Efficient CRISPR/Cas9-mediated genome modification of the glassy-winged sharpshooter Homalodisca vitripennis (Germar)

Microinjection of GWSS embryos on leaf discs results in high survival

We developed a platform for easy and efficient microinjection of GWSS embryos in situ. GWSS females oviposit their eggs side-by-side under the abaxial epidermis of leaves forming an egg mass²⁴. Within an egg mass, the anterior to posterior orientation of each egg is identical²⁵. In our system, females lay up to 30 eggs/mass (Fig. 1a–e), which is significantly higher than reported values of eight and twelve eggs/mass²⁴. Based on the timing of cellular blastoderm in A. pisum, N. lugens and O. fasciatus and absence of any detectable differentiation in GWSS embryos before 90 h post oviposition, we injected GWSS embryos one to two hours after egg deposition^18,19,20,25.

To date, the CRISPR/Cas9 machinery has been introduced into Hemiptera by microinjection of embryos removed from the leaf or by injection of the abdomen of gravid females. Our methods are distinct as we inject GWSS embryos in situ. Residing below the plant epidermis, embryo microinjections were simple to perform, as a mass with 20 eggs can be injected within ten minutes by a novice operator. To assess frequency and timing of egg hatch following in situ microinjection, 112 embryos were injected with water. One–two days post microinjection (dpi), a melanized scar developed at the injection site and served as a reliable indicator of embryo viability (Fig. 1b); four-five dpi, embryonic eye spots were clearly visible (Fig. 1c). There was near synchrony in egg hatch with 90.3% of the nymphs emerging at 7 dpi with 64.3% egg to nymph survival (Supplementary Table S1). These data are consistent with earlier reports of GWSS embryo development and nymph emergence on intact plants²⁵. In situ injection of embryos on leaves provides an efficient and simple platform for genome modification of GWSS using CRISPR/Cas9 technology.

High frequency mutagenesis at the cinnabar locus

The large red-brown eyes observed in early embryonic development (Fig. 1c) suggested that ommochrome and pteridine biosynthesis pathways control GWSS eye color; these pathways determine eye color in many insects including Drosophila melanogaster and the Hemiptera^22,26,27,28. Using the new assembly and annotation of the GWSS genome²², we identified GWSS orthologs for nine eye-color genes as potential targets. The cinnabar (cn) gene encodes the enzyme kynurenine 3-monoxygenase, which converts kynurenine to 3-hydroxykynurenin in the ommochrome biosynthesis pathway²⁹. The GWSS cn protein was most closely related to cn orthologs from three other hemipteran species and was chosen as a target for CRISPR/Cas9-mediated mutagenesis (Fig. 2a,c,e). sgRNAcn4-1 targets the conserved FAD-binding domain region of cn and was used in embryo microinjections with two concentrations of Cas9 (0 and 300 ng/µl) (Table 1). The presence of Cas9 in the injection mix caused a decrease in embryo survival by 31%.

When Cas9 and sgRNAcn4-1 were microinjected into GWSS embryos, we detected G0 late embryos, nymphs and adults with a spectrum of eye colors relative to wild-type GWSS (Figs. 1c,d, 3a,b, 4a). Six representative G0 adults illustrate that the eye-color mosaicism ranged from dark to bright orange with, in some cases, patches of colorless ommatidia (Fig. 4a). Based on phenotypes of G0 adults, the editing frequency was 58.9% (Table 1). In most cn G0 mutants, lines of cells with red–orange pigments were organized as arcs across the eye and red–orange ocelli were detected.

Initially, it was not clear if GWSS pair matings would be successful under our rearing conditions. Therefore, to establish cn lines, four newly emerged male and eight female G0 cn adults were pool mated. Genotypes of six representative G0 individuals (CnA-F) with cn mosaic eyes were determined (Fig. 4a–d, Supplementary Fig. S1). PCR amplification of the cn target region revealed a single PCR fragment indicating that the editing events did not cause large deletions or insertions (Fig. 4d; Supplementary Table S2). All six cn alleles contained small deletions of 5 bp (cn¹, cn², cn³), 2 bp (cn⁴), 6 bp (cn⁵) and 16 bp (cn⁶) at the sgRNAcn4-1 target site consistent with CRISPR/Cas9 mutagenesis (Fig. 4c). All alleles but cn⁵ resulted in frameshift mutations that are predicted to generate truncated proteins (Fig. 4c, Supplementary Fig. S1). We recovered more than 100 G1 nymphs and identified at least 54 with cn eye color. A mutant G1 male and mutant G1 female were selected to demonstrate transmission of cn alleles to the G2 generation. Sequencing indicated that the G1 female parent was transheterozygous (cn²cn⁴) and only the cn² allele was detected in the G1 male (Fig. 4b,c). All G2 progeny had cn mutant eyes and sequencing of the cn target region indicated that one G2 male was transheterozygous (cn²cn⁴) and the other was likely homozygous as only the cn² allele was detected (Fig. 4b,c).

The phenotypes of nymphs and adults in the G3 generation demonstrated that eye pigmentation patterns changed during GWSS development (Fig. 3e). In cn mutants, in which the brown ommochromes were not synthesized, the cells accumulating red pigments were revealed. In 1st- and 2nd-instar nymphs, the red-brown and red–orange pigmentation in wild-type and cn mutants was homogeneous across the eye, respectively. In later instars, patterning of pigments was different between wild-type and cn mutants. The lines of cells that form prominent brown horizontal stripes and pigmented arcs across the eyes were detected in the 3rd instar to adult in wild-type GWSS. In contrast, in cn mutants the horizontal pigment stripes were detected later and only in eyes of the 5th instar and adults, whereas the pigmented arcs were detected in the 3rd–5th instars and persisted through adulthood (Fig. 3e).

High frequency mutagenesis at the white locus

Using the new assembly and annotation of the GWSS genome²², we identified GWSS orthologs for the white, scarlet and brown proteins, which are ABC transporters that import pigment precursors into the developing eye (Fig. 2e). In D. melanogaster, the white (w) protein heterodimerizes with either the scarlet or brown proteins to import ommochrome and pteridine precursors, respectively^30,31. Two sgRNAs within exon 6 of the w gene that target the conserved AAA domain were designed³² (Fig. 2b). Phylogenetic trees indicated that the GWSS w protein was most closely related to w from three other hemipteran species and more distantly related to the brown and scarlet proteins (Fig. 2d).

To determine optimal conditions for editing the w gene, GWSS embryos were injected with two w sgRNAs (sgRNAw6-1 and sgRNAw6-2) and different amounts of Cas9 (Table 1). Relative to the zero Cas9 control, Cas9 (150 and 300 ng/µl) decreased embryo survival by 20.4% and 24.1%, respectively (Table 1). A higher frequency of mutagenesis was achieved with the lower Cas9 concentration. At four-five dpi, eyes were evident and developing embryos with wild-type and w mutant eye colors were easily discernible by the naked eye (Fig. 1e). These embryonic phenotypes were confirmed in G0 mutant nymphs and adults, where the degree of eye color mosaicism varied between individuals (Figs. 3c,f, 6a). In addition to mosaic eyes, G0 adults with strong w phenotypes had white ocelli.

We established four independent crosses (WhA-D) with pools of male and female G0 adults with mutant eye color (Figs. 3c, 6a). While all four crosses produced egg masses, only WhA, WhB and WhD produced nymphs; the WhC eggs did not hatch and were likely unfertilized. In WhA, we observed a mating couple and, following copulation, the male was collected for genetic analysis and the female placed in isolation for the establishment of the WhA line. Given that this line may be the result of a single-pair mating, we focused on the maintenance and genetic analysis of this line through subsequent generations.

The G0 female parent used to establish the WhA line had dark mosaic eyes and its mate had white eyes with a distinctive mosaic pattern and white ocelli (Figs. 3c, 6a). Mutant eye color was observed in G1 embryos at four-five dpi, in all instars, and the G1 through G4 adults (Figs. 3f, 6a). In eyes of some G1 w progeny, residual amounts of red pteridines were detected against a primarily white background; the red arcs mimicked the brown striations seen in wild-type eyes (Fig. 6a). The white eyes and ocelli were transmitted for four successive generations (Fig. 3f). In total, 24 mutant and nine wild-type G1 progeny were obtained from the WhA G0 cross.

In all generations (G0-G3), w mutant adults displayed notable differences in the color of their forewings relative to wild-type insects (Figs. 5a–c, 6a). The hindwing of GWSS was unpigmented with the exception of brown pigments (likely melanins) in the marginal regions. In contrast, as previously noted³³, the forewing possesses distinctive red pigmentation of the veins and interveinal spaces superimposed on the brown pigmentation of the forewing. Red pigments were detected in the partially confluent veins of the clavus (Fig. 5b). In addition, red veins flanked the interveinal spaces in the basal portion of the remigium but not in the veins surrounding the anteapical and apical regions of the forewing. Red pigments were also detected in the interveinal spaces including the basal portion of the inner anteapical and 5th apical spaces, the central portion of the central and outer anteapical spaces, and the apical portion of the outer discal and costal spaces; all of these red-pigmented regions were surrounded by a white margin (Fig. 5b). In w mutants, the red pigments of veins and the intervening spaces were absent and white unpigmented regions replaced the red domains (Fig. 5c).

To determine if the GWSS red pigments were pteridines, pteridines were extracted from forewings, hindwings and heads of wild-type, w and cn mutants (Fig. 5e–g). Pteridines were at very low to undetectable levels in wild-type, w and cn hindwings. In contrast, two classes of pteridines (with peak absorbances at 334 nm and 467 nm) were detected in heads and forewings. In both wild-type and cn GWSSs, the 334-nm pteridines had an absorbance close to 6-biopterin (standard), one of the most common pteridines in insects^34,35,36, and were more abundant in heads than forewings. In contrast, the 467-nm pteridines were more abundant in GWSS forewings (Fig. 5f). In w mutants, the 334-nm and 467-nm pteridines were reduced in the head. Even more striking was the minute quantities of both pteridine classes in forewings of the w mutant (Fig. 5f). These data suggest that the red pigments of the GWSS wings were pteridines.

To evaluate the number of mutations (inherited and not inherited) in G0 mosaic insects, we constructed w target-site amplicon libraries from the female and male parents of the WhA line, single G0 females from the G0 crosses WhB, WhC and WhD, and a non-injected, WT female. High rates of mutagenesis were detected in the G0 insects ranging from 11% (WhC female) to 99.94% (WhD female) (Supplementary Table S3). Mutations detected were consistent with Cas9 cleavage 3 bp upstream to the PAM sites and repair of the dsDNA breaks by non-homologous end joining (NHEJ) (Fig. 3, Supplementary Fig. S2a–f, Supplementary Table S3). In all five G0 GWSS adults sequenced, w sgRNAw6-1 was more efficient than sgRNAw6-2 in generating mutations. The number of unique w alleles detected in the five G0 mutant adults ranged from 227 to 1113 (Supplementary Table S3c). Deletions predominated being twice as frequent as insertions and > tenfold more frequent than substitutions (Supplementary Table S3). In contrast, the WhC female had low mutagenesis rates and substitutions predominated (Supplementary Fig. S2e).

To assess the parental origin and inheritance of the w alleles in insects derived from line WhA, we determined the sequence of the w target region in selected G1, G2 and G3 individuals, which were generated by single-pair matings (Fig. 6a). Three alleles (w^1–3) were detected in the WhA lineage (Fig. 6b,c). The G0 female parent carried the w¹ allele, which was an 83-bp insertion that began 3 bp upstream to the PAM site adjacent to the sgRNAw6-1 target; this generated a stop codon at residue 300, which would be predicted to produce a truncated and likely nonfunctional w protein (Fig. 6c, Supplementary Fig. S1). The second allele (w²) in the G0 female parent and G0 male parent had a 4-bp insertion 14 bp upstream to the PAM site adjacent to the sgRNAw6-2 target and a 3-bp deletion adjacent to the sgRNAw6-1 PAM site, as well as seven substitutions. The net outcome was a frameshift likely to result in a truncated peptide 323 amino acids (aa) in length (Fig. 6c, Supplementary Fig. S1). The w³ allele from the G0 male parent contained a frameshift due to a 4-bp insertion and 2 substitutions located 3 bp from the sgRNAw6-1 PAM site; this frameshift at aa 297 would be predicted to produce a protein prematurely terminated at residue 324 (Fig. 6c, Supplementary Fig. S1). Based on w target region sequence analysis and PCR products, all the G1, G2 and G3 w individuals carried two w mutant alleles, which was consistent with Mendelian inheritance of these three alleles (Fig. 6b–d, Supplementary Fig. 4). This is also consistent with the white eye color and the absence of red pigment in the wings, suggesting dysfunction of the proteins from the w¹, w², and w³ alleles (Figs. 5, 6a). The residual red-pigmented spots that were seen within eyes of some w mutants suggested that one or more of these proteins may be partially functional in selected cells within the GWSS eye (Figs. 3c,f, 6a).

The w and cn genes are located on autosomes

Our ability to establish w and cn lines and to perform pair matings with GWSS allowed us to perform reciprocal matings between mutant and wild-type adults to directly determine if w and cn were sex-chromosome linked or reside on autosomes. Four crosses were performed using w or cn mutants as male or female parents in crosses with wild-type insects (Fig. 7a). F1 progeny were phenotyped for eye color and sexed. All F1 progeny were wild-type indicating the recessive nature of the w and cn alleles and consistent with the location of both w and cn on autosomes (Fig. 7b). If either cn or w were X-chromosome linked, males would have inherited the mutant allele from their mothers in crosses 2 and 4 and displayed a mutant eye-color phenotype. These data were further supported by our genotypic data, acquired in parallel, that demonstrated two cn alleles and two w alleles were present in both sexes of G1, G2 and G3 adults (Figs. 4, 6).

Off-target analysis using the w and cn gRNAs

Genetic analysis and genetic control strategies require that mutations generated by CRISPR/Cas9 gene-editing strategies be target-site specific. Given the exceptionally high rates of CRISPR/Cas9-editing in GWSS, it was critical to assess sgRNA specificity in vivo. Cas-OFFinder was used to identify potential off-target sites³⁷. Four to five off-target sites for w and cn were chosen for analysis. Amplicon libraries for each off-target were prepared from genomic DNA from WhA-D or CnA-F G0 females. Of the 11 libraries analyzed, mean % reads mapped to the off-target ranged from 0 to 0.95%; however, sgRNAw6-1’s off-target 4 had a larger % reads (5.04%) (Fig. 8, Supplementary Table S4). These data indicated that off-target editing did not occur or occurred infrequently. Our data are comparable to the negligible off-target frequencies from Anopheles gambiae in which the impact of off-target effects on a gene drive strategy was determined³⁸.