Gene duplication and redundancy often pose challenges in attributing phenotypic effects to individual genes and discerning their contributions to biological processes of interest (Ritter, et al., 2013; Ewen-Campen, et al., 2017) . Commonly used approaches such as forward genetic screens often fail to identify the genes with redundant functions. To this end, we explored the possibility of disrupting many genes in a single animal with multiple rounds of CRISPR/Cas9 mediated genome editing. We utilized the previously outlined strategy to disrupt the genes by integrating a single strand DNA oligo (ssODN) via homologous recombination (Dokshin, et al., 2018; Ghanta & Mello, 2020) . To ensure the proper gene disruption, the integration of ssODN involved not only the insertion of in-frame stop codons but also the removal of 14 or 16 bases of coding sequence (Table 2 and 3). It also introduced a unique restriction enzyme cutting site for genotyping. We first sought to determine how many genes could be simultaneously disrupted in a single injection. A collection of GPCR genes were selected for the evaluation (Table 2 and 3). When one gene was targeted, we kept 16 transgenic F1 rollers for the downstream analysis. Overall, the editing efficiency was consistently high across 20 independent trials, exhibiting an average efficiency of 80% ( Figure 1A ). Similar to the earlier observations (Dokshin, et al., 2018) , we obtained both F1 heterozygotes and putative F1 homozygotes ( Figure 1B ). As previously indicated (Dokshin, et al., 2018) , it is likely that certain F1 homozygotes were trans -heterozygous, carrying two distinct types of insertions or a combination of an insertion and a deletion that removed the binding site of genotyping primers. Under our experimental conditions, we had an average efficiency of 52% in generating F1 heterozygotes, while the frequency of obtaining F1 homozygotes accounted for 28% in the total of 20 gene editing events ( Figure 1B ). To evaluated if the gene function was eliminated in the putative F1 homozygotes, we targeted at two genes npr-1 and odr-10 since their null mutants exhibit clear and robust phenotypes (de Bono & Bargmann, 1998; Sengupta, et al., 1996) . In each gene disruption, we retained 8 putative F1 homozygotes, and singled 12 F2s from each F1 homozygotes. Aggregation and chemotaxis assays were performed at F3 stage ( Figure 1C ). In both cases, no F3 offspring displayed either wild type or heterozygous phenotypes ( Figure 1D -G), suggesting that the putative F1 homozygotes are likely to be null mutants . However, opting for F1 heterozygotes is always advantageous in order to maintain a clear genotype of strains, particularly in cases where precise genome editing is required such as generating point mutations or inserting epitope tags (Dokshin, et al., 2018) .

When two genes were targeted simultaneously, we preserved 24 F1 rollers for subsequent analysis after each injection. In a total of 20 independent editing events, the efficiency of concurrent editing for both genes remained consistently high, with an average efficiency of 60% ( Figure 1H ). We also successfully recovered F1 animals that were heterozygous for both targeted genes in all our injections, exhibiting an average efficiency of 22% ( Figure 1H ). The simultaneous editing of three genes occurred less frequent but remained achievable, with an average efficiency of 43% ( Figure 1I ). Picking 24 transgenic F1 animals proved sufficient to obtain the triple mutants in each of our attempts ( Figure 1I ). In particularly, F1 animals containing the heterozygous form of all three targeted genes were obtained in all 20 trials, with an average efficiency of 13% ( Figure 1I ). Genome editing efficiency decreased substantially when four genes were simultaneously targeted in a single injection. We hardly recovered any quadruple mutants in all our attempts if less than 32 F1s were picked. In particular, the efficiency of obtaining F1 animals that were heterozygous for all four genes was very low in a total of 7 editing events ( Figure 1J ). Therefore, using our strategy, it is possible to pursue the editing of up to three genes with relatively high efficiency.

Under certain circumstances it is desirable to disrupt more than 3 genes in a single animal, which means that multiple rounds of gene editing are needed. We wondered if the repetitive gene editing would attenuate the efficiency of editing process. To probe this, we performed gene editing repeatedly in the same strain, with three genes targeted in each round of editing. We conducted three independent genome editing experiments in parallel, targeting a total of 45 genes with the aim of disrupting 15 genes in each animal (Table 2 and 3). Again, 24 F1 rollers were retained in each round of injection. In all three independent trials, we successfully achieved simultaneous editing of three genes in each of the five consecutive rounds of injections ( Figure 1K ). Importantly, we did not observe any noticeable reduction of editing efficiency for isolating F1 animals with triple editing or triple heterozygotes throughout the experiments ( Figure 1K and L ). These data suggest that repetitive genome editing in the same strain of C. elegans does not significantly affect the editing efficiency. We anticipate that this approach can be used to disrupt the redundant genes or a set of genes within a specific family in C. elegans .

C. elegans maintenance

C. elegans strains were maintained under standard conditions (Brenner, 1974) . The Bristol N2 were used as wild type. Strains used in this study were listed in Table 1.

CRISPR-based gene editing

The strategy involved the homology-directed integration of the single strand DNA oligo (ssODN) (Dokshin, et al., 2018) . The optimized ribonucleoprotein complexes containing Cas9 protein (IDT, #1081059), predesigned crRNA (https://eu.idtdna.com/site/order/designtool/index/CRISPR_PREDESIGN) and tracrRNA (IDT, #1072534) were mixed with ssODN donor template (synthesized by IDT) and a roller co-injection marker (pRF4:: rol-6(su1006) ) (Mello & Fire, 1995) , and injected into the gonad of C. elegans . The predesigned crRNAs targeted at the earliest possible exon, or the common exons if different splicing isoforms exist (Table 3). The rol-6 marker plasmid was prepared with midi-prep kit (QIAGEN, Cat. No.12143). The ssODN templates contained two 35-base homology arms flanking the targeted PAM sites. Between the homology arms, two in-frame stop codons were included. A unique restriction enzyme cutting site was also built in for genotyping. The insertion of ssODN introduced the stop codons and restriction enzyme sequence into the targeting site while simultaneously generated frameshift. The F1 roller animals were picked and genotyped for the integration of ssODN. Most of the genotyping primers amplified the fragments between 400 bp and 1000 bp surrounding the ssODN insertion sites. Restriction enzyme digestion of PCR products would generate two fragments of different sizes in the homozygous animals, three bands in the heterozygotes and only one band in the wild type. For many genotyping primers, longAMP Taq polymerase (NEB, M0323L) worked much better for the amplification. The injection mixtures for the disruption of different number of genes were prepared as the following:

i) One gene (Dokshin, et al., 2018) :

1) 0.5 μl Cas9 (10 mg/ml from IDT)

2) 5 μl tracrRNA (0.4 mg/ml in IDT duplex buffer)

3) 2.8 μl crRNA (0.4 mg/ml in IDTE pH7.5)

4) Thoroughly mix these three components and incubate the mixture at 37°C for 10 to 15 minutes.

5) 2.2 μl ssODN (1 mg/ml in nuclease free H ₂ O)

6) 2 μl rol-6 co-injection marker (600 ng/μl in nuclease free H ₂ O).

7) Use nuclease free H ₂ O to bring the volume to 20 μl.

8) Spin at 14 000 rpm for 10 minutes at room temperature, transfer 17 μl of the mixture to a new tube for the injection.

ii) Two genes:

1) 0.5 μl Cas9 (10 mg/ml from IDT)

2) 6 μl tracrRNA (0.4 mg/ml in IDT duplex buffer)

3) 2 μl of each crRNA (0.4 mg/ml in IDTE pH7.5)

4) Thoroughly mix these three components and incubate the mixture at 37°C for 10 to 15 minutes.

5) 2.5 μl of each ssODN (1 mg/ml in nuclease free H ₂ O)

6) 2 μl rol-6 co-injection marker (600 ng/μl in nuclease free H ₂ O).

7) Use nuclease free H ₂ O to bring the volume to 20 μl.

8) Spin at 14 000 rpm for 10 minutes at room temperature, transfer 17 μl of the mixture to a new tube for the injection.

iii) Three genes:

1) 0.5 μl Cas9 (10 mg/ml from IDT)

2) 6 μl tracrRNA (0.4 mg/ml in IDT duplex buffer)

3) 1.9 μl of each crRNA (0.4 mg/ml in IDTE pH7.5)

4) Thoroughly mix these three components and incubate the mixture at 37°C for 10 to 15 minutes.

5) 2.1 μl of each ssODN (1 mg/ml in nuclease free H ₂ O)

6) 2 μl rol-6 co-injection marker (600 ng/μl in nuclease free H ₂ O).

7) Spin at 14 000 rpm for 10 minutes at room temperature, transfer 17 μl of the mixture to a new tube for the injection.

iv) Four genes:

1) 0.5 μl Cas9 (10 mg/ml from IDT)

2) 6 μl tracrRNA (0.4 mg/ml in IDT duplex buffer)

3) 1.7 μl of each crRNA (0.4 mg/ml in IDTE pH7.5)

4) Thoroughly mix these three components and incubate the mixture at 37°C for 10 to 15 minutes.

5) 1.9 μl of each ssODN (1 mg/ml in nuclease free H ₂ O)

6) 2 μl rol-6 co-injection marker (600 ng/μl in nuclease free H ₂ O).

7) Spin at 14 000 rpm for 10 minutes at room temperature, transfer 17 μl of the mixture to a new tube for the injection.

Behavioral assays

Aggregation and bordering were assayed as described previously (Laurent, et al., 2015; de Bono & Bargmann, 1998) with minor alterations. L4 animals were picked to a fresh plate 24h before assay. Assay plates were seeded with a 1-cm diameter OP50 lawn two days earlier. Sixty day-one adults were picked to one assay plate, and bordering and aggregation scored 2h later. Chemotaxis assays were performed as previously described (Yoshida, et al., 2012) . Low concentration of diacetyl was prepared by diluting it with pure ethanol (1:2000). 1 μl of diluted diacetyl was placed on two spots at one side of the 9 cm assay plates, and 1 μl of ethanol was added on two spots on the other side. 1 μl of NaN ₃ (1M) was also added to those spots. About 150 synchronized day one adults were used in each assay, and were allowed to roam for 1 hour. The assay plates were stored at 4°C before counting. The chemotaxis indices were calculated as (the number of worms in the attractant area – the number of worms in the control area) / the total number of worms on the plate.

Table 1. Strains used in this study

Table 2. ssODNs used for genome editing in this study

Table 3. Genome editing related material used in this study