Author response:
Reviewer #1 (Public review):
Summary:
The authors use high-throughput gene editing technology in larval zebrafish to address whether microexons play important roles in the development and functional output of larval circuits. They find that individual microexon deletions rarely impact behavior, brain morphology, or activity, and raise the possibility that behavioral dysregulation occurs only with more global loss of microexon splicing regulation. Other possibilities exist: perhaps microexon splicing is more critical for later stages of brain development, perhaps microexon splicing is more critical in mammals, or perhaps the behavioral phenotypes observed when microexon splicing is lost are associated with loss of splicing in only a few genes.
A few questions remain:
(1) What is the behavioral consequence for loss of srrm4 and/or loss-of-function mutations in other genes encoding microexon splicing machinery in zebrafish?
It is established that srrm4 mutants have no overt morphological phenotypes and are not visually impaired (Ciampi et al., 2022).
We chose not to generate and characterize the behavior and brain activity of srrm4 mutants for two reasons: 1) we were aware of two other labs in the zebrafish community that had generated srrm4 mutants (Ciampi et al., 2022 and Gupta et al., 2024, https://doi.org/10.1101/2024.11.29.626094; Lopez-Blanch et al., 2024, https://doi.org/10.1101/2024.10.23.619860), and 2) we were far more interested in determining the importance of individual microexons to protein function, rather than loss of the entire splicing program. Microexon inclusion can be controlled by different splicing regulators, such as srrm3 (Ciampi et al., 2022) and possibly other unknown factors. Genetic compensation in srrm4 mutants could also result in microexons still being included through actions of other splicing regulators, complicating the analysis of these regulators. We mention srrm4 in the manuscript to point out that some selected microexons are adjacent to regulatory elements expected of this pathway. We did not, however, choose microexons to mutate based on whether they were regulated by srrm4, making the characterization of srrm4 mutants disconnected from our overarching project goal.
We are coordinating our publication with Lopez-Blanch et al. (https://doi.org/10.1101/2024.10.23.619860), which shows that srrm4 mutants also have minimal behavioral phenotypes.
(2) What is the consequence of loss-of-function in microexon splicing genes on splicing of the genes studied (especially those for which phenotypes were observed).
We acknowledge that unexpected changes to the mRNA could occur following microexon removal. In particular, all regulatory elements should be removed from the region surrounding the microexon, as any remaining elements could drive the inclusion of unexpected exons that result in premature stop codons.
First, we will clarify our generated mutant alleles by adding a figure that details the location of the gRNA cut sites in relation to the microexon, its predicted regulatory elements, and its neighboring exons.
Second, we will experimentally determine whether the mRNA was modified as expected for a subset of mutants with phenotypes.
Third, we will further emphasize in the manuscript that these observed phenotypes are extremely mild compared to those observed in over one hundred protein-truncating mutations we have assessed in previous and ongoing work. We currently show one mutant, tcf7l2, which we consider to have strong neural phenotypes, and we will expand this comparison in the revision. In our study of 132 genes linked to schizophrenia (Thyme et al., 2019), we established a signal cut-off for whether a mutant would be designated as having a neural phenotype, and we classify this set of microexon mutants in this context. Far stronger phenotypes are expected of loss-of-function alleles for microexon-containing genes, as we showed in Figure S1 of this manuscript in addition to our published work.
(3) For the microexons whose loss is associated with substantial behavioral, morphological, or activity changes, are the same changes observed in loss-of-function mutants for these genes?
We had already included two explicit comparisons of microexon loss with a standard loss-of-function allele, one with a phenotype and one without, in Figure S1 of this manuscript. We will make the conclusions and data in this figure more obvious in the main text.
Beyond the two pairs we had included, Lopez-Blanch et al. (https://doi.org/10.1101/2024.10.23.619860) describes mild behavioral phenotypes for a microexon removal for kif1b, and we already show developmental abnormalities for the kif1b loss-of-function allele (Figure S1).
Additionally, we can draw expected conclusions from the literature, as some genes with our microexon mutations have been studied as typical mutants in zebrafish or mice. We will modify our manuscript to include a discussion of these mutants.
(4) Do "microexon mutations" presented here result in the precise loss of those microexons from the mRNA sequence? E.g. are there other impacts on mRNA sequence or abundance?
See response to point 2. We will experimentally determine whether the mRNA was modified as expected for a subset of mutants with phenotypes.
(5) Microexons with a "canonical layout" (containing TGC / UC repeats) were selected based on the likelihood that they are regulated by srrm4. Are there other parallel pathways important for regulating the inclusion of microexons? Is it possible to speculate on whether they might be more important in zebrafish or in the case of early brain development?
The microexons were not selected based on the likelihood that they were regulated by srrm4. We will clarify the manuscript regarding this point. There are parallel pathways that can control the inclusion of microexons, such as srrm3 (Ciampi et al., 2022). It is well-known that loss of srrm3 has stronger impacts on zebrafish development than srrm4 (Ciampi et al., 2022). The goal of our work was not to investigate these splicing regulators, but instead was to determine the individual importance of these highly conserved protein changes.
Strengths:
(1) The authors provide a qualitative analysis of splicing plasticity for microexons during early zebrafish development.
(2) The authors provide comprehensive phenotyping of microexon mutants, addressing the role of individual microexons in the regulation of brain morphology, activity, and behavior.
We thank the reviewer for their support. The pErk brain activity mapping method is highly sensitive, significantly minimizing the likelihood that the field has simply not looked hard enough for a neural phenotype in these microexon mutants. In our published work (Thyme et al., 2019), we show that brain activity can be drastically impacted without manifesting in differences in those behaviors assessed in a typical larval screen (e.g., tcf4, cnnm2, and more).
Weaknesses:
(1) It is difficult to interpret the largely negative findings reported in this paper without knowing how the loss of srrm4 affects brain activity, morphology, and behavior in zebrafish.
See response to point 1.
(2) The authors do not present experiments directly testing the effects of their mutations on RNA splicing/abundance.
See response to point 3.
(3) A comparison between loss-of-function phenotypes and loss-of-microexon splicing phenotypes could help interpret the findings from positive hits.
See response to point 2.
Reviewer #2 (Public review):
Summary:
The manuscript from Calhoun et al. uses a well-established screening protocol to investigate the functions of microexons in zebrafish neurodevelopment. Microexons have gained prominence recently due to their enriched expression in neural tissues and misregulation in autism spectrum disease. However, screening of microexon functionality has thus far been limited in scope. The authors address this lack of knowledge by establishing zebrafish microexon CRISPR deletion lines for 45 microexons chosen in genes likely to play a role in CNS development. Using their high throughput protocol to test larval behaviour, brain activity, and brain structure, a modest group of 9 deletion lines was revealed to have neurodevelopmental functions, including 2 previously known to be functionally important.
Strengths:
(1) This work advances the state of knowledge in the microexon field and represents a starting point for future detailed investigations of the function of 7 microexons.
(2) The phenotypic analysis using high-throughput approaches is sound and provides invaluable data.
We thank the reviewer for their support.
Weaknesses:
(1) There is not enough information on the exact nature of the deletion for each microexon.
To clarify the nature of our mutant alleles, we will add a figure that details the location of the gRNA cut sites in relation to the microexon, its predicted regulatory elements, and its neighboring exons.
(2) Only one deletion is phenotypically analysed, leaving space for the phenotype observed to be due to sequence modifications independent of the microexon itself.
We will experimentally determine whether the mRNA is impacted in unanticipated ways for a subset of mutants with mild phenotypes (see the point 2 response to reviewer 1). We also have already compared the microexon removal to a loss-of-function mutant for two lines (Figure S1), and we will make that outcome more obvious as well as increasing the discussion of the expected phenotypes from typical loss-of-function mutants (see point 3 response to reviewer 1).
In addition, our findings for three microexon mutants (ap1g1, vav2, and vti1a) are corroborated by Lopez-Blanch et al. (https://doi.org/10.1101/2024.10.23.619860).
Unlike protein-coding truncations, clean removal of the microexon and its regulatory elements is unlikely to yield different phenotypic outcomes if independent lines are generated (with the exception of genetic background effects). When generating a protein-truncating allele, the premature stop codon can have different locations and a varied impact on genetic compensation. In previous work (Capps et al., 2024), we have observed different amounts of nonsense-mediated decay-induced genetic compensation (El-Brolosy, et al., 2019) depending on the location of the mutation. As they lack variable premature stop codons (the expectation of a clean removal), two mutants for the same microexons should have equivalent impacts on the mRNA.
Reviewer #3 (Public review):
Summary:
This paper sought to understand how microexons influence early brain function. By selectively deleting a large number of conserved microexons and then phenotyping the mutants with behavior and brain activity assays, the authors find that most microexons have minimal effects on the global brain activity and broad behaviors of the larval fish-- although a few do have phenotypes.
Strengths:
The work takes full advantage of the scale that is afforded in zebrafish, generating a large mutant collection that is missing microexons and systematically phenotyping them with high throughput behaviour and brain activity assays. The work lays an important foundation for future studies that seek to uncover the likely subtle roles that single microexons will play in shaping development and behavior.
We thank the reviewer for their support.
Weaknesses:
The work does not make it clear enough what deleting the microexon means, i.e. is it a clean removal of the microexon only, or are large pieces of the intron being removed as well-- and if so how much? Similarly, for the microexon deletions that do yield phenotypes, it will be important to demonstrate that the full-length transcript levels are unaffected by the deletion. For example, deleting the microexon might have unexpected effects on splicing or expression levels of the rest of the transcript that are the actual cause of some of these phenotypes.
To clarify the nature of our mutant alleles, we will add a figure that details the location of the gRNA cut sites in relation to the microexon, its predicted regulatory elements, and its neighboring exons.
We will experimentally determine whether the mRNA is impacted in unanticipated ways for a subset of mutants with mild phenotypes (see the point 2 response to reviewer 1).
