Insight into weevil biology from a reference quality genome of the boll weevil, Anthonomus grandis grandis Boheman (Coleoptera: Curculionidae)

Abstract

The boll weevil, Anthonomus grandis grandis Boheman, is one of the most historically impactful insects due to its near destruction of the US cotton industry in the early 20th century. Contemporary efforts to manage this insect primarily use pheromone baited traps for detection and organophosphate insecticides for control, but this strategy is not sustainable due to financial and environmental costs. We present a high-quality boll weevil genome assembly, consisting of 306 scaffolds with approximately 24,000 annotated genes, as a first step in the identification of gene targets for novel pest control. Gene content and transposable element distribution are similar to those found in other Curculionidae genomes; however, this is the most contiguous and only assembly reported to date for a member in the species-rich genus Anthonomus. Transcriptome profiles across larval, pupal, and adult life stages led to identification of several genes and gene families that could present targets for novel control strategies.

Introduction

Curculionidae (Coleoptera) or “true weevils” is a hyperdiverse family of insects with over 6,000 genera and 70,000 species, making it one of the most speciose animal taxa on the planet (https://www.gbif.org/species/search?rank=SPECIES&highertaxon_key=4239&status=ACCEPTED, accessed 26 May 2022). Many weevils are considered major agricultural and forest pests (Oberprieler et al. 2007; McKenna et al. 2009). The genus Anthonomus contains over 780 species and several key pests such as A. grandis grandis Boheman (cotton), A. eugenii Cano (pepper), A. quadrigibbus Say (apple), A. bisignifer Schenkling (strawberry), A. signatus Say (strawberry), and A. rubi Herbst (strawberry) (https://www.gbif.org/species/search?q=anthonomus&status=ACCEPTED, accessed 26 May 2022; Elmore et al. 1934; EFSA Panel on Plant Health 2017a, 2017b, 2018; Tonina et al. 2021). Despite this insect group's large diversity and impact on agriculture, no reference genomes have been reported for the Anthonomus genus.

The boll weevil (A. grandis grandis) is arguably the most impactful pest of cotton, Gossypium hirsutum L., in the United States because it nearly destroyed the cotton industry in the early 20th century. Widely considered to be native to Mexico, the boll weevil was first detected in the United States near Brownsville, TX, in 1892 (Hunter and Hinds, 1905). Boll weevil populations rapidly spread across the southern cotton belt, reaching the Carolinas and Virginia by 1922 and California by the late 1980s (Dickerson et al. 2001). The boll weevil's success can be attributed to its high fecundity in the field, dispersal capabilities, and ability to overwinter in a state of diapause between cotton growing seasons, which present management challenges. The boll weevil has been eradicated throughout most of the United States, except for the Lower Rio Grande Valley growing region along the US/Mexico border, where continued weevil detection and eradication efforts are focused. However, the boll weevil remains a key pest of cotton in parts of Mexico and has become a major cotton pest in South America (Ramalho and Wanderley 1996; Paula et al. 2013).

A high-quality genome for the boll weevil will enable greater understanding of its basic biology and assist in future development of alternative and sustainable controls. A fully assembled genome will be the primary resource for understanding genetic diversity, gene repertoire, and biological processes (BPs) that could yield novel gene targets. Furthermore, this first assembly of the Anthonomus genus will provide data for phylogenetic reconstruction among related weevils. Deeper understanding of weevil evolution and relatedness could support the development of boll weevil diagnostic tools, as well as highlight shared weevil genetic features among unrelated weevils to develop weevil-specific control strategies. In this paper, we present a high-quality reference genome for the boll weevil, A. grandis grandis, representing the first Anthonomus weevil to be fully sequenced. Additionally, we show differentially expressed genes (DEGs) and enriched gene ontology terms among 3 major life stages, representing the genetics underlying the development of this insect. We highlight several notable gene families that are relevant to boll weevil biology, host plant interaction, and insecticide detoxification, which may be exploited at the molecular level to sustainably control this cotton pest.

Methods

Weevil sources

Weevils used for genome assembly were obtained and reared from larval-infested prefloral buds (squares) of cotton plants, collected from a field in McAllen, TX, in August 2019. Infested squares were held at 29.4 ± 1°C and a photoperiod of 14:10 (L:D) hours and were checked daily for pupae. Pupae were harvested from squares, placed in Petri dishes with moistened vermiculite, and held under the same environmental conditions described above. Newly eclosed adults (<24 hours old) were sexed, and 8 males were flash frozen in liquid nitrogen and stored at −80°C until high molecular weight (HMW) DNA was extracted.

RNA from weevils at various life stages was sequenced to generate evidence for gene prediction and transcript expression analysis. A group of larvae (2nd and 3rd instar) were removed from squares and stored in RNAlater. All other larvae were held as described above to generate pupae and adults. Newly eclosed adult females were given squares for 3–7 days prior to being euthanized. Data for males, which consisted of pheromone-producing and nonproducing individuals, were used from Perkin et al. (2021). A total of 8 larvae, 4 pupae, 10 sexually mature but nonmated adult females, 10 sexually mature pheromone producing adult males, and 9 immature males not producing pheromone were used. All samples were put into individual microcentrifuge tubes with RNAlater and stored at −20°C until RNA was extracted.

HiFi data generation

HMW DNA was extracted from a single adult male boll weevil (ToLID icAntGran1), collected as described above, using the Qiagen MagAttract HMW DNA Kit (Qiagen, Hilden, Germany). DNA shearing was performed using the Diagenode Megaruptor 2 (Diagenode Inc., Denville, NJ, USA) with the 20 Kb fragment protocol. The sheared DNA was prepared for PacBio sequencing using the SMRTbell Express Template Prep Kit 2.0 (Pacific Biosciences, Menlo Park, CA, USA). The library was size-selected using a size-limiting solid phase reversible immobilization-bead cleanup to remove library molecules smaller than 3 Kb in length to generate the final libraries for sequencing. Sequencing was performed on a Sequel II System using Binding Kit 2.0, Sequencing Kit 2.0, and SMRT Cell 8M (Pacific Biosciences, Menlo Park, CA, USA). To target high fidelity (HiFi) reads, the library was sequenced using a 30-hour movie time on 1 SMRT cell. Raw subreads were converted to HiFi data using the circular consensus sequencing process to call a single high-quality consensus sequence for each molecule with a consensus accuracy cutoff of 99.5%.

Hi-C data generation

Tissue from another single adult male (ToLID icAntGran2) was crosslinked using the Arima genome-wide chromosome conformation capture (Hi-C) low input protocol, and proximity ligation was performed using the Arima Genomics Hi-C Kit (Arima Genomics, San Diego, CA, USA). After proximity ligation, the DNA was sheared using a Diagenode Bioruptor (Diagenode Inc., Denville, NJ, USA) and then size-selected to enrich DNA fragments between 200 and 600 base pairs. An Illumina library was prepared from the sheared and size-selected DNA using the Accel-NGS 2S Plus DNA Library Kit (Swift Biosciences, Ann Arbor, MI, USA). The final Illumina Hi-C library was sequenced on an NovaSeq 6000 System (Illumina Inc., San Diego, CA, USA) at the HudsonAlpha Genome Sequencing Center (Huntsville, AL, USA).

Genome assembly and organization

Adapter contamination was removed using HiFiAdapterFilt v1.0 (Sim et al. 2022), and adapter-free reads were used as input to HiFiASM v0.14.2-r315 (Cheng et al. 2021) to produce the preliminary primary pseudohaploid and corresponding alternate assemblies. Duplicate haplotigs were removed or trimmed using purge_dups v1.2.5 (Guan et al. 2020), with cutoffs automatically estimated from a generated histogram of read coverage. Purged primary sequences were added to the alternate haplotype contigs and purged of duplicates again for a final alternate assembly.

YaHS v1.0 (Zhou et al. 2022; https://github.com/c-zhou/yahs) with default error correction was used to scaffold the primary assembly after aligning the Hi-C data with the ArimaHi-C Mapping Pipeline (https://github.com/ArimaGenomics/mapping_pipeline). The Hi-C contact map was then manually curated with Juicebox v1.11.08 (Durand et al. 2016; Dudchenko et al. 2018). Autosomes were ordered according to size, and sex chromosomes were identified based on Hi-C data coverage. After manual curation of the Hi-C contact map, additional checks were performed for scaffolds containing contaminants with BlobTools v1.1.1 (Laetsch and Blaxter 2017), which were first summarized using BlobBlurb (https://github.com/sheinasim/blobblurb) and later removed from the assembly with FastaParser (https://github.com/sheinasim/FastaParser). The mitochondrial genome was identified with MitoHiFi v2.0 (https://github.com/marcelauliano/MitoHiFi) and removed from the nuclear assembly. Additional removal of contaminant sequences identified by the NCBI in the submission process was removed using the BEDtools maskfasta function (Quinlan and Hall 2010).

Quality assessment

GenomeScope 2.0 (Ranallo-Benavidez et al. 2020) k-mer analysis, with a coverage cutoff of 10 M, was used to estimate genome size using a 21-mer histogram generated by k-mer counting algorithm (Kokot et al. 2017) from the filtered HiFi reads. Additional quality assessments of the assembly were performed with KAT v2.4.1 (Mapleson et al. 2017) to assess assembly composition and quality, BUSCO v 5.3 (Manni et al. 2021) to assess completeness, and program used to estimate base accuracy (YAK) v0.1 (https://github.com/lh3/yak) to assess the base accuracy of the assembly.

RNA extraction, annotation, and RNA-seq

Total RNA was extracted from individual weevils in each life stage using the RNeasy extraction kit (Qiagen, Hilden, Germany). Samples were checked for quantity and quality using a Tapestation 4200 (Agilent, Santa Clara, CA, USA). Total RNA from 48 samples (8 larvae, 4 pupae, 4 adults of mixed sex, 10 adult females, 10 adult males, and 10 immature males) were submitted to Texas A&M AgriLife Genomics and Bioinformatics Service (TxGen) for purification, library preparation, and sequencing on an Illumina NovaSeq platform (Illumina, San Die-go, CA, USA) to generate paired end 150 bp sequence reads.

Gene prediction was carried out with the Braker v2.1.5 pipeline (Hoff et al. 2019). The draft assembly was soft-masked using a custom repeat library, produced by RepeatModeler v2.0.2a, RepeatMasker v4.1.0, and tandem repeat finder, TRF v4.09.1. Raw read fastq files were cleaned and trimmed via bbmap bbduk program and subsequently mapped to the soft-masked reference assembly using hisat2 and sorted with samtools v1.9 (Supplementary File 1). Braker determined gene models by incorporating multiple evidentiary data: the soft-masked assembly, protein hints from orthodb10 arthropoda database, as well as RNA sequence read data from 48 boll weevil individuals, representing the sequential developmental stages described above. The resulting gtf file was then converted to a multi-fasta with the Augustus programming suite perl script gtf2aa.pl. This amino acid multi-fasta was annotated with BLASTp-fast via OmicsBox (Valelncia, Spain). Basic local alignment search tool (BLAST) parameters limited hits to the Insecta database and those with E-values less than or equal to 10⁻⁵. Sequences were then classified into protein families using InterProScan and assigned gene ontology (GO) terms (Blum et al. 2021; Gotz et al. 2008).

Differential gene expression analysis was performed among all life stages using a gene-wise negative binomial generalized linear model and tested for significance with the quasi-likelihood F-test as described by Chen et al. (2014). Read counts per gene were normalized within each life stage and tested for significance with a glm contrast design of adult vs pupae and pupae vs larvae. Box plots were generated for the top 10 DEGs that were significant (normalized P value < 0.01) in both contrasts. The log₂ transformed counts per million for each sample and gene were visualized in a heatmap by life stage (7 larvae, 3 pupae, and 4 unsexed adults). Expression of gene patterns throughout development was determined using pooled counts of the log transformed counts/million matrix, grouped by life stage (Supplementary File 1). Gene ontology enrichment of DEGs by stage was measured using the Fisher exact statistic with daughter ontology term elimination using the TopGO package in R (Alexa and Rahnenfuhrer 2020).

Results and discussion

Genome assembly, annotation, and organization

A total of 15.75 Gb of HiFi data were generated from a single male boll weevil. The final primary pseudohaploid assembly icAntGran1.2 consisted of 306 scaffolds with 36.5 Mb N50 and 13.9 Mb N90 (Table 1, Fig. 1) with a total size of 706 Mb. Based on GenomeScope k-mer analyses of the HiFi reads, the genome size was estimated to be 703.6 Mb, which was very similar to the HiFi generated assembly length (Supplementary Fig. 1a). A total of 34,696 gene features were annotated from Braker (Hoff et al. 2019). Of those, 30,841 have at least one RNA read mapped to them, and 14,552 annotations have a gene ontology term. Genome completeness was assessed using Benchmarking Universal Single-Copy Orthologs (BUSCOs) from both the endopterygota and insecta databases. The BUSCO scores for complete single copies were nearly complete using both databases, 98.8 and 99.7%, respectively (Table 1). An estimate of base accuracy using YAK reported a raw quality value (QV) of 49.436 and an empirically adjusted QV of 52.243, which is less affected by differences in read coverage (Supplementary Fig. 1b).

Table 1.

Assembly statistics.

	Initial assembly	Finalized scaffold assembly
Assembly length	862 Mb	702 Mb
Contig count	2118	566
Contig N50	4.42 Mb	6.98 Mb
Contig L50	43	31
Contig N90	0.16 Mb	0.84 Mb
Contig L90	539	147
Merqury QV	51.23
Scaffold count		306
Scaffold N50		36.5 Mb
Scaffold L50		8
Scaffold N90		13.9 Mb
Scaffold L90		19
YAK QV		52.24
BUSCO endopterygota 10	C: 98.8% [S: 84.5%, D: 14.3%]	C: 98.8% [S: 96.8%, D: 2.0%], F: 0.3%, M: 0.9%, n: 2124
BUSCO insecta 10		C: 99.7% [S: 96.9%, D: 2.8%], F: 0.1%, M: 0.2%, n: 1367
Repeats, %
ȃTotal % of genome		57.42
ȃȃTotal interspersed repeats		56.33
ȃȃȃRetroelements		17.90
ȃȃȃDNA transposons		14.86
ȃȃȃUnclassified		23.57
Other (small RNA, satellites, simple repeats, low complexity)		0.92

Fig. 1.

a) Boll weevil karyotype consisting of 21 autosomes and an X chromosome. Gene density is plotted above each chromosome in 100 Kb windows. b) Hi-C contact map showing the 22 distinct linkage groups.

Flow cytometry estimated the genome to be 819–854 Mb (Barcenas-Ortega 1992; http://www.genomesize.com/result_species.php?id=3785, accessed 5 Aug 2022). The discrepancy between the flow cytometry genome size estimate and final pseudohaploid genome assembly size is a common phenomenon attributed to genomic regions that are difficult to assemble, e.g. very long repeat-rich stretches (Pflug et al. 2020). Consistent with this, repeat content of the assembly was approximately 57% of the genome, with most of these sequences (56%) annotated as total interspersed repeats. Within the interspersed repeats, 17.9% were retroelements and 14.9% were DNA transposons (Table 1). Total repeat content is greater in the boll weevil genome than those found in the genomes of many other Coleopteran pests [6% for red flour beetle, Tribolium castaneum (Tenebrionidae); 17% for the Colorado potato beetle, Leptinotarsa decemlineata (Chrysomelidae); and 21% for the endangered Dichotomius schiffleri (Scarabaeidea)]. However, the percentage of total repeats found in the boll weevil genome is within the range of those reported for several other Curculionidae, which tend to vary by subfamily and genome contiguity and quality: 76% for the Easter egg weevil, Pachyrhynchus sulphureomaculatus (Entiminae); 70% for the rice weevil, Sitophilus oryzae (Dryoptheroniae); 45% for the palm weevil, Rhynchophorus ferrugineus (Dryoptheroniae), 70% for the Argentine stem weevil, Listronotus bonariensis (Cyclominae); and 17% for both coffee berry borer, Hypothenemus hampeicontains and mountain pine beetle, Dendroctonus ponderosae (Scolytinae; Van Dam et al. 2021).

The boll weevil karyotype consists of 21 autosomes + Xyp (McNally et al. 2000; Fig. 1a). The contact map (Fig. 1b) is consistent with 22 distinct linkage groups; 21 autosomes and the X chromosome. The male-specific y chromosome was not determined due to discrepancies with HiC coverage and fragmentation (Supplementary Fig. 2). Among available true weevil genomes, A. g. grandis has nearly twice as many chromosomes as the red palm weevil, Rhynchophorus ferrugineus, with 11 chromosomes (Hazzouri et al. 2020); the Easter egg weevil, Pachyrhynchus sulphureomaculatus, with 12 chromosomes (Van Dam et al. 2021); the Argentine stem weevil, Listronotus bonariensis, the scolyntine coffee, Hypothenemus hampei, and the macadamia weevil, Hypothenemus obscurus, each with 7 chromosomes (Constantino et al. 2011); and the mountain pine beetle, Dendroctonus ponderosae, with 11 chromosomes (Keeling et al. 2022). The boll weevil karyotype is similar to the rice weevil, Sitophilus oryzae, which has 22 chromosomes (Silva et al. 2018). Weevil sex karyotype is mostly Xyp with a large X chromosome in both males and females and a minute y chromosome in males, which forms a parachute (p) structure during meiosis. The y chromosome can also be completely lost or gained. Within scolytine weevils, the X chromosome can fuse with an autosome and produce a neo-x, resulting in the loss of the ancestral yp chromosome. This results in the creation of an equally large, complementary neo-y chromosome (Lanier and Wood 1968; Lanier 1981; Keeling et al., 2013). Although some work has been done to describe specific sex determining genes for other coleopteran insects, including the tenebrionid red flour beetle, Tribolium casterneum (Shukla and Palli 2012), a sister taxon of weevil, the evolution of sex determination system in weevils has been understudied despite contemporary genomic advances.

Differential gene expression

Current control measures for the boll weevil mainly involve 3 components: (1) the use of pheromone-baited traps to detect incipient weevil populations; (2) applications of an insecticide such as Malathion, an organophosphate (IRAC mode of action 1b); and (3) cultural control practices such as timely crop destruction following harvest (Dickerson et al. 2001). When weevil numbers are high, some fields may be treated weekly with Malathion throughout the cotton growing season (Patrick B, personal communication). Repeated insecticide applications of the same organophosphate formulation have raised boll weevil's seasonal tolerance to Malathion (Pietrantonio and Sronce 2001). Although resistance as a heritable trait has not yet developed, the potential further motivates development of alternative control strategies that would reduce chemical application and improve specificity to the boll weevil. We surveyed gene expression across 3 of the major life stages (larvae, pupae, adult) and reported enriched GO terms and DEG patterns therein as an initial screen for potential target genes that may be useful for pest control in the field.

RNA-seq analysis across 3 life stages (larvae, pupae, and mixed sex adults) resulted in a total of 16,618 DEGs. Comparing the larval to pupal stage, 4,632 genes were up-regulated and 3,799 down-regulated, and 2,394 were up-regulated and 2,224 down-regulated from the pupal to adult stage (Fig. 2a; Supplementary Fig. 3). DEGs were partitioned into 7 distinct groups based on expression patterns (Fig. 2b; Supplementary Table 1). Among 7 differential expression patterns, patterns 3 and 4 showed higher expression in larvae and adults compared to pupae. We focused our analyses on the genes within these patterns as both life stages feed on cotton and are agriculturally and economically impactful. Adults and larvae are also suitable targets as they are more prone to exposure than immobile pupae. Interrupting gene function for host and mate identification as well as cotton feeding may provide sustainable crop protection. Patterns 3 and 4 as a combined set included 4,248 DEGs; of these, 2,864 genes were overexpressed at least 2-fold (Supplementary Table 2). GO enrichment analysis included 41 enriched GO terms, 28 BP, 8 cellular component (CC), and 5 molecular function (MF) terms (Fig. 3a–c). DEGs fell into multiple categories and 581 genes fell into all 3 GO categories (Supplementary Fig. 4). Manual curation of genes underlying enriched GO terms revealed many biologically relevant gene groups that may have potential for use in boll weevil control. Below, we highlight several of these groups and discuss their potential in novel pest control methods.

Fig. 2.

a) Differential expressed genes across 3 main boll weevil life stages (larvae, pupae, and mixed sex adults) shown as a heatmap. The number in parenthesis is the sample size of each group. Warmer colors show more highly expressed genes. b) The seven expression patterns among life stages. Patterns 3 and 4 are patterns where mobile life stages (larvae and adult) had increased expression compared to pupae.

Fig. 3.

Enriched GO terms in patterns 3 and 4. GO identification numbers around the outside of the circle correspond to the GO descriptions in the table. Dots in each pie piece indicate the number of genes underlying that GO term. Purple colored dots represent genes that are significantly up-regulated, orange are down-regulated. The center ring is color coded to indicate z-score; the height of the bar indicates overall significance of the group. a) MF, b) BP, c) CC.

Xenobiotic cytochrome P450s

The boll weevil has a long coevolutionary history with Malvaceae host plants that possess chemical and physical defenses to prevent weevil herbivory and fitness (de Moura et al. 2022). Since the domestication and commercialization of cotton, this insect has also been the target of synthetic chemicals, which function similarly to plant xenobiotics. Field-evolved resistance to the insecticide Malathion has not been observed in the boll weevil despite significant and chronic chemical exposure for decades. However, generational tolerance has been observed in the field (Kanga et al. 1995) and selection for resistance has been demonstrated in a laboratory setting for organophosphate and other classes of insecticides (Graves et al. 1967). These data suggest that induction (or lack thereof) of detoxifying genes, such as cytochrome P450s (CYP), esterases, and glutathione transferases, is associated with maintained susceptibility to synthetic chemicals in the field (Brattsten 1987a, 1987b).

Fold change of CYPs between active and inactive life stages helps inform what genes are associated with plant feeding and xenobiotic metabolism. Thus, we annotated boll weevil CYP genes to better understand the abundance of these genes and their role throughout development. We identified 86 CYPs in the boll weevil genome, but 9 CYP annotations lacked sufficient evidence to be further categorized (Supplementary Table 2). The remaining 77 CYPs were distributed among 4 major arthropod CYP clades (Feyereisen et al. 2006) as follows: mitochondria (9); CYP 2 clade (9); CYP 3 clade (28); and CYP 4 clade (31). Over 70% of these CYPs, 59 from CYP 3 and 4, are associated with xenobiotic metabolism and insecticide resistance and 28 of these are overexpressed (2–9.7-fold, P = 3.9E⁻⁸) in the adult and larval life stages compared to the pupal stage. The total number of identified CYP genes in the boll weevil is identical to that (86) found in the fruit fly, Drosophila melanogaster (Adams et al. 2000), and 16% higher than the number (74) found in the highly adaptive Colorado potato beetle, L. decemlineata (Wan et al. 2013). However, it is considerably less than the number (147) reported for the insecticide resistant red flour beetle, T. castaneum (Zhu et al. 2013). These findings suggest that insecticide resistance in the boll weevil is not limited by CYPs, which are frequently associated with chemical resistance in many other impactful systems but may be limited by other factors such as genetic diversity. It is also possible that CYPs are not associated with organophosphate insecticide metabolism or tolerance in the boll weevil.

Odorant receptors

Odorant receptors (ORs) are chemosensory membrane-bound neuronal receptors that, together with odorant binding proteins (OBPs), play a significant role in olfaction and influence insect behavior. They are the sensory genes most closely associated with host identification, mate detection, and predator avoidance (Fan et al. 2011). We annotated 60 ORs, several of which represent alternative splices, in the boll weevil assembly including the odorant receptor coreceptor (ORCO). This co-receptor is required for smell chemosensation as it helps form a binding complex with other ORs. Only 3 ORs appear to be up-regulated (>4-fold) in larvae and adults compared to pupae (differentially expressed [DE] patterns 3 and 4) and are annotated to 3 distinct genomic regions on 2 separate contigs. There are 32 OBPs annotated in this assembly with over half (17) having at least 2-fold up-regulation (Supplementary Table 2). Furthermore, 12 of these 17 OBPs reside in a gene-rich 150 Kb region on scaffold 12. This scaffold also includes 5 ORs and ORCO, which are not differentially expressed, but reside approximately 4 Mb upstream of the OBP suite. Additionally, there are approximately 40 other coding genes in that region. Two of these genes are annotated as transcription regulators: protein fork head, which regulates RNA polymerase II and is located ∼200 Kb upstream of the OBPs; and TF Sox-14, which is identified 1.9 Mb upstream of this OBP region. This gene-rich region, replete with ORs, ORCO and many tandem OBPs, is also an active genic region for boll weevil pheromone recognition and cotton detection and is therefore an interesting region to develop molecular technologies such as targeted gene therapy, via CRISPR-cas9 (clustered regularly interspaced short palindromicrepeats and associated protein 9), RNA interference, or transgenic cotton.

Pheromone biosynthesis

The boll weevil pheromone is an aggregation pheromone, produced by adult males, but attractive to both sexes (Hardee et al. 1972). It is composed of 4 pheromone components: 2 terpene alcohols [components I and II; (+)-cis-2-isopropenyl-1-methyl cyclobutaneethanol and cis-3,3-dimethyl-Δ1,β-cyclohexaneethanol, respectively] and 2 terpene aldehydes (components III and IV; cis-3,3-dimethyl-Δ1,α-cyclohexaneacetaldehyde and trans-3,3-dimethyl-Δ1,α-cyclohexaneacetaldehyde; Tumlinson et al. 1969). Pheromone is produced via the mevalonate pathway, which also includes juvenile hormone production. This pathway is well characterized in insects and in other weevil species (Bellés et al. 2005; Tittiger and Blomquist 2016). A previous report highlighted DEG within the mevalonate pathway between pheromone-producing male boll weevils and those not producing pheromone which included several up-regulated intermediate chemicals leading to isopentenyl synthesis (Perkin et al. 2021). In the current study we found 3 genes up-regulated in the mevalonate pathway that produce major pheromone precursors, 1 isopentenyl-PP (3.87-fold change, P = 0.17) and 2 farnesyl-PP (5.14-fold change, P = 0.10 and 3.24-fold change, P = 0.13) in larvae and adults compared to pupae (Supplementary Table 2). We also found a farnesol dehydrogenase, which is a precursor to juvenile hormone, that was one of the top 10 most highly expressed genes in the dataset (Supplementary Fig. 5). Pheromone is produced only in adult males, but juvenile hormone is needed in larvae for molting and other developmental processes. Pheromone plays a key role in host location and mating, making both pheromone production and larvae development important candidate processes to target for pest management.

Digestive proteinases

Serine and cysteine proteinases have a variety of roles in insects including apoptosis, embryonic development, immune responses, and digestion (Cerenius et al. 2010). Some pest management strategies have incorporated molecular techniques to target digestive proteinases in the gut by feeding insects proteinase inhibitors or inhibiting proteinase gene expression using RNA interference. These strategies have led to smaller larvae, longer developmental times, and increased mortality (Medel et al. 2015; Zhou et al. 2008; Pitino et al. 2011; Zhu and Palli 2020). Similarly, neonate boll weevil larvae reared on diet with trypsin and chymotrypsin inhibitors had reduced weight, increased mortality, and developmental deformities (Franco et al. 2004). Early work annotating boll weevil gut proteinase genes showed they are dominated by serine proteinases, specifically trypsin and chymotrypsin (Franco et al. 2004). Fourteen serine proteinases were identified using reverse transcription PCR, 11 trypsin and 3 chymotrypsin-like sequences, with 4 of them expressed in the boll weevil gut (Oliveira-Neto et al. 2005). A more recent study using pyrosequencing identified both serine and cysteine proteinases (12 and 98, respectively) in the boll weevil larval gut, suggesting digestion in the boll weevil may require both types of proteinases (Salvador et al. 2021). In this study, we identified 59 serine, serine-like, cysteine, and cysteine-like proteinase transcripts up-regulated at least 2-fold in adults and larvae compared to pupae (Supplementary Table 2). Oliveira-Neto et al. (2005) identified 1 chymotrypsin that was induced by feeding, expressed in both larvae and adults, and concentrated in the adult gut. The same chymotrypsin is highly up-regulated in larvae and adults in our study; in fact, it is one of the top 10 most highly expressed genes in the dataset (Supplementary Table 2) for adults and larvae (11.77-fold, P = 1.48e⁻⁰⁹). More work is needed to fully annotate boll weevil gut proteinases and to determine specific targets for proteinase inhibitors.

Conclusion

A high-quality reference genome was presented for the boll weevil, A. grandis grandis, representing the first Anthonomus weevil to be fully sequenced. This genome will be a primary resource for genome comparisons to understand the genetic and evolutionary relationships across weevils. Additionally, DEGs and enriched gene ontology terms were identified for 3 major life stages, representing the genetics underlying the development of this insect. Several notable gene families were highlighted that are relevant to boll weevil biology, host plant interaction, and insecticide detoxification. This high-quality reference genome for the boll weevil will make definitive annotations possible and potentially lead to the development of reliable templates to design gene-specific management tools that are safe, eco-friendly, and target-specific.

Data availability

All raw reads used to assemble and scaffold this Anthonomus grandis grandis genome assembly were deposited at DDBJ/ENA/GenBank within BioProject PRJNA767408. These include Sequence Read Archive (SRA) accession SRR16132310 containing HiFi data used to generate primary and alternate haplotig assemblies and accession SRR18146191 containing short-paired end reads used for Hi-C scaffolding. The primary icAntGran1.2 and icAntGran1.2 alternate haplotype assembly versions discussed in this manuscript are available in GenBank accessions JAKYJU000000000.2 (BioProject PRJNA767408) and JAKYJV000000000.2 (BioProject PRJNA807391), respectively. Both assemblies are under the Ag100Pest umbrella project, BioProject PRJNA555319. The primary assembly and the official gene set are also available at the i5k Workspace@NAL (Poelchau et al. 2015), and annotations used for DEG in this study are available upon request. RNA sequences for life stages were deposited to SRA BioProject PRJNA734329.

Supplemental material available at G3 online.

Funding

This work was supported by the US Department of Agriculture, Agricultural Research Service (USDA-ARS) and supported under CRIS Project 3091-22000-038-00D.