RNAi-based knockdown of candidate gut receptor genes altered the susceptibility of Spodoptera frugiperda and S. litura larvae to a chimeric toxin Cry1AcF

Insect rearing

S. frugiperda and S. litura egg clusters were obtained from infested plots (IARI, New Delhi) of maize and okra followed by initiation of nucleus colony culture in our laboratory. Rearing of neonates was performed in sterilized castor leaves in a growth chamber; sixth-instar larvae were pupated in sterilized sand followed by disinfection of pupae via rinsing in 0.02% formaldehyde. Adults were fed with artificial diets comprising honey, multivitamin supplements, ascorbic acid and methyl-p-hydroxybenzoate (Shankhu et al., 2020; Dutta et al., 2021). A continuous culture was maintained on castor leaves for recurrent insect life cycle progression. Surface-sterilized (with 70% ethanol) fourth-instar larvae (0.4–0.5 g body mass) were used for subsequent experiments.

Cry1AcF toxicity assessment in test insects

Larvae were randomly assigned to each treatment. Larvae were starved for 6 h followed by oral ingestion with 10 µL of phosphate-buffered saline (PBS, pH 7.4) containing Cry1AcF toxin in different doses using a sterilized 26-gauge hypodermic needle (Hamilton syringe; Sigma-Aldrich St. Louis, MO, USA). Negative control consisted of PBS only. A known toxin TcaB (Mathur et al., 2019; Dutta et al., 2021) was used as the positive control. Larvae were incubated in sterile six-well polystyrene plates containing artificial diet at 28 °C, 60% relative humidity. Insect mortality data was recorded at 24 h after oral ingestion. The whole experiment was repeated at least thrice (n = 50 per treatment). The details of Cry1AcF protein production and purification are described in our previous studies (Rathinam et al., 2017, 2019).

Histopathology study

At 24 h after oral ingestion, toxin-treated and control larvae were inoculated with 100 µL of 10% formalin via intra-hemocoel injection followed by dipping the larvae in formalin (Shankhu et al., 2020). After 24 h, midgut tissue was dissected out using a sterilized blade under the microscope, and suspended in ice-cold fixative (4% paraformaldehyde + 2.5% glutaraldehyde in 0.1 M PBS, pH 7.2) followed by transfer to fresh fixative and overnight storing at 4 °C. Next, samples dehydrated through a graded ethanol series. Dehydrated samples were immersed in xylene for clearing of tissues. Embedded (in paraffin wax) samples were processed into 6 µm cross-sections using Leica RM2165 microtome (Dutta et al., 2021; Santhoshkumar et al., 2021). Sections fixed in glass slides via egg albumin solution and dried overnight. Slides washed using graded ethanol series and sections dewaxed via xylene treatment. Sections were periodically stained with Cole’s haematoxylin and eosin (Sigma-Aldrich, St. Louis, MO, USA). Samples were mounted in Dibutylphthalate Polysterene Xylene (DPX; Sigma-Aldrich St. Louis, MO, USA). More detail about the procedure can be found in Dutta et al. (2021) and Santhoshkumar et al. (2021). Photomicrographs were obtained in an Zeiss Axiocam MRm microscope.

Cry1AcF cytotoxicity analysis in test insects

Total hemocyte counts (THC) were obtained from toxin-treated and control larvae at 24 h after oral ingestion. Sample of fresh hemolymph (10 µL) was extracted (by piercing through the integument above head using a sterilized needle) from each larvae and diluted three-fold with ice-cold anticoagulant containing phenylthiourea (4 mg mL⁻¹ in PBS) to suppress the melanization (Mathur et al., 2019). To examine hemocyte viability, 0.4% Trypan blue (w/v in PBS) was mixed to the hemolymph, which was not treated with the anticoagulant (Mathur et al., 2019). On both occasions, hemolymph sample was applied into a hemocytometer (Neubauer) for cell enumeration in Zeiss Axiocam MRm microscope. Experiment was repeated thrice (n = 10 per treatment).

Cry1AcF immune-stimulatory activity measurement in test insects

Phenoloxidase (PO) enzyme activity was measured in the hemolymph of toxin-treated and control larvae at 12 and 24 h after inoculation. A total of 10 µL hemolymph was extracted from each larvae as described above and 100 µL PBS was added to it by centrifugation at 1,006 g for 2 min at 4 °C. Cell-free supernatant was used for downstream analyses. A total of 100 µL of hemolymph sample (cell-free supernatant) was added to 100 µL of 20 mM L-3,4-dihydroxyphenylalanine or L-DOPA (4 mg mL⁻¹ in PBS) in a 96-well ELISA plate and incubated at 28 °C in dark for 30 min (Mathur et al., 2019; Dutta et al., 2021). Absorbance of the sample was examined at 490 nm in a microplate reader (BioTek, Winooski, VT, USA). PO activity was expressed as the increase in absorbance at 490 nm per minute per mg of protein. As control, PBS and L-DOPA were mixed equally and its absorbance value was subtracted from each sample. Experiment was repeated thrice (n = 10 per treatment).

RNA extraction from S. frugiperda and S. litura

To analyze the transcription dynamics of different Cry receptor-encoding genes in different developmental stages and tissues of S. frugiperda and S. litura, RNA was extracted from different samples using TRIzol reagent (Invitrogen, Waltham, MA, USA) by following the manufacturer’s protocols. In order to minimize genomic DNA contamination, extracted RNA was digested with DNase I enzyme (TakaRa, Kusatsu, Japan). Purity of the RNA was assessed in a Nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), and RNA integrity was evaluated by electrophoresing samples on 1% (w/v) agarose gel. RNA (~1 µg) was reverse transcribed to cDNA by using a first-strand cDNA synthesis kit (Superscript VILO; Invitrogen, Waltham, MA, USA).

RT-qPCR analysis

RT-qPCR-based transcription profile analysis of Cry receptor-encoding genes was performed in a CFX96 thermal cycler (BioRad, Hercules, CA, USA). qPCR reaction volume (10 μL) consisted of 1.5 ng cDNA, forward and reverse primers (750 nM each), and 5 μL SYBR Green PCR master-mix (BioRad, Hercules, CA, USA). qPCR cycling conditions included a hot start phase of 95 °C for 30 s, followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s. Additionally, a melt curve program (95 °C for 15 s, 60 °C for 15 s, followed by a slow ramp from 60 °C to 95 °C) was included to ascertain the amplification specificity. Quantification cycle (Cq) values were obtained from CFX Maestro software (BioRad, Hercules, CA, USA). Housekeeping genes of S. frugiperda (ribosomal protein S3 or rps3) and S. litura (β-actin and GAPDH) were included as the internal reference for normalizing the expression value of target genes. qPCR fold change data was calculated using 2^−ΔΔCq method. qPCR was run with five biological and three technical replicates for each samples. Primer designing was carried out in OligoAnalyzer tool (https://eu.idtdna.com/). PCR reaction efficiency of different primers was determined by constructing a standard curve (where Cq values were plotted against cDNA concentrations) across the five-fold dilutions of cDNA samples followed by calculating the slope using a linear regression equation: E = (10^(−1/slope) – 1) × 100. Primer details and PCR efficiency data are given in Tables S1 and S2.

DsRNA synthesis

Targeted region for dsRNA synthesis from the coding sequences of different receptors was designed using different in silico tools such as siDirect (http://sidirect2.rnai.jp/; predicts siRNA biogenesis probability across the query sequence), Dharmacon (http://horizondiscovery.com/) and dsCheck (http://dscheck.rnai.jp/; helps in avoiding off-target sites by comparing sequence homology of predicted siRNAs with that from non-target organism). DsRNA sequences were PCR-amplified (according to the standard protocol) from the cDNAs of S. frugiperda and S. litura (extracted from the midgut tissues of fourth-instar larvae) using gene-specific primers that harbored SacI and HindIII restriction endonuclease sites (Tables S1 and S2). PCR products were cloned into SacI and HindIII-digested L4440 plasmid (Addgene) that contained two T7 polymerase promoter sequences (aids in driving DNA to RNA transcription) in inverted orientation flanking the multiple cloning site. DsRNA molecules were bacterially expressed in Escherichia coli HT115 (DE3) cells, which were transformed with recombinant L4440 plasmid. HT115 cells were cultured in Luria Bertani medium (Sigma, St. Louis, MO, USA) containing ampicillin (50 µg mL⁻¹) and tetracycline (12.5 µg mL⁻¹) at 37 °C for 12 h with constant agitation at 200 rpm. In order to induce T7 polymerase synthesis, 0.4 mM IPTG was mixed into the culture and incubated for an additional 4 h at 37 °C. Transcribed dsRNAs were isolated from bacterial aliquots (dsRNAs purified from E. coli cells by phenol/chloroform/isoamyl alcohol extraction via shaking at 65 °C for 30 min followed by centrifugation at 12,000 × g for 15 min, upper phase containing dsRNA transferred to a new tube, precipitated with isopropanol, washed with ethanol, followed by re-suspending the nucleic acid pellet in 0.05 M PBS) and electrophoresed on a 1% (w/v) agarose gel (Ahn et al., 2019). DsRNA of a gfp gene (Genbank ID: HF675000) ligated into L4440 plasmid was used as the non-native control (in order to check whether dsRNA itself has any negative effect on insects).

RNAi bioassay

Bacterially-expressed dsRNAs (10 µL suspension amounting to ~10 µg of dsRNA) were orally administered into the 6 h starved fourth-instar larvae of S. frugiperda or S. litura using a sterilized needle as described above. GFP dsRNA and 0.05 M PBS served as the non-native and negative control, respectively. Inoculated larvae were incubated in sterile 6-well plates as described above. Post 24 h of dsRNA treatment, larvae were orally ingested with LD₅₀ or LD₉₀ doses of Cry1AcF toxin and incubated in six-well plates as depicted above. After 24 h insect mortality data was recorded. The whole experiment was repeated at least thrice (n = 50 per treatment).

RNAi-induced downregulation (fold change in expression of a target gene in dsRNA-treated larvae was subtracted from that in control larvae) of target receptor-encoding genes was validated by RT-qPCR assay. RNA was isolated from the midgut tissues of ten random larvae belonging to each dsRNA treatment groups. RNA was reverse-transcribed to cDNA as depicted previously. qPCR run conditions were followed as explained above.

Statistical analysis

LD₅₀ and LD₉₀ doses of Cry1AcF was determined by using probit analysis in SPSS v. 21 software (IBM Corp., Armonk, NY, USA). Data of different bioassay (THC and PO assay data were checked for normality using Shapiro–Wilk test) and qPCR experiments was subjected to one-way ANOVA followed by Tukey’s honest significant difference (HSD) test (for multiple comparison among different treatments) in SAS v. 14.1 software (SAS, Inc. Cary, NC, USA).

Biological activity of Cry1AcF against S. frugiperda and S. litura

Oral ingestion of Cry1AcF (Fig. 1A) caused toxicity in the fourth-instar larvae of S. frugiperda and S. litura. Phenotypic changes such as dead-like posture with extended prolegs and hemolymph melanization was evident in Cry1AcF-treated larvae compared to normal phenotypes in PBS-ingested control larvae at 24 h after inoculation (Fig. 1B). Cry1AcF conferred a gradual and significant (P < 0.01) decrease in mean percent survival of S. frugiperda (F (4, 45) = 56.82, P = 0.0021) and S. litura (F (4, 45) = 47.38, P = 0.0032) with increasing Cry1AcF doses, i.e., 25–200 ng/larva (Figs. 1C, 1D). While compared with a gut-active reference toxin TcaB (characterized from an insect parasitic bacterium Photorhabdus akhurstii in our laboratory; Mathur et al., 2019; Santhoshkumar et al., 2021), Cry1AcF caused significantly (P < 0.01) greater larval mortality across the dose range in both the test insects (Figs. 1C, 1D). Consequently, Cry1AcF exhibited lower LD₅₀ and LD₉₀ values than TcaB in both the test insects (Table 1). The LD₅₀ and LD₉₀ values for Cry1AcF in S. frugiperda was derived as 29.78 and 99.77 ng/larva, respectively. The LD₅₀ and LD₉₀ values for Cry1AcF in S. litura was calculated as 38.76 and 117.32 ng/larva, respectively (Table 1). When larval body mass was included in calculation, Cry1AcF caused greater toxicity in S. frugiperda in terms of lower LD₅₀ (53.60 ng/g) and LD₉₀ (179.59 ng/g) values compared to S. litura (LD₅₀: 69.77 ng/g; LD₉₀: 211.18 ng/g).

Cry1AcF causes gut leakiness in S. frugiperda and S. litura

Presumably, orally delivered Cry1AcF catalytically disrupted the midgut architecture of test insects leading to leaky gut and consequent hemolymph melanization. In order to validate this hypothesis, comparative histopathology experiment of PBS- and Cry1AcF-ingested (in LD₅₀ doses) S. frugiperda and S. litura midgut transverse sections was carried out at 24 h after inoculation. Midgut tissues of PBS-ingested larvae exhibited normal morphology in terms of regular arrangement of epithelial cells (with intact nucleus) which were resting on the basement membrane. Gut tissue damage was negligible because the protective barrier provided by visceral muscle layer (interface between gut epithelium and hemocoel or body cavity) was intact in both the test insects (Figs. 2A, 2C). By contrast, in Cry1AcF-ingested larvae, epithelial cell lining was displaced, cells were disintegrated and sloughed off into the gut lumen. Additionally, a discontinuous lining of visceral muscle layer was evident suggesting the structural damage to gut-body cavity barrier in the test insects (Figs. 2B, 2D).

Cry1AcF reduces hemocyte numbers and viability in S. frugiperda and S. litura

Interestingly, the cytotoxic effect of Cry1AcF was observed in both the test insects at 24 h after oral ingestion. Total circulatory hemocyte counts (THC) of S. frugiperda was significantly (F (2, 27) = 24.44, P = 0.0072) reduced from 5.99 × 10⁶ mL⁻¹ in PBS to 3.48 × 10⁶ mL⁻¹ and 2.27 × 10⁶ mL⁻¹ in Cry1AcF-treated insects at LD₅₀ (30 ng/larva) and LD₉₀ (100 ng/larva) doses, respectively. Similarly, THC of S. litura was significantly (F (2, 27) = 29.37, P = 0.0058) declined from 9.88 × 10⁶ mL⁻¹ in PBS to 6.89 × 10⁶ mL^-1 and 4.36 × 10⁶ mL⁻¹ in Cry1AcF-treated insects at LD₅₀ (40 ng/larva) and LD₉₀ (120 ng/larva) doses, respectively (Fig. 3A).

In addition, viability of hemocyte cells was examined by Trypan blue staining assay. In this assay, viable cells avert the stain penetration while dead cells dye dark blue because of stain penetration via disintegrated cell membrane. Approximately 100 cells were sampled from each treatment (n = 10), and live cells were counted and expressed as percent viability. Cell viability of S. frugiperda was significantly (F (2, 27) = 16.77, P = 0.0091) reduced from 100% in PBS to 65% and 43% in Cry1AcF-treated insects at LD₅₀ and LD₉₀ doses, respectively. Likewise, cell viability of S. litura was significantly (F (2, 27) = 14.88, P = 0.0062) decreased from 100% in PBS to 69% and 49% in Cry1AcF-treated insects at LD₅₀ and LD₉₀ doses, respectively (Fig. 3B). Comparative photomicrographs (generated via Trypan blue exclusion assay) represent the live and dead hemocyte cells of S. frugiperda treated with PBS and Cry1AcF, respectively (Figs. 3C, 3D).

Cry1AcF induces immune-stimulatory activity in S. frugiperda and S. litura

The disruption of hemocyte or immunocyte cell morphology due to Cry1AcF-induced toxicity is indicative of Cry1AcF immune-stimulatory activity in test insects. Additionally, larval melanization due to Cry1AcF catalytic activity exemplifies the elevated PO enzyme activity in the hemolymph resulting from the conversion of hemocyte-bound proPO to PO. In view of this, hemolymph PO activity was measured in the test insects at 12 (in order to detect the early responses in host insect) and 24 h after Cry1AcF ingestion. Compared to PBS (0.05 OD₄₉₀/min/mg protein), a significantly (F (2, 27) = 19.24, P = 0.004) higher amount of PO induction (0.42–0.68 OD₄₉₀/min/mg protein) was documented in S. frugiperda larvae ingested with LD₅₀ and LD₉₀ doses of the toxin, at 12 h after inoculation (Fig. 4A). At 24 h, PO activity further increased (F (2, 27) = 17.56, P = 0.003) to 0.62–0.92 OD₄₉₀/min/mg protein in S. frugiperda (Fig. 4B). An identical trend was observed in S. litura larvae (Fig. 4).

Cry1AcF intoxication altered the expression of gut receptors in S. frugiperda and S. litura

Initially, stage-specific and tissue-specific expression profiles of different Cry receptor-encoding genes was analyzed in S. frugiperda and S. litura via RT-qPCR. Transcripts of CAD, ABCC2, ALP1 and APN receptor were most abundantly expressed (P < 0.01) in the fourth- and fifth-instar stage compared to other life stages of S. frugiperda (Fig. S1). All the receptor gene mRNAs were greatly upregulated (P < 0.01) in the midgut tissues than other body parts including foregut, hindgut, Malpighian tubules, head and fat body of S. frugiperda fourth-instar stage (Fig. S2). Likewise, CAD, ABCC2, ALP1 and APN transcripts were abundantly expressed (P < 0.01) in both fourth- and fifth-instar stages of S. litura (Fig. S3). All these receptor mRNAs were highly upregulated in the midgut tissues compared to other body parts of S. litura (Fig. S4).

Intriguingly, CAD, ABCC2, ALP1 and APN receptor-encoding genes were significantly (P < 0.01) overexpressed in the midgut tissues in S. frugiperda fourth-instar larvae at 12 h after oral administration of Cry1AcF toxin in LD₅₀ dose, compared to the expression levels of identical genes in PBS ingested larvae. At 24 h after toxin ingestion, expression level of these genes were further upregulated by several folds (P < 0.001) compared to their expression level during 12 h time period (Fig. 5A). A similar trend was observed with S. litura fourth-instar larvae. Most importantly, CAD transcripts were greatly expressed (P < 0.001) in S. litura at both 12 and 24 h after Cry1AcF ingestion (Fig. 5B). Herein, we assume that one or few of these receptor proteins may play crucial role during Cry1AcF intoxication process in S. frugiperda and S. litura. However, the possibility that the upregulation of these receptor genes was an indirect response to the action of the Cry toxin in the midgut tissue cannot be dismissed because of our inability of using a non-receptor midgut protein as a control.

RNAi of gut receptor-encoding genes altered S. frugiperda and S. litura susceptibility to Cry1AcF

Target dsRNA molecules (ideally 400–500 bp long sequences are readily processed by RNaseIII or Dicer enzyme) from the coding sequences of S. frugiperda (CAD: 411 bp, ABCC2: 406 bp, ALP1: 440 bp, APN: 410 bp) and S. litura (CAD: 427 bp, ABCC2: 447 bp, ALP1: 419 bp, APN: 442 bp) were designed according to the prediction of greater siRNA generation probabilities in the targeted region compared to untargeted stretches in the same sequence. Next, possible RNAi off-target effect was analyzed by using our target dsRNA sequences as query against Drosophila melanogaster siRNA database in DsCheck server (http://dscheck.rnai.jp/). Results indicated minimal homology of D. melanogaster siRNAs with that of siRNAs generated from dsRNAs of CAD, ABCC2, ALP1 and APN (data not shown). DsRNAs (cloned in L4440 vector) were expressed via E. coli HT115 (RNase III deficient) cells since ingestion of this recombinant bacteria can produce intact dsRNA molecules in the insect gut.

CAD, ABCC2, ALP1 and APN expression in the midgut tissues of S. frugiperda were considerably (P < 0.01) reduced by 75%, 61%, 79% and 65% in CAD, ABCC2, ALP1 and APN dsRNA-ingested larvae, respectively, compared to their expression in PBS-ingested larvae at 24 h after dsRNA treatment (Fig. S5; fold change in expression of a target gene in dsRNA-treated larvae was subtracted from that in control larvae and converted into percentage values). Similarly, CAD, ABCC2, ALP1 and APN expression in S. litura gut were markedly (P < 0.01) downregulated by 76%, 58%, 68% and 72% in CAD, ABCC2, ALP1 and APN dsRNA-ingested larvae, respectively, compared to control (Fig. S6). Ingestion of GFP dsRNA did not change (P > 0.01) the expression of receptor genes mRNAs in either of the insect species. In addition, transcript abundance of CAD was unchanged (P > 0.01) in ABCC2, ALP1 and APN dsRNA-treated insects. Vice versa was observed with other receptor mRNAs as well in either of the insect species (Figs. S5 and S6). These results has indicated the target-specific induced silencing of receptor-encoding genes in the midgut tissues. However, S. frugiperda CAD, ABCC2, ALP1 and APN shared 98%, 99%, 99% and 80% identity in the dsRNA-binding regions with homologous transcripts XM_035584870, OL955484, XM_035587634 and XM_022979826, respectively (Fig. 6A). S. litura CAD, ALP1 and APN shared 99%, 100% and 99% identity in the dsRNA-binding regions with homologous transcripts XM_022970523, XM_022967284 and XM_022969246, respectively (Fig. 6B). This suggests that RNAi of receptor-encoding genes might have silenced their allelic variants as well. Nevertheless, this could not be proved experimentally because of the limitations associated with designing specific primers, which would distinguish the two highly homologous gene transcripts.

RNAi-induced downregulation of CAD, ABCC2, ALP1 and APN transcripts in the midgut of S. frugiperda or S. litura did not confer any off-target effect on larval development and metamorphosis process as evidenced by the percent pupation and adult emergence data obtained at 7–10 days after dsRNA ingestion (Fig. S7).

Post 24 h of dsRNA ingestion, larvae were force-fed with Cry1AcF toxin in LD₅₀ and LD₉₀ doses followed by incubated in artificial diet. After another 24 h, larval mortality data was recorded. Interestingly, in CAD and ABCC2 silenced S. frugiperda, Cry1AcF (in both sub-lethal and lethal doses)-induced toxicity did not differ significantly (P > 0.01) with that of PBS- or GFP dsRNA-treated larvae (Fig. 7A). On the contrary, compared to control, Cry1AcF (in both LD₅₀ and LD₉₀ doses)-induced mortality in ALP1 and APN dsRNA pretreated S. frugiperda were significantly (P < 0.01) reduced by 42.85–43.39% and 46.15–49.06%, respectively (Fig. 7A). However, in ALP1 and APN silenced S. litura, Cry1AcF (in LD₅₀ and LD₉₀ doses)-induced toxicity did not differ significantly (P > 0.01) with that of PBS- or GFP dsRNA-treated larvae (Fig. 7B). Compared to control, a marginal yet significant (P < 0.01) reduction (28.57% and 21.05% in LD₅₀ and LD₉₀ doses, respectively) in S. litura mortality was documented when larvae were pretreated with ABCC2 dsRNAs (Fig. 7B). When CAD transcripts were knocked down in S. litura, a greatest (P < 0.01) reduction (55.36% and 52.63% in LD₅₀ and LD₉₀ doses, respectively) in larval mortality was documented in comparison to control larvae (Fig. 7B). Taken together, our data putatively indicates ALP1/APN and CAD as potential receptors for Cry1AcF intoxication in S. frugiperda and S. litura, respectively, because induced suppression of these receptor-encoding genes disrupted the larval susceptibility to Cry1AcF at both LD₅₀ and LD₉₀ doses.