Large-scale F0 CRISPR Screens in vivo using MIC-Drop

Abstract

The zebrafish is a powerful model system for studying animal development, modeling genetic diseases, and for large-scale in vivo functional genetics. Because of its ease of use and its high efficiency in targeted gene perturbation, CRISPR-Cas9 has recently gained prominence as the tool-of-choice for genetic manipulation in zebrafish. However, scaling up the technique for high-throughput in vivo functional genetics has been a challenge. We recently developed a method, Multiplexed Intermixed CRISPR Droplets (MIC-Drop), that makes large-scale CRISPR screening in zebrafish possible. Here we outline the step-by-step protocol for performing functional genetic screens in zebrafish using MIC-Drop. MIC-Drop uses multiplexed single-guide RNAs (sgRNAs) to generate biallelic mutations in injected zebrafish embryos, allowing genetic screens to be performed in F0 animals. Combining microfluidics and DNA barcoding enables simultaneous targeting of tens to hundreds of genes from a single injection needle, while also enabling retrospective and rapid identification of the genotype responsible for an observed phenotype. The primary target audiences for MIC-Drop are developmental biologists, zebrafish geneticists, and researchers interested in performing in vivo functional genetic screens in a vertebrate model system. MIC-Drop will also prove useful in the hands of chemical biologists seeking to identify targets of small molecules that cause phenotypic changes in zebrafish. Using MIC-Drop, a typical screen of 100 genes can be conducted within 2–3 weeks by a single user.

INTRODUCTION

CRISPR-Cas9 technology has revolutionized biomedical sciences by enabling facile and rapid genetic perturbation in multiple organisms^1-5. Genome wide CRISPR screens have proven invaluable in identifying novel gene functions and discovering gene targets of small molecules and drugs^6-11. However, most large-scale CRISPR screens thus far have used simple cell culture systems^6,7,9,10. Although these are powerful models, in vitro cell cultures are phenotypically limited to observations at the level of single cells, monolayers, or organoids¹². These systems cannot fully recapitulate the wide array of animal physiologies and behaviors relevant to human health. Whole animals, on the other hand, model a plethora of complex biological processes such as gastrulation, tumorigenesis, infection and immune response, social interactions, sensory response behaviors, and many more. However, scaling up CRISPR-Cas9 technology for in vivo functional screens has been extremely challenging due to the cost, labor, and time required to generate mutant animals. A handful of large-scale screens in whole animal models have been completed by sheer brute-force scaling of established single-gene approaches^12-19. A streamlined platform that enables large-scale CRISPR-based perturbation in vivo will open new avenues for whole animal functional genetic screens. Here, we outline the protocol for such a platform. MIC-Drop combines microfluidic technology, DNA barcoding, and single-needle embryo injection to enable large-scale CRISPR screens in zebrafish with speed and efficiency previously available only to in vitro systems²⁰.

Zebrafish are an excellent model system for large scale in vivo genetic screens^16,21. Zebrafish are genetically tractable — over 70% of human genes have zebrafish orthologues²². Additionally, around 80% of human genes with known disease-causing variants have zebrafish orthologues making them a great vertebrate model for disease modeling²². One can study a range of complex phenotypes in zebrafish including embryonic development^23-26, regeneration^27-29, cognition^30-35, diseases such as cancer^36,37, metabolic disorders³⁸, and various neurodevelopmental and neuropsychiatric disorders^17,33,39,40. Zebrafish’s high fecundity, external fertilization, rapid development, and small size also make them well-suited for genetic screens^21,41.

DEVELOPMENT OF THE MIC-DROP PLATFORM

Current strategies to perform CRISPR-Cas9 mutagenesis in zebrafish require loading a mix of Cas9 protein or mRNA and target-specific sgRNA in an injection needle, and subsequently injecting this mix into single-celled embryos^15,42,44. For each gene of interest, a separate injection needle must be loaded with gene-specific Cas9-sgRNA mix and injected into a new batch of embryos. These embryos are raised separately, genotyped to identify ‘founders’, and out-crossed and in-crossed multiple times to generate homozygous mutants. Unsurprisingly, scaling up this method for large-scale CRISPR screens is challenging. Recent studies have used multiplexed sgRNAs with high mutation efficiencies to generate biallelic F0 mutants that recapitulate germline mutant phenotypes thereby opening the doors to F0 CRISPR screens^45-49. Nonetheless, the current F0 screening approach still requires targeting one gene from a single injection needle, thus limiting its potential for large-scale screens (Figure 1a).

Figure 1:

Comparison between a traditional and MIC-Drop-based F0 CRISPR screen in zebrafish. (a) In a traditional screen, an sgRNA-Cas9 mRNA or RNP mix targeting different genes is injected from separate needles. The injected embryos are maintained separately and are phenotyped to identify genetic perturbations that cause the desired phenotype. (b) MIC-Drop uses microfluidics to generate nanoliter-sized droplets containing multiplexed RNPs targeting a single gene as well as a unique DNA barcode associated with that gene. Droplets targeting hundreds of different genes are intermixed, loaded into a single injection needle, and individual droplets are injected into zebrafish embryos. The injected embryos are raised together, screened for phenotype, and the causative genotype retrospectively identified by barcode amplification and sequencing.

The MIC-Drop platform uses innovative methods to obviate the shortcomings of traditional screening approaches (Figure 1b)²⁰. First, a microfluidic device is used to generate uniform, nanoliter-sized droplets containing Cas9, multiplexed sgRNAs targeting a single gene-of-interest, and a unique DNA oligonucleotide barcode associated with each gene. Droplets targeting hundreds of different genes are intermixed, loaded into a single injection needle, and a single droplet is injected per embryo. The injected embryos are then raised together en masse. At a desired timepoint, embryos are screened for phenotypic changes resulting from a specific gene perturbation. The gene target that resulted in the phenotype is rapidly identified by barcode amplification and its subsequent sequencing.

Three unique advantages make MIC-Drop scalable: (1) Multiplexed sgRNAs (typically four sgRNA targeting each gene) significantly increase the efficiency of biallelic mutation, thereby allowing us to screen for phenotypes in F0 animals, also referred to by many as “crispants”²⁰. We and others have shown that crispants successfully recapitulate germline mutant phenotypes^46-49. (2) Use of microfluidics enables generation of nanoliter-sized droplets containing Cas9, multiplexed sgRNAs, and a unique DNA barcode associated with each gene. Droplets targeting hundreds of different genes are intermixed and injected from a single needle, thus bypassing the need to reload separate injection needles for each targeted gene. (3) Finally, the use of a unique DNA barcode associated with each gene allows us to raise the droplet-injected embryos en masse without the need to track the genotype of each individual embryo. The barcode also enables rapid identification of the genotype of any embryo that displays a desired phenotype by barcode amplification and sequencing.

APPLICATIONS OF THE METHOD

Application of the MIC-Drop platform will vary widely based on the needs and desires of those employing it. One of the strengths of the method is its versatility and adaptability to nearly any situation where large-scale genetic targeting and/or high throughput microinjection are employed. In zebrafish we have demonstrated its use in both a reverse-genetics screen of 188 poorly characterized genes, as well as demonstrating the ability to identify genetic targets of small molecules²⁰. While these applications both utilized CRISPR-Cas9 mediated mutagenesis, other CRISPR technologies such as CRISPRi and CRISPRa, which have been utilized in zebrafish previously^50,51, can be substituted for Cas9 mutagenesis when desired experimentally. Additionally, while the protocol published here is for zebrafish screening purposes, the technique should be easily applied to other organisms where microinjection is commonly used^52-56. We would also like to stress that any technique that could benefit from high throughput droplet microinjection containing retrievable barcodes could be readily adapted for the MIC-Drop platform. For example, some drugs or compounds are not readily transportable into zebrafish larvae in standard bath applications; in such cases microinjection of these compounds is common^57,58. A library of these compounds could be readily incorporated with the MIC-Drop platform screen, with each candidate compound simply replacing the Cas9 and gene-specific sgRNAs.

The protocol we outline here is focused on CRISPR-Cas9 mutagenesis screens and thus some of the detailed methods such as sgRNA design and selection will be specific to such a screen. Adaptations may require additional optimization and validation, but the general workflow will remain similar for any screens using the MIC-Drop platform.

COMPARISON WITH OTHER METHODS

Large-scale forward genetic screens firmly established zebrafish as a powerful model system for functional genetics. These early screens utilized N-ethyl-N-nitrosourea (ENU) and retroviral insertional mutagenesis to generate random mutations and required monumental efforts from a large group of researchers^24-26,59-62. With over 1000 developmental mutants generated, these screens fueled developmental research using zebrafish for years thereafter. Notably, some of the mutants identified have yet to be mapped. To this day forward genetic screens continue to be employed to identify the genetic regulators of various biological processes⁶³.

Similarly, advances in reverse genetic techniques such as transcription activator-like effector nucleases (TALENs)^64,65, zinc finger nucleases (ZFNs)^66,67, and most recently CRISPR-Cas9⁴², have resulted in their own distinct boons to zebrafish research. Since our lab used zebrafish to demonstrate the first instance of CRISPR-Cas9 gene editing in an animal⁴², thousands of sgRNAs have been generated to target as many genetic loci in zebrafish and to interrogate the functional consequence of their perturbation⁶⁸.While both forward and reverse genetic approaches have been invaluable to biological and biomedical sciences, they each serve distinct purposes. We consider the MIC-Drop technique as a platform that utilizes the various strengths of both approaches. Most large-scale mutagenesis screens in whole animals have relied on forward genetics. This is largely due to the high costs and labor associated with performing a large-scale reverse genetic screen in whole animals. The MIC-Drop platform helps to alleviate some of those issues, making reverse genetic screens feasible in whole animal models. One of the main benefits of the MIC-Drop platform is the time needed to complete a screen. The Tübingen zebrafish screen took ~3 years to complete⁶⁹. A MIC-Drop screen of 100 genes can be completed within a single month. Of course, such a MIC-Drop screen is at a much smaller scale than a forward genetic screen. In principle, a whole-exome MIC-Drop screen could be performed, although the cost of library generation would be considerable.

While MIC-Drop offers the first highly scalable method of reverse genetic screening available to an animal model, several smaller-scale reverse genetic screens have been conducted in animal models^{13,14,17,45,70-72}. The largest zebrafish screen of this nature targeted 128 genes and focused on generating germline alleles¹⁴. The breeding required for generation of stable germline mutants contributes significantly to the time and space required for a screen of this size. Additionally, many injection sessions would be required for such screening. MIC-Drop screens offer substantial improvements in these areas.

MIC-Drop also provides substantial improvement over the existing F0 CRISPR screening method, which requires targeting different genes from separate injection needles. Although the initial effort to design and generate sgRNAs is the same for both methods, MIC-Drop enables targeting hundreds of genes from a single injection needle, thereby significantly reducing the time and effort required to perform large-scale F0 screens. Additionally, MIC-Drop sidesteps the need to maintain different mutants separately, allowing raising of all mutants en masse followed by isolation of mutants with the desired phenotype, thus reducing cost and effort needed for maintenance and phenotyping. There is an initial set up cost for generating the sgRNA library (same for a traditional F0 screen) and purchasing the droplet generator. However, both the sgRNA library and the instrument are reusable multiple times. In its current format, the commercially purchased Cas9 is the most expensive reagent for a MIC-Drop screen. The cost can be significantly reduced by using user-made recombinant Cas9. The largest and arguably most important difference in cost between a MIC-Drop based screen and a more traditional F0 screen is the associated cost of labor. The ability to target multiple genes using a single injection needle drastically decreases the amount of labor required. Take for example even a relatively small screen of 50 genes. In our experience a well-practiced researcher can target a maximum of 10 genes in a single injection session using conventional methods. Using the MIC-Drop protocol that same researcher can inject at least 50 genes in as much time. As the number of genes targeted increases the labor related savings increases even further.

LIMITATIONS

The protocol described here relies on CRISPR-Cas9 mutagenesis, and thus many of the limitations are directly related to this technique. One of the well-recognized limitations is the on-target vs. off-target activity of CRISPR-Cas9 genome editing. This is particularly relevant for MIC-Drop-based screens which rely on F0 phenotyping and utilize up to 4 distinct sgRNAs for each gene targeted. For this reason, we heavily prioritize use of sgRNAs with zero predicted off-target sites, and strongly recommend mRNA co-injection rescue experiments as well as subsequent generation of germline mutants where outcrossing can segregate any phenotypes not associated with mutations in the gene of interest.

Another CRISPR-Cas9 related limitation of this method is that its efficiency is directly tied to the efficiency of the chosen sgRNAs. In standard CRISPR-Cas9 injections targeting a single gene, this efficiency can be estimated based on subsequent analysis of the injected embryos’ genomic DNA and the specific region being targeted. T7 endonuclease assays²⁰ or standard gel or capillary electrophoresis following amplification of target region^15,16, can offer direct measurements. However, for large-scale genetic screening it is not feasible to directly analyze each individual sgRNA’s cutting efficiency. Thus MIC-Drop screens rely on the efficiency calculations estimated by algorithms included in the programs used for sgRNA selection. This is of minimal concern when considering false positive hits, as downstream validation will confirm the targeted mutagenesis through the techniques described here. Of greater concern is the possibility that, for a small number of genes, the corresponding sgRNAs display low cutting efficiencies and do not result in the efficient biallelic mutation needed for F0 phenotyping. In our experience, the steps outlined here for sgRNA selection, and the inclusion of 4 multiplexed sgRNAs has led to efficient mutagenesis of many previously described genes, and accurate recapitulation of their germline mutant associated phenotypes²⁰. Nonetheless, if needed, sgRNA efficiency testing can be performed in a random sample of droplet-injected embryos. Additionally, there have been recent efforts to curate lists of validated sgRNAs used for CRISPR-Cas9 mutagenesis in zebrafish. As these lists grow, fully validated libraries of sgRNAs targeting the zebrafish genome/exome will be invaluable to future MIC-Drop users¹⁶.

Phenotyping of F0 generation larvae is a distinct advantage of this technique considering the time it saves; however, it carries limitations as well. As mentioned above, high efficiency of biallelic mutation is necessary for F0 phenotyping. Even with high efficiency editing, F0 larvae will be mosaic mutants with multiple insertions and deletions of varying lengths each potentially conferring a differing effect on gene function. This may introduce higher degrees of phenotypic variability than what is seen in stable germline knockout fish. In some cases, this variability may present as incomplete penetrance of the phenotype of interest which has the potential to “mask” biologically relevant hits in a screening format.

EXPERIMENTAL DESIGN

Overview and timeline of MIC-Drop

The flowchart in Figure 2 outlines the steps of MIC-Drop and the time required for completing each step by a single user . The timing needed to perform some steps will vary dependent on the number of genes the user is targeting in their screen. We have attempted to include time needed per gene when possible.

Figure 2:

Flowchart detailing the steps of a MIC-Drop screen and expected time to complete each step.

Target selection and sgRNA synthesis

The MIC-Drop platform relies on F0 phenotyping. As such, efficient biallelic loss-of function-mutagenesis in injected embryos is essential. This makes target selection and sgRNA design one of the most critical steps in the MIC-Drop protocol. We recommend selecting four target sites for each gene targeted. For genes with ohnologs, two sgRNAs targeting each ohnolog can be selected in an effort to avoid phenotype masking by genetic compensation. Alternatively, each ohnolog can be targeted separately by four sgRNAs. To the best of our knowledge, the optimal number of guides used to target each gene in an F0 based screen has not been determined. We chose four sgRNAs to maximize the likelihood of generating biallelic loss-of-function mutations. However, we acknowledge that increasing the number of sgRNAs used will also increase the probability of off-target activity. In our published screen, four sgRNAs per gene worked well to identify both novel and previously described phenotypes with limited false positives due to off-target activity²⁰. The MIC-Drop protocol is compatible with lower numbers of sgRNAs per gene if users wish to reduce potential off-target activity or decrease the cost of library generation.

We use CHOPCHOP⁷³ for sgRNA design, however, alternative design tools such as CRISPRscan⁷⁴, CRISPOR⁷⁵, and the ZebrafishGenomics track on UCSC Genome Browser can be used^15,16. For our screens, we select sgRNAs targeting early coding exons to introduce nonsense mutations as early as possible and thus increase the likelihood of loss-of-function mutations. To avoid downstream translation, we also include sgRNAs with target sequences downstream of alternative translation initiation sites. We also prioritize exons shared among all known splice variants. This targeting strategy can be amended for the individual purposes of each screen. For example, if performing a screen of a family of kinases, an investigator may choose to target exons coding for conserved functional domains critical for kinase activity. In any case, we urge those considering this platform to prioritize efficiency for an F0 based screen. It is also equally important to select sgRNAs with no predicted off-target sites that contain 3 or fewer nucleotide mismatches with the target site (Figure 3a). sgRNAs with off-target activity may confound F0 screening results. In cases where sgRNAs with zero off-target sites cannot be designed, we select guides with off-targets containing at least 3-base mismatches with the target site, of which at least 1 mismatch is in the seed region (12 bases 5’ to the PAM) (Figure 3b)⁷⁶. Guides starting with the nucleotide sequence “GN” are prioritized to optimize SP6-dependent in vitro transcription, as SP6 will be the polymerase we use. Note, that while we have chosen to use SP6 (which requires guides starting with “GN”) for transcription, other polymerases such as T7 are suitable as well. Users should be sure to understand any varying sequence requirements needed for their choice of polymerase. Picking guides that start with “GN” avoids the need to add supernumerary “G” or “GG” to the gRNA sequences for efficient in vitro transcription by SP6 and T7 polymerases, respectively. Addition of supernumerary “GN” significantly reduces sgRNA targeting efficiency and is best avoided^15,49.

Figure 3:

CHOPCHOP enables sgRNA design for a MIC-Drop screen. Sample output results from CHOPCHOP when targeting (a) rx3 and (b) tbx16 genes. Only sgRNAs starting with ‘GN’ are selected. The selected sgRNAs are non-overlapping, have high predicted efficiency (> 40), and target early coding sequences. The ‘NGG’ at the 3’ end of the target sequence is the protospacer adjacent motif (PAM). MM0, MM1, MM2 and MM3 columns show the number of sequences with 0, 1, 2 or 3 bp mismatches, respectively. CHOPCHOP uses green, yellow and red to label sgRNAs with no predicted off-targets, sgRNAs with a few off-targets with 3 bp mismatches (MM3), and sgRNAs with several off-targets containing 3 or less mismatches, respectively. In (a) sgRNA with rank 1 is not selected because it targets the last exon of the rx3 coding sequence. sgRNAs with no predicted off-targets (green) are selected. (b) When sgRNA with zero off-targets cannot be designed in the desired region, sgRNAs with off-target containing 3 mismatches (MM3), at least one of which is in the seed region (highlighted in inset), is selected.

For sgRNA synthesis, we anneal gene-specific forward oligos and an sgRNA universal reverse oligo and perform fill-in PCR as reported previously^16,44 to generate the DNA template for in vitro transcription (Figure 4a). The four DNA templates for each gene are pooled in equal ratio and used for in vitro transcription using SP6 RNA polymerase (Figure 4b). If a user desires, synthetic gRNAs (either sgRNAs or dual gRNAs) can be substituted for in vitro transcribed sgRNAs in a MIC-Drop screen. Synthetic guides are not limited by “GN” requirement and therefore offer a wider selection of gRNA targets^49,77,78. Nevertheless, gRNAs should still be prioritized for their on-target efficiency and minimal predicted off-targets. A drawback of using synthetic guides is that they are significantly more expensive compared to in vitro transcribed guides. Given that oligo libraries for in vitro transcription are reusable, the cost of four in vitro transcribed guides is a fraction of the cost of synthetic guides even after including the reagent and labor cost associated with in vitro transcription.

Figure 4:

Assessment of DNA and RNA quality using agarose gel electrophoresis. (a) Sample gel electrophoresis of DNA templates used to generate sgRNAs targeting 2 different genes (4 templates for each gene). The DNA templates of length 116 base pairs (bp) were assembled using fill-in PCR of gene-specific forward oligos and the sgRNA universal reverse oligo. The DNA templates contain an SP6 RNA polymerase site for in vitro transcription (IVT) and unique spacer sequences targeting each gene. (b) The four DNA templates targeting each gene are pooled and in vitro transcribed. The quality of the sgRNAs is assessed by agarose gel electrophoresis. The sgRNAs should migrate as a single band of 96 bases on a 2% agarose gel. (c) The spacer sequences are unique to each targeted gene and therefore act as a barcode. The DNA barcode for each gene is generated using a standard set of biotinylated primers to extend and end-modify any of the four DNA templates in (a) for efficient recovery and unambiguous sequencing. The DNA barcode should migrate as a single band of 137 bp. (d) The DNA barcode is amplified and sequenced using either M13F or M13R primers. Even though we have used biotinylated DNA template as the barcode, unmodified DNA template is also stable and can be successfully amplified from zebrafish embryos after 5 days post injection. All procedures related to zebrafish studies were approved by the Institutional Animal Care and Use Committee at the University of Utah. Uncropped blot for Figure 4d is provided as Source Data.

Barcoding of MIC-Drops

Barcoding is essential for MIC-Drop screens. It allows intermixing of droplets targeting several genes and retrospective identification of the targeted gene for each droplet-injected embryo. In our case, we extend and end-modify any one of the four DNA templates for sgRNA synthesis and use this short double stranded DNA as the barcode²⁰. The 19–20 bp spacer sequence is unique to the targeted gene and is used as the diagnostic for the perturbed gene. The template extension allows for unambiguous and efficient amplification of the barcode and adds the universal primer site utilized in downstream sequencing (Figure 4c). The 5’-biotin modification was added in anticipation that the barcode may need to be enriched prior to amplification. However, we have found that unmodified barcodes are sufficiently stable and can be used in MIC-Drop screens (Figure 4d). Even though we used unique barcodes for each targeted gene, if desired, a standard set of barcodes can be used for all screens. A standard set of barcodes can also be used when using synthetic guides in lieu of in vitro transcribed sgRNAs.

The short dsDNA barcode is stable for up to 7 days post fertilization (dpf)²⁰. This allows for screening of most morphological, physiological, and some behavioral phenotypes. However, if a desired phenotype appears later than 7 dpf, a barcode capable of genome integration may be used for amplification at juvenile or adult stages. 5’-phosphate-modified short dsDNA capable of genome integration has been previously used to detect the on-target and off-target activities of CRISPR-Cas9 (ref. 79). Similar strategies can be co-opted for genome-integratable barcodes that persist in juvenile and adult F0 zebrafish.

MIC-Drop generation and injection of droplets

The use of microfluidics to generate nanoliter-sized droplets encapsulating the Cas9-sgRNA ribonucleoprotein and the dsDNA barcode is another key feature of MIC-Drop²⁰. We repurposed a BioRad droplet generator, typically used for digital droplet PCR, to generate the droplets. We tested several oil and surfactant combinations and discovered 3% 008-FluoroSurfactant (wt/v) dissolved in Novec HFE-7500 fluorinated oil (3% FS-HFE, hereafter) to be optimal for generating droplets that are stable for prolonged storage and are non-toxic upon injection in zebrafish embryos. Droplets generated using BioRad’s proprietary fluorinated oils (Droplet generation Oil for EvaGreen and Droplet generation Oil for Probes) are suboptimal, as they frequently both coalesced and broke apart, and thus, are not recommended.

Each droplet generated contains sgRNAs and barcodes for a single gene; as such, researchers can target a distinct gene with each embryo injection. As an experienced user can regularly inject 300–500 embryos in a single morning, it is possible to target 300–500 distinct genes in a single injection session. However, in a MIC-Drop screen we typically aim to have a coverage of 6–10 embryos injected for each gene targeted. Assuming a 100% phenotypic penetrance, a coverage of 6–10 embryos ensures 94–99% probability (Poisson distribution) of injecting ≥3 embryos targeting each gene, providing phenotypic redundancy. Phenotypes identified across multiple embryos are more likely to be caused by true gene perturbation rather than spurious effects, such as spontaneous mutations or injection artifacts. Additionally, many genetic perturbations may display incomplete phenotypic penetrance, and would be lost as false negatives without sufficient redundancy. Finally, injecting multiple droplets targeting each gene is helpful to account for random distribution of the droplets upon mixing and to ensure droplets targeting each gene are injected. Taking all of this into consideration, we typically pool droplets targeting up to 50 genes (including controls). This allows for 6–10 embryos injected with droplets targeting each gene. As noted, these suggestions are based on a single-user regularly injecting 300–500 embryos each morning and may be adjusted if multiple users are simultaneously injecting or if fewer embryos can be regularly injected. The same pool of droplets can also be injected in multiple clutches of embryos and on different days to account for clutch-to-clutch variability, and to maximize repeatability of hits, and minimize false positives.

Zebrafish line

We perform our screens in TuAB or Tu zebrafish lines. The danRer10/11 (GRCz10/11) reference database used for sgRNA design by CHOPCHOP is based on sequencing the Tu line. Should other lines be used, one should be aware of polymorphisms that may affect sgRNA targeting efficiency. Gene perturbation efficiency will likely also be affected depending on whether the droplets are injected into the yolk sac or in the cell itself. We inject droplets directly into cells to maximize target editing efficiency.

Phenotyping

As with any high-throughput screening methodology, the final readout or output is key to the success of a MIC-Drop screen. In general, screens can be performed to (i) discover the phenotypic effects of targeting a list of genes (no a priori knowledge of phenotype) or (ii) to identify genes (from a list of candidate genes) responsible for a very specific phenotype. The phenotype to be screened can be any phenotype that is possible to study in a zebrafish model. Therefore, when considering a MIC-Drop screen, thoughtful consideration of the phenotyping step is crucial.

In the screen where no a priori phenotypic knowledge or assumptions are made, what steps will be taken to identify possible phenotypic changes? A morphological stereoscopic analysis may be sufficient for dramatic developmental defects, but subtler changes may be missed. Researchers may wish to employ similar strategies as those conducted in early zebrafish forward genetic screens where a wide range of analyses by multiple individuals was performed to capture the full gamut of phenotypes^24-26.

Researchers whose aim is to identify genes responsible for a specific phenotype should also give careful consideration as to whether the MIC-Drop platform will be effective to achieve their goal. Consider the following when deciding on phenotypes to study: (a) How obvious is the phenotype? (b) How distinct is the phenotype from non-specific phenotypes commonly observed in injected embryos due to injection-associated toxicity²⁰? (c) Is there significant inter-individual variability of the phenotype of interest? Less obvious phenotypes may require user expertise, special equipment for imaging, and transgenic zebrafish lines for clear identification of the phenotype being studied. The phenotype should be sufficiently distinct from spurious gross morphological defects commonly observed in around 10% of injected zebrafish embryos²⁰. Finally, the phenotype being studied should not vary greatly from embryo-to-embryo. This is especially true when performing behavioral assays as they show high inter-individual variability. To test whether your phenotype of interest is suitable for a MIC-Drop screen, one should consider performing a mock screen utilizing sgRNAs targeting a positive control gene to estimate false positive rate before investing too heavily in the sgRNA library needed for the screen. We describe the design of such a mock screen in an Experimental Design section below.

Barcode retrieval and sequencing to identify “hits”

Barcode retrieval is performed by lysing the embryos and amplifying the DNA barcode from the lysate. A critical step to ensure unambiguous sequencing results is to thoroughly wash the embryos multiple times to get rid of residual contaminating barcodes in the media. It is also advisable to dechorionate embryos prior to lysis to remove any barcodes non-specifically sticking to the chorion. Finally, to minimize barcode cross-contamination, we amplify the barcode with 30% dUTP. In this way, any residual, cross-contaminating, uracil-containing DNA amplicons from previous PCR amplifications are degraded by incubating with uracil DNA glycosylase (UDG) prior to subsequent barcode amplification. Subsequent Sanger sequencing of the amplicon reveals the unique sequence associated with the targeted gene.

Following sequencing, initial candidate hits are determined by examining the frequency with which each barcode is recovered. “Hits” are genes whose barcodes are identified at a frequency significantly higher than what would be expected from random sampling. We determine the binomial probability and consider barcodes that occur with a p < 0.05 as “hits” and candidates for follow-up and validation. This cutoff limit does not guarantee the elimination of false positives or false negatives; and in certain circumstances it may be appropriate to raise or lower this cutoff. For example, if a screen revolves around a phenotype that overlaps with a common injection related phenotype, a more stringent cutoff may be required to avoid multiple false positives.

Conversely, it is important to note that the observed frequency used in the binomial probability calculations is reliant on (1) distribution of droplets in the mix and (2) phenotypic penetrance. So, in some screens (without a predefined phenotype) it might be reasonable to treat a gene whose barcode occurs with low frequency, and thus does not meet the probability cutoff, as a “hit” if the phenotype observed is unique and consistent across multiple embryos injected.

Secondary validation and rescue

Because our cutoff is based on a probability estimate, we expect that with large-scale experiments a small number of “hits” will prove to be false positives based on random chance. Increased cutoff stringency and repeated screening of multiple clutches on multiple days should ensure that very few “hits” are false positives. However, even with these precautions, secondary validation is needed to ensure that the targeted gene of interest does result in the expected phenotype when perturbed. This initial validation includes repeated injection, mRNA rescue experiments, and eventual generation of germline mutant lines.

The first validation step is simple repeated injection using standard methods to examine larger sample sizes. With the smaller number of candidate hits from initial screening it is much more feasible to perform direct injections of 30–50 embryos for each candidate hit and assess phenotypic penetrance. Additionally, with a smaller number of candidate hits, assessing the cutting efficiency of the selected guides is feasible. During the repeat injections, individual sgRNAs or combinations of the four sgRNAs can be used to ensure that the observed phenotype is due to targeted gene perturbation and not due to sgRNA-specific off-target activity. Furthermore, the phenotype can be validated by performing injections with a new set of sgRNAs targeting the gene.

Rescue of phenotypes is another quick and efficient way to confirm that the observed phenotype is a result of on-target gene perturbation and not a consequence of an off-target effect. We typically co-inject the mRNA for the gene being targeted along with the sgRNA and Cas9 RNP mix and look for loss of phenotype in the co-injected embryos relative to RNP-only injected embryos. The amount of mRNA co-injected can be varied to further confirm a dose-dependent rescue effect.

If direct injections and mRNA rescue experiments confirm the genotype-phenotype correlation and on-target effects of injected sgRNAs, a candidate hit can be confirmed as a validated hit. However, we also recommend the generation of germline mutant alleles where successive rounds of out-crosses and in-crosses can confirm that the mutant alleles generated do indeed cause the phenotype of interest. As the strategy for generating germline mutants has been described in detail previously¹⁶, we will not repeat it in this protocol.

Mock/Pilot Screening and Essential Controls

Prior to beginning a MIC-Drop based screen we recommend that investigators estimate the expected false positive and false negative rates related to the experimental readout. False positive rates can be determined by performing a mock/pilot screen with scrambled sgRNAs or no sgRNAs in the droplets, or even by assessing the frequency of the phenotype in the background of the wild-type zebrafish being used for the screen. If the phenotype being screened for is observed in appreciable numbers in the wildtype background or in a mock screen, it is likely the screen will have a high false positive rate. False positives can be eliminated by retesting, but they reduce the efficiency of the screen and should be minimized when possible.

False negatives can occur when gene knockdown is inefficient, for example when MIC-Drop design, production, or injection are suboptimal. If an investigator wishes to confirm the efficacy of their technique prior to conducting a screen, they may conduct a mock/pilot screen to assess whether known genetic regulator(s) produce the expected effect. We have found that droplets targeting rx3 (a gene causing an easily identifiable missing eyes phenotype) is an outstanding control to test for technical efficiency of droplet injection.

One useful mock screen might involve combining rx3 droplets with negative control droplets (scrambled sgRNA sequence) in a ratio of 1:4 (rx3:scrambled). For the most accurate quantification of false positive and negative rates, we suggest injection of ~100 embryos using this mixture, followed by phenotyping and barcode retrieval of all embryos. Following droplet injection and phenotypic screening one would expect 20 embryos with the rx3 phenotype per 100 embryos injected, assuming random droplet distribution and 100% phenotypic penetrance. Assessing the number of barcodes retrieved in comparison to the phenotypes observed in each embryo will yield estimates of the false positive and false negative rates expected in a larger experimental screen.

Additionally, we recommend the following controls are included in each screen.

Injection control –

sgRNAs targeting a gene in which F0 mutagenesis results in an easily identifiable/quantifiable phenotype. This controls for the effectiveness and efficiency of each injection. The F0 phenotypic penetrance when targeting this gene should be near 100% when standard microinjections are utilized. We commonly include droplets targeting rx3 as an injection control. After injection, if the proportion of embryos displaying the rx3 related phenotype (missing eyes) does not match the proportion of rx3 droplets, then injection efficiency may be suboptimal. (In some cases, this control may not be needed as the positive control may serve the same purpose; however, we find that the presence of an unrelated phenotype serves as an effective injection control).

Positive control –

sgRNAs targeting a gene in which F0 mutagenesis results in the phenotype-of-interest. This positive control is helpful to ensure that the MIC-Drop procedure and phenotyping assay are working properly.

Un-injected sibling controls –

controls for overall health in the clutch of embryos injected. Health and survival of injected embryos is expected to be affected slightly by injection; however, excessive decreases in survival or widespread morphological defects associated with injection suggest that improved injection technique is needed. Common issues include: over-injection (injecting too high a volume causing nucleic acid related toxicity) and excessive mechanical damage caused by oversized needle or rough handling of embryos. Keeping un-injected sibling controls allows the investigator to set a baseline survival rate and assess for any morphological abnormalities present in wildtype larvae possibly due to background mutations in the strain.

MATERIALS

BIOLOGICAL MATERIALS

• Any wildtype zebrafish strain such as Tu, AB, TuAB, and NHGRI-1 can be used. The wildtype strains can be obtained from Zebrafish International Resource Center (ZIRC). Note that the choice of wildtype strain may affect sgRNA design. The GRCz11 reference genome was obtained from a Tu background.

REAGENTS

• Gene-specific forward oligo (5’-ATTTAGGTGACACTATA GN_(18/19) GTTTTAGAGCTAGAAATAGC-3’), where GN_(18/19) refers to the sgRNA target/spacer sequence (Supplementary Table 1) (Integrated DNA Technologies (IDT))

• sgRNA universal reverse oligo (5’-AAAAGCACCGACTCGGTGCCACTTTTTCA AGTTGATAACGGACTAGCCTTATTTTAACTTGCTATTTCTAGCTCTAAAAC-3’) (IDT)

• Barcode generation and amplification primers (Supplementary Table 1)(IDT)

• Phusion HS Flex DNA polymerase and 5x HF buffer (New England Biolabs; cat. no. M0535L)

• 100 mM dNTP Set (Invitrogen, cat. no. 10297018) (see Reagent Setup)

• Agarose (Fisher Scientific, cat. no. BP160-500)

• UltraPure Ethidium bromide, 10 mg/mL (Invitrogen, cat. no. 15585011)

  ! CAUTION Ethidium bromide is a carcinogen. Wear gloves when handling.

• DNA Clean & Concentrator-5 (Zymo Research; cat. no. D4014)

• ZR-96 DNA Clean & Concentrator-5 (Zymo Research; cat. no. D4024)

• MegaScript SP6 transcription kit (Invitrogen; cat. no. AM1330)

• RNase Away (Thermo Scientific, cat. no. 7003)

• RiboLock RNAse Inhibitor (40 U/μL) (Thermo Scientific, cat. no. EO0381)

• RNA Clean & Concentrator-5 (Zymo Research; cat. no. R1014)

• RNA Clean & Concentrator-96 (Zymo Research; cat. no. R1080)

• EnGen Spy Cas9 NLS (New England Biolabs, cat. no. M0646M) and 10x NEBuffer r3.1 (cat. no. B6003S)

  ▲ CRITICAL Cas9 protein from different vendors may have different stability in droplets depending on the buffer being used to store them.

• 0.5% Sterile Phenol Red solution (Sigma-Aldrich, cat. no. P0290-100 mL)

• Novec HFE-7500 Engineered Fluid (3M, cat. no. Novec 7500) (see Reagent Setup)

• 008-Surfactant-1g (Ran Biotechnologies, cat. no. 008-Fluorosurfactant)

• E3 embryo medium (see Reagent Setup)

• Proteinase K, recombinant PCR grade (Roche, cat. no. 03115828001)

• 100 mM dUTP solution (Thermo Scientific, cat. no. R0133) (see Reagent Setup)

• Uracil-DNA glycosylase (UDG) (New England Biolabs, cat. no. M0280S)

• GoTaq DNA polymerase (Promega, cat. no. M3008)

• Exonuclease I (New England Biolabs, cat. no. M0293)

• Shrimp alkaline phosphatase (New England Biolabs, cat. no. M0371)

• Methylene blue (Fisher Scientific, cat. no. M291-25)

• Sodium chloride (Fisher Scientific, cat. no. S271-3)

• Potassium chloride (Fisher Scientific, cat. no. P330-3)

• Calcium chloride dihydrate (Fisher Scientific, cat. no. C70-500)

• Magnesium chloride hexahydrate (Sigma-Aldrich, cat. no. 63068-250G)

• Tris-base (Sigma-Aldrich, cat. no. T6066-1KG)

• Ethylenediaminetetraacetic acid (EDTA) (Sigma-Aldrich, E6758-500G)

• Triton X-100 (Sigma-Aldrich, cat. no. T8787-100)

• Acetic Acid, glacial (Sigma-Aldrich, cat. no. A6283)

• Tricaine-S (Syndel)

EQUIPMENT

• Standard equipment for zebrafish husbandry and breeding

• Disposable transfer pipettes (Thermo Scientific Samco, cat. no. 204-1S)

• Petri dish

• Standard equipment for nucleic acid gel electrophoresis

• FluorChem M Imaging system (Protein Simple)

• T100 Thermal Cycler (BioRad)

• PCR strip tubes with domed caps (Thermo Scientific, cat. no. AB-1113)

• 96-well PCR plates (Thermo Scientific, cat. no. AB-0600)

• Adhesive PCR plate sealing sheets (Thermo Scientific, cat. no. AB-0558)

• Square well 96-well plates (Cytiva, cat. no. 7701-1651)

• Nanodrop 2000C spectrophotometer (Thermo Scientific)

• Multichannel pipettes (Rainin) and tips

• QX200 Droplet Generator and cartridge holder (BioRad, cat. no. 1864002)

• DG8 Cartridges and Gaskets (BioRad, cat. no. 1864007)

• Capillary tubes (World Precision Instruments, cat. no. TW100F-3)

• Flaming/Brown Micropipette Puller (Sutter Instrument, cat. no. P-1000)

• Microloader tips (Eppendorf, cat. no. 5242956003)

• Dumont Tweezer, style 5 (Electron Microscopy Sciences, cat. no. 72700-D)

• Microinjection mold (Adaptive Science Tools, cat. no. TU-1)

• Picoliter injector (Harvard Apparatus PLI-100)

  ▲ CRITICAL Other injector systems may be used. However, the pressure settings specified in Step 44 will vary with other injectors.

• Micromanipulator (Narishige)

• Surgical Probe for positioning zebrafish embryos in injection mold

• Microscope for microinjection (Nikon SMZ645)

• Microscope light source (Nikon Transformer Model XN)

• Stage micrometer slide (Azzota, cat. no. TS-M7)

• Incubator set at 28.5 °C

• Zebrabox (ViewPoint Behavior Technology)

REAGENT SETUP

• 10 mM dNTP mix: Mix 50 μL each of 100 mM dATP, dCTP, dGTP and dTTP. Bring volume to 500 μL with nuclease-free water to make 10 mM dNTP stock. Mix thorughly and aliquot. Aliquots can be stored at −20 °C for up to 6 months.

• dNTP with 30% dUTP: Mix 50 μl each of 100 mM dATP, dCTP, and dGTP. Add 35 μL of 100 mM dTTP and 15 μL of 100 mM dUTP. Bring volume to 500 μl with nuclease-free water to make 10 mM stock. Mix and store aliquots at −20 °C for up to 6 months.

• Oligos: Resuspend oligos to 100 μM with nuclease-free water. Make a 10 μM solution of gene-specific forward oligos ordered in 96-well plates at 500 pmol scale by resuspending in 50 μL water. Resuspended oligos can be stored at −20 °C for > 6 months. Avoid multiple freeze-thaw.

• 3% FS-HFE: Make a 5% FS-HFE solution by adding 20 mL Novec HFE-7500 to 1g of 008-surfactant. Vortex vigorously for 5–10 min to dissolve surfactant. In another glass vial, add Novec FIFE-7500 and 5% FS-HFE in a 2:3 ratio to make the desired volume of 3% FS-HFE. 3–5% FS-HFE can be stored at room temperature (23°C in our lab) for up to 6 months.

• Methylene blue: Make a 1000x stock by dissolving 0.5 g of methylene blue powder in 1 L of milliQ water. Methylene blue solution can be stored at room temperature for up to 6 months.

• E3 medium: Make 1 L 60x E3 media by dissolving 17.4 g NaCl, 0.8 g KCl, 2.9 g CaCl₂.2H₂O, and 4.89 g MgCl₂.6H₂O in milliQ water. Bring pH to 7.2 with NaOH. Prepare 1x medium by adding 16.5 mL of the 60x stock to 1 L of milliQ water. To make 1x E3 with methylene blue, add 1 mL of 1000x methylene blue stock to 1x E3 media. Store E3 medium for up to 1 month at room temperature.

• Tris-Acetate EDTA (TAE) buffer: Make 1 L 50x TAE buffer by dissolving 242 g Tris base, 57.1 mL of Acetic acid, and 100 mL of 0.5 M EDTA (pH 8.0) in milliQ water. Prepare 1x medium by adding 20 mL of the 50x stock to 1 L of milliQ water. Store TAE buffer for up to 6 months at room temperature.

• Tricaine methane sulfonate solution: Dissolve 400 mg of Tricaine in 100 mL of E3. Aliquot and store at −20 °C for up to 1 month. To make 1x, add 1 mL of tricaine solution to 25 mL E3.

  ▲ CRITICAL Only use freshly made 1x Tricaine solution.

• 2x PCR lysis buffer: Add 20 mL of 1 M Tris-HCl (pH 8.0) and 8 mL of 0.5 M EDTA (pH 8.0). Add 4 mL Triton X-100. Bring volume to 1 L with milliQ water. Stir to dissolve Triton X-100. Store lysis buffer for up to 6 months at room temperature.

EQUIPMENT SETUP

• Injection mold: Dissolve 3 g agarose in 100 ml 1x TAE. Microwave until agarose is fully dissolved and cool to touch. Pour agarose into petri dishes (~3/4 full) and place a microinjection mold on top of the agarose taking care not to trap any air bubbles. Leave to solidify. Remove the mold from the solidified gel. Add 1x E3 with methylene blue to the gel. Same injection gel can be reused multiple times. Store injection gel with 1x E3 with methylene blue at 4 °C in between use. Seal petri dish with parafilm to avoid drying out the gel.

• Microinjection needles: Pull needles using a Sutter Instrument set at the following settings: Heat: 565, Pull: 64, Velocity: 77, Time: 80, and Pressure: 500.

  ▲ CRITICAL Only use freshly made needles for droplet injection. Dust and debris in injection needle will clog needle and/or break droplets.

• Droplet Generator setup: Set up the droplet generator per manufacter’s instruction (Supplementary Figure 1 and Supplmentary Video 1).

• Injection setup: Set up the injector per manufacter’s instruction (Supplementary Figure 2).

PROCEDURE

CRITICAL:

All procedures related to zebrafish studies were approved by the Institutional Animal Care and Use Committee at the University of Utah.

Design and selection of sgRNAs targeting Gene(s)-of-Interest ● TIMING 5–10 min/gene

1. Access the CHOPCHOP web gRNA design tool (https://chopchop.cbu.uib.no/). CRISPR/Cas9 sgRNAs can be designed using multiple alternative tools, including CRISPRSCAN and CRISPOR.

2. Enter the ‘ENSEMBL/gene ID’ of the target gene-of-interest. Use ‘Danio rerio (danRer11/GRCz11)’ as the reference genome. Design gRNAs using ‘CRISPR/Cas9’ for gene ‘knock-out’.

3. Under ‘Options’, select the following: ‘GN or NG’ for 5’ requirement for sgRNA. ‘sgRNA length without PAM’ can be customized to either 19 or 20 nt. Keep the remaining parameters at the default settings. Search to find ‘Target Sites’. ?TROUBLESHOOTING

4. Pick 4 sgRNAs that meet the following criteria: (i) Non-overlapping sgRNAs of length 19 or 20 nt that target the first-half of the entire coding sequence (ii) sgRNAs that start with GN (iii) sgRNAs with high ‘Efficiency’ score (>40) (iv) sgRNAs with GC content between 45%-75% (v) sgRNAs with 0 off-targets. (vi) sgRNA with low self-complementarity (<3) (vi) sgRNAs without poly Ts (TTTT) in the sequence.?TROUBLESHOOTING

5. If sgRNAs with 0 off-targets cannot be designed, select sgRNAs with minimum possible off-targets. The off-targets should have at least 3 base mismatches (MM3) with the target sequence. Prioritize sgRNA with at least 1 mismatch in the seed region.

6. If targeting genes with known ohnologs, select 2 sgRNAs targeting each ohnolog.

7. Repeat Steps 2-6 to design sgRNAs for all gene(s)-of-interest.

8. Copy the target sequences (without NGG PAM), add SP6 promoter sequence to the 5’ end of the target sequence, and a 20 nt sequence at the 3’ end that complementary base pairs with the sgRNA universal reverse oligo. The sequence of the final gene-specific forward oligo should be: ATTTAGGTGACACTATAgnnnnnnnnnnnnnnnnnnnGTTTTAGAGCTAGAAATAGC where GN_(18/19) refers to the sgRNA target/spacer sequence (Supplementary Table 1)

9. Order the gene-specific forward oligos either as desalted 10 mM scale in individual tubes or 500 pmol scale in 96-well plates. Order the sgRNA universal reverse oligo (Supplementary Table 1) in bulk as HPLC-purified oligo.

sgRNA synthesis ● TIMING 2 d

• 10.Add nuclease-free water to lyophilized gene-specific forward oligos ordered at 10 mM scale to make 100 μM stocks. For 500 pmol oligo order, add 50 μL of nuclease-free water and resuspend to make 10 μM stocks.
• 11.Resuspend sgRNA universal reverse oligo in nuclease-free water to make a 100 μM stock.
• 12.
  Set up the annealing reaction as shown below:

  	Volume	Final concentration
  Gene-specific forward oligo (100 μM or 10 μM)	1 μL or 10 μL	2 μM
  100 μM sgRNA universal reverse oligo	1 μL	2 μM
  5x HF buffer	10 μL	1x
  10 mM dNTPs	1 μL	200 μM
  DMSO	1.5 μL	3%
  Phusion HS Flex polymerase	0.5 μL	1 U
  Nuclease free water	35 μL or 26 μL

CRITICAL STEP:

Separate synthesis of individual DNA templates is only required if those templates will subsequently be used for barcode generation in Step 25. However, using these templates for barcode generation is not a necessity. For example, some users may choose to create a standard number of reusable barcodes rather than unique barcodes for every gene they wish to interrogate. If users elect such a strategy, then DNA template synthesis can be done in a pooled format by adding 0.5 μM of each gene-specific forward oligo (corresponding to the same target gene) to the annealing reaction.

• 13.
  Perform a fill-in PCR in a thermal cycler at the following settings:

  Cycle number	Denature	Anneal	Extend
  1	98 °C, 2 min
  2		50 °C, 10 min
  3			72 °C, 10 min

• 14.Perform cleanup using a PCR cleanup kit per manufacturer’s instructions. Use a ZR96 DNA Clean and Concentrator-5 cleanup kit for cleaning up larger number of samples.
• 15.Elute the PCR product in 15–20 μL nuclease-free water. Determine concentration using a nanodrop. ?TROUBLESHOOTING

  ■ PAUSE POINT Isolated PCR product can be stored at −20 °C for up to 6 months. Store the individual DNA templates in separate tubes/wells of a 96-well plate. Any of the four DNA templates for each gene will be used in Step 25 to generate the DNA barcodes.

• 16.In a separate tube/ 96-well plate, pool template DNAs targeting the same gene in an equimolar ratio. Typically, equal volumes of the four DNA templates can be combined together.

  ■ PAUSE POINT Pooled DNA templates can be stored at −20 °C for up to 6 months.

• 17.
  Set up an in vitro transcription reaction using MEGAscript SP6 kit as follows:

  	Volume	Final concentration
  ATP	1 μL	5 mM
  CTP	1 μL	5 mM
  GTP	1 μL	5 mM
  UTP	1 μL	5 mM
  10x Reaction buffer	1 μL	1x
  RNAse inhibitor (40 U/μL)	0. 25 μL	1 U
  SP6 enzyme mix	1 μL	1x
  Pooled template DNA	200 ng	20 ng/ μL
  Nuclease-free water	to 10 μL

• 18.Incubate at 37 °C overnight (16 h) in a thermal cycler.
• 19.Add 10 μL of water to bring sample volume to 20 μL. Add 1 μL of 2 U/ μL Turbo DNase I. Mix and then incubate at 37 °C for 20 min.
• 20.Cleanup RNA using an RNA Clean and Concentrator-5 kit or a ZR96 RNA Clean and Concentrator-5 as per manufacturer’s instructions. Elute in 15 μL nuclease-free water.
• 21.In a separate tube, add 1 μL of RNA and 9 μL of water. Mix and measure RNA concentration of each sample using a Nanodrop. Calculate the stock RNA concentration. ?TROUBLESHOOTING

  ▲ CRITICAL STEP Dilute the RNA before concentration measurement

• 22.Heat the diluted sgRNA samples at 70 °C for 5 min. Transfer the tubes to ice to rapidly cool down the samples.
• 23.Run 400–800 ng of the RNA on a 2% agarose gel containing ethidium bromide to determine RNA integrity. A single band of ~100 bp is expected (Figure 4b). RNA forms secondary structures which may occasionally result in more than one band. This should not affect the sgRNA editing efficiency. Low molecular weight smear is indicative of degraded sgRNAs. ?TROUBLESHOOTING

  ! CAUTION Ethidium bromide is a carcinogen.

• 24.Flash-freeze stock sgRNAs in liquid nitrogen and store at −80 °C.

  ■ PAUSE POINT Pooled sgRNAs can be stored at −80 °C for up to 6 months. Avoid multiple freeze-thaws.

Barcode generation ● TIMING 3 h

CRITICAL:

The DNA template generated in Step 15 is used as barcode. Any one of the 4 DNA templates targeting gene(s)-of-interest is extended to include a universal primer for sequencing and end-modified with 5’-biotin for better stability and to aid in barcode retrieval.

• 25.
  Set up a PCR reaction to extend and end-modify the template DNA as follows:

  	Volume	Final concentration
  10 μM Barcode forward	2.5 μL	0.5 μM
  10 μM Barcode reverse	2.5 μL	0.5 μM
  5x HF buffer	10 μL	1x
  10 mM dNTPs	1 μL	200 μM
  DMSO	1.5 μL	3% v/v
  Template DNA from Step 15	50 ng	1 ng/μL
  Phusion HS Flex polymerase	0.5 μL	1 U
  Nuclease free water	To 50 μL

• 26.
  Perform a PCR reaction in a thermal cycler using the following settings:

  Cycle number	Denature	Anneal	Extend
  1	98 °C, 30 s
  2-31	98 °C, 10 s	53 °C, 30 s	72 °C, 10 s
  32			72 °C, 2 min

• 27.Isolate the barcode after PCR cleanup using a Zymo Clean and Concentrator-5 or ZR96 Clean and Concentrator-5. Elute in 30 μL nuclease-free water.
• 28.Determine DNA barcode concentration using a Nanodrop. ?TROUBLESHOOTING

  ■ PAUSE POINT DNA barcode can be stored at −20 °C for up to 6 months.

MIC-Drop generation ● TIMING 20 min – 1 h

• 29.
  Make the RNP complex for each gene by combining the following:

  	Volume	Final concentration
  Multiplexed sgRNAs (from Step 24)	5000 ng	200 ng/μL
  20 μM EnGen Cas9	4.2 μL	3.36 μM
  10x NEBuffer r3.1	2.5 μL	1x
  Nuclease free water	To 19 μL

• 30.Incubate the RNP mix at room temperature for 5–10 min. Add 2.5 μL of 100 ng/μL DNA barcode from Step 28 and 3.5 μL of 0.5% Phenol Red dye.
• 31.Switch on the BioRad QX200 droplet generator. Set up the cartridge. Droplets can be generated for up to 8 samples in a single cartridge. Mix and add 20 μL of RNP and barcode mix from Step 30 to each well of the cartridge labeled “sample” (Supplementary Video 1). If making droplets for fewer than 8 samples, fill the remaining wells with 1x NEBuffer r3.1.

  ▲ CRITICAL STEP Ensure no air bubbles are trapped in between the sample and the cartridge.

• 32.Transfer 70 μL of 3% FS-HFE to the cartridge wells that are labeled “oil”. Seal cartridge using the rubber gasket.
• 33.μLace the cartridge in the droplet generator and close the lid of the instrument. Droplet generation should commence once the lid is closed.
• 34.Once droplet generation is done, carefully transfer the droplets (40 μL) to another PCR tube containing 30–50 μL of 3% FS-HFE (Supplementary Video 1).

  ▲ CRITICAL STEP Avoid shaking of droplets. When pipetting droplets, retract and depress the pipette plunger slowly to avoid breaking or fusing of droplets. If tubes containing droplets are dropped, breakage or fusion of droplets is likely.

• 35.Examine droplet integrity by viewing under a stereoscope (Supplementary Figures 3a-b). This can be achieved while droplets are in transparent tubes or by pipetting a small volume of droplets onto a glass microscope slide. Droplets should appear identical in size. If substantial numbers of fused or split droplets are observed, droplets should be re-generated to ensure proper droplet integrity is maintained for subsequent injections. ?TROUBLESHOOTING
• 36.If generating droplets for more than 8 samples, repeat Steps 31–35 for the remaining samples. Once generated, maintain droplets at 4 °C by placing the PCR tubes on ice.

  ■ PAUSE POINT Droplets can be stored at 4 °C for 2–3 weeks, however, for best results, inject droplets within 1 week of making.

  ▲ CRITICAL STEP Do not freeze droplets. Freezing and thawing results in droplet disintegration.

• 37.To intermix droplets targeting multiple genes: In a separate PCR tube, add 50 μL of 3% FS-HFE. Transfer equal volumes (2–5 μL) of each droplet to this tube. Also include droplets targeting a gene with known phenotype as an injection control. Mix the droplets gently.

  ▲ CRITICAL STEP Because 3% FS-HFE density is higher than the aqueous droplets, the droplets float on the surface of the oil; this creates a visible layer of droplets on top of the 3% FS-HFE base volume (Supplementary Figure 3a). When intermixing droplets, make sure to pipette only from this droplet layer to ensure an equal number of droplets is transferred from each separate tube of droplets. Pipetting from below the droplet layer will lead to an under-representation of the droplets in question.

  ▲ CRITICAL STEP When transferring droplets from one tube to another, always ensure the receiving tube has 3% FS-HFE.

  ■ PAUSE POINT Droplets targeting up to 50 genes can be intermixed and stored at 4 °C for 2–3 weeks. However, for best results, inject intermixed droplets within 1 week of making.

Microinjection of droplets ● TIMING 1 h – 3 h

• 38.Prepare the intermixed droplets (Steps 29-37), pull needles, and make the injection mold prior to the day of injection (see Equipment Setup).
• 39.Set up paired zebrafish crosses, with a divider separating the male and the female, the evening before planned injections. Set up 8–10 paired crosses to ensure at least a few of the pairs will lay eggs. Alternatively, set up group crosses if a large number of eggs is desired.
• 40.On the morning of injection, transfer zebrafish to fresh water. Pull dividers from a portion of the crosses and allow zebrafish to lay eggs. The time from removing dividers to egg laying will vary from cross to cross. For best results, pull the divider shortly after facility lights turn on. Steps 41-44 below (which involve loading and prepping the injection needle) can be performed while fish are breeding or while waiting for fish to breed after removing dividers. Stagger pulling the dividers such that separate batches of embryos may be collected at different times to provide a continuous source of single-cell stage embryos for injection.
• 41.Pipette 3–5 μL of droplet mixture into a microloader tip. Be sure to keep the microloader tip in the droplet layer to pipette a majority of droplets rather than 3% FS-HFE (even doing so, 3% FS-HFE will make up to 30% of the total volume). Transfer the droplets to a microinjection needle, by placing the microloader tip fully into the microinjection needle and slowly depressing the pipette plunger while gradually lifting the tip out of the needle to avoid generating air bubbles. (Supplementary Video 2).

  ▲ CRITICAL STEP Pipetting 3–5 μL of droplets should result in transfer of around 500 droplets.

• 42.If air bubbles have been trapped in the needle, the needle may be flicked gently to rid any such bubbles. Attach the needle to the injector.

  ▲ CRITICAL STEP Avoid vigorous shaking or dropping the microinjection needle, which may lead to droplet coalescence. Do not re-pipette droplets from the microinjection needle.

• 43.Trim needle using a pair of tweezers such that the tip width is ~10–20 μm (Supplementary Video 3).

  ▲ CRITICAL STEP For best results, trim the thin and flexible part of the microinjection needle. Make finer trims while looking under the microscope to gradually increase the tip width.

• 44.Because of the density difference between the oil and the droplets, the droplets will float on the surface of the oil (Supplementary Figure 3b). Use “Clear” setting on the instrument to get rid of the excess oil at the tip of the microinjection needle (Supplementary Video 3).

  ▲ CRITICAL STEP Typical “Clear” pressure on the instrument is set at 50–60 psi. Continue pressing “Clear” until all the excess oil is pushed out of the needle. At 10–20 μm tip width, when pressed “Clear”, the oil should flow out smoothly at a rate of 1–2 droplet/s. Stop clearing once a few droplets come out of the microinjection needle.

• 45.Collect clutches of embryos from successful breeding pairs or groups using husbandry techniques standard to your lab or animal facility. Set aside a portion of embryos (we recommend ~50 if possible) to serve as un-injected controls. Arrange embryos in the injection gel from Equipment Setup; use a number of embryos that the user is comfortable injecting in ~30 minutes (before they reach the 2-cell stage). Tuck them gently in the grooves of the gel using a probe, and orient them in the desired position for injection. Ensure embryos are covered with a small volume of E3 media to avoid drying out.
• 46.Push the foot pedal until part of a droplet is ejected into the E3 media on the injection gel. Insert the injection needle into an embryo and inject the remaining droplet in the embryo (Supplementary Videos 4-6). Pull the needle out of the embryo just before the subsequent oil reaches the tip of the needle, and eject any remaining portion of the droplet into the E3 media. Aim to inject approximately ½ - ¾ of droplet.

  ▲ CRITICAL STEP It should take 2–3 taps of the foot pedal to inject one droplet. Adjust injection time and injection pressure if needed. Typical settings are 40–70 ms for injection time and 6–9 psi for injection pressure. Although a small volume of the oil, if injected, is inconsequential to the developing embryos, we recommend avoiding injecting oil in the embryos. This can be achieved by adjusting the injection time and pressure.

• 47.Push out the oil in between two adjacent droplets in the injection mold. Repeat injection of the subsequent droplets in more embryos. ?TROUBLESHOOTING

  ▲ CRITICAL STEP It should take 1–2 taps of the foot pedal to clear out the oil in between two neighboring droplets.

  ▲ CRITICAL STEP There will be more oil in between the first 10–30 droplets and therefore will require more than 1–2 taps to clear it out (Supplementary Video 4). If the gap between two adjacent droplets is large, use the “clear” setting to push out more of the excess oil. As more droplets are injected, the space between two adjacent droplets will decrease (Supplementary Videos 5-6). Adjust the pressure settings (reduce the injection pressure and/or time) to ensure single droplet injection per embryo.

• 48.Inject 300–500 embryos from different clutches (an experienced researcher can inject ~500 embryos in ~3 hours). CRITICAL STEP: For best results, only inject fertilized single-cell embryos.
• 49.Transfer the injected embryos to a petri dish in E3 medium with methylene blue. Transfer should occur after all embryos in the injection gel have been injected, or careful steps should be taken to ensure that only injected embryos are transferred to dishes meant for injected larvae used in subsequent phenotyping/screening.

  Wash the injected embryos 1x with E3 to get rid of residual 3% FS-HFE, and RNP and barcode mix.

  ▲ CRITICAL STEP Washing the embryos and transferring them to separate petri dish is critical to minimize carryover contamination of residual barcode from the injection mold.

• 50.Transfer and split 30–50 embryos per petri dish containing E3 with methylene blue.
• 51.Save un-injected embryos from the same clutch as controls. Incubate injected and un-injected embryos at 28.5 °C.

Phenotyping ● TIMING 1 h – 4 h

• 52.12–24 hours post fertilization (hpf), remove and count any dead embryos. Also, remove and count embryos that show gross deformities resulting from nucleic acid toxicity. Assess the survival and general morphology of injected embryos compared to un-injected siblings. Injection should not cause more than a 5–10% decrease in survival when compared to un-injected controls. Place the embryos back in incubator. ?TROUBLESHOOTING
• 53.Screen embryos for phenotype(s)-of-interest at desired timepoint. The screening method and the development time/stage will vary based on the interested phenotype (see Experimental Design). ▲ CRITICAL STEP If sample fixation is required for determining the phenotype-of-interest, investigators should perform preliminary experiments to determine barcode retrieval is compatible with their required fixation method. ?TROUBLESHOOTING

Barcode retrieval and sequencing ● TIMING 1 d

• 54.Isolate embryos displaying the phenotype(s)-of-interest. If necessary, dechorionate the embryos using a pair of tweezers.
• 55.Wash the embryos 1x with E3, and then transfer embryos to a new petri dish using a transfer pipette. Wash embryos 3x with E3, each time aspirating off most of the media and then refilling.

  ▲ CRITICAL STEP Extensive washing is necessary to get rid of any carryover residual DNA barcode sticking to embryos.

• 56.Transfer each embryo to a PCR strip tube or a 96-well plate along with 10 μL of E3 media.
• 57.Add Proteinase K at a concentration of 0.2 mg/mL to the desired volume of 2x PCR lysis buffer. Add 10 μL of PCR lysis buffer containing Proteinase K to each of the samples.
• 58.Centrifuge the samples at 3000xg for 2 min at room temperature. Incubate at 50–55 °C overnight (16 h) to lyse the embryos. Heat inactivate Proteinase K at 95 °C for 10 min.

  ■ PAUSE POINT Heat inactivated samples can be stored at −20 °C for > 1 month.

• 59.Centrifuge the samples at 3000xg for 5 min at room temperature to pellet the debris.
• 60.
  In a microcentrifuge tube, prepare a PCR master mix to amplify the DNA barcode.

  	Volume for 1 sample	Final concentration
  100 μM Barcode sequencing forward primer (Supplementary Table 1)	0.075 μL	0.5 μM
  100 μM Barcode sequencing reverse primer (Supplementary Table 1)	0.075 μL	0.5 μM
  5x GoTaq buffer	3 μL	1x
  10 mM dNTPs with 30% dUTP	0.3 μL	200 μM
  GoTaq polymerase	0.075 μL	0.375 U
  UDG	0.075 μL	0.375 U
  Nuclease free water	9.4 μL
  Total	13 μL

• 61.Aliquot 13 μL per tube/well of the PCR master mix into PCR tubes or PCR plates. Add 2 μL of the cleared embryo lysate containing DNA barcode from Step 58. Include a negative control samples that contains 2 μL water instead of lysate. Mix by pipetting.
• 62.
  Cap/Seal the tubes and set up a PCR reaction using the following settings:

  Cycle number	UDG activity	UDG inactivation	Denature	Anneal	Extend
  1	37 °C, 10 min
  2		95 °C, 10 min
  3-36			95 °C, 30 s	55 °C, 30 s	72 °C, 15 s
  37					72 °C, 5 min

• 63.Ensure proper barcode amplification by running a few lysate containing samples and the negative control on a 2% agarose gel. Load 5 μL of the PCR reaction samples in the gel. Resolve the gel at 120 V for 30 min. A single band of 215 bp is expected in samples containing the lysate. Negative control should have no amplification product (Figure 4d). ?TROUBLESHOOTING
• 64.To perform enzymatic cleanup, transfer 5 μL of each sample to another PCR tube/plate. Add 0.5 μL of Exonuclease I and 1 μL of shrimp alkaline phosphatase to each sample.
• 65.Mix and incubate samples at 37 °C for 15 min, and then heat inactivate at 80 °C for 15 min.
• 66.Dilute the sample as needed for sequencing, we typically dilute 3–5 fold with nuclease free water. Submit 2–5 μL for sequencing. Barcodes contain sites for universal primers M13F and M13R. ?TROUBLESHOOTING

  ▲ CRITICAL STEP It is important to submit the correct amount of sample for sequencing. Sample concentration can be estimated by running them on a 2% agarose gel and comparing the band intensities against a known amount of molecular weight marker. Alternatively, samples can be cleaned up using a column-based cleanup method and concentration determined using Nanodrop.

Identifying hits and secondary validation ● TIMING variable based on the phenotype being studied

• 67.Tally the number of barcodes for each gene and list the phenotypes observed for each gene knockout.
• 68.Prioritize genes with high barcode frequency among the sequenced samples and those that yielded consistent phenotypes across multiple embryos.
• 69.To perform secondary validation of hit candidate genes, prepare a mix of sgRNAs targeting the gene-of-interest (100 ng/μL), and EnGen Cas9 protein (2 μM) in 1x NEBuffer r3.1. Incubate at room temperature for 10 min. Add phenol red dye (0.05% v/v).
• 70.Calibrate the injection needle such that the injection volume is 1–1.5 nL (ref. 16). Inject sgRNAs and Cas9 RNP mix into single-celled embryos.
• 71.At the desired time point, count number of injected embryos with the desired phenotype to assess phenotypic penetrance. ?TROUBLESHOOTING

Rescue of phenotype with mRNA coinjection ● TIMING variable based on the phenotype being studied

CRITICAL:

To ensure that the observed phenotype is due to on-target gene perturbation, one can perform phenotype rescue by co-expressing the gene being targeted.

• 72.Clone the zebrafish gene(s)-of-interest in pcs2+ vector. CRITICAL STEP: We use zebrafish coding sequences that are codon optimized and optimized to remove BamHI and EcoRI restriction enzyme sites for more efficient cloning in pcs2+ vector.
• 73.Linearize the vector and perform in vitro transcription to generate mRNA using an mMessage mMachine SP6 kit, as per manufacturer’s instructions.
• 74.Prepare a mix of sgRNAs targeting the gene-of-interest (100 ng/μL), and Cas9 protein (2 μM) in 1x NEBuffer r3.1. Incubate at room temperature for 10 min. Add the mRNA encoding the targeted gene (200–500 ng/μL) and phenol red dye (0.05% v/v). As control, make a separate mix of only sgRNAs and Cas9 protein.
• 75.Calibrate the injection needle and inject 1–1.5 nL of the mix in single-celled embryos.
• 76.Score phenotypes at the desired time point. ?TROUBLESHOOTING

Troubleshooting

Troubleshooting advice can be found in Table 1.

Table 1:

Troubleshooting

Step	Issue	Possible reason	Solution
3	Gene ID not found in database	Gene is labeled differently in GRCz10 database	Use the transcript ID or Ensembl ID
4	No common sgRNAs targeting all isoforms of the gene	Several identified isoforms with differing lengths	Use a specific isoform as target; Use “Union” instead of “Intersection” on CHOPCHOP under Options → General → Isoform consensus determined by. The default setting on CHOPCHOP is “Intersection”, which searches for sgRNAs only in regions present in all isoforms of the gene. “Union” searches for sgRNAs in regions present in any isoform.
15	DNA template concentration low	Inefficient PCR assembly; Not enough templates	Redo the PCR reaction; Increase concentration of forward and reverse oligos
21	sgRNA concentration is low	Poor in vitro transcription	Repeat in vitro transcription using RNAse-free conditions; Set up multiple in vitro transcription reactions, combine during sample cleanup
23	sgRNA degraded	sgRNAs are susceptible to degradation by RNAses	Repeat in vitro transcription using RNAse-free conditions
28	DNA Barcode concentration is low	Inefficient PCR	Redo the PCR reaction; Set up multiple reactions, combine samples during cleanup
35	Droplets are not uniform	Clogged instrument	Do not use samples with high viscosity. Use recommended buffers
47	High backpressure during droplet injection	Excess oil in the needle	Use “clear’ to push out excess oil.
47	Oil is being accidentally injected in embryos (Supplementary Video 7)	High injection pressure; needle pore is too wide	Reduce injection pressure so that each droplet is injected in 2-3 foot pedal pushes; Ensure needle is trimmed to the right bore size.
47	Droplets have a lot of oil in-between	Excess oil in the needle	Use “clear’ to push out excess oil.
47	Droplets are too close to each other	High injection pressure; No oil in the needle	Reduce injection pressure so that each droplet is injected in 2-3 foot pedal pushes
47	Needle clogged (Supplementary Video 8)	Debris in the needle	Increase pressure to clear debris; Press “Clear”.
52	Low embryo survival	Poor injection technique; Unhealthy clutch of embryos	Optimize injection technique; Redo injection in a separate clutch
53	High incidence of gross deformity in injected embryos	Unhealthy clutch of embryos; High nucleic acid concentration	Redo injection in a separate clutch; Remeasure DNA barcode and RNA concentration; Reduce DNA barcode amount, if needed.
63	Negative control also shows barcode amplification	Ambient DNA barcode contamination	Setup barcode amplification reaction in a PCR workstation; Cleanup workstation after use
63	No barcode amplification	Barcodes have degraded; PCR amplification did not work	Perform barcode amplification at an earlier time point; Avoid adding tissue debris in PCR reaction; Set up a positive PCR control (DNA barcode spiked-in in lysate from un-injected sample)
66	Sanger sequencing results show overlapping signals	Multiple droplets injected in the embryo; Carryover contamination	Wash embryos thoroughly before lysis and DNA amplification
66	Poor sanger sequencing quality	Suboptimal DNA concentration	Submit the recommended amount for sequencing; Column cleanup samples before sequencing
71	“Hits” from screen do not show phenotype during secondary validation	Inefficient target editing during validation	Ensure efficient target gene editing; Re-inject droplets targeting the gene and assess phenotype
76	Phenotype is not rescued upon mRNA coinjection	Poor expression of the protein	Ensure the protein is being expressed using western blot of whole embryo lysate or fluorescent imaging (if tagged with a fluorescent marker)
76	Phenotype is not rescued upon mRNA coinjection	Phenotype is a result of off-target sgRNA activity	Inject individual sgRNAs in separate embryos and measure phenotype penetrance. If phenotype still observed, ensure targeted gene is being edited.

Timing

Steps 1–9: Design and selection of gRNAs targeting Gene(s)-of-Interest 5–10 min/gene

Steps 10–24: sgRNA synthesis 2 d

Steps 25–28: Barcode generation 3 h

Steps 29–37: MIC-Drop generation 20 min–1 h

Steps 38–51: Microinjection of droplets 1 h–3 h

Steps 52–53: Phenotyping 1 h–4 h

Steps 54–66: Barcode retrieval and sequencing 1 d

Steps 67–71: Identifying hits and secondary validation variable based on the phenotype being studied

Steps 72–76: Rescue of phenotype with mRNA coinjection variable based on the phenotype being studied

Anticipated Results

Assess successful oligo assembly by agarose gel electrophoresis of the DNA templates. A single band with a size of 115/116 bp is expected (Figure 4a). If one is making DNA templates at a large scale, it is not feasible to assess oligo assembly for each sample. Spot check a few samples from each plate. The expected concentration of DNA template is 80–200 ng/μL.

sgRNA integrity should be checked using gel electrophoresis as specified in Steps 21–23. A single band of ~100 bp is expected (Figure 4b). RNA forms secondary structures which may occasionally result in more than one band. This should not affect the sgRNA editing efficiency. Low molecular weight smear is indicative of degraded sgRNAs. The expected concentration of pooled sgRNAs is 500–3000 ng/μL.

DNA barcode quality should also be evaluated using gel electrophoresis. The barcode should run as a single band of 137 bp (Figure 4c).

Ensure MIC-Drop generation went smoothly by visually inspecting the droplets. Droplets should be uniform, and ~100 μm in diameter, each droplet containing ~0.7–0.9 nL of the sgRNA-Cas9 and DNA barcode mix⁷⁹ (Supplementary Figure 3a). The approximate amounts of each component per droplet are as follows: DNA barcode (10 pg), Cas9 protein (540 pg), and sgRNAs (200 pg). Occasional coalescing of a couple droplets is normal (Supplementary Figure 3b). Similarly, visually check that droplets have not coalesced during transfer into the injection needle (Supplementary Figure 3c). Accidental fusion/breaking of a couple droplets is normal but should be avoided by handling the droplets with care.

A single user can inject 300–500 embryos in a single morning. For best results, ensure injections are done in undivided embryos in the single cell. Injection of divided embryos or injection in the yolk sac may result in reduced phenotypic penetrance. Gene perturbation efficiencies can be determined using a variety of techniques described elsewhere^16,20. A 5–10% reduction in viability is expected in injected embryos compared to un-injected controls. Dramatically reduced viability may suggest poor injection technique and/or bad embryo batch. 5–10% of the viable embryos may show gross morphological defects due to nucleic acid toxicity²⁰. A significant increase in deformed embryos could be a result of inaccurate RNA/DNA barcode concentration measurement. An increased number of dead or deformed embryos in injected samples could also be a result of targeting genes that cause early lethality (before 24 hpf) or are important for early somitogenesis.

Barcodes should be efficiently retrieved from larva up to 7 dpf. The typical amplicon concentration is around 50–150 ng/μL. We dilute the sample post enzymatic cleanup and submit 10–20 ng of the amplicon for sequencing.

Supplementary Material

Supplementary Table 1

Supplementary Table 1. A list of sgRNA target sequences and primers used in the manuscript. sgRNA target/spacer sequences of control scrambled sgRNAs, sgRNAs targeting rx3 and tbx16 genes, and primers used for barcode generation and barcode amplification.

Video 1

Supplementary Video 1. Droplet generation using a QX200 droplet generator. Droplets targeting up to 8 genes are generated during a single run of the droplet generator. A single run takes ~2 min.

Video 2

Supplementary Video 2. Transfer of MIC-Drops into a microloader tip. Colored droplets (used as proxies for droplets targeting different genes) being loaded into a microloader tip.

Video 3

Supplementary Video 3. Trimming of the microinjection needle for droplet injection. After transferring droplets to a microinjection needle, the needle is trimmed to the desired width. Excess oil is cleared out until the first droplet reaches the tip of the injection needle.

Video 4

Supplementary Video 4. Injection of the first MIC-Drops in zebrafish embryos. The first 10–30 droplets are separated by larger volume of 3% FS-HFE and require several pushes of the foot pedal (5–10 pushes) to clear out the excess oil. In this video, food coloring is used in the droplets for better visualization. Food coloring may be toxic to developing embryos and should not be used in experiments.

Video 5

Supplementary Video 5. Injection of subsequent MIC-Drops in zebrafish embryos. After the first 10–30 droplets have been injected, the remaining droplets are closer together, and require only 2–3 pushes of the foot pedal to clear out the excess oil. In this video, food coloring is used in the droplets for better visualization. Food coloring may be toxic to developing embryos and should not be used in experiments.

Video 6

Supplementary Video 6. Injection of droplets containing phenol red, as used in experiments.

Video 7

Supplementary Video 7. Video illustrating accidental injection of an oil droplet in an embryo. Although injection of a small volume of oil is non-consequential to embryo development, it is best to avoid injecting oil.

Video 8

Supplementary Video 8. Clearing a clogged needle. Occasionally, injection needle can get clogged. The needle can be unclogged by pressing the “clear” button or trimming the needle.

SI

Supplementary Figure 1. Set up for droplet generation. Droplets are generated using a QX200 droplet generator per manufacturer’s instruction.

Supplementary Figure 2. Instrument setup for microinjection. A picoliter injector is connected to an air flow and an injector mounted on a micromanipulator. The picoliter injector is also connected to a foot pedal (not shown) that delivers a calibrated amount of air pressure when pressed. A dissecting microscope with a light source is used for observing and injecting zebrafish embryos.

Supplementary Figure 3. Cas9-sgRNA containing droplets are stable during handling. (a) Aqueous droplets are less dense than 3% FS-HFE, and hence float on the surface of the oil. In this figure, aqueous droplets containing food coloring are used as proxies for droplets targeting different genes. In MIC-Drop screens, phenol red dye is used to track droplet injection. Food coloring may be toxic to developing embryos and should not be used in experiments. (b) Droplet generation and handling results in occasional coalescing of a couple of droplets. (c) Droplets transferred to a microinjection needle are intact.

Source data

Source Data for Figure 4d. Uncropped blot.

Data availability statement

All data pertaining to this manuscript are included in figures and tables. Raw data for Figure 4 is provided as Source Data. Please direct reasonable material requests to R.T.P.