3. Results
3.1 Novel candidate receptors of Bt toxins in the DBM midgut identified by on-membrane capture
To investigate proteins at DBM midgut targeted by Bt toxins, we developed on-membrane capture method that is modified from far-western blotting (Wu et al. 2007, Figure 1 ). Briefly, three Bt proteins with different degree of toxicity to DBM (low activity Cry1Ac, middle activity Cry1Ab, and high activity Cry1Bd) were separated by an SDS-PAGE gel. In parallel, same amount of Bt proteins was separated by another SDS-PAGE gel. Then, proteins were transferred to two nitrocellulose membranes, respectively. Proteins on one NC membrane were denatured and renatured by gradually reducing the guanidine-HCl concentration (Wu et al. 2007). The membrane was blocked with protein-free buffer, then was incubated with total proteins isolated from DBM midgut epithelium. Bt protein on the membrane captures BBMV proteins if they form a complex. The Bt proteins on the NC membrane were positioned by aligning to Bt proteins on the other NC membrane that visualized with Ponceau S, then were cut for trypsin digestion. Digested peptides were precipitated in acetone that dissolves the NC membrane. Peptides were dried and were analyzed by nano liquid chromatography-tandem mass spectroscopy (nano LC-MS/MS) coupled with the mass spectrometer (Figure 1 ).
MS/MS spectra were queried against a combined protein database including DBM protein sequences and amino acid sequences of Cry1Ac, Cry1Ab, and Cry1Bd. A total of 520 peptides were detected in three bands. In each band, the peptides of corresponding Bt toxin were most abundant (Figure 2A). 38.2% of peptides were identified as Cry1Ac in the band of Cry1Ac, 26.5% as Cry1Ab in the band of Cry1Ab, and 27.6% as Cry1Bd in Cry1Bd band (Figure 2A ). The rest of the peptides corresponded to 81 unique proteins. Of these, 35 proteins were identified in the Cry1Ac band, 40 were in the Cry1Ab band, and 39 were in the Cry1Bd band (Figure 2A ). Molecular weights of these proteins were dominantly at 50 kd and 200 kd, isoelectric points of proteins were approximately 5.5 (Figure 2B ). These suggested that Bt toxins interacted with specific proteins at DBM midgut.
17 proteins were identified in all three bands (Figure 2C ), indicating these proteins are unlikely associated with toxicity of Bt to DBM. Expectedly, two known receptors cadherin and APN2 previously excluded from DBM resistance to Bt toxins were found to interact with all three Bt toxins (Nakanishi et al. 2002, Chang et al. 2012, Guo et al. 2015). Some DBM proteins were identified uniquely from one of the bands, and these proteins are more likely to be associated with different degree of Bt toxicity (Figure 2C, Supplemental Table 2 ).
3.2 DBM glucosinolate sulfatases (GSSs) are receptors of Bt toxin Cry1Bd
We paid closer attention to the proteins uniquely interacted with Cry1Bd. The field population of DBM still remains highly susceptible to Cry1Bd but has evolved cross-resistance to other four Bt toxins including Cry1Ac and Cry1Ab (Ferre et al. 1991, Tabashnik et al. 1993, Tabashnik et al. 1994, Tabashnik et al. 1996, Kuo et al 2000, Liu et al. 2001, Zhao et al. 2007). Therefore, if Cry1Bd is to be used to control this pest, it is important to investigate potential Cry1Bd receptors.
13 DBM proteins were candidate receptors of Cry1Bd. We analyzed these candidates to identify the transmembrane helices and the GPI-anchor sequences which are known characteristics of Bt receptors, like the primary receptor cadherin and the secondary receptors ALP and APN. It resulted in identification of four proteins with GPI-anchor sites ATP synthase F0 subunit 8, β-1,3-glycosyltransferase 5, and two glucosinolate sulfatases. ATP synthase F0 subunit 8 is a mitochondrial protein that has been eliminated as a Bt toxin target, likewise, β-1,3-glycosyltransferase 5 has been ruled out as contributing to Bt resistance in Plutella (Baxter et al. 2008). Two glucosinolate sulfatases (GSS), GSS1 and GSS2, matched the peptide244RIFAAMVK252 (Supplemental Figure 1 ), have predicted GPI-anchor sites at the position C485 (Supplemental Figure 2 ). In addition, GSS2 contains an N-terminal transmembrane helix. Likely, GSS2 is a membrane anchor protein with a C-terminal extending outside of the cell membrane (Supplemental Figure 3 ). These characteristics suggested GSSs are candidate receptors of Cry1Bd in the DBM.
To further investigate GSSs are the receptors of Cry1Bd, we tested whether GSSs directly interact with Cry1Bd. We carried out pull-down experiments and found that both GSS1 and GSS2 pulled down Cry1Bd, suggesting direct interactions between Cry1Bd and GSSs (Figure 3A ). In contrast, either Cry1Ac or Cry1Ab were pulled down by GSSs, is consistent with the results that GSSs are absence in proteins captured by Cry1Ac or Cry1Ab (Figure 3A , Supplemental Table 1 ). The interactions of GSSs with Bt proteins were further validated by far-western blot. Similar to on-membrane capture, His-tagged GSS1 or His-tagged GSS2 were incubated with NC membrane containing Bt proteins. Consistently, both GSSs interact with Cry1Bd not Cry1Ac or Cry1Ab (Figure 3B ). Altogether these results suggested GSSs bind with Cry1Bd directly and specifically, serving as receptors in the DBM to Cry1Bd toxin.
3.3 Reduction in GSSs expression increases tolerance of DBM Bt toxins
To investigate roles of GSSs in the toxicity of Bt to DBM, we generated transgenic lines expressing dsRNA corresponding to the GSSssequence for plant-mediated RNAi of GSSs in DBM. The transgenicA. thaliana lines successfully knocked down GSSsexpression compared to lines producing dsRNA corresponding to GFP (dsGFP ) in DBM (Figure 4A and 4B ). As results of reduction in expression of GSSs gene, GSSs -silenced larvae became more tolerant to Cry1Bd compared with controls. The LC50 of Cry1Bd to larvae were increased 69-fold from 0.05 mg/L of control larvae to 3.16 mg/L of GSSs -silenced larvae (Figure 4C ). These results demonstrated that GSSs are susceptible factors that are critical for Cry1Bd toxicity to DBM. Interestingly, GSSs -silenced larvae also became more tolerant to Cry1Ac and Cry1Ab. Therefore, knock down of the expression of GSSs is correlated with increased tolerance of DBM to Bt toxins.
3.4 GSSs are susceptible factors to Cry1Bd
To confirm that DBM GSSs are causative agents of susceptibility to Cry1Bd, we introduced DBM GSS1 and GSS2 into a Cry1Bd-resistant silkworm strain Nistari (Chen et al. 2014). Two transgenic silkworm lines expressing dsRed marker and one of the DBMGSSs were established (Figure 5A ). Inverse PCR of genomic DNA confirmed the presence of transgenes, that GSS1 lines had one copy of the transgene on chromosome 23 and GSS2 lines had one copy of GSS2 on chromosome 11 (Supplemental Figure 4 ). As GSSs are more likely to be susceptible factors to Cry1Bd, a single allele of the gene would convert the resistance of transgenic Nistari. Therefore, the individuals with dsRed fluorescent were selected from the offspring of transgenic and wildtype Nistari, then were tested for tolerance to Cry1Bd. Remarkably, the silkworm larvae with one of GSSs became more sensitive to Cry1Bd compared to controls, as LC50 of Cry1Bd dropped to 1.86 mg/L of GSS1 larvae and 1.30 mg/L ofGSS2 larvae from 33.90 mg/L of wildtype ones (Figure 5B ). These results demonstrated DBM GSSs are susceptible factors of Cry1Bd toxin.