Slo2 potassium channel function depends on RNA editing-regulated expression of a SCYL1 protein

Reviewer #1:

1) My major comment is on the work that extrapolates the nematode findings to the human Slack channel. The latter has previously been shows to differ from the C-elegans channel in several ways. This interaction with the human channel is an important aspect of the presented work. Examination of the data in Figure 11, however, suggest that the gating of the human channel is quite different from the Slo2 channel records shown in Figure 8. Specifically (if the time bars shown in Figure 11A are correct, the overall open probability of the human channel is dominated by the presence of very long closed states (lasting many seconds). Examination of the X-axis scale bars in Figure 11B and C suggest these may have been excluded from the analyses. The conclusions on the effects of SCYL1 on open probability stated in the last paragraph of the Results section may not be completely valid until these are taken into account.

We thank the reviewer for the comment. The open probability in the original Figure 11 was calculated from all the open and closed events. As the reviewer pointed out, events of the very long closed state had a major impact on the open probability. However, they had little contribution to the dwell time histograms in the original figure because these events account for a tiny (nearly negligible) portion of the total event counts and the majority of them fell outside of the x-axis displayed in the figure. In the revised manuscript, we added a new panel (panel D) to Figure 11 to show dwell time histograms of the very long closed events (>30 ms, chosen arbitrarily as the threshold) for both groups. Because these events were scarce, our approach used for comparing τ and A values in panels B and C of this figure and of Figure 8 could not be applied to them. Therefore, we compared the average duration and frequency of these very long closed events between hSlo2.2 and hSlo2.2+SCYL1 (Figure 11D, bottom). We also have replaced the traces in Figure 11A with longer and more representative ones. In addition, the dwell time histograms in panels B and C of both Figures 8 and Figure 11, which were based on a single sample recording trace in the original figures, have been revised to include events from all the recordings to be consistent with the histogram in the new panel D. In the revised manuscript, we describe the presence of the very long closed state and its impact on the open probability (–subsection “scyl-1 expression depends on RNA editing at a specific 3’-UTR site”). The legends of Figure 8 and Figure 11 have also been updated accordingly.

2) The bimolecular fluorescence complementation assay presented does make a case that there may be a physical interaction of SCLY-1 with SLO-2 but is not completely definitive. It would be good to have some other indicator of physical interaction to support this claim. A more conventional coimmunoprecipitation experiments would be a good addition, one that could perhaps be carried out using the co-expression Xenopus oocyte heterologous expression system.

We performed coimmunoprecipitation assays with transgenic worms expressing HA-tagged SCYL-1 and GFP-tagged SLO-2. Worms instead of a heterologous expression system were used for this experiment because SLO-2 is poorly expressed in both Xenopus oocytes and HEK293 cells even with a codon-optimized version (our unpublished observation). Our results of the co-IP assays were in agreement with those of the BiFC assays (Figure 9), which provides further evidence for physical interactions between SLO-2 and SCYL-1.

Reviewer #2:

Comments

1) Please clarify the functional relationship between adr-1 and adr-2. Indeed, in the Introducution the authors seem to indicate that ADAR proteins could compete ("altering the accessibility") for certain binding sites. In this model, wouldn't one expect that adr-1(lf) would "free" access to the scyl-1 site, which would be inconsistent with the similarity of adr-1 and adr-2 mutant phenotypes?

Do the authors think that ADR-1 promotes the recruitement of ADR-2 to the scyl-1 3'UTR?

We thank the reviewer for the comment. Indeed, our description about the functional relationship between ADR-1 and ADR-2 in the Introduction was inaccurate. ADR-1 promotes RNA editing by binding to target mRNAs and recruiting ADR-2 to some specific editing sites. Our results with adr-1(lf) and adr-2(lf) mutants are compatible with this knowledge. In the revised manuscript, a sentence in the Introduction has been changed to “ADR-1 is catalytically inactive but can promote RNA editing by binding to selected target mRNA and tethering ADR-2 to RNA substrates (Ganem et al., 2019; Rajendren et al., 2018; Washburn et al., 2014).” (Introduction)

2) I was very intrigued by the results described in Figure 10, and specifically Figure 10D/E. The result is not what I had intuitively expected. Since it was performed in an adr-1/2 wild-type background as far as I could determine from the Materials and methods section, I would have thought that ADAR activity would have edited the wild-type sequence and I would have expected to see GFP in both cases.

I would have performed this experiment in an adr-1(lf) background. Have the authors attempted this experiment in this background? How many independent lines were tested with the wp1923 construct? Is expression seen anywhere else in the nervous system (outside de VNC) to make sure that GFP can indeed be expressed from this transgene?

We obtained five wp1923 lines and three wp1924 lines. While GFP signal was observed in all the wp1924 lines, no GFP signal was detected anywhere in any of the wp1923 lines. We were also surprised by the lack of GFP expression in the wp1923 lines. As suggested, we performed the same experiments as those in Figure 10E with the adr-1(zw96) mutant. Specifically, adr-1(zw96) mutant strains expressing the wp1923 transgene were made by injecting the wp1923 plasmid into the mutant whereas that expressing the wp1924 transgene by first integrating the wp1924 transgene into the wild-type genome and then crossing it into adr-1(zw96). As expected, GFP signal was not detected in the adr-1(zw96) mutant worms harboring the wp1923 transgene (not shown). Interestingly, GFP signal from the wp1924 transgene was much weaker (by ~50%) in the mutant strain than the wild-type strain (Figure 10 F and G). Because scyl-1 3’-UTR had already been “edited” at the ADR-1-dependent editing site in the wp1924 transgene, the difference in GFP signal between wild-type and adr-1(zw96) strains suggests that ADR-1 can also regulate scyl-1 mRNA level through a post-editing effect, perhaps through interacting with some other proteins. In agreement with this possibility, we were able to isolate several mutants showing decreased GFP signal from the wp1924 transgene in a pilot genetic screen (Figure 10—figure supplement 1).

3) Following on this previous point, a direct way to demonstrate the functional importance of this editing site would be to generate the following genotype by engineering the A>G mutation by CRISPR/Cas9 gene editing:slo-2(gf); adr-1(0); scyl-1(A>G)

My prediction would be that the mutation in scyl-1 would restore the slo-2 gain-of-function locomotor impairement, by bypassing the adr-1 requirement. To me this experiment would very strongly support this new and exciting functional regulation and I would encourage the authors to perform this rather simple experiment.

As suggested by the reviewer, we tried to engineer the A>G mutation at the ADR-1-dependent editing site in scyl-1 3’-UTR in the worm genome by CRISPR/Cas9. Unfortunately, our attempt was unsuccessful, probably because this editing site is located within an inverted sequence (Figure 10D). We then tried to address the reviewer’s comment by expressing Prab-3::scyl-1::unc-10 3’-UTR in slo-2(gf);adr-1(lf) double mutant and wild type (as a control) in the hope that expression of this transgene in the double mutant would restore the slo-2(gf) sluggish locomotion by bypassing the ADR-1 requirement. However, expression of the transgene caused larval arrest in the double mutant and a sluggish phenotype in wildtype worms, which prevented us from performing further analysis. Finally, we built a scyl-1(zw99);adr1(zw96) double mutant, and recorded whole-cell currents from VA5 motor neuron. We observed similar VA5 outward currents between the double mutant and the scyl-1(zw99) single mutant (Figure 7A), suggesting that SCYL-1 and ADR-1 likely contribute to SLO-2 function through a common pathway. This result provides further evidence for the putative role of ADR-1 in regulating SCYL-1 expression.

4) – In wormbase, the annotation of the scyl-1 locus shows a rather sizable 3'UTR. I was curious whether the authors have any comments on that point, and whether this is a common feature of ADAR-edited 3'UTRs.

Does the human SCYL1 3'UTR have a similar hair-pin structure, which could suggest a similar regulation mechanism?

In Wormbase, there are four different scyl-1 transcripts that differ only in the 3’-UTR sequence and length (167 nt, 169 nt, 564 nt, and 1862 nt). The identified ADR-1-dependent editing site exists only in the variant with the longest 3’UTR, which we speculate could be a mechanism for cell-specific expression of scyl-1. It is difficult to comment on whether RNA editing by ADARs at 3’-UTRs is a common mechanism of regulating mRNA expression based on the rather limited literatures. The 3’-UTR of human SCYL1 transcripts (NM_020680.4) also has a high probability of forming hair-pin structures based on software prediction (https://rna.urmc.rochester.edu/RNAstructureWeb/Servers). It remains to be determined whether human SCYL1 transcripts are also edited at the 3’-UTR, and if so, whether the editing regulates SCYL1 transcript expression. We added these comments to the Discussion section in the revised manuscript –.

5) I was a bit surprised about the strategy used to generate the scyl-1 expression pattern. Why was the in vivo recombination approach used? What happens when the 0.5kb promoter-GFP construct is injected alone?

The previous gene is only approx 2kb upstream and there is significant sequence conservation to C. remanei and C. briggsae DNA less than 1kb upstream of the ATG (see UCSC genome browser for example). Did the authors test such a 2kb promoter fragment?

We initially tried to examine scyl-1 expression pattern using a 2-kb Pscyl-1 because, as the reviewer pointed out, another gene (lap-2) resides approximately 2 kb upstream of scyl-1. However, GFP signal was not detected in transgenic worms expressing the Pscyl-1(2 kb)::gfp transcriptional fusion. We therefore resorted to the commonly used in vivo homologous recombination approach to include potential distant upstream regulatory elements in the Pscyl-1::gfp transcriptional fusion. A description about the unsuccessful experiment with the 2-kb Pscyl-1 has been added to the manuscript (–subsection “ADR-1 regulates SLO-2 function through SCYL-1”). As suggested by the reviewer, we also created transgenic worms expressing only the Pscyl-1(0.5 kb)::gfp transcriptional fusion but were unable to detect any GFP signal from them (not shown).

6) Figure 6: I'm not convinced that scyl-1::GFP labels vm1 or vm2 muscles based on this image. I could be wrong, but higher magnification images would need to be checked.

We thank the reviewer for pointing out our mistake. In Figure 6, Pscyl-1::GFP labels uterine ventral cells rather than vulval muscle cells.