A Saliva Protein of Varroa Mites Contributes to the Toxicity toward Apis cerana and the DWV Elevation in A. mellifera

Varroa destructor mites express strong avoidance of the Apis cerana worker brood in the field. The molecular mechanism for this phenomenon remains unknown. We identified a Varroa toxic protein (VTP), which exhibited toxic activity toward A. cerana worker larvae, in the saliva of these mites, and expressed VTP in an Escherichia coli system. We further demonstrated that recombinant VTP killed A. cerana worker larvae and pupae in the absence of deformed-wing virus (DWV) but was not toxic to A. cerana worker adults and drones. The recombinant VTP was safe for A. mellifera individuals, but resulted in elevated DWV titers and the subsequent development of deformed-wing adults. RNAi-mediated suppression of vtp gene expression in the mites partially protected A. cerana larvae. We propose a modified mechanism for Varroa mite avoidance of worker brood, due to mutual destruction stress, including the worker larvae blocking Varroa mite reproduction and Varroa mites killing worker larvae by the saliva toxin. The discovery of VTP should provide a better understanding of Varroa pathogenesis, facilitate host-parasite mechanism research and allow the development of effective methods to control these harmful mites.


Effects of purified E-VTP on honeybees.
We found that the mortality rates of A. cerana worker larvae and pupae challenged with purified E-VTP were significantly higher than those of the controls and increased with the concentrations of injected E-VTP at 2 days (  Table S1). The cadavers also exhibited the same traits as those resulting from infesting A. cerana worker L5 larvae with mites or injecting these insects with crude mite saliva (Fig. 1e). The survival rates of A. cerana worker L5 larvae injected with E-VTP were significantly lower than those of the controls (Fig. 1d, df = 7, 16, F = 96.533, P = 0.000). No deformed wings were found in A. cerana worker adults.
When A. mellifera worker larvae and pupae were injected with purified E-VTP at different concentrations, we found that their mortality rates were similar to those of the controls at 2 days ( Fig. 1b- (5), L3 larvae: df = 6, 14, F = 0.238, P = 0.956; (6), L5 larvae: df = 7, 16, F = 0.295, P = 0.946; (7): pupae: df = 6, 14, F = 1.653, P = 0.205); the survival rates of the worker larvae and pupae were not significantly different among the treatments and the controls. However, 40-60% of the resulting adults from the injected worker larvae had deformed wings (Fig. 1d). The DWV loads of A. mellifera wing-deformed adults were significantly higher than those of normal-winged adults ( Supplementary Fig. S4), indicating that the E-VTP injection of A. mellifera worker L5 larvae significantly increased the DWV titers in the resulting adults. Similar wing traits were found in the adults after A. mellifera worker L5 larvae were challenged with E-VTP or with mites or injected with crude mite saliva (Fig. 1e). When the control materials were injected into A. mellifera worker L5 larvae, none of the resulting adults had increasing DWV loads or deformed wings. These results indicated that introducing VTP into A. mellifera worker larvae stimulated DWV proliferation and that high DWV titers caused deformed wing adults.

Discussion
The co-adaptation of V. destructor and A. cerana results from an interesting dynamic balance between the host and parasite populations and accounts for the resistance of A. cerana to these mites. Effective grooming and hygienic behavior, non-production of the mites in the worker brood, and the "entombing" of the infested drone brood are reportedly the major factors ensuring this resistance [10][11][12][13]15,16 . Grooming and hygienic behavior reduces the susceptibility of A. cerana colony to V. destructor infestation 16-19 but seems to be highly variable 20   susceptibility of mite-infested worker brood of A. cerana leads to more efficient hygienic behavior 21 . However, the removal of the Varroa-infested worker brood does not necessarily induce the death of the mites, as most mites may escape from brood cells that were opened during removal 10 . Interestingly, V. destructor mites or the injection of mite saliva can kill A. cerana worker larvae 21,22 . A. cerana larvae block Varroa mite reproduction in the infected worker brood 9 and are sacrificed due to mite saliva toxin (similar to a mutual destruction strategy). Lack of production of mites in the worker brood was not caused by the death of A. cerana larvae, as these mites, fed with fresh A. cerana or A. mellifera larvae, were also infertile (unpublished results). The reason why these mites produce a toxic protein to kill A. cerana larvae remains unknown. However, in the field, V. destructor mites exhibit a marked preference for the drone brood and are rarely found in the worker brood of A. cerana colonies as long Figure 2. RNAi-mediated knock down of vtp expression in adult V. destructor and its effects on toxicity toward honeybees. Eight live mites were collected from each treatment group, for 2 repeats. The mites were treated by overnight (approximately 15 hours) immersion in different solutions at 16 °C and subsequently placed on A. mellifera and A. cerana worker L5 larvae. (a) CK, V. destructor mites that were not treated; 0.9% NaCl, mites treated with 0.9% NaCl; dsRNA-gfp, mites treated with dsRNA-gfp (2.5 μg/μL); dsRNA-vtp, mites treated with dsRNA-vtp (2.5 μg/μL). 0 d, mites collected immediately after immersion; 3 d, treated mites placed on L5 larvae for 3 days; 7 d, treated mites placed on L5 larvae for 7 days; 13 d, treated mites placed on L5 larvae for 13 days, when eclosion was observed. One-way ANOVA with post hoc multiple comparison tests, NS, not significant; ***P < 0.001, the values were expressed as the mean values ± S.D. (b) Mortality rate of A. cerana worker L5 larvae infested by Varroa mites; (c) Percentage of adult eclosion and percentage of eclosed adults with deformed wings resulting from A. mellifera worker L5 larvae infested by Varroa mites. CK, worker L5 larvae without mite infestation; Vd, worker L5 larvae exposed to Varroa mites that were not treated; 0.9% NaCl, worker L5 larvae exposed to Varroa mites treated with 0.9% NaCl; dsRNA-gfp, worker L5 larvae exposed to Varroa mites treated with dsRNA-gfp (2.5 μg/μL); dsRNA-vtp, worker L5 larvae exposed to Varroa mites treated with dsRNA-vtp (2.5 μg/μL). One-way ANOVA with post hoc multiple comparison tests, *P < 0.05, ***P < 0.001, n = 6 L5 larvae in each of three replicates per treatment. The data are presented as the mean values ± S.D. The experiment was repeated more than two times.
SCientifiC REPORTS | (2018) 8:3387 | DOI:10.1038/s41598-018-21736-9 as a sufficient drone brood is available 13 . This phenomenon is considered crucial for the balanced host-parasite relationship in A. cerana during the long co-evolution of these species 13 . Based on the results of the present study, we suggest a mechanism to explain why Varroa mites express the strong avoidance of the worker brood. Under mutual destruction stress, Varroa mites likely evolved to protect themselves in the field by avoiding the worker brood, instead by preferring drone brood where they do not kill drones by the saliva toxin and subsequently reproduce. This strategy significantly reduces mite death and non-reproduction risks. The mite infestation of drone brood typically does not induce a significantly negative effect on colony size 23 , therefore almost the entire A. cerana colony can be protected. Thus, we propose a modified mechanism for the mite avoidance of the worker brood, which is probably due to mutual destruction stress.
In contrast to A. cerana worker larvae, infected A. mellifera workers and drones support the fertility of V. destructor mites, and VTP is neither toxic to the workers nor to the drones of A. mellifera. This finding indicates that A. mellifera individuals are less susceptible to this mite. However, social behaviors, such as grooming, and hygienic behaviors in the A. mellifera colony were much weaker than those in the A. cerana colony 4 , resulting in damage to the A. mellifera colony by the mites. Furthermore, two unique tolerance factors in the A. cerana colony, non-reproduction in the worker brood and entombing of drone larvae, do not occur in A. mellifera. These findings imply that A. mellifera colony is more susceptible to this mite than A. cerana, the original host of this mite. It is reasonable to imagine that without strong avoidance of the worker brood and the toxic effect on A. mellifera larvae, V. destructor successfully parasitizes the worker brood in A. mellifera. However, why A. mellifera individuals are less susceptible to VTP is unclear. This VTP contains neither known domains nor repeated sequences, but 4-6 α-helices in its secondary structure. The function of VTP receptors and/or detoxification factors in A. mellifera hemolymph deserves further research.
Insects are generally attacked by many types of parasites. These parasites may secrete venomous proteins and non-proteinaceous compounds 24-28 that alter the immunity and/or modulate the development and physiology of the hosts or, in few cases, kill the hosts 28 . However, these proteins or compounds do not directly kill the hosts, and knowledge of the characteristics of individual salivary components is limited 28 . Unexpectedly, V. destructor mites deliver a toxic protein from the saliva to kill the infected host, which in turn prevents the reproduction of these mites in worker broods. This mutual destruction strategy may be too costly for the mites and A. cerana bees. However, from the point of view of evolutionary pressure, we can explain why these mites are rarely found in the worker brood cells of A. cerana colonies in the field 4 . VTP is not toxic to the drone of A. cerana, potentially explaining why the reproduction of Varroa mites in the original host A. cerana is limited to drone brood, although the reasons why the drones are resistant to the Varroa mites deserve further research. VTP is also safe to the worker adults of both bee species, which may reflect the importance of evolutionary adaptation, as the mites need the worker adults for transportation into other colonies 29 , and the bee colony may be collapsed if the infected worker adults could be killed by the mites.
DWV is prevalent in A. mellifera and V. destructor but is not common in A. cerana [30][31][32] . Although the introduction of higher titers of recombinant DWV into bee pupae caused A. mellifera adult wing deformities 33 , in absence of V. destructor infestation or with lower titers of DWV in the mite-infested bees 33 , A. mellifera adults did not show wing-deformities 6 . Interestingly, Varroa mites have been feeding on European bees on an isolated Brazilian island for decades, but do not activate DWV or cause deformed bees or colony losses 34 . The reasons for these results remain unknown. The following factors should be considered: the bees are not sensitive to DWV infection; the mites may lose or contain less toxin protein; DWV is a weakly pathogenic isolate. In the present study, although A. cerana adults from E-VTP-injected larvae showed no deformities wings, we found that introducing E-VTP into A. mellifera larvae increased DWV loads in the adults, which subsequently developed the deformed-wing trait. Although VTP together with DWV did not generally cause the death of the larvae of normal A. mellifera colonies, we observed that infestation with Varroa mites decreased the emergence rate of a Varroa-sensitive A. mellifera colony in presence of DWV. The RNAi-mediated suppression of vtp gene expression in the mites significantly increased the emergence rate of the larvae in this A. mellifera colony. Thus, these results would promote the development of anti-mite products for the protection of A. mellifera colonies, based on RNAi-mediated and/or gene-editing (such as CRISPR/cas9) knockdown of vtp receptor(s) in A. mellifera.

Methods and Materials
Mites and insects. Using a camel hair brush, mature female V. destructor mites were collected from the worker pupae in A. mellifera hives that had not been treated with acaricides in an apiary in Conghua, Guangzhou, China. These mites were placed in sterile Petri dishes (diameter = 9 cm; 20 mites per dish) and used for bioassays within one hour.
The 3 rd stage larvae (L3 larvae) and 5 th stage larvae (L5 larvae; spinning phase) from freshly capped cells, and the pupae (white-eyed, unpigmented cuticle pupae) and adults of drones and workers of A. mellifera or A. cerana derived from apiaries without chalkbrood and foulbrood symptoms in Conghua, Guangzhou were collected according to Kanbar and Engels 35 and used for bioassays in the laboratory 36 . The developmental stages of the honeybees were defined according to Bitondi et al. 37,38 . Viral and microsporidia loads. We used RT-PCR assays [30][31][32][39][40][41][42][43][44][45][46][47][48][49] to examine the mites and mite saliva collected according to Richards et al. 24 (Nosema apis and N. ceranae). Total RNA was isolated from honey bees (larvae, pupae and adults of workers and drones), mite saliva and mites using RNAqueous kits (Ambion, Austin, TX, USA) according to the manufacturer's instructions. The isolated RNA was quantified by spectrophotometry. DNA was removed from the samples by incubation with DNase I (5 units of DNase I in an appropriate buffer containing the RNase inhibitor RNAsin [Invitrogen, Carlsbad, CA, USA]) at 37 °C for 45 min. The primers for detecting each of the 16 viruses and two microsporidia parasites (Supplementary Table S2) were selected based on previous reports 31,32,[39][40][41][42][43][44][45][46][47][48][49] . The cDNA synthesis was performed using a PrimeScript TM 1st Strand cDNA Synthesis Kit (TaKaRa, Dalian, China). The PCR amplification program involved an initial denaturation cycle at 95 °C for 5 min, followed by 35 cycles at 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 1 min, with a final cycle at 72 °C for 7 min. The PCR products were sequenced using the corresponding specific primers.
Amplification of the full-length sequence of the vtp gene. We obtained the partial amino acid sequence of a metalloprotease homolog (herein called VTP) based on the Varroa mite salivary secretome (Zhang and Han, unpublished data). To obtain the full-length sequence of the vtp gene, we isolated total RNA from pooled samples of adult females Varroa mites using RNAiso Plus reagent To determine the full-length sequence of the vtp gene, we also isolated genomic DNA from six sterilized V. destructor eggs (sterilized using 70% ethanol) using the KAPA Express Extraction Kit (No. KK7101, Kapa Biosystems, Boston, USA). The vtp gene was PCR amplified from genomic DNA using PfuUltra II Fusion HS DNA Polymerase (Stratagene, Heidelberg, Germany) and the primers VTP-F-EcoRI (5′-GAATTCATGTTCAAACTTCTCGTTATCG-3′) (EcoRI) and VTP-R-HindIII (5′-AAGCTTTTAGGAGGCGAGCGCCTGCTGGA-3′) (HindIII). The PCR product was purified and cloned into the pEASY-Blunt cloning vector (Transgen BioTech, Beijing, China), after which the resulting plasmid was transformed into E. coli Trans T1 (Transgen BioTech, Beijing, China). The start codon (ATG) and stop codon (TAA) in the primers are indicated in bold, and the restriction sites are indicated in italics. Two clones were sent to Invitrogen LTD. (Shanghai) for DNA sequencing. The resulting sequence was analyzed using DNASTAR software and the ClustalW program and was submitted to GenBank under accession number (KU647280). The translated protein sequence was analyzed by Expasy (http://web.expasy.org/protparam/). The signal peptide, secondary structure and subcellular localization of the translated protein were predicted by the SignalP 4.1 Server (http://www.cbs.dtu.dk/ services/SignalP), Interpro (http://www.ebi.ac.uk/interpro/sea-rch/sequence-search) and TargetP (http://www.cbs.dtu. dk/services/TargetP/), respectively.

Western blotting analysis.
To confirm the VTP protein in the saliva of the mites, we prepared mite saliva, purified E-VTP and TAG protein as described above and analyzed these proteins via Western blotting. The proteins were transferred to a polyvinylidene difluoride (PVDF) membrane. The membrane was incubated overnight at 4 °C with primary rabbit polyclonal anti-VTP antibodies (diluted 1:1000) produced by AB Clonal Technology (Yingji Biotechnology, Shanghai, China). Following three washes, the membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (Boster Biological Technology, Wuhan, China) at a 1:1000 dilution for 1 hour. The immunoreactive protein bands were detected using DAB Western Blotting substrate (Boster Biological Technology, Wuhan, China).

Expression of VTP in Escherichia coli and bee bioassay.
We cloned the amplified gene cDNA into the pEASY-T1-Simple vector and transformed the resulting plasmid into E. coli TransT1 cells (Transgen BioTech, Beijing, China). DNA sequencing was performed by Invitrogen (Shanghai, China). The pEASY-T1-Simple-vtp plasmid was digested using EcoRI and HindIII. The resulting vtp fragment was purified and subsequently ligated into the expression vector pET-32a(+) (Novagen, Darmstadt, Germany), which has an N-terminal His-Tag/ thrombin/S-Tag enterokinase configuration plus an optional C-terminal His-Tag sequence. This plasmid was digested using EcoRI and HindIII to yield pET-32a(+)-vtp.
The resulting plasmid was transformed into E. coli Transetta cells (Transgen BioTech, Beijing, China) to create Transetta/pET-32a(+)-vtp for expression of the recombinant protein VTP (=E-VTP). The vector without insert sequence was transformed into the same E. coli strain to express the tag protein (approximately 20.2 kDa) as a control (=TAG). A single bacterial colony was grown overnight in 15 mL of LB-carbenicillin (100 μg/mL) broth. Tokyo, Japan). Larvae and pupae were injected into the hemocoel of each insect near the end of the abdomen. The bled larvae were discarded after injection. The adults of workers were anesthetized using nitrogen for 8 min prior to injection 50 . After injection, the larvae were reared with 30 μL of larval food 51 and the adults were fed 3 mL of sugar solution (w/v, 50%) in a culture plate in a growth cabinet at 34 °C and 80% RH. Fresh food was provided to the larvae at 12-h intervals. After 60 hours, the larvae were transferred to a culture plate sealed with Parafilm. Various negative controls, such as uninjected bees (=CK) and bees injected with only PBS (=PBS) or only the TAG protein, were established, with 3 replicates per treatment. The L5 larvae were also injected with crude saliva as a positive control. The bees were assessed daily for mortality, emergence or deformed wing rate (for A. mellifera).

RNAi-mediated knock down of vtp gene expression.
To determine the functions of VTP, we knocked down the expression of vtp using the RNAi method 52 . Briefly, dsRNA-vtp was prepared using the MEGAscript kit (AM1334, Ambion, Carlsbad, CA, USA) according to the manufacturer's instructions. The dsRNA-gfp was prepared for use as a negative control. A 405-bp (bases 1-405) fragment of the vtp gene was amplified from the cDNA transcripts of a female V. destructor using vtp-specific primers (VTP-F/R), and the GFP coding region was amplified from the control GFP-plasmid using GFP-specific primers (GFP-F, TCAAGAAGGACCATGTGGTC and GFP-R, TTCCATGGCCAACACTTGTCC). The dsRNAs were generated using the Ambion MEGAscript T7 Kit and subsequently purified using MEGAclear (AM 1908, Ambion, Carlsbad, CA, USA). The resulting dsRNAs were ethanol-precipitated, resuspended in 0.9% NaCl at a working concentration of 2.5 μg/μL 52 , and subsequently stored at −80 °C until further use. Adult mites collected from capped brood cells were soaked overnight in 500-μL microfuge tubes containing 15 μL of dsRNA-vtp, dsRNA-gfp (both at 2.5 μg/μL) or 0.9% NaCl at 16 °C, and subsequently transferred to cell culture plates containing L5 larvae of A. mellifera or A. cerana. A Varroa-sensitive A. mellifera population from Gaozhou, Guangdong was employed to determine the effect of the RNAi-mediated knock down of vtp expression on the mortality of A. mellifera L5 larvae. Two mites and one L5 larva were placed in each culture plate, and the plates were maintained at 33 ± 1 °C and 80% RH in a growth cabinet (SANYO, Tokyo, Japan) for 12 days until the adults emerged 52 . Emerged adults with normal or deformed wings were frozen using liquid N2 and stored at −80 °C for subsequent RNA extraction to determine their DWV titers 32,33 .
We also employed qRT-PCR to determine the effect of knocking down the expression of vtp using RNAi. The total RNA was extracted from mites at 0, 3, 7 and 13 d after the various treatments described above to evaluate the persistence of the effect of vtp RNAi. After DNase I treatment, the RNAs were reverse-transcribed using a TransScript All-in-One First-Strand cDNA Synthesis SuperMix for PCR Kit (Transgen BioTech, Beijing, China). The PCR reactions were conducted in triplicate in an Mx3000P Real-Time PCR System (Stratagene, California, USA) using SYBR Green (SYBR Premix Ex Taq II [Tli RNaseH Plus], TaKaRa, Dalian, China). A control without template was included in all PCR batches. The vtp primer pairs used were VTP-qF (AACGCATTCAAGACTACATCACCAA) and VTP-qR (CTTTGACAACGTTCTCCTTCTGCT). The V. destructor actin gene was used for normalization and amplified using Actin-F (CATCACCATTGGTAACGAG) and Actin-R (CGATCCAGACGGAATACTT) primers. The PCR program included a single cycle at 95 °C for 5 min and 40 cycles at 95 °C for 15 s, 58 °C for 30 s and 72 °C for 30 s. To obtain dissociation curves, the PCR products were heated to 95 °C for 15 s, cooled to 58 °C for 60 s and subsequently heated to 95 °C for 15 s. The mRNA levels relative to that of actin mRNA were calculated using Mx3000P Software (version 4.1) (Agilent, Palo Aito, CA, USA). The fold-differences in the mRNA levels were calculated using the 2 −∆∆Ct method. Data analysis. The data were analyzed using a normal one-way analysis of variance (ANOVA) using SPSS statistical software (16.0), and the significance of the between-treatment differences was evaluated using Duncan's multiple range test, with significance indicated at a P value of <0.05. All data were expressed as percentages and arcsine-square root transformed prior to conducting ANOVA. The values are expressed as the mean values ± S.D. A P value of <0.05 indicated statistical significance.