SnoRNA guide activities: real and ambiguous

In eukaryotes, rRNAs and spliceosomal snRNAs are heavily modified post-transcriptionally. Pseudouridylation and 2′-O-methylation are the most abundant types of RNA modifications. They are mediated by modification guide RNAs, also known as small nucleolar (sno)RNAs and small Cajal body-specific (sca)RNAs. We used yeast and vertebrate cells to test guide activities predicted for a number of snoRNAs, based on their regions of complementarity with rRNAs. We showed that human SNORA24 is a genuine guide RNA for 18S-Ψ609, despite some noncanonical base-pairing with its target. At the same time, we found quite a few snoRNAs that have the ability to base-pair with rRNAs and can induce predicted modifications in artificial substrate RNAs, but do not modify the same target sequence within endogenous rRNA molecules. Furthermore, certain fragments of rRNAs can be modified by the endogenous yeast modification machinery when inserted into an artificial backbone RNA, even though the same sequences are not modified in endogenous yeast rRNAs. In Xenopus cells, a guide RNA generated from scaRNA, but not from snoRNA, could induce an additional pseudouridylation of U2 snRNA at position 60; both guide RNAs were equally active on a U2 snRNA-specific substrate in yeast cells. Thus, post-transcriptional modification of functionally important RNAs, such as rRNAs and snRNAs, is highly regulated and more complex than simply strong base-pairing between a guide RNA and substrate RNA. We discuss possible regulatory roles for these unexpected modifications.

The ASEs in this type of modification guide RNA are formed by internal loops in 5' and 3' stems associated with box H (ANANNA) and box ACA, respectively; these loops are usually called pseudouridylation pockets. Typically, the distance from the box H or box ACA to the position of the target uridine inside the pocket is 15±1 nucleotide (reviewed in Bachellerie et al. 2002;Yu et al. 2005;Watkins and Bohnsack 2012). These are the basic rules for functional modification guide RNAs, yet they are neither sufficient nor absolutely essential (Deryusheva and Gall 2018).
Pseudouridylation and 2'-O-methylation are the most abundant and functionally important posttranscriptional modifications in eukaryotic rRNAs and spliceosomal snRNAs (Yu et al. 1998;Dönmez et al. 2004;Lapeyre 2005;Liang et al. 2007Liang et al. , 2009Jack et al. 2011). With a few exceptions, positioning of these modifications depends on the guide RNA activities. Recent studies identified transcriptome-wide spread of pseudouridines and 2'-O-methylated residues Schwartz et al. 2014;Dai et al. 2017). The patterns and levels of RNA D e r y u s h e v a , T a l r o s s a n d G a l l 4 modifications are highly regulated, and they may be significantly modulated under different physiological and pathological conditions (Wu et al. 2011;Schwartz et al. 2014;Krogh et al. 2016;Sharma et al. 2017;Taoka et al. 2018).
Usually, computational or visual search for regions of complementarity between guide and substrate RNAs is used to predict specific targets for modification guide RNAs. However, such predictions may turn out to be false when they are tested experimentally (Xiao et al. 2009;Deryusheva and Gall 2018;Deryusheva et al. 2020). In a number of studies of eukaryotic snoRNA modification guide activities, in vitro reconstitution systems showed robust efficiency using cell nuclear extracts (Jády and Kiss 2001;Ma et al. 2005;Deryusheva and Gall 2009;Xiao et al. 2009;Deryusheva and Gall 2013) or recombinantly expressed and purified protein complexes (Kelly et al. 2019;Yang et al. 2020). At the same time, some snoRNAs that are inactive in cell free in vitro assays can function efficiently as modification guide RNAs in living cells Gall 2013, 2019b).
In our recent analysis of human and Xenopus snoRNA sets we found snoRNAs with regions complementary to 18S and 28S rRNAs at positions that have never been found modified in any species (Deryusheva et al. 2020 D  e  r  y  u  s  h  e  v  a  ,  T  a  l  r  o  s  s  a  n  d  G  a  l  l   5 conditions such as stress. Alternatively, these domains may trigger RNA degradation when they are exposed for modification; that is, the modified RNA molecules are undetectable because they are unstable.

Human SNORA24 is a genuine guide RNA for position 609 in 18S rRNA
Since its discovery, human SNORA24 has been assigned to the known pseudouridine at position 609 in 18S rRNA (Kiss et al. 2004). Yet, in a later study SNORA24 activity on this position was not confirmed using yeast nuclear extract (Xiao et al. 2009). Since false-negative results are possible in this type of in vitro assay Gall 2013, 2019b) we repeated the test using living cells. First, we made a construct to express human SNORA24 in vivo in S.
cerevisiae. For some reason this snoRNA did not accumulate in yeast. Therefore, we switched to a vertebrate cell system. In our previous study we performed comparative analysis of mammalian and amphibian snoRNAs and modification patterns of their predicted targets (Deryusheva et al. 2020 rRNA. Furthermore, we demonstrated again that in vitro reconstitution assays are less compatible with modification guide RNA activities than methods based on living cells.

Artificial substrate RNAs, but not endogenous rRNAs undergo modification
Xenopus SNORA24, which lacks the pseudouridylation pocket for 18S-574 (equivalent to human 18S-609), shows complementarity to 18S rRNA at position 1607 (equivalent to human 18S-1649) ( Figure 1B). Yet, neither position 574 nor 1607 is modified in Xenopus 18S rRNA (Deryusheva et al. 2020 rRNA, an additional construct was made to assay Xenopus SNORA24 modification activity in yeast. This construct was used to express a fragment of vertebrate 18S rRNA as an artificial substrate RNA. Intriguingly, when this construct was expressed in the wild type yeast strain, the SNORA24 substrate RNA became modified at the tested position even without co-expression of exogenous guide RNA ( Figure 1C, top black trace).
We screened several yeast mutant strains deficient for different pseudouridine synthases and found that yeast Pus7p was responsible for this modification ( Figure 1C, blue trace), even though D e r y u s h e v a , T a l r o s s a n d G a l l 7 the modified sequence does not contain the canonical motif recognized by yeast Pus7p ( Figure   1B). It is important to clarify here that yeast ( Figure 1C, black trace), but not Xenopus, Pus7p ( Figure 1C, pink trace) can modify that sequence. It seems that Pus7p synthases from different species, like Pus1p itself (Behm-Ansmant et al. 2006), differ in their activities on certain RNA substrates. In these experiments, modification of yeast U2 snRNA at position 35 served as an internal control for Pus7p modification activity ( Figure 1D). Ultimately, in the mutant pus7Δ strain the chimeric SNORA24-SCARNA4 guide RNA could induce pseudouridylation of the artificial substrate RNA at the position corresponding to Xenopus 18S-1607 ( Figure 1C, green trace The modification of the SNORA24 substrate RNA by yeast Pus7p is not the only case in which the endogenous modification machinery is active on an artificial substrate RNA but not on endogenous yeast rRNA. Another example is a potential substrate for one of the two copies of X. tropicalis SNORA15 (Supplemental Figure S1). Here again, no modification is found at the predicted target position in 18S rRNA from any species. In addition to xtSNORA24 and yeast Pus7p, we identified several other vertebrate snoRNAs that were functional in vivo on artificial substrate RNAs, but not on endogenous rRNAs. One such snoRNA is the orphan vertebrate SNORA18. It has an evolutionarily conserved ASE that can base-pair with 18S rRNA to convert U1720 to pseudouridine ( Figure 2A). Yet, this position is not modified in any vertebrate species tested so far (Deryusheva et al. 2020). When we expressed SNORA18 along with the corresponding artificial substrate RNA in yeast cells, the artificial substrate RNA became pseudouridylated at the predicted position ( Figure 2B, pink trace).
However, like vertebrate 18S rRNA, yeast 18S rRNA was not modified by SNORA18, even if its ASE was yeast-optimized and it was then overexpressed (

Heat shock does not induce 18S rRNA pseudouridylation
All snoRNAs that functioned on artificial substrates but not on endogenous rRNAs have somewhat imperfect base-pairing between their ASEs and the predicted target sequences (Figures 1B,2A). For example, yeast Pus7p recognized and modified a sequence that is different from the canonical Pus7p-specific motif ( Figure 1B). We hypothesized that such unusual activities might play regulatory roles and corresponding modifications might be induced on endogenous RNAs by stress. In fact, under stress conditions yeast snR81 interacts with a slightly diverged sequence to mediate stress-inducible modification of U2 snRNA (Wu et al. 2011). It has been shown that in response to heat shock, Pus7p induces RNA pseudouridylation at certain positions in U2 snRNA and mRNAs (Wu et al. 2011;Schwartz et al. 2014). However, pseudouridylation of 18S rRNA at position 1585, a modification we expected to be catalyzed by yeast Pus7p, was not reported for yeast exposed to a short one-hour heat shock at 45°C Begik et al. 2021). We wondered if newly synthesized rRNA will gain additional modifications upon long exposure to heat. We compared pseudouridylation patterns of the 3'-terminal 400 nucleotides in yeast 18S rRNA isolated from wild type cells cultured at 30°C and at 37°C for over 24 hrs. We did not detect any differences between these samples ( In the same 400-nt 3'-terminal region of 18S rRNA, we predicted the potential target for SNORA18. We grew yeast cells transformed with the expression construct for a yeast optimized variant of SNORA18 (Figure 2A, C-to-U mutant) at 37°C to a high cell density (OD 600 ~ 5-6).
These growing conditions did not induce modification activity of the yeast optimized SNORA18 rRNAs). X. laevis SNORA28 contains the corresponding ASE, yet we observed the lack of this modification in X. laevis 18S rRNA. In fact, this ASE is somewhat imperfect in xlSNORA28.
When this snoRNA was tested in the yeast cell system, it showed modification activity on a fragment of rRNA inserted in an artificial backbone RNA, but could not modify yeast 18S rRNA at position 808 (Deryusheva et al. 2020). We noticed that the lack of pseudouridylation of 18S rRNA at position 828 in three amphibian species, X. laevis, axolotl and newt, but not in X.
tropicalis (much warmer climate), correlated with the environmental conditions in which these D e r y u s h e v a , T a l r o s s a n d G a l l 1 1 species normally live (Deryusheva et al. 2020). We speculated that higher temperature might enhance modification activity of the imperfect pseudouridylation pocket of xlSNORA28.
We grew yeast cells transformed with xlSNORA28 expression constructs at 37°C. In these conditions xlSNORA28 still could not induce pseudouridylation of yeast 18S-U808 ( Figure 3B).
Then we cultured the X. laevis cell line XTC at 29.5°C; importantly, above this temperature these cells start dying in a few hours. 18S rRNA from the heat stressed XTC cells did not have additional pseudouridine at position 828, the potential target for xlSNORA28 ( Figure 3C). Additionally, we analyzed 18S rRNA from axolotl embryos raised at different temperatures between 16° and 29°C until they were about to hatch; embryos that developed at 29°C never hatched. As we mentioned above, in this species 18S rRNA is missing pseudouridylation at position 828. Axolotls are more temperature sensitive than frogs, but their 18S rRNA modification patterns still showed no pseudouridylation at position 828 in the heat shock conditions ( Figure 3D). Thus, taken together our data suggest that heat does not facilitate pseudouridylation activity of imperfect snoRNAs on 18S rRNA in yeast and vertebrate cells.  (Schattner et al. 2006). The configuration of the pseudouridylation pocket for this modification is not canonical: it has three target nucleotides within the pocket instead of two ( Figure 4A). We tested SNORA57 modification activity on yeast 18S rRNA, which does not have pseudouridine at the equivalent position 947 ( Figure 4B, black trace). When we expressed human SNORA57 in yeast cells, we did not detect pseudouridylation of position 947 in yeast 18S rRNA ( Figure 4B, red trace). Perhaps the yeast 18S rRNA sequence is slightly diverged at the target position, which makes SNORA57 nonfunctional on endogenous rRNA. Therefore, we made a yeast optimized version of SNORA57 ( Figure

Another layer of complexity in higher eukaryotes
In yeast, a single guide RNA, snR81 modifies both 25S rRNA and spliceosomal U2 snRNA (Ma et al. 2005;Wu et al. 2011). Yet, in higher eukaryotes, a specialized class of modification guide RNAs is involved in endogenous spliceosomal U snRNA modification. These are the so-called scaRNAs, concentrated in the nuclear Cajal bodies (CBs). In our previous study we showed that both snoRNAs and scaRNAs can modify certain positions in rRNAs (Deryusheva and Gall 2019a). However, it is still an open question whether scaRNAs and snoRNAs are interchangeable in the case of endogenous snRNA modification in higher eukaryotes.
To assess the ability of a typical snoRNA to modify U2 snRNA in a vertebrate cell system we took advantage of the differences between Xenopus and mammalian U2 snRNA modification patterns. Pseudouridine at position 60 has been reported in human ( Figure 5B, bottom grey trace) and rodent U2 snRNA (Deryusheva et al. 2012;Deryusheva and Gall 2017), although in Xenopus this position is not modified ( Figure 5B, green trace). We generated artificial U2-Ψ60 guide RNAs from X. tropicalis SCARNA4 and SNORA28 by ASE replacement in their U2-Ψ41 and 18S-Ψ778 pockets, respectively. Importantly, the configuration of U2-Ψ60 pockets was identical in both constructs ( Figure 5A). These two artificial guide RNAs, which could target As we mentioned above, non-canonical ASEs are typical for snoRNAs that showed guide RNA modification activities on artificial substrates but not on the same sequences within endogenous RNA molecules. Such imperfect snoRNA interaction with substrate RNA is essential for stress inducible modification of U2 snRNA by yeast snR81 (Wu et al. 2011

Expression constructs for snoRNAs and artificial substrate RNAs
To express human SNORA24 and modified X. tropicalis SNORA28 and SCARNA4 in the X.
laevis XTC cell line, fragments of these snoRNA host genes were amplified from genomic DNAs and cloned into the pCS2 vector. Overlap extension PCR was used to replace pseudouridylation pockets in SNORA28 and SCARNA4. Constructs for expression of vertebrate snoRNAs in yeast were generated by cloning corresponding snoRNA coding sequences into the YEplac181 vector, which contains a GPD promoter, an RNT1 cleavage site and an snR13 terminator (Huang et al. 2011). This vector was generously provided by Yi-Tao Yu, University of Rochester Medical Center. PCR-based mutagenesis was used to make yeast-optimized versions of vertebrate snoRNAs.
Artificial substrate RNA constructs were made as previously described Gall 2013, 2019a The annotated protein coding sequence for X. tropicalis pseudouridine synthase Pus7 was amplified from total RNA using the OneTaq One-step RT-PCR kit (New England Biolabs).
XmaI and XhoI restriction sites were added with oligonucleotides and the resulting fragment was cloned into the p415Gal1 vector.

RNA extraction and ectopic expression analysis
RNA was extracted from control and transfected cells using the Trizol reagent in the case of vertebrate cells or hot acid phenol in the case of yeast cells. RNA was purified using the Directzol RNA miniprep kit (Zymo Research). Proper processing and expression levels of exogenous RNAs were analyzed using northern blotting as described (Deryusheva and Gall 2013). To detect pseudouridines and 2'-O-methylated residues, we used fluorescent primer extensionbased techniques as described (Deryusheva and Gall 2009;Deryusheva et al. 2012