Casting CRISPR-Cas13d to fish for microprotein functions in animal development

Summary

Protein coding genes were originally identified with sequence-based definitions that included a 100-codon cutoff to avoid annotating irrelevant open reading frames. However, many active proteins contain less than 100 amino acids. Indeed, functional genetics, ribosome profiling, and proteomic profiling have identified many short, translated open reading frames, including those with biologically active peptide products (microproteins). Yet, functions for most of these peptide products remain unknown. Because microproteins often act as key signals or fine-tune processes, animal development has already revealed functions for a handful of microproteins and provides an ideal context to uncover additional microprotein functions. However, many mRNAs during early development are maternally provided and hinder targeted mutagenesis approaches to characterize developmental microprotein functions. The recently established, RNA-targeting CRISPR-Cas13d system in zebrafish overcomes this barrier and produces potent knockdown of targeted mRNA, including maternally provided mRNA, and enables flexible, efficient interrogation of microprotein functions in animal development.

Graphical abstract

Genetics; Developmental biology; Complex system biology

Pioneering the discovery of eukaryotic microproteins

At the onset of the genomics era, stringent sequence-based gene definitions were adopted to avoid annotating biologically irrelevant open reading frames (ORFs)—especially short ORFs (≤ 100 codons long) that exhibit a high frequency based on probability alone. For example, in zebrafish (Danio rerio) there are over 425,000 ORFs with canonical (AUG) start codons between 10 and 100 codons within messenger RNAs (mRNAs) and long non-coding RNAs alone. Only approximately 1,000 (0.02%) of these short ORFs are currently described as protein coding, leaving countless short ORFs with the potential to be translated as “beautiful needles in the haystack”¹ (Figure 1A).

Figure 1.

Finding needles in the haystack: Short open reading frame translation is widespread across diverse taxa

(A) Rendition of a histogram representing all transcribed open reading frames (ORFs) in a single genome binned by length and categorized into short ORFs with translation potential (yellow haystack, left axis) and annotated proteins (blue books, right axis). Needle illustrates a short ORF that encodes a biologically active microprotein. Note: Left axis has a 100x scale compared to the right axis for ease of visualization. Based on¹

(B) Ribosome profiling, proteomics, and functional genetics have cooperated to reveal widespread short ORF translation.

(C) 1. Translation of upstream (in mRNA sequence once called the 5′ UTR) short ORFs (uORFs) represses translation of the canonical ORF, but the list of uORFs that encode functional microproteins continues to grow. 2. Downstream (within sequence once called the 3′ UTR) short ORFs (dORFs) translation enhances canonical ORF translation. It remains unknown whether dORFs encode active microproteins. 3. Short ORFs that overlap with a canonical ORF (oORFs) are intrinsically tricky to interrogate, but a handful produce active microproteins. 4. RNAs that were initially annotated as non-coding (e.g., long non-coding RNAs, circular RNAs) contain short, translated ORFs (microprotein ORFs) that produce microproteins with diverse functions.

Functional genetics in budding yeast first confirmed the existence of eukaryotic “needles” upstream of the canonical ORF encoding GCN4,² a transcription factor that activates genes involved in amino acid biosynthesis.³ Translation of four short, upstream (within mRNA sequence once defined as the 5′ untranslated region (UTR)) ORFs (uORFs) in GCN4 is required for nutrient-dependent regulation of GCN4 translation but relies on the uORF’s ability to engage ribosomes rather than a biologically active peptide product. However, subsequent functional genetic studies revealed short ORFs that encode biologically active peptide products (from here on: microproteins) in diverse organisms from budding yeast⁴ to legumes⁵ and insects^6,7,8 (Figure 1B). Notably, all initially discovered microproteins are encoded on polycistronic mRNAs, and parallel bioinformatic analyses revealed that many mRNAs contain uORFs,^{9,10,11,12,13,14,15} including a population that shows amino acid conservation in mammals.⁹ This evidence collectively questioned the fundamental assumption that eukaryotic mRNAs are translated into a single peptide product.

Further supporting initial functional genetics and bioinformatics, the advent of ribosome profiling^16,17,18 and targeted proteomic approaches^{19,20,21,22,23,24,25} over the last decade has revolutionized our ability to detect short ORF translation. Ribosome profiling, for example, leverages that ribosomes protect short fragments of bound mRNAs which are amenable to isolation and next-generation sequencing. Remarkably, some methods yield ribosome protected fragments that are consistent enough to define the reading frame¹⁸ (Figure 1B). Complementary detection of short ORF encoded peptide products with targeted proteomics has supported widespread translation of short ORFs on polycistronic mRNAs across many eukaryotic taxa^{16,17,21,23,24,26,27,28,29,30,31,32,33,34,35,36,37,38} (Figure 1B). Omics-based approaches have consequently completed the paradigm shift in RNA biology—many eukaryotic mRNAs exhibit multiple, distinct translation events rather than producing a single peptide product.

Furthermore, omics techniques have defined diverse classes of translated short ORFs and polycistronic mRNA structures (Figure 1C, reviewed extensively in^39,40,41,42). For instance (and like those in GCN4), translation of uORFs often acts to suppress translation of the canonical coding sequence.^{2,12,43,44,45,46} On the other hand, translation of downstream short ORFs (dORFs) (within mRNA sequence once called the 3′ UTR) enhances canonical ORF translation.³⁸ Although many uORFs and dORFs regulate translation independently of microprotein activity, the catalog of uORFs that encode functional microproteins continues to grow.^4,33,47,48 In fact, even short ORFs nested within canonical coding sequences (overlapping ORFs) are translated into functional microproteins.^49,50 In addition, many short ORFs are translated from novel mRNAs (previously annotated as non-coding) and circular RNAs, with a handful known to encode biologically active microproteins.^{27,28,29,33,51,52,53,54,55,56,57,58,59,60,61,62} However, although techniques continue to evolve and shine light on new short, translated ORFs, the potential functions encoded within the vast majority of identified short ORF peptide products remain unknown.

Microproteins are integral in diverse eukaryotic biology, including animal development

The arginine attenuator peptide (AAP) in budding yeast is one of the first functional microproteins identified in a eukaryote. AAP is a 25 amino acid microprotein produced from a uORF in the Saccharomyces cerevisiae CPA1 gene, which encodes an enzyme involved in arginine synthesis. AAP is required for free arginine to repress CPA1 translation, enabling arginine to negatively regulate its own biosynthesis.⁴ However, with a few notable exceptions^5,6,7,8, most functional microproteins have been discovered in the last ten years with the aid of ribosome profiling^16,17 and advances in proteomic and biochemical tools.^21,24,25,63 Although microproteins encoded in animal genomes will be the main focus of this review, microproteins also play critical and emerging roles in fungi,^4,12,64 bacteria^65,66,67 (reviewed in⁶⁸), and plants^5,69,70 (reviewed in^36,71).

In animals, microproteins play integral roles in diverse biological processes including transcription^56,72,73 (Figure 2A), ribosome biogenesis,⁴⁹ cell growth and viability,³³ cancer cell survival,³⁵ embryonic stem cell pluripotency⁷⁴ (Figure 2B) and differentiation,⁷⁵ immune responses^48,60,62 and both cardiac^34,53,54,76 (Figure 2B) and skeletal^27,28,29 muscle function and development (Figure 2C). In fact, developmental contexts have provided significant insights into microprotein functions^{6,7,8,29,52,53,74,75,76,77} (Figures 2A–2C). For instance, parallel developmental studies in Tribolium castaneum and Drosophila melanogaster identified the first animal microprotein gene called mille-pattes or tarsal-less/polished rice, respectively, that produces a polycistronic mRNA that encodes functionally redundant 11–32 amino acid microproteins.^6,7,8 Tarsal-less microproteins interact with Ubr3, an E3 ubiquitin ligase, and enable it target the transcription factor shavenbaby.^72,78 Consequent proteosome-mediated cleavage of the shavenbaby N-terminus switches it from a transcriptional repressor to a transcriptional activator, inducing the expression of genes critical for distal limb patterning in Drosophila^6,7 and abdominal segment identity in many other insects^8,73 (Figure 2A). Similarly, characterized microproteins often function as essential regulators of protein activity or key signaling molecules, supporting that uncharacterized short ORF peptide products likely have additional functions in orchestrating animal development.

Figure 2.

Established microproteins play critical roles in animal development

(A) Tarsal-less/polished rice/mille-pattes is conserved throughout insects and produces a polycistronic mRNA that encodes functionally redundant microproteins critical for regulating the activity of the transcription factor shavenbaby for its role in specifying distal limb identity in Drosophila melanogaster and abdominal segment identity in the larvae of many insects, including Tribolium castaneum.

(B) apela/elabela/toddler encodes a signaling molecule that is conserved from zebrafish to humans that is critical for mesodermal migration and heart formation in both zebrafish and mouse, resulting in a lethal phenotype. Loss of Apela/Elabela/Toddler function in mouse also affects placental development and maternal health. Furthermore, APELA/ELABELA/TODDLER is required for maintaining pluripotency in human embryonic stem cells.

(C) Myomixer/Minion encodes a conserved, membrane-tethered ligand that is both necessary and sufficient for the cellular fusion required for muscle cells to develop into multinucleated muscle fibers in mouse. Because Myomixer/Minion is necessary for the development of the diaphragm muscle, knockout mice fail to inflate their lungs, resulting in perinatal lethality

Short ORF translation is prevalent during vertebrate development

Early stages of vertebrate development and vertebrate pluripotent stem cells show widespread short ORF translation.^{17,32,33,38,54,79,80} Among many vertebrate model systems, zebrafish (D. rerio) has emerged as a premier model system because of its high offspring number, ex-utero development, and genetic tractability. Not to mention, zebrafish exhibit many conserved cellular, molecular, and genetic foundations of development,⁸¹ and more than 70% of human genes have at least one ortholog in zebrafish, making zebrafish studies highly informative for human biology.⁸² Critically, zebrafish exhibit hundreds of translated short ORFs across their first five days of development.^32,54 Although first identified in 2014, only twelve zebrafish developmentally translated short ORFs are currently characterized.^53,54,77,83 Targeted mutagenesis has revealed that four of these short ORFs produce functional microproteins within the first seven days of development. The pioneer vertebrate functional microprotein called apela (elabela, toddler) is a conserved signaling molecule that is critical for zebrafish mesodermal migration, heart formation in zebrafish and mouse,^53,54,76 pluripotency in human embryonic stem cells,⁷⁴ and placental development in mouse⁷⁶ (Figure 2B). Specifically in zebrafish, the microprotein bouncer localizes to the oocyte membrane and is necessary and sufficient for fertilization.⁸³ Two additional zebrafish microproteins encoded in linc-mipep and linc-wrb have been recently reported to coordinate to regulate locomotor activity in larvae (4–7 days post-fertilization), directly or indirectly changing chromatin accessibility in CNS cells called oligodendrocytes.⁷⁷ Still, hundreds of developmental short ORF encoded peptide products in zebrafish alone remain without insights into their functions.

Maternally provided RNAs impede microprotein characterization in early development

Animal development presents unique challenges to elucidating microprotein functions. During maturation, animal oocytes are packed with maternal RNAs and proteins. Immediately after fertilization, maternally provided RNAs and proteins control the developmental program while the zygote remains transcriptionally silent. As the zygote awakens its genome and ramps up production of its own transcripts, it is weaned from maternal control in a process known as the maternal-to-zygotic transition (reviewed extensively in⁸⁴). Still, the majority of RNAs during the maternal-to-zygotic transition are maternally provided and present a significant barrier to elucidating microprotein functions in early animal development (Figure 3A). Indeed, a developmental function for a microprotein encoded on a maternally provided mRNA has yet to be described. apela mRNA is transcribed de-novo from the zygotic genome, which facilitated its discovery with a zygotic mutant.^53,54 Similarly, linc-mipep and linc-wrb zygotic mutants show phenotypes that do not involve contributions from any maternal mRNA.⁷⁷

Figure 3.

Maternally provided RNA presents a significant barrier to targeted mutagenesis approaches in early animal development that is overcome with CRISPR-Cas13d

(A) Representation of RNA dynamics during early animal development (Based on⁸⁴). Traces show individual contributions from maternal (red) and zygotic (blue) RNA, whereas the rectangle above represents the cumulative RNA population over time. As the zygote generates its own RNAs, it takes control of its development through a process known as the maternal-to-zygotic transition. However, maternal RNAs still constitute a significant proportion of all RNAs throughout the maternal-to-zygotic transition.

(B) In a routine, targeted mutagenesis strategy, parents are maintained as heterozygotes for the loss-of-function allele (denoted by a yellow box with an orange star). With this approach, homozygous mutant offspring will possess wildtype maternal RNA in early development—known as zygotic mutants (left panel). To eliminate both maternal and zygotic RNA, the mother must be homozygous for the loss-of-function allele, but biallelic deletion of early developmental genes is often lethal. It is therefore necessary to generate a maternal line that is homozygous mutant in only its germline—a maternal mutant. If fertile, crossing this maternal mutant with a heterozygous male generates homozygous mutant offspring without wildtype maternal RNA—called maternal/zygotic mutants (right panel).

(C) Injection of CRISPR-Cas13d protein or mRNA with guide RNA in zebrafish produces potent degradation of both maternal and zygotic targeted RNAs, eliminating the need to generate maternal/zygotic mutants for loss-of-function analyses during the maternal-to-zygotic transition.

Elucidating microprotein functions in early development therefore requires either an over-expression approach or a loss-of-function strategy that targets both maternal and zygotic gene products to prevent maternal mRNA from masking any effect of a deletion in the offspring alone (Figure 3B). Targeted mutagenesis approaches (e.g., CRISPR-Cas9) therefore require the time-consuming generation of a dual maternal/zygotic mutant line, making it an inherently risky and expensive approach to interrogate the function of even a few developmental microproteins (Figure 3B). Alternatively, conventional RNAi is generally ineffective in zebrafish and other teleosts (e.g., medaka, killifish, Mexican tetra), which has led to widespread use of morpholinos as the preeminent knockdown technology. Although morpholinos have correctly elucidated gene function in select studies, the general utility of morpholinos has been questioned over off-target effects and general toxicity.⁸⁵ In some cases, loss-of-function mutant and morpholino-treated animals exhibit discordant phenotypes.^86,87 More strikingly, additional phenotypes have been observed on morpholino treatment in loss-of-function mutants,⁸⁸ indicating cellular effects unrelated to the targeted RNAs. In fact, some morpholinos can trigger cellular immune responses and off-target mis-splicing.^89,90 On top of these concerns, morpholinos are expensive and their validation can be cumbersome.

CRISPR-Cas13d knocks down barriers to studying maternally provided mRNAs

Following the discovery and widespread application of DNA-targeting CRISPR systems (e.g., CRISPR-Cas9), microbial (meta)genomic mining revealed a unique CRISPR-associated (Cas) protein—Cas13a (originally C2c2).⁹¹ Instead of DNA-nuclease domains (HNH and/or RuvC-like domains like Cas9 and Cas12a), Cas13a contains two RNA-nuclease domains (HEPN domains⁹²) that enable Cas13a to function as an autonomous, programmable (through its CRISPR (guide) RNA) RNA-targeting nuclease.^93,94,95,96 More Cas13 protein families—denoted Cas13b,⁹⁷ Cas13c,⁹⁸ and Cas13d^99,100—were subsequently identified. The Cas13d family exhibits the smallest proteins (average length = 930 amino acids), and Cas13d from Ruminococcus flavefaciens produces potent mRNA knockdown in mammalian cell culture.¹⁰⁰

To overcome the limitations in targeting maternally provided mRNAs, we established the RNA-targeting CRISPR-Cas13d system in zebrafish where it autonomously and potently degrades target mRNAs through at least the first 48 hours of development (see system development in¹⁰¹ and detailed methods in¹⁰²). Because it targets both maternally provided and zygotically transcribed RNA (Figure 3C), CRISPR-Cas13d is a straightforward tool for loss-of-function studies of early developmental RNAs, including microprotein mRNAs and long non-coding RNAs. RT-qPCR offers a simple first-pass knockdown validation and a diagnostic tool for guide RNA design. For targets with ≥ 60% knockdown and a phenotype, it is simple to generate a high number of knockdown embryos to analyze the onset and the spectrum of the developmental effect. Further, this high offspring number is amenable to interrogating the molecular phenotype (e.g., genes and processes affected) and transcriptome-wide CRISPR-Cas13d targeting specificity with RNA-sequencing (Figure 4A).

Figure 4.

CRISPR-Cas13d is an efficient and versatile tool for interrogating microproteins in vertebrate development

(A) Standard pipeline for targeting microprotein mRNAs with CRISPR-Cas13d in zebrafish embryos. Cas13d protein or mRNA with guide RNA(s) are injected and developmental phenotypes are assessed at multiple timepoints. RT-qPCR is employed as a first-step knockdown validation. For targets with ≥ 60% knockdown and with consistent phenotypes between guide RNAs, RNA-seq assesses the molecular phenotype and further interrogates CRISPR-Cas13d targeting.

(B) When microprotein mRNA knockdown results in a phenotype, synthetic microprotein or synthetic microprotein mRNA is co-injected to test for microprotein activity. Synthetic mRNA provides insights into either microprotein and/or non-coding RNA activity with simple frameshift and synonymous mutations induced in the in vitro transcription vectors.

(C) When rescue experiments support microprotein activity, tagged microproteins can be generated within in vitro transcription vectors (and without the need for endogenous tagging). Rescue of the CRISPR-Cas13d knockdown phenotype with any tagged microprotein validates its activity and enables imaging to determine microprotein localization and co-immunoprecipitation to define microprotein interaction partners.

Further, CRISPR-Cas13d guide RNAs are sequence-unconstrained, which enables effective targeting for any RNA of interest. Similarly, cross-validation of observed phenotypes with either multiple cocktails of guide RNAs or individual guide RNAs is feasible because their synthesis is simple and relatively inexpensive. Depending on the application, Cas13d guide RNAs can be co-injected with either Cas13d protein (quick knockdown, long production) or mRNA (long knockdown, quick production). Regardless of approach (see detailed methods in¹⁰²), CRISPR-Cas13d is an effective and efficient knockdown system. A CRISPR-Cas13d novice could reasonably perform a dozen unique CRISPR-Cas13d knockdowns in six months. Finally, although constitutive, heterologous expression systems in cell culture have revealed non-specific CRISPR-Cas13d RNA degradation when targeting abundant transcripts,¹⁰³ the injection-based tools in zebrafish are powerful assets to interrogate any potential off-target effects. For example, co-injection of mRNA with the tbxta coding sequence rescues the “No-Tail” phenotype observed under CRISPR-Cas13d tbxta knockdown (targeting the endogenous 3′ UTR of tbxta), eliminating off-target concerns in this knockdown paradigm.¹⁰¹ CRISPR-Cas13d therefore opens the door to the interrogation of developmental functions for microproteins in zebrafish.

In addition to microprotein mRNAs, CRISPR-Cas13d enables loss-of-function studies for any maternally provided RNA of interest. For example, analyses of single zygotic mutants for many zebrafish long non-coding RNAs failed to find developmental functions.^88,104 Although zygotic mutants enable maternally provided RNA to mask any developmental effect, CRISPR-Cas13d eliminates this maternal RNA and reduces the likelihood of false negatives in screens for non-coding RNA functions in development (Figure 3C). Outside of maternally provided RNA, double (or triple, etc.) mutants are required to uncover the impact of long non-coding RNAs with complementary functions. In contrast, CRISPR-Cas13d facilitates simple targeting of multiple RNAs to test for overlapping functions. Intriguingly, many microproteins are encoded on RNAs once described as non-coding, raising the possibility that a notable subset of RNAs have both coding and non-coding functions⁶⁰ or translation-dependent, non-coding functions (somewhat analogous to uORFs and dORFs). CRISPR-Cas13d, alongside established injection-based approaches in zebrafish, provides a truly feasible system in vertebrates to simultaneously interrogate both non-coding and coding functions within diverse RNAs.

Unraveling microprotein and non-coding RNA functions with CRISPR-Cas13d

On top of its potent loss-of-function effects, the flexibility of CRISPR-Cas13d guide RNAs enables routine injection-based approaches to provide key insights into developmental microprotein mRNA and/or non-coding RNA functions. When CRISPR-Cas13d knockdown of a microprotein mRNA produces a developmental phenotype, the paramount question is whether the phenotype is a result of microprotein activity (Figure 4B). Because microproteins are short, a chemically synthesized microprotein can be co-injected in knockdown embryos to test its ability to rescue the phenotype. However, RNA-based approaches are more flexible and efficient. In-vitro transcription vectors in zebrafish enable the generation of a “rescue” mRNA with the microprotein coding sequence flanked by exogenous 5′ and-3′ UTRs,¹⁰⁵ allowing guide RNAs targeting the microprotein UTRs to knockdown only the native zebrafish mRNA (Figure 4B). Microprotein activity is supported when a “rescue” mRNA alleviates the knockdown phenotype, but a frameshifted version fails to do so. Likewise, synonymous mutations in the microprotein sequence in the “rescue” mRNA should also alleviate the phenotype.

Rescue experiments can be leveraged for additional insights into conservation of function and critical amino acid residues. If microprotein function is conserved, homologous microprotein sequences (or syntenic non-coding RNAs) from other species in “rescue” mRNAs should similarly alleviate the developmental effect. In parallel, for any critical amino acid residue, a mutation in the “rescue” mRNA will fail to alleviate the phenotype. In a like manner, any developmentally critical non-coding RNA discovered with CRISPR-Cas13d can be mutated to assess for functional sequences and/or structures. Importantly, fluorescent or epitope tags can be added to microprotein coding sequences in the transcription vector alone, without the need for endogenous tagging. Although tagging has the potential to disrupt microprotein function, rescue of its CRISPR-Cas13d knockdown phenotype validates the activity of any tagged microprotein. Tag-based imaging can define microprotein subcellular localization whereas co-immunoprecipitation experiments based on the tag can identify microprotein interaction partners (Figure 4C).

Novel combinations of CRISPR-Cas13d will provide additional insights

RNA-seq is very effective at defining the transcriptome-wide effects in CRISPR-Cas13d knockdowns. However, a plethora of techniques are eagerly waiting to work in concert with CRISPR-Cas13d to provide further insights into microprotein and/or non-coding RNA functions. For example, v3 HCR in situ hybridization¹⁰⁶ in knockdown embryos can provide spatiotemporal readouts for transcripts of interest from RNA-seq to determine if their localization(s) is changing (Figure 5A). In-situs for cell-type specific markers (e.g., endoderm, ectoderm, and mesoderm) in knockdown embryos allow interrogation of potential cell fate decision and differentiation defects. Further, a key process during the maternal-to-zygotic transition is the activation of transcription from the zygotic genome (Figure 3A), which involves major changes to chromatin accessibility. CRISPR-Cas13d is amenable to genomics-based techniques (e.g., ChIP-seq,¹⁰⁷ATAC-seq¹⁰⁸) to interrogate chromatin states under microprotein knockdown because it is straightforward to generate many, homogeneous knockdown embryos (Figure 5B).

Figure 5.

Future combinations of CRISPR-Cas13d will paint a more complete picture of developmental microprotein functions

(A) Spatiotemporal gene expression in CRISPR-Cas13d knockdown embryos can be visualized for transcripts of interest from RNA-seq or for cells of interest using v3 HCR in situ hybridization.

(B) Chromatin accessibility during zygotic genome activation can be assessed in knockdown embryos using omics-based techniques such as ATAC-seq.

(C) CRISPR-Cas9 is highly efficient at generating bi-allelic deletions in F0 zebrafish. In combination with CRISPR-Cas13d, analyzing F0 zygotic mutants will provide insights into the individual role of zygotically transcribed RNA. Similarly, a germline targeted Cas9 potentiates the creation of F0 maternal mutants to test the individual role of maternally provided RNA.

Although CRISPR-Cas13d enables the dissection of gene function during the maternal-to-zygotic transition without maternal/zygotic mutants, parallel targeted mutagenesis (e.g., CRISPR-Cas9) enables investigations into the individual phenotype contributions from maternally provided and/or zygotically transcribed RNA. In zebrafish, CRISPR-Cas9 is highly optimized and can produce over 90% biallelic deletions in F0 mutants,^109,110 enabling rapid comparisons with CRISPR-Cas13d knockdowns. For instance, if zygotic RNA is the major contributor, zygotic F0 mutants will recapitulate the CRISPR-Cas13d knockdown phenotype. Alternatively, if maternal RNA is the major contributor, zygotic mutants will fail to recapitulate the phenotype (Figure 5C). In this case, a germline targeted Cas9 construct is available in zebrafish that can effectively generate maternal mutants in F0¹⁰⁹. Importantly, crossing F0 maternal mutants to wildtype males would validate the essential role of maternally provided RNA in a single generation (Figure 5C).

Further, simultaneous targeting of multiple microprotein or non-coding RNA genes is feasible with CRISPR-Cas13d. In fact, multiple targeting is especially useful in zebrafish where at least 26% of genes have a paralog from the teleost-specific genome duplication that could retain overlapping function.⁸² Moreover, CRISPR-Cas13d is effective in mouse, killifish, and medaka, enabling interrogation of conserved microprotein and non-coding RNA functions in development.¹⁰¹ Although this review has focused on development, adult zebrafish afford a large repertoire of tissue and cell-type specific drivers and human disease models. Cas13d could therefore be employed in a selective manner to interrogate microprotein or non-coding RNA functions in homeostasis or in zebrafish models of human disease.

Future perspectives for CRISPR-Cas13, microproteins, and non-coding RNAs

In all, CRISPR-Cas13d in zebrafish not only expands a solid vertebrate developmental model system but also opens novel angles for diverse developmental questions about microprotein and non-coding RNAs, particularly those that involve the maternal-to-zygotic transition. Alongside CRISPR-Cas13d, RNA and protein injection-based tools in zebrafish enable relatively simple dissection of diverse microproteins (e.g., encoded in overlapping ORFs, 2–10 codon ORFs) and non-coding RNAs (e.g., coordinate functions, and complementary roles). The utility of CRISPR-Cas13 across teleosts and mammals potentiates powerful comparative studies that can test conservation of function and provide insights into the evolution of both microproteins and non-coding RNAs. Similarly, implementation of CRISPR-Cas13d in plants could provide insights into microprotein and/or non-coding RNA function(s) during early plant development, which may involve contributions from both maternally and paternally provided RNAs.^111,112,113 One caveat to the CRISPR-Cas13d system is that it degrades maternally provided microprotein mRNAs, not the microproteins themselves. Any maternally provided microproteins whose function depends entirely on their maternal microprotein pool would escape detection in a CRISPR-Cas13d loss-of-function screen.

Beyond its current role in RNA knockdown, the CRISPR-Cas13 system has exciting potential. For instance, catalytically dead Cas13d fused to translation initiation factors was recently shown to increase translation in E. coli.¹¹⁴ If established in zebrafish, this would enable over-expression studies for microproteins without the need for injection of in vitro transcribed RNA that may fail to recapitulate the native microprotein mRNA expression pattern. Furthermore, application of the dead Cas13-based RNA imaging approaches (developed in mammalian cell culture) to the accessible live imaging provided by developing zebrafish would illuminate the spatiotemporal dynamics of microprotein mRNA and/or non-coding RNA in a living organism.¹¹⁵ Finally, implementing novel microprotein tagging approaches (e.g., MicroID,²⁴ non-canonical amino acid labeling⁶³) in zebrafish would complement functional studies with CRISPR-Cas13d to better elucidate microprotein functions in their native contexts.