CRISPR/Cas9‐based functional characterization of the pigmentation gene ebony in Plutella xylostella

Body pigmentation is an important character of insects in adapting to biotic and abiotic environmental challenges. Additionally, based on the relative ease of screening, several genes involved in insect melanization have been used in classic genetic studies or as visual markers in constructing transgenic insects. Here, a homologue of the Bombyx mori melanization‐inhibiting gene ebony, associated with the conversion of dopamine to N‐β‐alanyl dopamine, was identified in a global pest, Plutella xylostella. The CRISPR/Cas9 system was applied to generate multiple Pxebony knockout alleles which were crossed to produce a Pxebony knockout strain, showing darker pigmentation in larvae, pupae and adults, compared with wildtype. Interestingly, we observed that Pxebony heterozygotes displayed an intermediate darkened phenotype, indicating partial dominance between the knockout and wildtype alleles. The fitness costs of Pxebony deficiency were also assessed in the mutant strain, indicating that embryo hatchability and larval survival were significantly reduced, while the eclosion rate was not obviously affected. Our work provides a potential target for exploring CRISPR‐based genetics‐control systems in this economically important pest lepidopteran.


Introduction
The diamondback moth (DBM), Plutella xylostella, is one of the most destructive global pests of agriculture (You et al., 2013;You et al., 2020). It causes significant economic losses by feeding on Brassica crops and reducing crop yields. However, the control of DBM is challenged by its extreme insecticide resistance, exacerbated by the high levels of chemical-based control used against it (Furlong et al., 2013). Although genetic control techniques (e.g., gene drive systems) have been proposed as promising alternatives for agricultural pest management, proof-of-principle systems in insects are limited to dipterans, mainly built in the model Drosophila melanogaster and several human disease vectors (Gantz et al., 2015;Scott et al., 2017;Kyrou et al., 2018;L opez Del Amo et al., 2020). To explore the possibility of applying genetics-based population control tools in DBM, endogenous melanization/pigmentation genes (of which null mutations cause various visual phenotypes) can be used as the assessable primary targets for testing the functional efficiency of constructed prototypes. Therefore, genes regulating insect body pigmentation may be ideal candidates as molecular markers.
There are diverse insect pigments, among which melanin is the predominant class manipulating insect cuticle colouration and sclerotization (Wittkopp and Beldade, 2009). Genes involved in melanin metabolism have been well studied in D. melanogaster (Wittkopp et al., 2003a). The primary precursor tyrosine is converted to dihydroxyphenylalanine (dopa) by tyrosine hydroxylase, followed by the production of dopamine from dopa by dopa-decarboxylase. Dopa and dopamine can be further used to produce black or brown melanin by yellow or phenol oxidase proteins. Alternatively, dopamine can be conjugated with β-alanine by β-alanyl-dopamine synthase (encoded by ebony) to generate N-β-alanyl dopamine (NBAD), which is the precursor of yellowish-tan NBAD-sclerotin, while the protein encoded by tan is involved in the reverse reaction hydrolysing NBAD back to dopamine. In addition, dopamine can be modified by arylalkylamine-Nacetyltransferase to generate N-acetyl dopamine (NADA) and the unpigmented NADA-sclerotin ( Fig. 1A) (Wittkopp et al., 2003a).
Among these melanin pathway genes, ebony has attracted interests from researchers because of its conserved role in limiting melanization among different insect species. It has been documented that the body melanization of D. melanogaster adults was enhanced in the absence of ebony function, while the ectopic expression of Ebony suppressed melanin formation (Wittkopp et al., 2002). In varied field-caught D. melanogaster populations, the genomic diversity inside or flanking ebony loci can be linked to differences in its expression resulting in regional light or dark body pigmentation variations (Pool and Aquadro, 2007). Additionally, the dramatic pigmentation differences between Drosophila americana and Drosophila novamexicana has been linked to the divergence of Ebony expression levels, indicating that Ebony is also responsible for pigmentation variation between closely related insect species (Wittkopp et al., 2003b). In the model lepidopteran Bombyx mori, it has been reported that ebony loss-of-function mutants showed higher levels of melanism at larval and pupal stages (Futahashi et al., 2008), while the ectopic overexpression of ebony altered wildtype black pigmentation into lighter body colour (Osanai-Futahashi et al., 2012). Related research in the butterfly, Vanessa cardui, showed that ebony knockout was sufficient to cause over-melanized wing phenotypes (Zhang et al., 2017). Based on the ease of visual screening, the ebony gene has been utilized as a molecular marker in the construction of gene drive systems in D. melanogaster by integrating sgRNA-expressing cassettes into the ebony coding sequence (L opez Del Amo et al., 2020). However, the functions of ebony remain unclear in DBM, and the elucidation of ebony function may provide potential genomic tools for building gene drives targeting this economically important pest.
In this study, the homologue of B. mori ebony in DBM was identified and CRISPR/Cas9 was utilized to generate an ebony knockout strain for characterizing its gene functions. Although some fitness costs were observed in the ebony mutants, our work furthers the potential of applying ebony as a molecular marker in the construction and testing of gene drive systems in DBM, which may provide future population-engineering solutions.

Identification and phylogenetic analysis of Pxebony
To identify putative ebony orthologs in DBM, we used B. mori Ebony protein sequence as a query to blast against the DBM genome database. Protein g36009, which showed the highest score (total score = 1169.07, E value = 0), was searched for conserved domains using the CD-search programme in NCBI. The blast result confirmed the existence of an adenylation domain of nonribosomal peptide synthetases (A_NRPS, 17-509 aa) and a Phosphopantetheine attachment site (PP-binding site, 542-608 aa), which were conserved in other insect Ebony Figure 1. Illustration of the insect melanin biosynthesis pathway (A) and phylogenetic analysis of Ebony proteins across different species (B). (A) pathway was adapted from a previous review (Wittkopp et al., 2003a). Genes involved in the melanin pathway are indicated in italics and the corresponding colours of melanin pigments are highlighted (dopa melanin and dopamine melanin are the main sources of black and brown pigments while NADA-sclerotin is colourless). The ebony gene studied here is marked with a dashed box; (B) Evolutionary relationships among different insect species comparing putative Ebony homologues. Maximum Likelihood Method was applied to build the tree with 1000 bootstraps. Numbers at nodes represent bootstrap values (given in percentage). The scale bar indicates 0.1 amino acid substitution per site. aaNAT, arylalkylamine-N-acetyltransferase; DDC, dopa-decarboxylase; NADA, N-acetyl dopamine; NBAD, N-β-alanyl dopamine; PO, phenol oxidases; TH, tyrosine hydroxylase. proteins. No other predicted DBM proteins showed close similarity to the query sequences. Therefore, g360009 was considered as the Ebony ortholog in DBM, named as Pxebony hereafter. Based on the DBM genome database, the total length of Pxebony predicted primary transcript is 15 787 bp while its coding region is 2499 bp, which contains 16 exons and encodes 833 amino acids ( Fig. 2A).
Ebony sequences of other insects were collected from NCBI database and aligned with Pxebony using the Clus-terW tool in Mega 7. A phylogenetic tree was subsequently constructed, where Pxebony closely clustered with other lepidopteran Ebony proteins, consistent with the evolutionary relationship among insects tested in the current study. This result further confirmed that Pxebony is the Ebony ortholog in DBM and likely to exhibit conserved functions reported in other insects.

CRISPR/Cas9-mediated knockout of Pxebony
The sgRNA target site used in the previous report of Spodoptera litura ebony (Slebony) knockout (Bi et al., 2019) was also found in Exon 5 of Pxebony. However, the cleavage efficiency (probability of generating frameshift mutations) of this target was estimated as only 54.52% with CHOPCHOP. To explore other potential targets as well as improve editing efficiency in Pxebony, we used CHOP-CHOP to search the 512 bp Pxebony genome sequence surrounding the Slebony target site. Two sgRNA targets, e-sgRNA1 and e-sgRNA2, predicted with the highest cutting efficiency (72.07% and 79.64%, respectively), were selected for subsequent CRISPR targeting.
In total, 315 eggs were microinjected with the complex of Cas9 protein and two in vitro transcribed sgRNAs, with 156 G 0 pupae surviving (survival rate = 49.5%). Patchy melanization was observed in some G 0 adult wings (Fig. 2B). Although this mosaic phenotype was not very distinct against a wildtype (non-mutated) background via direct visual screening, significant differences were found between them by quantifying the pigmentation intensity of the same regions on the wings with an image analysis tool ImageJ (Fig. 2C). G 0 adults were subsequently inbred in pools to produce the G 1 generation, where substantially melanized adults were observed. A single male and a female melanized G 1 adult (G 1 founders) were randomly collected and crossed to generate a Pxebony knockout line, which thus contained up to four different mutant alleles. The cloning and sequencing of a mosaic G 0 and the two G 1 founders revealed frameshift mutations in either one or both sgRNA target sites (Fig. 2D,E), resulting in premature stop codons and likely disrupting protein function ( Figure S1). Our results confirmed the editing events induced by CRISPR/Cas9 in Pxebony.
Mutant phenotypes of the Pxebony knockout strain Darker pigmentation was observed at different developmental stages in the Pxebony knockout strain, including late larvae, pupae and adults, but not embryos or early larvae, compared to wildtype individuals. The head capsule and thoracic legs of the third and fourth instar larvae were brown in wildtypes but black in the Pxebony strain ( Fig. 3A-C). There are two different, naturally occurring pigmentation patterns in DBM, that is, with (Fig. 3D) or without (Fig. 3E) vertical stripes through the whole body. However, the yellow pigmentation shown in both patterns of wildtype pupae was basically lost in Pxebony-deficient individuals (left most of each pair). In addition, both Pxebony mutant male and female adults have melanized heads, abdomens, legs and wings, which were distinctly darker than their wildtype counterparts ( Fig. 3F-N). By crossing Pxebony null mutants with wildtypes, an intermediate melanized phenotype was observed in the heterozygous offspring. The phenotype variation was more distinguishable in adult wings than other tissues, thus, we took photos (Fig. 4A,B) and subsequently quantified the pigmentation intensity of adult wings with ImageJ. The mean integrated density of the same areas in the three lines were then compared. Our result showed significant differences among Pxebony À/À, Pxebony À/+ and WT +/+ lines in both male and female adults, validating the existence of intermediate phenotype between darkpigmented null-mutation and lighter wildtype colouration (Fig. 4C). Such a partially dominant marker allowing discrimination of homozygous and heterozygous mutants from each other as well as from wild type would potentially be very useful. However, the intermediate melanization of heterozygotes was not always easy to distinguish from wildtype or homozygotes in adults, due to the loss of wing scale debris during adult growth, which affects the appearance of adults. Therefore, it was not easy to count the ratios of each phenotype (wildtype, intermediate, or dark pigmentation) in a larger population, where individuals grew at inconsistent developing stages, for analysing the segregation and inheritance patterns of wildtype and mutant alleles in their progeny.
Although the Pxebony homozygotes selected for taking photos in Figs 3 and 4 happened to look generally smaller than wildtypes, this difference was not consistent among rearing populations and generations. It could be explained by natural variation in the genetic background or the existence of four mutant alleles in the knockout line (fitness of body size might be linked to some alleles). However, more experiments are needed to investigate the actual reasons contributing to this body size variation.

Fitness analysis of the Pxebony mutant strain
To detect potential fitness costs associated with the Pxebony disruption, the embryo hatchability, larval developmental period, survivability and pupal eclosion rate of the  . Fitness analysis of Pxebony null mutants and wildtypes. Eggs were collected from Pxebony and wildtype counterparts (WT) moths (n = 228 and 259, respectively). The proportion of eggs hatching ('hatchability') was determined. For each genotype, 228 larvae (76 per replica, 3 replicas) were monitored for survival to pupation ('larval survival rate'), with the time to pupation additionally recorded for one replica ('larval longevity'). In addition, the survival to adult was determined (5 Â 15 pupae per genotype, recorded as 'eclosion rate'). All data were presented with mean AE SEM. The t-test analysis was conducted with GraphPad Prism 8.3.1 software to compare fitness differences, which is represented with '**' (P value < 0.01), '***' (P value < 0.001) or 'n.s.' (not significant). Actual values and corresponding SEMs used in upper figure are listed in table below.
Pxebony knockout line were analysed. We found that the embryonic hatching rate and larva-to-pupa survival rate of Pxebony mutants were significantly lower than those of their wildtype progenitor strain (Fig. 5). Additionally, Pxebony individuals needed significantly more time to complete larval development than the wildtype cohorts. Nevertheless, no significant difference was observed in the pupal eclosion rates (pupa-to-adult survival) between the two lines tested. These findings indicate a role for Ebony in embryonic and larval development, potentially due to an effect on levels of dopa, dopamine or their derivatives, with loss of Ebony having an adverse effect on development (Fig. 5).

Discussion
Melanin is the major group of pigments forming insect body colour patterns, which play an important role in responding to biotic and abiotic environmental challenges. Investigating endogenous genes associated with melanin biogenesis, metabolism and transport will therefore shed light on mechanisms of insect adaptation and evolution. Additionally, these essential genes controlling observable phenotypes can be applied in insect genome engineering as valuable molecular markers to qualify or quantify editing events.
In this study, we investigated the melanin pathway gene ebony by using the genome editing tool CRISPR/Cas9 to build ebony knockout strains, showing that Pxebony null mutations caused darker pigmentation in older DBM larvae, pupae and adults. This phenotype was consistent with previous reports in D. melanogaster, Aedes aegypti, B. mori, V. cardui and S. litura (Wittkopp et al., 2002;Futahashi et al., 2008;Li et al., 2017;Zhang et al., 2017;Bi et al., 2019), revealing the conserved function of ebony in maintaining light body colouration across various insect species. In contrast to B. mori, where the ebony-mutant phenotype was less obvious in adults than in larvae and pupae (Futahashi et al., 2008) homozygous Pxebony mutant adults were clearly distinguishable from wildtypes. We also found that Pxebony disruption had minor effects on the growth and development of DBM, reducing embryonic hatchability and larval survivability as well as extending larval developmental period, without affecting adult emergence. These results indicate that Pxebony could be utilized not only as a molecular marker but also a regulator of persistence in gene drive system development. For example, the Pxebony locus could be used as a location site for a Cas9 expressing transgene in a split/drive design, as this allelewith its associated disrupted copy of Pxebony may eventually disappear in the target or non-target populations due to slight fitness cost of Pxebony mutation, in turn restricting the spread of the sgRNA target allele (which can only be driven in the presence of Cas9) in the population. This design is potentially beneficial for constructing confinable population suppression systems and mitigating potential ecological risks. However, further investigations (e.g., with mathematical modelling) would be needed to analyse such possibilities in detail for specific gene drive designs. In contrast, previous report of another pigmentation gene Pxyellow showed no significant impacts on moth fitness caused by Pxyellow defect (Wang et al., 2020), making it an ideal neutral marker for testing drive systems but without the additional limiting function shown in Pxebony. Therefore, our research on Pxebony provides more options and expands the genetic tool box for developing different gene drives in DBM.
It is interesting that, in contrast to the recessive ebony mutation in B. mori (Futahashi et al., 2008), but similar to D. melanogaster (True, 2003), Pxebony loss-of-function alleles caused a partial dominant phenotype, meaning that the heterozygous mutations resulted in incomplete melanization while homozygous deficiency was responsible for a completely dark pigmentation phenotype. Theoretically, such a characteristic could prove beneficial in an applied context, for example through providing a non-molecular means to estimate the allele frequency of a transgene (such as Cas9, above) inserted into that locus as part of a drive system. As described in our results, however, the intermediate phenotype was not always easily distinguishable from full knockout or wildtype individuals by simple microscope screening. Hence, rapid and automated phenotypic screening based on sophisticated image analysis [which has been demonstrated for the identification of mosquito genotypes (Crawford et al., 2020;Koskinioti et al., 2021)] are required if such a benefit is to be realized in a larger lab population or field setting. Pigmentation variation, due to divergence of ebony sequences among different D. melanogaster populations, may be involved in adaptation to various geographic environments (Pool and Aquadro, 2007). However, it remains unclear whether Pxebony expression also contributes to DBM adaptation to diverse environments, which helped DBM to become a cosmopolitan pest. In addition to melanin synthesis, evidence supporting other biological roles of ebony has been found in the model insect D. melanogaster. For example, ebony might regulate cuticular hydrocarbon (CHC) composition, where the lack of ebony biased the formation of long chain instead of short chain CHCs, correlated with pigmentation changes (Massey et al., 2019). Additionally, Ebony was detected in the visual and nervous system, participating in the regulation of fly vision, rhythm activity as well as mating behaviour (Suh and Jackson, 2007;Takahashi, 2013;Ziegler et al., 2013). This demonstrates that ebony is a pleiotropic gene, at least in Drosophila, harbouring multiple biological functions, which require further studies to be investigated in DBM.

Insect rearing
All CRISPR-treated mutants were generated from the Vero Beach wildtype strain and reared under the same conditions (relative temperature = 25 C, relative humidity = 50%, and light: dark cycle = 16 h: 8 h) unless specified. Larvae were fed with beet armyworm artificial diet (Frontier Biosciences, Germantown, MD, USA) while adults with 10% sugar solution .

Phylogenetic analysis
For identifying the putative ebony homologue in DBM, B. mori Ebony protein sequence (accession number: BAH11147.1) was used to BLAST against the P. xylostella Pacbiov1 genome databases (http://lepbase.org/). Ebony protein sequences of other insect species, including A. aegypti, Anopheles gambiae, D. melanogaster, Apis mellifera, Tribolium castaneum, Leptinotarsa decemlineata, S. litura, Papilio xuthus and Papilio machaon, were also downloaded from NCBI databases by applying B. mori Ebony as the BLAST query. Evolutionary relationships were further analysed with Mega 7. To be specific, the sequence alignment was completed with ClustalW, and a phylogenetic tree was created using the Maximum Likelihood Method with 1000 bootstraps.
Eggs collected within 30 min post oviposition were injected with the sgRNAs/Cas9 mixture and then maintained in Petri dishes containing wet cotton balls for maintaining humidity. Hatched larvae were reared with artificial diet until pupation. Surviving pupae were sexed and set up in five deli pots for crossing, with 10 G 0 females and 10 G 0 males in each pot. Eclosed adults were allowed to randomly mate and produce G 1 s, of which adults with distinguishable dark pigmentation were pair-crossed to generate Pxebony knockout strains. Since G 1 knockouts exhibited the same phenotype, one of these strains was taken forward for further analysis, and named as Pxebony hereafter.

Identification of mutations
Both G 0 and G 1 adults were screened and photographed with a camera-integrated Leica E24 HD stereo microscope (Leica Biosystems, Milton-Keynes UK). Genomic DNA of obviously darker pigmented moths was extracted using the NucleoSpin Tissue Kit (Macherey-Nagel, Düren, Germany). Additionally, the 516 bp fragment containing both sgRNA target sites was amplified and cloned using the pJET1.2 Cloning Kit (Thermo Fisher). Positive amplicons were sent for Sanger sequencing to confirm mutations.

Fitness analysis of Pxebony mutants
To investigate potential fitness effects caused by Pxebony knockout, the Pxebony null mutant strain and their wildtype counterparts were assessed for hatchability (proportion of egg hatching), larval survival rate, time to pupation and pupal eclosion rate.
Initially, 40 pupae of each strain (wildtype and Pxebony) were separately set up in cross pots and allowed to mate freely once eclosed. Two days after mating, parafilm sheets coated with cabbage extract were put in each pot for 1 h oviposition after which time the parafilm sheet was discarded and replaced with a fresh sheet. This step aimed to remove partially developed embryos carried in female adults and ensure that eggs collected on the replacement sheet were at a very similar developmental stage (i.e., freshly fertilized). The new parafilm sheets were placed in cross pots for 30 min to collect fresh eggs, which were then kept in Petri dishes to allow them to hatch, and the hatch rates were subsequently recorded. Additionally, 76 neonate larvae of the same line hatched on the same day were moved from Petri dishes to a deli pot as one biological replicate, and reared with artificial diet till pupation. The total developmental period from neonate larvae to pupae was recorded as larval longevity, and the number of successfully pupated individuals was also counted for analysing larval survival rate. All the experiments above were conducted with three biological replicates. In addition, to investigate eclosion ability, 15 pupae were pooled as one biological replicate and five replicates in total were carried out, after which the number of successfully emerged adults were counted.
Data collected in this section were presented with mean-AE SEM. The t-test analysis was conducted with GraphPad Prism 8.3.1 software to compare fitness differences, which were sorted into '*' (P < 0.005), '**' (P < 0.001) and 'n.s.' (not significant).
In conclusion, our work showed the possibility of using Pxebony as an accessible target in constructing DBM genetic control systems, as well as proving the material for further studies on lepidopteran epidermis formation, circadian rhythms and mating behaviours.
supported by the CSC Scholarship from the Chinese Government, and THS was supported by a UK Biotechnology and Biological Sciences Research Council (BBSRC) Impact Acceleration Account grant (BB/S506680/1). LA was supported by core funding from the BBSRC to The Pirbright Institute (BBS/E/I/00007033, BBS/E/I/00007038 and BBS/E/I/00007039).

Data availability statement
The data that support the findings of this study are available on request from the corresponding author upon reasonable request.