Characterization of E93 in neometabolous thrips Frankliniella occidentalis and Haplothrips brevitubus

Insect metamorphosis into an adult occurs after the juvenile hormone (JH) titer decreases at the end of the juvenile stage. This generally coincides with decreased transcript levels of JH-response transcription factors Krüppel homolog 1 (Kr-h1) and broad (br), and increased transcript levels of the adult specifier E93. Thrips (Thysanoptera) develop through inactive and non-feeding stages referred to as “propupa” and “pupa”, and this type of distinctive metamorphosis is called neometaboly. To understand the mechanisms of hormonal regulation in thrips metamorphosis, we previously analyzed the transcript levels of Kr-h1 and br in two thrips species, Frankliniella occidentalis (Thripidae) and Haplothrips brevitubus (Phlaeothripidae). In both species, the transcript levels of Kr-h1 and br decreased in the “propupal” and “pupal” stages, and their transcription was upregulated by exogenous JH mimic treatment. Here we analyzed the developmental profiles of E93 in these two thrips species. Quantitative RT-PCR revealed that E93 expression started to increase at the end of the larval stage in F. occidentalis and in the “propupal” stage of H. brevitubus, as Kr-h1 and br mRNA levels decreased. Treatment with an exogenous JH mimic at the onset of metamorphosis prevented pupal-adult transition and caused repression of E93. These results indicated that E93 is involved in adult differentiation after JH titer decreases at the end of the larval stage of thrips. By comparing the expression profiles of Kr-h1, br, and E93 among insect species, we propose that the “propupal” and “pupal” stages of thrips have some similarities with the holometabolous prepupal and pupal stages, respectively.


Introduction
The manner of insect metamorphosis is classified into three major classes, i.e. ametaboly, hemimetaboly, and holometaboly. Among these classes, hemimetabolous insects develop directly from the nymph to the adult; the external morphology of hemimetabolous nymphs is similar to that of adults except for the absence of wings, and successive growth of wing pad is observed in the nymphal stage. Formation of the adult wings and genitalia is completed at the end of the nymphal stage. By contrast, holometabolous larvae develop to the adult through several larval instars and the pupal stage; degeneration of larval tissues and formation of adult structure proceed rapidly during the pupal stage. In addition to these categories, there is a unique type of metamorphosis called Neometaboly. Neometaboly, observed in thrips (Thysanoptera) and a part of hemipteran species such as whiteflies and male mealybugs, is defined by a life cycle in which feeding larvae develop to the adults via non-feeding stages with external wing primordia [1][2][3]. As is observed in other hemimetabolous species, the external morphology of the neometabolous nymphs resembles that of adults, but there are no external wing pads. Rapid proliferation of wing primordial cells starts at the end of the nymphal stage, then the wings become visible externally in the subsequent non-feeding stage. In the case of thrips, the inactive stage is called "propupa" and "pupa", and they complete adult development through molting; the number of "propupal" and "pupal" instars varies among thrips species. The "propupa" and "pupa" of thrips look somewhat similar to pupae in holometabolous species, but unlike holometabolous pupae, "propupa" and "pupa" of thrips are mobile.
Insect molting and metamorphosis are generally regulated by molting hormone (ecdysone) and juvenile hormone (JH) [4][5][6]. In the post-embryonic development of holometabolous insects, JH is synthesized and secreted continuously through the penultimate larval instar, and pulses of ecdysone induce larva-larva molts in the presence of abundant endogenous JH. Once JH biosynthesis ceases in the final larval instar, ecdysone triggers larva-pupa and pupa-adult metamorphosis. A similar role of these hormones has been observed in hemimetabolous insects: ecdysone induces nymph-adult metamorphosis after JH titer decreases [7].
Transcription factors that mediate hormonal signaling in insect metamorphosis have been identified and characterized mainly in model insect species such as the fruit fly Drosophila melanogaster, the red flour beetle Tribolium castaneum, the silkworm Bombyx mori, and the cockroach Blattella germanica. Krüppel homolog 1 (Kr-h1) was identified as an early response gene in JH signaling [8][9][10] downstream of JH receptor Methoprene-tolerant (Met)/Taiman (Tai) complex [11,12]. Kr-h1 was reported as a repressor of metamorphosis activated by JH in a Met-dependent manner in hemimetabolous Pyrrhocoris apterus [13]. Subsequently, downstream target genes of Kr-h1 have been identified, including broad (br) [14] and Ecdysoneinduced protein 93F (E93) [15]. br is a transcription factor that directs holometabolous pupal development [16][17][18][19] and promotes growth of wing pads in hemimetabolous species [19][20][21]. In general, the expression of br is induced by 20-hydroxyecdysone (20E), and this induction could also be affected by JH [19]. br expression in pupae could be induced by exogenous JH via Met and Kr-h1 (Konopova and Jindra 2008;Minakuchi et al., 2009) [10,16]. In the silkworm B. mori, it was revealed that JH represses the 20E-mediated induction of br through the binding of Kr-h1 to its binding site (Kr-h1 binding site, KBS) in the br promoter [14]. Meanwhile, E93, an ecdysone-responsive, helix-turn-helix transcription factor with a Pipsqueak DNA-binding motif [22,23], was identified as a key factor that promotes adult development [24]. The role of E93 in insect metamorphosis was first reported in D. melanogaster: E93 is involved in salivary gland cell death in the prepupal stage in response to the ecdysteroid peak [22]. Functional analyses revealed that E93 is essential for adult development in hemimetabolous cockroach B. germanica, holometabolous D. melanogaster and T. castaneum [24], as well as programmed cell death of nymphal tissues [25]. In B. mori, analysis of the promoter region of E93 revealed that JH suppresses ecdysone-inducible E93 transcription via Kr-h1 [15]. The genetic interaction between Kr-h1 and E93 has also been clarified in the hemimetabolous B. germanica and in the holometabolous T. castaneum [7,26]: Kr-h1 suppresses the upregulation of E93 in the juvenile stage, whereas E93 represses Kr-h1 expression in adult metamorphosis. Recently, expression analyses of E93 revealed that the differential expression of E93 is involved in sexual dimorphic development in the strepsipteran Xenos vesparum [27] as well as in the mealybug Planococcus kraunhiae [28]. The regulatory interaction among Met, Kr-h1, and E93 is considered as the center of JH signaling, and called as 'Met-Kr-h1-E93 (MEKRE93) pathway' [7,29], or 'Metamorphic Gene Network' [26,30], which is thought to exist both in hemimetabolous and holometabolous species.
Thrips, categorized as a member of neometabolous species, develop through inactive nonfeeding stages referred to as "propupa" and "pupa". To understand the mechanisms of hormonal regulation in thrips metamorphosis, we previously examined the developmental expression profiles of Kr-h1 and br in the western flower thrips Frankliniella occidentalis (Thripidae), which has one "propupal" instar and one "pupal" instar, and in Haplothrips brevitubus (Phlaeothripidae), which has one "propupal" instar and two "pupal" instars, i.e. pupa I and pupa II [31]. We revealed that the expression profiles of Kr-h1 and br in the post-embryonic development are similar to those reported in holometabolous insect species: the transcript levels of Kr-h1 and br decreased in the "propupal" and "pupal" stages, and that their transcription was upregulated by the presence of exogenous JH [31].
To understand the hormonal regulation of metamorphosis in thrips in more detail, we here analyzed the developmental expression profiles of E93 in F. occidentalis and H. brevitubus by quantitative RT-PCR, and compared them with those of Kr-h1 and br. We also analyzed the effects of treatment with an exogenous JH mimic (JHM) at the onset of metamorphosis on E93 expression.

Thrips rearing
Frankliniella occidentalis were provided by Professor T. Murai in Utsunomiya University. Larvae and adults were fed with germinated broad bean seeds (Kokusai Pet Food, Kobe, Japan) in plastic containers in accordance with a previously reported method [32], at 23±1˚C with a long-day photoperiod (16L-8D). Haplothrips brevitubus were provided by Ishihara Sangyo Company (Osaka, Japan). They were raised with eggs of the Mediterranean flour moth Ephestia kuehniella (commercially available as Ento-food, from Arysta LifeScience, Tokyo, Japan) at 25±1˚C with a long-day photoperiod (16L-8D) as reported previously [31].

cDNA cloning
Total RNA was isolated from the whole body of F. occidentalis "propupae" and "pupae" using TRIzol reagent (Thermo Fisher Scientific, MA) with nuclease-free glycogen (Thermo Fisher Scientific) as a carrier, and first strand cDNA was synthesized using PrimeScript II reverse transcriptase with random hexamers (Takara Bio, Shiga, Japan). Primers for F. occidentalis E93 (FoE93) were designed from our unpublished transcriptome database (RNA-seq). A part of FoE93 cDNA was amplified based on an RT-PCR-approach using Tks Gflex DNA Polymerase (Takara Bio). Primer sequences are listed in S1 Table. To amplify part of H. brevitubus E93 (HbE93) cDNA, total RNA was extracted from the whole body of H. brevitubus using TRIzol reagent and nuclease-free glycogen as a carrier, and reverse-transcribed using a PrimeScript 1st strand cDNA Synthesis Kit with oligo-dT primers, as described previously [31]. Degenerate primers were designed based on the conserved amino acid sequences in the Pipsqueak DNA-binding motif [23] of other insects' E93. PCR with HbE93-F1 and HbE93-R1 primers was performed at an annealing temperature of 50˚C, followed by nested PCR with HbE93-F2 and HbE93-R2 primers at an annealing temperature of 40˚C. Primer sequences are listed in S1 Table. These PCR products were purified, cloned into a pGEM-T Easy vector (Promega) and sequenced. DNA sequence data have been deposited in the DDBJ/EMBL-Bank/GenBank International Nucleotide Sequence Database with the following accession numbers: LC415027 (HbE93) and LC415028 (FoE93).

Quantitative RT-PCR analysis
To analyze the developmental expression profiles and the transcript level after JHM treatment using quantitative RT-PCR, cDNA synthesis was performed as reported previously [31]. Briefly, a bunch of staged animals were pooled for RNA isolation at each time point. To analyze the transcript level in JHM-treated F. occidentalis or H. brevitubus (see below), 8-14 individuals were pooled for RNA isolation. Total RNA was isolated from the whole body using TRIzol reagent with RNase-free glycogen (Thermo Fisher Scientific) as a carrier, treated with DNase I (Takara Bio), and reverse-transcribed using a PrimeScript 1st Strand cDNA Synthesis Kit with an oligo-dT primer. Alternatively, some RNA samples were reverse transcribed using a PrimeScript RT reagent Kit with gDNA Eraser (Takara Bio; results shown in Fig 1C-1H).
The transcript levels were quantified using a real-time thermal cycler (Thermal Cycler Dice TP800, Takara Bio). Quantitative RT-PCR was carried out in a 14-μl reaction volume containing SYBR Premix Ex Taq (Takara Bio), 0.2 μM of each primer (see S1 Table), and 1 μl of template cDNA or standard plasmids. PCR conditions were 95˚C for 30 s, followed by 40-45 cycles at 95˚C for 5 s and 60˚C for 30 s. After the thermal cycling, the absence of unwanted byproducts was confirmed by melting curve analysis. Serial dilutions of a plasmid containing a part of the ORF of each gene were used as standards. Analysis was performed two or three times for each sample, and the average was calculated. The transcript levels of br, Kr-h1, and E93 were normalized to that of rpL32 in the same samples. Primer sequences are listed in S1 Table.

Hormonal treatments
Juvenile hormone mimic (JHM) treatment was performed as reported previously [31]. Briefly, a JHM, pyriproxyfen, was dissolved in methanol at a concentration of 3.2 μg/ml (10 mM). To examine the effect of JHM on the expression of E93, newly molted F. occidentalis "propupae" within 8 h after ecdysis were anesthetized with ether for 1.5-2 min, were dipped into 10 mM pyriproxyfen or methanol as a control for 10 sec, and total RNA was extracted 48 h later when they were on Day 0 of the pupal stage. Four to 7 animals were pooled for RNA isolation in each replicate, and treatments were performed in 4 biological replicates.
Newly molted H. brevitubus "propupae" within 5 h after ecdysis were anesthetized on ice, dipped in 10 mM pyriproxyfen or methanol as a control, and total RNA was extracted 72 h later when they were on Day 1 of the Pupa II stage, as described previously [31]. Eight to 11 animals were pooled for RNA isolation in each replicate, and treatments were performed in 2 biological replicates.
In both species, a few animals that died immediately after the treatment, due to the toxicity of the solvent (less than 5% of total animals tested), were excluded from the experiments. Quantitative RT-PCR was conducted as described above.

Statistical analyses
Multiple comparison was performed by one-way analysis of variance (ANOVA) and Tukey-Kramer method by using the software GraphPad Prism version 8.

cDNA sequences of E93 of F. occidentalis and H. brevitubus
Tblastn searches of the transcriptome database of F. occidentalis identified a contig homologous to B. germanica E93 (accession number, CCM97102), and RT-PCR primers were designed to amplify part of the E93 cDNA. A 788-bp fragment was amplified by RT-PCR with FoE93-fwd1 and FoE93-rev1 primers. The nucleotide sequence of this fragment was identical

PLOS ONE
Characterization of E93 in neometabolous thrips to that obtained from the transcriptome. Blastx searches of this partial sequence identified a protein sequence of F. occidentalis E93 (XP_026276121.1, annotated as "mushroom body large-type Kenyon cell-specific protein 1"), which was not yet registered when we started cDNA cloning. Multiple alignment of E93 sequences revealed that F. occidentalis E93 (XP_026276121.1) was homologous to E93 of other insects, being the Pipsqueak DNA binding motif specially well conserved (S1 Fig). A 142-bp partial fragment for E93 was amplified by nested RT-PCR with HbE93-F2 and HbE93-R2 primers from the cDNA pool of H. brevitubus. Blastx searches showed that this was homologous to counterparts in other insects. Multiple alignment is shown in S2 Fig.

Developmental expression profile of E93 mRNA
The transcript levels of E93 were measured by quantitative RT-PCR. In F. occidentalis, FoE93 mRNA remained low until the end of the larval stage, increased rapidly in the "propupal" stage, and decreased in the adult stage (Fig 1A). In H. brevitubus, HbE93 mRNA remained low from the embryonic stage until the "propupal" stage, and increased in the "pupa I" and "pupa II" stages ( Fig 1B).
We showed previously that the expression profiles of Kr-h1 and br in post-embryonic development in thrips have some similarities to those reported in holometabolous insect species [31]. In thrips, the transcript level of Kr-h1 is maintained in the larval stage, whereas br is highly expressed from the beginning of the last larval instar until the "propupal" stage (diagram of the expression profile is shown in Fig 3B). Although JH titer has not been measured in thrips due to their small size, we estimate based on the expression profile of Kr-h1 that JH titer decreases in the "propupal" stage. In this study, we compared the expression profiles of br, Kr-h1, and E93 in the "propupal" and "pupal" stages. In both species, the transcript levels of Kr-h1 and br decreased at the end of the "propupal" stage as E93 increased (Fig 1C-1H). This observation was also consistent with the findings in holometabolous insects: the peak of Kr-h1 and br decreased at pupation, whereas E93 increased at the beginning of the pupal stage [26].

Effects of JHM treatment on the expression of br, Kr-h1, and E93 in adult development
We reported previously that treating "propupae" of F. occidentalis or H. brevitubus with exogenous JHM prevented pupa-adult metamorphosis; treatment of F. occidentalis "propupae" at 0-18 h after ecdysis with 10 mM pyriproxyfen resulted in lethality in "propupa" (6%), "pupa" (91%) and newly-ecdysed adults (3%), while treatment of "propupae" with 10 mM pyriproxyfen caused lethality in "propupa" (17%), "pupa I" (17%), "pupa II" (62%) and newly-ecdysed adults (4%) [31]. Moreover, JHM treatment caused prolonged expression of Kr-h1 and br [31]. Here, we examined the transcript level of br, Kr-h1, and E93 after JHM treatment of newlymolted "propupae". In F. occidentalis, Fobr was upregulated by JHM treatment compared with solvent-treated individuals (Fig 2A), whereas the upregulation of FoKr-h1 was not statistically significant ( Fig 2B). The level of the FoE93 transcript in control animals increased 48 h after solvent treatment compared with that in 0-24-h old "propurae", whereas that in JHM-treated "pupae" was slightly lower (Fig 2C). Similarly, we examined the transcript level of br, Kr-h1, and E93 in H. brevitubus. Hbbr and HbKr-h1 were upregulated by JHM treatment compared with solvent-treated individuals (Fig 2D and 2E). The transcript level of HbE93 in JHM-treated "pupae" was suppressed compared with that in solvent-treated "pupae" (Fig 2F). Thus, JHM treatment to the "propupa" resulted in upregulation of br and Kr-h1, and suppression of E93 in H. brevitubus. Although similar effects were observed in F. occidentalis, it seemed to be less effective compared with H. brevitubus. It is possible that the response to exogenous JHM is different between these two species. Alternatively, this might be due to difference in time points for qRT-PCR: transcript levels were examined at 48 h after treatment in F. occidentalis, while they were examined at 72 h after treatment in H. brevitubus.

Possible regulatory mechanism of E93 transcription in thrips
The regulatory mechanism of E93 transcription was first elucidated in B. germanica, T. castaneum, D. melanogaster, and B. mori [7,15,24,26,30]. E93 has also been characterized in other species including paraneopteran brown planthopper Nilaparvata lugens [33] and common bed bug Cimex lectularius [34]. In hemimetabolous B. germanica, endogenous JH, through Kr-h1, prevented 20E-induced upregulation of E93 in the nymphal stage. Once JH titer decreased at the onset of nymph-adult metamorphosis, increasing levels of E93 induced adult differentiation, and downregulation of Kr-h1 and br expression levels. A similar interaction was observed in holometabolous T. castaneum and D. melanogaster [24,26]. In T. castaneum, a transient peak in Kr-h1 in the prepupal stage blocked precocious upregulation of E93 in the larva-pupa transition [26]. After pupation, with low JH titer, 20E-inducible E93 directed adult differentiation and suppressed Kr-h1 and br [24]. A similar interaction among Kr-h1, br, and E93 has been confirmed in D. melanogaster [24,26]. In B. mori, JH suppressed 20E-inducible E93 expression, and this suppression was mediated by Kr-h1 through its binding to a Kr-h1 binding site located in the E93 promoter region [15]. In this study, we were not able to clarify the regulatory interaction among Kr-h1, br, and E93 in thrips, because RNAi-mediated  knockdown was not applicable. However, judging from the expression profiles of Kr-h1, br, and E93 (Figs 1 and 2) [31], we propose that their regulatory interaction may be conserved in thrips as well. A small peak of Kr-h1 exists in the "propupal" stage of thrips, which coincides with the peak of br (Fig 3B). We propose that JH in the "propupal" stage of thrips prevents precocious induction of E93 through Kr-h1 and br, thus preventing precocious adult differentiation. Normally, adult differentiation is induced by E93 with low levels of Kr-h1 and br in the absence of JH. If JHM was applied at the beginning of the "propupal" stage, the expression of Kr-h1 and br was prolonged and the expression of E93 was downregulated, which suppressed adult development (Fig 3C). Thus, although the regulatory interactions among Kr-h1, br, and E93 remain unclear, it is very likely that the MEKRE93 pathway [7] is conserved in thrips, too.
Thrips develop through an inactive non-feeding stage referred to as "propupa" and "pupa", after the larval stage. The significance of these inactive, pupa-like stages has been discussed. A part of larval tissues degenerates at the larva-pupa transition, whereas the adult structure is formed by newly-proliferating cells. Degeneration of the larval muscles was observed in the "propupal" and "pupal" stages of Frankliniella fusca and Haplothrips verbasci [35]. As stated above, the overall developmental expression profiles of Kr-h1 and br in thrips (Fig 3B) are similar to that of holometabolous species rather than that of hemimetabolous species [7,30]: the expression of br is maintained in the nymphal stage in Hemimetabola, whereas br is specifically expressed at larva-pupa transition of Holometabola and thrips. In this study, we showed that the transcript level of E93 increased as those of Kr-h1 and br decreased in the "propupal" stage of thrips (Fig 3B). This observation was also consistent with the findings in holometabolous insects: the peak of Kr-h1 and br decreased at pupation, whereas E93 increased at the beginning of the pupal stage. We propose that the "propupal" stage of thrips is similar to the holometabolous prepupal stage just before pupation, and that the "propupal" stage of thrips is essential for the formation of "pupa". Also, we propose that the "pupal" stage of the thrips has similarities with the pupal stage in holometabolous insects. In Hemimetabola, it has been suggested that the final nymphal instar is ontogenetically homologous to holometabolous pupal stage; this was first proposed by Hinton [36] and has been supported by others [7,13,37]. From the expression profiles of Kr-h1, br, and E93, we propose that the "propupal" and "pupal" stages in thrips is similar to the final nymphal instar in hemimetabolous species.
In conclusion, we revealed the developmental expression profiles of E93 in two thrips species. The results indicated that E93 induces adult differentiation after JH titer decreases at the end of the larval stage of thrips. We propose that the role of E93 in adult differentiation and hormonal regulation of its transcription are conserved among a variety of insect species including thrips. In addition, we propose that the "propupal" stage of thrips has similarities with the holometabolous prepupal stage, whereas the "pupal" stage of the thrips is similar to the pupal stage in holometabolous insects. Our findings will contribute significantly to solving the mystery of how metamorphosis in thrips evolved.