Reviewer 1: Dr. Eugene V. Koonin
Reviewer's comments: Very interesting, solid work. The extensive alternative splicing yielding nonsense mRNAs of clock genes is demonstrated beyond doubt. The direct involvement of this phenomenon in clock regulation remains a hypothesis, and the authors are quite careful about. The only substantial question that I have is: would it be possible to compare the level of alternative splicing in clock genes to the overall background in Arabidopsis? Is it the case that the clock genes are particularly prone to alternative splicing in general and to accumulation of PTC in particular? An answer in the affirmative would extremely interesting and supportive of the regulation hypothesis but even a negative answer would be interesting.
Data from our previous transcriptome-wide RNA-seq study [REF [2
] was incorporated into the Arabidopsis TAIR10 genome annotation and estimates that approximately 42
% of Arabidopsis genes are alternatively spliced. Currently, there are about 29 characterized core clock and circadian-associated genes in the TAIR10 annotation. Approximately half of them (15) are represented by two or more gene models. Therefore, at first glance the extent of alternative splicing in circadian genes appears similar to transcriptome-wide estimates. However, in this study we did not seek to explore statistics of alternative splicing among circadian genes per se
. More importantly, we found that the patterns of alternative splicing in some key circadian genes can be very complex and in most cases generates nonsense isoforms. Moreover, circadian genes have a broad repertoire of alternative splicing including intron retention, alternative donor/acceptor sites, cassette exons, and alternatively spliced 5' and 3' UTRs. Thus, the central components of the clock oscillator CCA1
are represented by two and five gene models, respectively. Interestingly, we found that only one intron 4 retention event (I4R) introducing PTCs downstream of their DNA binding domains is conserved between CCA1
homologs. In addition, both the 5' and 3' UTRs of LHY
are alternatively spliced suggesting that some of its isoforms may be regulated by NMD. Indeed, after this paper was submitted we found that the LHY
isoform retaining an intron in its 3' UTR is up-regulated in NMD mutant (data not shown).
Concerning the question of whether the alternative splicing of the circadian genes generates more PTCs than average across the transcriptome, we have no statistically meaningful answer. However, the notion is that most of the alternative splicing events in circadian genes introduce PTCs and these events are likely to be at least as frequent as they are on average across the transcriptome. Not all PTCs will elicit mRNA degradation. Currently we are trying to determine which of the PTCs in circadian genes elicit NMD.
Reviewer 2: Dr. Chungoo Park
In this manuscript, the authors showed that Arabidopsis thaliana circadian clock genes tend to be extensively alternatively spliced, and such isoforms especially retaining intron are subject to contain premature termination codons and to be conserved. However, these findings have been already observed by Filichkin et al. (2010) [Ref [2
]. In this study, which is different from previous study?
Authors' response: Even though the results reported in this paper are broadly related to our previous studies they are clearly unique. Further, this study addresses a different set of fundamental biological questions related to the role of unproductive alternative splicing (UAS) in circadian clock regulation for the following reasons.
First, unlike our previous study, here we interrogated novel splice junctions (SJs) in a subset of circadian gene transcripts at specific times of day. To pinpoint the peak expression phase for each transcript we used our recently obtained diurnal microarray datasets [REF [2
]. In contrast to the pooled sampling scheme this approach added an important advantage of enabling detection of minor splice variants at peak expression times, which can be missed if timing is not correct.
Second, using a diurnal time course sampling we showed that splice isoforms harboring in-frame premature termination codons (PTCs) can accumulate in a cyclical manner and in many (but not all) instances mirror the oscillations of the productive transcripts. Importantly, our time course data suggests a temporal shift in isoform ratios depending on the time of day or environmental conditions such as temperature changes.
Third, here we demonstrated that the conservation of similar AS events can be observed among circadian-regulated homologs within the same species. In the previous study [REF [2
] we experimentally validated the majority of novel SJs predicted by RNA-seq for hundreds of randomly selected genes. Among these genes, CCA1
, a central component of the circadian clock presented an interesting case because its intron 4 retention event was conserved across different plant phyla. Here we show that such conservation can be extended to the homologs within the same species with overlapping but distinct functions such as in the cases of CCA1/LHY
. This notion further reinforces a hypothesis of the functional significance of UAS.
Finally, we showed that some circadian-associated genes such as RVE2
employ a strategy of introducing a PTC via a poison cassette exon (PCE) – a conserved UAS event previously shown to be involved in homeostatic regulation of some mammalian SR splicing factors [REF [8
]. To our knowledge this is the first example of such an AS event among circadian genes suggesting that the inclusion of a PTC via a PCE event may be a widespread mechanism, regulating in this instance homeostatic expression of a key circadian gene. Importantly, a time course analysis of the accumulation of RVE2 transcripts (Figure
) showed that the PCE isoform may not always mirror precisely the accumulation of its full-length protein-coding counterpart and suggests a possibility that the production of the PTC
transcript may undergo a phase shift under specific diurnal conditions. It is possible that similar to the case of some SR splicing factors, RVE2 expression is regulated by coupling of UAS with mRNA decay.
Combined, the approaches described above allowed us to detect novel AS events in circadian genes which otherwise could fall below statistical significance in RNA-seq analysis because of low isoform abundance at specific times of day and/or environmental conditions resulting in insufficient read coverage. A systematic interrogation of the putative novel splicing events allowed us to validate experimentally that AS of key circadian genes in most cases generates isoforms harboring PTCs. Future studies that explicitly demonstrate regulatory role(s) of UAS in clock function are likely to have major biological implications because the circadian clock orchestrates global gene expression in plants.
Reviewer's comments: One of main assertions in this study is that most circadian clock genes have alternative splicing forms. Whether this pattern is circadian gene-specific or most plant genes are extensively alternatively spliced is unclear.
Authors' response: Please see our response to the similar question from reviewer #1.
Reviewer's comments: To rule out noise hypothesis, abundance of unproductive AS among eukaryotes and conservation of alternative splicing events in relative species were mentioned. To this end, the authors should show direct evidences using plants and circadian genes tested.
The "stochastic splicing noise” versus regulated unproductive AS hypothesis is mentioned in the Discussion section of our manuscript. Rapidly accumulating evidence [for examples see references [8
] indicates that unproductive alternative splicing (UAS) coupled with NMD is a regulatory gene expression mechanism that is widespread across eukaryotes. Even though our study suggests that UAS could be an important regulatory mechanism for several circadian genes it is not designed to directly resolve this overarching biological question. The references with the examples supporting "conservation of AS events between different species" have been added to the Discussion section.
Reviewer 3: Dr. Marcelo Yanovsky
Reviewer's comments: Recent work has provided evidence of widespread occurrence of alternative splicing (AS) in plants. Interestingly, genetic approaches have revealed an important role for AS in the proper regulation of circadian rhythms in plants. The manuscript by Filichkin and Mockler contributes to expand our knowledge of the interplay between AS and circadian networks through a thorough evaluation of AS of plant circadian clock genes using RNA-seq. The authors found that many circadian clock-associated genes in Arabidopsis thaliana are alternatively spliced. The results from RNA-seq data were confirmed using several approaches including RT-PCR, qRT-PCR and/or Sanger sequencing. Most AS events led to the incorporation of in-frame premature stop codons (PTC) through full or partial intron retention. This was a widespread phenomenon among mRNAs from the CCA1/LHY family of MYB transcription factors. Interestingly, an in-frame PTC also resulted from the inclusion of a "poison cassette exon" in one of the homologues of CCA1 known as REVEILLE 2 (RVE2). Intron retention events in the MYB family of transcription factors associated with the clock also took place in the 3´utr region, suggesting they may contribute to regulate mRNA stability. Interestingly, daily oscillations with different phases were observed for the different isoforms of the RVE2 gene. Finally, the authors confirmed and extended previous analysis of the effect of stress treatments on AS of CCA1, showing that the full length transcripts increases in response to cold treatments, whilst the isoform retaining intron 4 increases in response to heat, and these responses are sustained under prolonged stress conditions. The dependence of relative abundance of certain isoforms of clock associated genes on time of day or stress treatments suggest that AS of these genes may contribute to fine-tuning the regulation of physiological processes by the clock, as well as regulation of the clock in response to environmental signals.
Reviewer's comments: The manuscript is well written, clear and concise. The data is interesting and presented in a timely fashion for the field, reinforcing the importance of the interplay between AS and the regulation of circadian networks, through changes in AS of core-clock and clock-output genes.
Reviewer's comments: As far as data presentation is concerned, it would have been nice to see as part of the main figures, in addition to read coverage graphs that allow the identification of intron retention or exon skipping events, graphic data on exon/exon junctions, which should allow the identification of alternative donor and/or acceptor sites.
The graphical illustrations and criteria for coverage of exon/exon junctions by Illumina reads used for the identification of alternative donor and/or acceptor sites are shown in Supplementary Figure 7 of our previous publication [Ref [2
Reviewer's comments: Finally, in the near future, it will be important to start testing the functional significance of the different AS isoforms of the clock genes, and the role of AS in the proper physiological regulation of the circadian network.
Authors' response: Indeed, this is very important point. Unproductive alternative splicing (UAS) in circadian clock genes presents a very special case study because the daily oscillations of transcripts levels require rapid and steep adjustments in mRNA levels. Currently, studies of the role of AS in regulation of the circadian network are under way in many labs worldwide using an array of mutant and transgenic lines.
The authors mention at the beginning that “To maximize the sensitivity for minor spliced products and to avoid a possible light/dark-dependent shift in isoform ratios, the sampling was done at the phase of peak expression for each tested genes as shown in Figure
”. Although this approach may simplify the analysis, I do not think it will maximize the sensitivity for minor spliced products. It is possible for instance that the relative abundance of different isoforms changes throughout the day, as shown by the authors for RVE2
, and therefore this approach may lead to reduced rather than increased sensitivity.
Authors' response: We found that this approach works in most cases but we agree that for some genes under specific conditions the effect could be opposite: the unproductive isoform may exhibit different behavior than its productive counterpart. This is why we conducted a time course with limited points here (i.e. RVE2) and employ a full time course strategy in the follow up studies.
Reviewer's comments: Indeed, I believe the present description of AS events among plant clock associated genes underestimates the total number of events. For example, additional isoforms of PRR9 have been reported, besides those associated with retention of intron 3, which were not detected/reported here.
We agree and clearly state in the paper that our study is likely to underestimate the total number of AS and UAS events in circadian genes. The Supplemental Figure
, for example, demonstrates that differential intron expression analysis suggests additional intron retention events under heat stress. Undoubtedly, future studies of abiotic or biotic stress treatments, or specific mutations in splicing-related genes will reveal more novel AS patterns.
Finally, in Figure
, the authors should add the primer that allows amplification of the fully spliced isoform, in the 3´-5´direction.
the primer pairs used for amplification of the fully spliced isoforms have been added.