaillé’s amazing gifts as an experimental RNA biologist and some of our lab's cutting-edge ac⁴C analysis methods.

How did you decide to work together as co-first authors?

This collaboration arose from an email conversation between the corresponding authors Dr. Cavaillé and Dr. Meier. As mentioned, the Cavaillé lab's focus is dissecting the mode of action and the biological roles of small regulatory RNAs. Dr. Cavaillé had been interested in SNORD13's base-pairing interactions with 18S rRNA since his PhD work; however, at that time the modifications present in that region were not known, and there were also no tools to analyze them. The characterization of a functional homolog of SNORD13 in yeast together with our group's development of tools for profiling ac⁴C reanimated this project that had been dormant for many years in the Cavaillé laboratory. The groups quickly established a great rapport which saw us through many samples, an encounter with a space telescope, two papers, and the onset of a pandemic!