INTRODUCTION

As a pest, the European corn borer (Ostrinia nubilalis) is responsible for substantial yield losses in corn and other crops worldwide.

Although Bacillus thuringiensis (Bt) maize adoption in the United States led to regional suppression of O. nubilalis (Dively et al., 2018), the first case of practical resistance to Bt Cry1F maize was identified in Nova Scotia, Canada, in 2018 (Smith et al., 2019).

Since then, resistance to Bt Cry proteins has been documented in other regions of Canada and observed on a sweet corn plot producing Cry1A.105 and Cry2Ab2 proteins in Connecticut, USA (NC 246, 2024). Although gene regions segregating with resistance have been identified (Coates et al., 2011;Coates & Siegfried, 2015;Farhan et al., 2023), functional validation by gene knockout is needed to enhance sequence-based resistance monitoring (Pezzini et al., 2024) and provide mechanistic insight into cross-resistance between pyramided Bt toxins and foliar sprays (Abdelgaffar et al., 2021).

Beyond Bt resistance, prior quantitative trait locus mapping and genomic approaches have identified candidate genes associated with other traits of agroeconomic importance. For example, population differences in O. nubilalis female pheromone composition (Dopman et al., 2004;Lassance et al., 2010) and male preference (Unbehend et al., 2021) could alter the efficacy of passive monitoring techniques (i.e., pheromone lures) and affect the rate at which resistance alleles evolve by gene flow. Additionally, life cycle differences in generation number (voltinism; Kozak et al., 2019;Levy et al., 2015) determine the distribution and abundance of pests across space, time and host plants. Consequently, there is a need for robust genetic techniques to probe the genetic basis of these pest traits. Considering that the function (Mackay & Anholt, 2024) and evolution (Dopman et al., 2024;Tigano & Friesen, 2016) of a gene depends on its genetic background, greater understanding of the genetic mechanisms underlying behavioural and ecological differences among corn borer populations is critical to effective integrated pest management and insect resistance management for this insect (Coates et al., 2018).

The advent of CRISPR/Cas9 genome editing has revolutionized the study of gene function in insects (Taning et al., 2017). This system utilizes a single guide RNA (sgRNA) to direct the Cas9 nuclease to specific, complementary genomic loci, where it induces a doublestrand break (DSB) upstream of the protospacer adjacent motif (PAM). Aberrant repair of these breaks allows for targeted disruption, insertion, or replacement of genes (Taning et al., 2017). Particularly in O. nubilalis and other Lepidoptera, where conventional methods like RNA interference (RNAi) are inefficient and often falter (Cooper et al., 2021;Guan et al., 2018;Khajuria et al., 2011;Terenius et al., 2011), CRISPR/Cas9 has paved the way for numerous studies of gene function. These endeavours have illuminated key mechanisms underlying diverse phenotypic traits, ranging from pigmentation and wing patterning to development and reproductive behaviour (Li et al., 2021).

In this study, we demonstrate that CRISPR/Cas9 is a highly efficient and robust tool for generating heritable mutations in O. nubilalis.

We used both single and dual sgRNAs to generate LOF mutants for three genes. Across targets, editing efficiency remained high and sufficient for streamlined studies of gene function in F 0 mutants. We also describe an approach for quantifying interactions between mutations and their genetic background. This methodology should accelerate future genetic studies in O. nubilalis and facilitate similar screens in other related insects.

RESULTS AND DISCUSSION

Efficient CRISPR/Cas9-mediated mutagenesis

To assess the capacity for CRISPR/Cas9-mediated genome-editing to produce LOF mutations in O. nubilalis, we targeted two circadian clock genes, period (Konopka & Benzer, 1971) and pigment dispersing factor receptor-like ( pdfr; Hyun et al., 2005;Lear et al., 2005;Mertens et al., 2005). PERIOD is involved in the main transcriptional repression by CRYPTOCHROME 2 of the Lepidopteran circadian clock (reviewed in Brady et al., 2021), oscillating in abundance every $24 h in both entrained (light:dark, LD) and free-running (continuous darkness, DD) conditions (Hardin et al., 1990(Hardin et al., , 1992)). Meanwhile, pdfr encodes the extracellular receptor for the neuropeptide pigment dispersing factor (PDF, Renn et al., 1999). PDF/PDFR signalling synchronizes transcriptional oscillations between clock neurons, helping maintain robust behavioural rhythms in DD (Lin et al., 2004;Peng et al., 2003;Renn et al., 1999) and adjusting these rhythms to seasonal changes in LD cycles (Ruf et al., 2021;Yoshii et al., 2009). In addition to their roles in the circadian clock network, in O. nubilalis, genetic differences at Z-linked period and pdfr are associated with the 14-21 day delay in spring emergence that contributes to the life cycle difference in the number of generations per growing season between univoltine ( period U pdfr U ) and bivoltine ( period B pdfr B ) populations (Kozak et al., 2019). Univoltine prepupae also exhibit a relative increase in their propensity for seasonal diapause (Ikten et al., 2011;Yu, 2022) and an earlier peak adult activity under DD (Dayton & Owens, 2024;Kozak et al., 2019).

For the CRISPR/Cas9 system, gene-specific sgRNAs were designed on the O.nubilalis RefSeq assembly (GCF_963855985.1) to target the coding sequences of period (GenBank: LOC135086880) and pdf receptor-like (GenBank: LOC135087024; Figure 1, Table S1).

Following oviposition by univoltine ( period U pdfr U )and bivoltine (period B pdfr B ) females, clusters of embryos were injected with Cas9: sgRNA ribonucleoprotein (RNP) complexes in 2022 (2.5 μM Cas9, 150 ng/μL sgRNA). To test the effects of different sgRNA concentrations on somatic editing efficiency, eggs were injected with a higher sgRNA concentration in 2023 (2.5 μM Cas9, 180 ng/μL sgRNA). Typically, 7-12 embryos were injected per cluster. After hatching, injected F 0 was tracked throughout development. Most injected F 0 individuals that hatched survived to adulthood (Table 1), with an average of six F 0 per injected cluster (95% confidence interval [CI]: 3.5-9.0). Mating success did not differ significantly across sgRNAs (binomial GLM; LR χ 2 = 2.83, df = 2, p = 0.243) or injection year (χ 2 = 0.50, df = 1, p = 0.481; Table 1). After mating, successful F 0 parents were screened for CRISPR/Cas9-induced mutations by polymerase chain reaction (PCR) and Sanger sequencing. Individuals were classified as somatic mutants if they exhibited >20% indel frequency, as estimated by deconvolution of Sanger sequence electropherograms (DeLay et al., 2018;Synthego Performance Analysis, 2019). CRISPR/Cas9 was highly effective. Across years and loci, 74% (95% CI: 60%-89%) of F 0 individuals were somatic mutants. Somatic editing efficiency was robust to the sgRNA (binomial GLM; LR χ 2 = 1.02, df = 2, p = 0.601) and injection year (LR χ 2 = 2.57, df = 1, p = 0.11; Table 1), and the overall efficiency was comparable to or higher than reports from other Lepidoptera (see Supporting Results in Supporting Information S1). In fact, editing efficiency was so high that nearly 100% of mutations in the F 0 were heritable and transmitted to F 1 progeny (Table 1). Although the classification as 'somatic mutant' is binary, the F 0 injected in 2023 (86% indels) exhibited significantly greater conversion of wildtype alleles than the F 0 from 2022 (78% indels; Wilcoxon Sign-Ranked W = 143, p = 0.049), underscoring a possible impact of the increase in sgRNA concentration (1.2-fold) on editing efficiency (Bassett et al., 2013;Perera et al., 2018;Wang et al., 2013). Diverse mutations were observed at period (Figure 1a) and pdfr (Figure 1b). Deletions were more frequent than insertions (χ 2 = 14.52, df = 1, p < 0.001) and represented 82% (95% CI: 37%-100%) of all sequenced mutations. 78% (95% CI: 64%-92%) of mutations were frameshifts that produced a premature stop codon, significantly more than expected for indels within a coding sequence (i.e., two out of every three; χ 2 = 4.08, df = 1, p = 0.043; Wu et al., 2018). These frameshift mutations were considered LOF null alleles, as they were predicted to encode extensively truncated proteins of PERIOD (wildtype: 1159 amino acids vs. sgRNA1: 67-89 amino acids, sgRNA2: 139-152 amino acids) and PDFR (wildtype: 431 amino acids; sgRNA1: 224 amino acids).

Mutagenesis of period but not pdfr disrupts rhythmic eclosion

To assess the phenotypic effect of period and pdfr on circadian behaviour in the corn borer, we examined the timing that adult moths eclosed from their pupal case under a summer-like photoperiod (LD 16:8 h; e.g., Markert et al., 2016). Eclosion was described in Arbitrary Zeitgeber Time (AZT) when AZT0 was the time that lights turned on. The rhythmic strength of eclosion timing was quantified by the proportion of moths who eclosed within a peak 8-h eclosion window/ gate (e.g., Ikeda et al., 2021;Liu et al., 2023;Winfree, 1970). Genetic background (univoltine/bivoltine) of wildtype and mutants did not influence rhythmic strength (binomial GLM; χ 2 = 0.08, df = 1, p = 0.78; AIC = 64.5; Figure S1), so strains were combined for subsequent analyses (Figure 2; AIC = 59.0). Most wildtype adults eclosed within an 8-h gate from AZT15 to AZT22 in 2022 (0.83, 95% CI: 0.77-0.89) and in 2023 (0.75, 95% CI: 0.70-0.80; Figure 2, Figure S2).

Eclosion within this gate was significantly affected by mutagenesis of per/pdfr (binomial GLM; LR χ 2 = 40.9, df = 2, p < 1 Â 10 À9 ) and T A B L E 1 CRISPR/Cas9-induced mutagenesis in O. nubilalis. Single Cas9:sgRNA mixtures targeting the period and pigment-dispersing factor receptor-like (pdfr) genes were microinjected into O. nubilalis eggs.

1) 12 67 (81%) 49% (51) 58% (19) 86% (7) period (2) 12 50 (65%) 48% (42) 80% (15) 100% (4) pdfr (1) 4 11 (52%) 78% (9) 71% (7) 100% (3) 2023 period (1) 8 56 (nd) 63% (27) 93% (15) 100% (5) period (2) 8 79 (nd) 51% (43) 84% (19) 100% (5) pdfr (1) 4 33 (nd) 64% (14) 60% (5) 100%

Grand mean 58% (47-71%) 74% (60-89%) 98% (92-100%)

Note: Percent survival was not determined (nd) in 2023. The percentage of successful matings was determined from a subset (n) of the total individuals that survived to adulthood. Similarly, the percentage of somatic mutants is based on the number of genotyped (n) fertile F 0 adults that presented somatic mosaicism >20%. Germline mutation rate corresponds to the percentage of F 1 progeny with a mutated allele of the total progeny genotyped (n). Grand means with 95% confidence interval are provided in the bottom row. Figure 2). The more pronounced phenotypic effect of period mutation in 2023 was consistent with the higher concentration of sgRNAs and greater observed conversion of wildtype to null alleles (see above).

This reduction in rhythmic strength was concordant with significantly different distribution of eclosion across the entire 24-h distribution (Figures S1 and S2). Overall, the aberrant gating and distribution of eclosion observed for our period F 0 mutants recapitulated phenotypes of stable germline clock mutants in other insects. For instance, in Drosophila melanogaster, rhythmic strength in period knockouts was reduced in LD and, in the absence of light entrainment, was entirely disrupted in DD (Konopka & Benzer, 1971;Qiu & Hardin, 1996;Ruf et al., 2021). In Lepidoptera, clock gene mutation similarly reduced eclosion rhythms in LD and completely abolished them in DD (Bombyx mori: Ikeda et al., 2021, Nartey et al., 2021; Helicoverpa armigera: Liu et al., 2023). Although expression of target clock genes was not measured in our study, aberrant behavioural rhythms in eclosion have been linked to disrupted molecular rhythms in circadian transcription in the brain (Ikeda et al., 2021;Markert et al., 2016;Nartey et al., 2021). Together, our mutants provide one more layer of genetic evidence highlighting the conserved essentiality of the circadian clock for the normal timing of insect eclosion (Truman & Riddiford, 1974).

Unlike period, mutations in pdfr did not alter the rhythmic strength of eclosion in 2022 (0.10, 95% CI: À0.09 to 0.28; p = 0.575) or 2023 (À0.11, 95% CI: À0.30 to 0.07; p = 0.230; Figure 2). These observations were similar to those in D. melanogaster, wherein persistent coupling of pacemaker neurons by short LD cycles (Vaze & Helfrich-Förster, 2021;Yoshii et al., 2009) masks the effects of pdf/ pdfr loss and behavioural rhythms remain largely unaffected (Hyun et al., 2005;Myers et al., 2003;Renn et al., 1999;Ruf et al., 2021).

Instead, the consequences of pdf/pdfr mutations in Drosophila are apparent in DD, when transcription oscillations among pacemaker neurons become desynchronized (Lin et al., 2004;Peng et al., 2003;Renn et al., 1999) and output behavioural rhythms rapidly dampen (Hyun et al., 2005;Myers et al., 2003;Renn et al., 1999;Ruf et al., 2021). Additionally, pdf/pdfr mutants in Drosophila are unable to entrain and adjust their locomotor activity to longer photoperiods (Vaze & Helfrich-Förster, 2021;Yoshii et al., 2009). Future studies on F 1 mutants in O. nubilalis will need to investigate the effects of pdfr loss on behavioural rhythms in DD and multiple LD photoperiods.

Targeted deletion of prothoracicotropic hormone by dual sgRNA injections

We expanded our investigation on the utility of CRISPR/Cas9 by simultaneously injecting multiple sgRNAs that targeted different regions of the prothoracicotropic hormone (ptth) gene. While the use of a single sgRNA successfully generated small indels at period and pdfr (Table 1 and Figure 1), we predicted that co-injecting two co-directional sgRNAs could induce large-scale LOF mutations, increase the frequency of null allele conversion within an individual (Kroll et al., 2021), and facilitate more rapid screening by PCR-based fragment size differences (e.g., Han et al., 2024;Markert et al., 2016;Zhao et al., 2023). Two co-directional sgRNAs were designed to target the 5 0 UTR and coding exon 1 of ptth (GenBank: LOC135088612;

F I G U R E 2 period but not pdfr mutagenesis reduces rhythmic strength in wildtype eclosion timing. Wildtype Ostrinia nubilalis were injected with Cas9:sgRNA targeting period and pdfr in 2022 and 2023. Eclosion of uninjected wildtype and injected F 0 adults exposed to 16L:8D hour were monitored in Arbitrary Zeitgeber Time (AZT). Rhythmic strength was quantified as the proportion of adults that eclosed from their pupal case between AZT15 and AZT22, the 8 h window encapsulating the time most wildtype adults emerge. The main effects of treatment and injection year on the rhythmic strength were modelled by a binomial logistic regression (AIC = 59.0, df = 5). Within each year, the difference in rhythmic strength (a proportion) between wildtype and the period versus pdfr F 0 was determined. Significant differences between groups were measured by Wald's test and do not encompass zero (i.e., no difference). Letters denote significant differences among groups (p < 0.05), with B-Y adjustment for multiple comparisons. Error bars represent 95% confidence intervals.

Figure 3), immediately upstream of the region encoding the cleaved bioactive PTTH peptide (reviewed in W. Smith & Rybczynski, 2012).

Cas9:dual sgRNA complexes were injected into early-stage embryos and a random sample of larvae and adults were screened for mutations (2.5 μM Cas9, 180 ng/μL sgRNA). Consistent with the period and pdfr injections, 72% of F 0 were somatic mutants (95% CI: 56%-88%, n = 32), and 59% (95% CI: 36%-77%, n = 23) of these mutants bore large deletions (>200 bp; Figure 3). These deletions eliminated the original ptth start codon and were expected to be LOF alleles (0-24 amino acids) for PTTH (221 amino acids). Notably, three ptth loss alters development in O. nubilalis

In insects, binding of the neuropeptide PTTH to its receptor on the prothoracic gland, TORSO (Rewitz et al., 2009), activates MAPK/ERK signalling and promotes synthesis of ecdysone, the developmental hormone essential for moulting and metamorphosis (reviewed in W. Smith & Rybczynski, 2012). A seasonally regulated shutdown in ecdysone synthesis, attributed to the absence of PTTH secretion from the brain, is broadly considered to be the neuroendocrine basis for prepupal and pupal diapause (Denlinger et al., 2012). Furthermore, considering that both secretion of PTTH is governed by a photosensitive circadian clock (e.g., Sakurai, 1983;Selcho et al., 2017;Truman, 1972) and a functioning circadian clock is essential for seasonal responsiveness (Liu et al., 2023), it is predicted that these two pathways converge to regulate diapause. However, the nature of this relationship between the circadian clock network, PTTH and diapause incidence has never explicitly been validated by gene knockout.

Often, the phenotypic effect(s) of mutations at a given locus depend on their genetic background. Whereby, a general framework to identify gene interactions is to screen paired wildtype versus mutant individuals across multiple backgrounds (Figure 4; Mackay & Anholt, 2024;Turner, 2014). To measure interactions between the circadian clock and ptth, we explored how functionally defined circadian clock alleles in bivoltine and univoltine genetic backgrounds, specifically on the Z chromosome (i.e., period & pdfr; Kozak et al., 2019;Yu, 2022), might interact with mutant ptth to alter development (Figure 5). For example, reciprocal crosses between each period background generate two sets of F 0 offspring that differ based on whether daughters inherited a bivoltine (Z B ) or univoltine (Z U ) Z chromosome from their father (note: sons are all heterozygous for Z B /Z U ; Figure 5,

x-axis). Since only F 0 females will differ by their genotype (e.g., Z U /W or Z B /W), the two populations will exhibit divergent allele frequencies on the Z chromosome (i.e., Z U $ 0.75 vs. 0.25, Z B $ 0.25 vs. 0.75; Figure 5). We injected early-stage embryos from these two populations with either Cas9:dual sgRNAs (RNPs) targeting ptth (see However, there was no significant evidence of an interaction between ptth mutation and the genetic background of the Z chromosome (period, pdfr) on either mass (two-way analysis of variance: F = 1.46, p = 0.229; Figure 5a) or diapause incidence (LR χ 2 = 1.40, df = 1, p = 0.237; Figure 5b). Instead, mass was significantly affected et al., 2009;Shimell et al., 2018). These conserved responses to PTTH loss were consistent with low ecdysteroid titers and past work demonstrating that low ecdysteroidogenesis by the prothoracic gland delays developmental timing and induces overgrowth (Colombani et al., 2005). Although ecdysone itself was not measured here, a reduced ecdysone titre in ptth F 0 mutants would be in line with the significant effect of ptth mutation on diapause incidence (binomial GLM; LR χ 2 = 10.4, df = 1, p = 0.001; Figure 5b). Whereby, the incidence of prepupal diapause by ptth mutants was 24% (95% CI: 11%-37%) higher than wildtype individuals (Wald's Z = 3.60, p < 0.001; Figure 5b).

The significant additive effects of ptth mutation support the classic model that seasonal inhibition of PTTH secretion contributes to prepupal diapause induction (Denlinger et al., 2012). The lack of apparent epistasis between ptth and period could suggest that PTTH operates independently of or downstream of the circadian clock's role in diapause induction. However, considering that we were unable to detect known background-specific differences in diapause incidence between paired wildtype populations (Figure 5b), our study may have been underpowered and benefitted from multiple replicate populations. Additionally, genotyping to remove the genetically invariant (Z B /Z U ) male offspring from each population would have increased the magnitude of any effect by making Z chromosome allele frequencies in each F 0 population (x-axis, Figure 5) fixed from each other.

CONCLUSIONS

We have demonstrated that the CRISPR/Cas9 system is a highly efficient and robust tool for generating LOF mutants in O. nubilalis. In the absence of visible markers or transposition reagents, we were able to generate precise mutations in three genes, recover a variety of mutant alleles, and capture phenotypic differences among F 0 mutants.

From start to finish, F 1 -bearing germline mutations at a geneof-interest can be isolated in one generation (35 days). When establishing stable mutant O. nubilalis lines, we recommend that researchers plan to outcross and screen three (95% CI: 2-5) injected F 0 adults per desired LOF mutant (Table 1 and Figure 3c). This recommended number of injected F 0 adults (n Injected Adults ) to screen per desired LOF mutant (n LOF ) is based on the observed (Table 1) proportion of injected adults that mated ( p mated ), bore somatic mutations ( p somatic ) that were frameshifts ( p frameshift ), and were transmitted to F 1 offspring (n germline ): Alternatively, for streamlined studies of gene function, CRISPR/ Cas9 is efficient enough for immediate developmental and behavioural phenotyping of F 0 knockouts (Figures 2 and 5). To support these screens and increase the conversion of wildtype to null LOF alleles in F 0 , two to three sgRNAs can be designed per coding sequence (e.g., Kroll et al., 2021;Zhu et al., 2020). See Kroll et al. (2021) for additional considerations when designing F 0 knockout screens.

Finally, although we did not identify significant evidence of twolocus epistasis, an approach that screens paired wildtype/mutant F 0 for quantitative traits from multiple backgrounds merits further exploration (Figure 4; Mackay & Anholt, 2024;Turner, 2014). Whereas this design has traditionally been limited to model organisms (e.g., S. cerevisiae, C. elegans, D. melanogaster), pairing modern geneediting systems (e.g., CRISPR/Cas9) with the natural variation observed in non-model insects could revolutionize the study of genes and their modifiers on quantitative traits of agroeconomic and evolutionary interest (e.g., resistance, diapause, sexual communication).

EXPERIMENTAL PROCEDURES

Insect stocks

European corn borer (O. nubilalis) eggs were collected from laboratory populations maintained at Tufts University (Medford, MA). These were originally derived from populations in northeastern USA, exhibit divergent allele frequencies at period and pdfr, and have repeatedly been selected and studied for different univoltine ( period U pdfr U ) versus bivoltine ( period B pdfr B ) phenotypes (e.g., Dayton & Owens, 2024;Kozak et al., 2019). Adults were mated in cages containing ad libitum water and covered with parchment paper for oviposition. Egg clusters were cut out, suspended over artificial corn borer diet (Southland Products, USA), and reared under LD 16:8 h in a climate-controlled room (25.5 C, 50% RH). S3). Under a dissecting microscope, embryos were injected within 30-60 min. of oviposition. Injections were made at a $45 -60 angle and the plate was rotated to access eggs. Because eggs located in the middle of clusters were difficult to access with the micromanipulator, typically only eggs located on the perimeter were injected. Following microinjection, eggs were stored on artificial European Corn Borer Diet (Southland Products, USA) and maintained under LD 16:8 h at 25.5 C and 50% RH. After 24 h, eggs were visually checked under a dissecting microscope; those without scar tissue evidence of injection were punctured to kill (Figure S3). Approximately 7-12 eggs were typically injected per egg cluster. Individuals were weighed on day 30 and larvae who failed to pupate by day 38 were considered in diapause (Beck & Hanec, 1960;Mutchmor & Beckel, 1959).

sgRNA design and synthesis

Phenotyping period/pdfr injected mutants

Screening mutants

Analyses

All statistical analyses were conducted in R (R Core Team, 2023). For eclosion phenotyping, the 8 h window of the day containing the greatest number of emergences was considered the main eclosion gate (Liu et al., 2023;Winfree, 1970). The main effects of sgRNA target and background on the probability of eclosing within this 8 h gate were quantified by a binomial logistic regression. Differences in the distribution of adult eclosion were compared using Watson's U 2 test within the circular package (Agostinelli & Lund, 2023). To evaluate evidence for epistasis between genetic background and ptth mutation, a linear model included the main effects of treatment ( ptth vs. control sgRNA) and genetic background (Z B vs. Z U ) on body mass. A binomial logistic regression tested these effects on diapause incidence. For all regression models, Anova() from the car package (Fox & Weisberg, 2019) determined whether the main effects of treatment, background, and/or their interaction were significant. Significant differences between group proportions and means were respectively evaluated by Wald's Z test and Welch's t-test.