Silk Gland Gene Expression during Larval-Pupal Transition in the Cotton Leaf Roller Sylepta derogata (Lepidoptera: Pyralidae)

The cotton leaf roller, Sylepta derogata, is a silk-producing insect pest. While young larvae feed on the underside of leaves, the older ones roll cotton leaves and feed on the leaf edges, which defoliates cotton plants. The larvae produce silk to stabilize the rolled leaf and to balloon from used to new leaves. Despite the significance of silk in the biology of pest insect species, there is virtually no information on the genes involved in their silk production. This is a substantial knowledge gap because some of these genes may be valuable targets for developing molecular pest management technologies. We addressed the gap by posing the hypothesis that silk gland gene expression changes during the transition from larvae to pupae. We tested our hypothesis using RNA-seq to investigate changes in silk gland gene expression at three developmental stages, 5th instar larvae (silk producing; 15,445,926 clean reads), prepupae (reduced silk producing; 13,758,154) and pupae (beyond silk producing; 16,787,792). We recorded 60,298 unigenes and mapped 50,158 (larvae), 48,415 (prepupae) and 46,623 (pupae) of them to the NCBI database. Most differentially expressed genes in the 5th instar larvae/prepupae libraries were relevant to nucleotide synthesis and maintenance of silk gland function. We identified down-regulated transcriptional factors and several genes involved in silk formation in the three libraries and verified the expression pattern of eight genes by qPCR. The developmental- and tissue-specific expression patterns of the fibroin light chain gene showed it was highly expressed during the larval silk-producing stage. We recorded highest expression of this gene in the larval silk gland, compared to other tissues, including midgut, hindgut, epidermis, Malpighian tubes, hemolymph and fat body. These data are a genetic resource to guide selection of key genes that may be targeted for in planta and other gene-silencing technologies for sustainable cotton agriculture.


Introduction
Insects representing at least 20 orders produce silk thread. Some produce silk only in the larval or adult stages and others in both [1]. In a narrow sense, the term 'silk-producing insects' refers to domesticated and wild species used in commercial silk industries, however, the silks of many insects, particularly pest species, are not commercially useful. Some silk-producing insects inflict tremendous damage on crops and forests. For example, neonates of the lackey moth, Malacosoma neustria testacea Motsch, a forest and fruit tree pest, produce silk, form a larval web, and then feed on new buds and leaves. All the leaves can be destroyed during population outbreaks [2]. The rice leaf folder, Cnaphalocrocis medinalis Guenee (Lepidoptera: Pyralidae) folds rice leaves vertically with silk, creating a sort of tent in which larvae feed on the rice epicuticle and leaf tissue. The resulting damage leads to substantial rice production losses [3]. The cotton leaf roller (CLR), Sylepta derogata Fabricius, rolls cotton leaves into conical structures and feeds on the edges of the rolled leaf, resulting in serious defoliation. The larvae use silk to stabilize their leaf cones and to balloon to other leaves. High pest populations significantly decrease cotton production [4].
Although silk plays fundamental roles in the biology of these and other pests, information on their silk glands and mechanisms of silk production remains relatively scarce, particularly compared to the abundant literature on silk production in the domesticated silk worm, Bombyx mori [5]. This is a crucial lacuna because silk glands of insects in various orders develop from different cell types and they are associated with different anatomical structures [6]. More to the point, due to their actions in the biology of pest species, genes involved in silk production are potential targets for crop-and pest-specific molecular technologies designed to manage pests by silencing specific genes. We addressed the lacuna by posing the hypothesis that silk gland gene expression changes during the developmental transition from larvae to pupae. Here, we report on the outcomes of our transcriptomic analyses designed to test our hypothesis.

Insects
CLRs were collected from local velvetleaf, Abutilon theophrasti Medic (Malvaceae), a cosmopolitan weed which grew naturally near the field in the Yangzhou University experimental farm located at 32°42 0 N, 119°4 0 E. Neither the CLRs nor the velvetleaf are endangered and protected species. Thus, no specific permissions were required for this research). The insects were maintained on fresh velvetleaf leaves ab lib at 27±1°C, 70%±7% relative humidity (RH), and 14L:10D in the laboratory. Fifth-instar larvae, prepupae and pupae (day 2) were anesthetized and the silk glands were isolated (20 glands/preparation, two biologically independent preparations) for transcriptome sequencing. The resulting transcriptomes represent phases of the CLR life cycle characterized by high, reduced and no silk production.

cDNA library construction and Illumina sequencing
Total RNA was extracted with the SV Total RNA Isolation System (Promega, Madison, WI, USA) following the manufacturer's protocol. Quality and quantity of the RNA was determined on a Nanodrop spectrophotomter (Thermo Fisher, Massachusetts, USA). cDNA library construction and Illumina sequencing were performed at the Beijing Berry Genomics Co. Ltd. (Beijing, China). 20 μg of pooled total RNA was fragmented in a Biorupter Cracker for 2 min following the manufacturer's protocol (Diagenode, Liège, Belgium). Fragments of 250-300 bp were recovered and purified using the Oligotex mRNA MiniKit (Qiagen, Dusseldorf, Germany). Using these fragments as templates, random hexamer primers (6 bp) were employed to

Analysis of transcript expression differences between developmental stages
The transcript expression abundances were calculated by Reads per Kilobase of exon model per Million mapped reads (RPKM), which eliminates the influence of different gene lengths and sequencing discrepancy on the calculation of expression abundance. The formula is: RPKM ¼ total exon reads mapped reads ðmillionsÞ Â exon length ðKBÞ We set the q-value at<0.05 as the minimum false discovery rate for the statistically significance differences in gene expression.

qPCR and data analysis
We isolated larvae, prepupae and pupae as described, collecting 30, 45, 60 silk glands from each group, with three independent biological replicates. RNA was isolated and cDNA prepared as described. qPCR was performed using the Mastercycler ep realplex (Eppendorf, Germany). Gene specific primers (Table 1) were designed using Beacon Designer 7.6 and synthesized by Sangon Biotech Co., Ltd (Shanghai, China). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a stably expressed reference gene [7]. The reaction program was: 2 min at 95°C for 15 s, 55°C for 30 s, and 72°C for 30 s. Melting curves confirmed a single gene-specific peak and the absence of primer dimmer peaks. GoTaq qPCR Master Mix (Promega, Madison, WI, USA) was used to measure the mRNA levels according to the manufacturer's instructions. A five-fold dilution series was used to construct a relative standard curve to determine the PCR efficiencies and for quantification. Each reaction was run in triplicate (technical replications) with three independent biological replicates. Relative quantification of 8 genes was calculated by the comparative 2 -ΔΔCT method [8] to identify the relative mRNA levels of the samples from different life stages.

Development-and tissue-specific expression of a gene encoding a fibroin light chain (FLC)
Larvae, instars 1-5, prepupae, pupae, and adult females and males were collected (about 0.08 g tissue/sample) with 3 independent biological replicates, frozen in liquid nitrogen and kept under -70°C until analysis.
The midgut, hindgut, silk gland, fat body, epidermis, and Malpighian tubes were isolated from 5 th instar larvae (5 tissues per pool, 3 independent biological replicates). The larvae were washed with distilled water twice and anesthetized on ice. The pleopod was cut with dissecting scissors and hemolymph was collected with pressure on the body using a pipettor. All of the tissue samples were frozen in liquid nitrogen and kept under -70°C for analysis [9]. R NA was isolated and cDNA prepared as described. Development-and tissue-specific qPCR analysis of FLC expression was performed as described.

Transcriptome sequencing and sequence assembly
Before mapping the tags, the transcriptomes were preprocessed, yielding 42,040,007 clean reads and 60,298 unigenes. All the raw tags were filtered with reference sequences, leaving 15,445,926, 13,758,154, and 16,787,792 clean reads from the larval, prepupal and pupal libraries, respectively ( Table 2). Analysis of sequencing saturation showed that the number of detected genes increased until the sequencing reads reached 3 million or more, indicating the identified expressed reads were sufficient to represent the entire transcriptional information of the CLR genome.

Gene identification and annotation
All distinct sequences longer than 200 bp were searched against the NR, Swiss-prot, and KEGG protein databases by BLASTX with a cut-off E-value of 1e -10 . We recorded 19,090 (41.1% of all distinct sequences; by NR), 12,538 (27.0%; Swiss-prot), 11,260 (24.3%; NT) and 1,567 (3.4%; KEGG) transcripts (Fig 1). Identities of many genes (about 37%) were not found in available databases, as expected due to the lack of transcriptomic information in tissues within pest species.

Gene expression profiles
We identified 50,158, 48,415, and 46,623 transcripts from the silk gland libraries. Among these, 37,846 transcripts were expressed in all three libraries, 44,897 were expressed in larval and prepupal libraries, and 39,873 were expressed in prepupal and pupal libraries. We recorded 40,771 transcripts expressed in larvae and pupae (Fig 2). Expression of 38 genes changed (27", 11#) during the larval/prepupal transition. The prepupal-pupal transition led to expression changes in 355 genes (137", 218#). The broader excursion from larvae to pupae included up-regulation of 140 and down-regulation of 169 genes ( Table 3; specific genes listed in S1 Table). Genes related to silk formation We selected seven silk gland genes that changed in expression during the developmental excursion as directly or indirectly involved in silk formation ( Table 4; partial sequences of specific genes listed in S2 Table). Expression of genes encoding the heavy-and light-chain fibroin proteins declined as the larvae moved beyond the silk-producing larval stage. A gene encoding a fibroin-modulator-binding protein-1 is a transcriptional regulatory factor operating in the B. mori silk gland. Expression of this gene increased by almost 3-fold as the larvae went into the pre-pupal stage. Three genes act in endocrine signaling, which we take to indirectly influence silk production via their influence on silk gland developmental physiology as the insects moved from larvae to prepupae to pupae. We included a gene encoding a phosphatidylethanolamine binding protein (PEBP) because it acts in cell signaling and may indirectly influence silk production. Expression of this gene declined by a wide extent as the larvae advanced to prepupae.  Genes encoding transcription factors Development from larvae to pupae was attended by changes in expression of genes encoding a range of transcription factors. Expression of 17 genes changed expression (14", 3#) going from larvae to prepupae; 21 genes (14", 7#) changed expression in the transition from prepupae to pupae. We listed these genes in a separate table (Table 5; specific gene sequences listed in S2 Table) to highlight the actions of their cognate proteins as transcription factors. Some genes underwent very large changes in expression, 20 of which are listed in Table 6 (specific gene sequences listed in S2 Table). A gene encoding PEBP decreased in expression by about 1400-fold and expression of a gene encoding a possible serine protease inhibitor increased by over 1200-fold. Apparent development regulation of several other genes, including genes encoding a sugar transporter protein, a peroxidase, and an aldehyde oxidase I, went through similar large-scale expression changes. We list these in Table 6 to emphasize the scale of expression changes.

Confirmation of gene expression
Expression analysis of 8 genes demonstrated parallel expression changes recorded by RNA-seq and qPCR (Fig 3). Of particular interest, genes encoding a xanthine dehydrogenase (XDH), a possible CRCT domain protein BRC1 (Brc1), and ribonuclease P protein subunit p40 (RPP0) were expressed in parallel.

Discussion
The data reported in this paper strongly support our hypothesis that silk gland gene expression changes during the developmental transition from larvae to pupae. Three points form a solid argument. One, we generated transcriptomes from silk glands prepared from larvae, prepupae and pupae, from which we identified 355 silk gland genes that changed expression during development from 5 th instar larvae to pupae, understanding that not all 355 genes act directly in silk production. Two, we confirmed expressed changes in 8 genes by qPCR, indicating our RNA-seq data reliably represents changes in gene expression. Three, 38 of the expressionaltered genes encode transcription factors, which would influence expression of other genes.  Taken together, these points support our inference that gene expression in CLR silk glands is under strong developmental control. In silk gland biology, the transition represents a period of high silk production in 5 th instar larvae, to reduced production in pre-pupae to atrophy of the silk gland in pupae. We surmise developmental regulation of gene expression makes up the proximal mechanism of these changes in silk gland biology. The biological significance of regulating silk gland gene expression touches on several areas. As seen in many areas of biology, such as development of reproductive systems, silk production is restricted to the life stages in which silk contributes to CLR biology. Overall CLR fitness would be substantially reduced due to biological costs of allocating resources toward maintaining unnecessary silk glands and to protein-costly silk production. Perhaps more to the point, unnecessary silk could be directly deleterious to CLR and many other silk-producing insect species. We infer that silk gland gene expression is tightly regulated.
The major developmental hormones, juvenile hormone (JH) and 20-ecdysteroid (20E), signal the transition from larvae to pupae and also influence silk gland biology [11]. The JH analog, methoprene, is applied to silkworms, B. mori, to extend the last larval instar and enhance silk production [6]. JH esterase is responsible for the developmentally-regulated decline in JH titers associated with pupation [12]. Our data show that JH esterase is expressed at low levels in 5 th instar larval silk glands, then increases by 13-fold in pre-pupal silk glands before declining to near-zero in pupae. This is consistent with high silk production during cocoon formation and subsequent atrophy of the silk gland. Similarly, ecdysone oxidase (EO) reduces 20E titers  by converting 20E into inactive 3-dehydroecdysone [13]. Expression of the gene encoding EO in silk glands declined by about 1100-fold to virtually zero in the larval-pre-pupal transition. We infer the decline in silk gland EO expression is consistent with increased 20E titers during silk production and preparation for the pupal molt. Altered expression of silk gland genes encoding enzymes that regulate JH and 20E titers aligns with the actions of these hormones in silk production during development. Gene expression is influenced by a large number of transcription factors. We recorded substantial changes in expression of genes encoding transcription factors in the larval-pupal transition. Among these, expression of 10 genes increased by > 6-fold and expression of another 10 decreased by >3-fold. Genes encoding sequence-specific DNA-binding transcription factor  activities, MMP14 (35673-35676) increased by > 8-fold. These large changes in expression of an array of genes indicate the dynamic metabolic and developmental changes in the silk gland during silk production and silk gland atrophy during subsequent development into pupae.
In particular, expression of the molt-regulating transcription factors HR3 (37748-37751) increased by 6-to 7-fold in the larval-pre-pupal transition, then decreased by circa 3-fold in the pre-pupal-pupal period. These proteins act in gene regulation during metamorphosis. Silk gland-specific changes in expression of these genes are associated with the changes in the structure and function of the gland during the larval-pupal metamorphosis. HR3 genes are crucial to molting physiology at the organismal level and they trigger hypotheses about the actions of these and other transcription factors in the silk gland [14]. Fibroin and sericin are the main silk proteins. Fibroin is made of heavy-and light-chain polypeptides incorporated with the 25 kDa protein, P25. Sericin is a glycoprotein responsible for agglomerating the silk proteins [15]. Silk gland expression of gene 8810, encoding a fibroin heavy chain, was highest in larvae, lower by about 4.3-fold in pre-pupae and still lower, by about 9-fold, in pupae. We recorded similar data for a FLC, high expression in larvae, slightly lower in pre-pupae and reduced by about 21-fold in pupae. This pattern is consistent with the silk-producing biology of the silk gland.
Our qPCR analysis shows FLC expression gradually increased during larval development to a peak at the 5 th instar. Because older larvae depend on silk, we infer the FLC expression pattern is consistent with the silk-producing and leave-rolling behavior of CLR larvae. Similarly, silk gland FLC expression was much higher compared to other tissues. Again, our interpretation is FLC acts in silk production.
Developmental regulation of silk gland gene expression led to very large changes in some genes. These rather striking changes are not unexpected because the silk glands undergo fundamental changes from metabolically active, silk-producing tissues to regulated atrophy in a limited time frame. The functions of some genes, such as yellow-d (30647), are not yet known well enough to generate meaningful commentary. Changes in other genes lead to useful hypotheses about their specific functions in the context of silk gland biology. The purpose of this first description of silk gland transcriptomes is to report on the large changes in gene expression. Detailed discussion of each gene would amount to speculations and we limit our discussion of individual genes to a phosphatidylethanolamine binding protein (PEBP), a peroxidase and a chemosensory protein.
PEBPs form a protein family, members of which occur in a very wide range of organisms, including bacteria, plants and animals [16]. These proteins are mostly signal moieties with substantial importance in biomedicine [17]. Silk gland gene 33585 encodes a PEBP that declined in expression (by almost 1400-fold) during the larval-pre-pupal excursion. The biological functions of PEBP in insects are not fully appreciated, but they probably act in intracellular signaling roles, as reported in biomedical research. Reumer et al. [18] discovered the Drosophila PEBP1 (CG18594) is expressed in larval fat bodies and in the Drosophila larval hemocyte cell line, l(2)mbn. The authors investigated the idea that PEBP conveys immune protection against bacterial infections. They showed that larvae overexpressing PBEP1 synthesized and released increased anti-microbial proteins into the hemolymph and that infection stimulated a 6.5-fold increase in PBEP1 expression. They inferred that Drosophila PEBP1 acts in immunity. We do not yet appreciate the specific role of this gene in silk glands, but surmise that PEBP acts in various signaling functions.
Expression of a peroxidase (POX; 46154) was down-regulated by nearly 1,000-fold in the transition from prepupae to pupae. POXs act in many areas of insect biology, including cuticle tanning and anti-oxidant metabolism. A new role of POXs has recently come to light. POXs also generate prostaglandins (PGs) that mediate follicle maturation [19] and coordinate expression of eggshell genes in Drosophila [20]. Most recently, Park et al. [21] reported that POXs act in PGE 2 -mediated immune reactions to infection. We speculate that PGs act in silk gland physiology and down-regulation of POX is congruent with the loss of function and break down of the silk gland.
Insects sense chemicals via soluble proteins, which act by binding air-borne chemicals, such as sex pheromones and plant volatiles. The soluble proteins, bound with their signal ligands, interact with specific chemoreceptors located on sensory nerve membranes, which leads to the transduction from chemical to electrical signals and subsequent integration within the central nervous system [22]. Insects express two families of soluble olfactory proteins, odorant binding proteins (OBPs) and chemosensory proteins (CSPs). CSPs are smaller than OBPs (100-120, compared to 150-160 amino acids), although they act in similar functions [22]. CSPs are expressed in antennae, but also in other, non-sensory, tissues and body parts, such as antennae-removed heads, thoraces, abdomens, gonads, wings and pheromone glands [23]. The broad distribution of these proteins led to suggestions that CSPs act in additional roles, possibly female reproduction, limb regeneration and embryonic development. We report that silk glands express one gene encoding a CSP, 31019, which declined in expression by about 1,000-fold during the larval-pre-pupal transition. We infer this protein acts in the silk gland during the silk-producing phase of pupation, possibly directly in silk production.
Arthropods, including pest insects, use silk in a very wide range of biological functions, including food capture, nests, guidelines, dispersal, reproduction, egg coatings and protective cocoons. Some silk-producing insects influence the structure of their environments. For an unusual example, the aquatic insect Hydropsyche siltalai (Tricoptera, Hydropsychidae, netspinning caddisflies) produces silk cup-like nets to capture small prey organisms. Statzner et al. [24] reported that their silk is fixed to gravel pieces that results in gravel consolidation in stream beds, which influences sediment erosion. The concept that insect silk production can influence local environments prompts our suggestion of substantial research into pest insect silk production.
Supporting Information S1