Honey Bee Parasitic Mite Contains the Sensilla-Rich Sensory Organ on the Foreleg Tarsus Expressing Ionotropic Receptors With Conserved Functions

Honey bee parasitic mites (Tropilaelaps mercedesae and Varroa destructor) detect temperature, humidity, and odor but the underlying sensory mechanisms are poorly understood. To uncover how T. mercedesae responds to environmental stimuli inside a hive, we first identified the sensilla-rich sensory organ on the foreleg tarsus. The organ appeared to correspond to Haller’s organ in ticks and contained four types of sensilla, which may respond to different stimuli based on their morphology. We searched for differentially expressed genes between the forelegs and hindlegs to identify mRNAs potentially associated with the sensory organ. The forelegs were enriched with mRNAs encoding sensory proteins such as ionotropic receptors (IRs) and gustatory receptors, as well as proteins involved in ciliary transport. We also found that T. mercedesae IR25a and IR93a were capable of rescuing temperature and humidity preference defects in Drosophila melanogaster IR25a and IR93a mutants. These results demonstrate that the structures and physiological functions of ancient IRs have been conserved during arthropod evolution. Our study provides insight into the sensory mechanisms of honey bee parasitic mites, as well as potential targets for methods to control the most serious honey bee pest.


INTRODUCTION
The number of managed honey bee colonies has declined across North America and Europe in recent years (Goulson et al., 2015). Pollination by honey bees is critical for maintaining ecosystems and producing many agricultural crops (Klein et al., 2007;Aizen and Harder, 2009). Prevention of honey bee losses has, therefore, become a major issue in apiculture and agriculture. Although there are many potential causes for the observed declines, ectoparasitic mites are considered to be major threats to the health of honey bees and their colonies (Evans and Schwarz, 2011;Goulson et al., 2015). Varroa destructor is present globally (except Australia) and causes both abnormal brood development and brood death in honey bee colonies (Rosenkranz et al., 2010). The mites feed on hemolymph and also spread honey bee viruses, particularly deformed wing virus (DWV) (de Miranda and Genersch, 2010;Martin et al., 2012). In many Asian countries, another honey bee ectoparasitic mite, Tropilaelaps mercedesae, is also prevalent in Apis mellifera colonies (Anderson and Morgan, 2007;Luo et al., 2011). These two emerging parasites of A. mellifera share many characteristics (Anderson and Roberts, 2013). For example, they have similar reproductive strategies (Sammataro et al., 2000) and both are vectors for DWV (Yue and Genersch, 2005;Dainat et al., 2009;Forsgren et al., 2009;Wu et al., 2017). As a result, T. mercedesae or V. destructor infestations have similar negative impacts on A. mellifera colonies (Dainat et al., 2012;Khongphinitbunjong et al., 2016;Nazzi and Le Conte, 2016). Although T. mercedesae is currently restricted to Asia, it has the potential to spread and establish worldwide due to the global trade in honey bees.
Varroa destructor prefers temperatures of 32 ± 2.9 • C, reproduces best at 32.5-33.4 • C, and has been shown to discriminate temperature differences of 1 • C (Le Arnold, 1987, 1988;Le Conte et al., 1990). Furthermore, its reproduction also depends on humidity of 55-70% . These results demonstrate thermo-and hygrosensation of V. destructor play important roles to adapt to the honey bee hive environment; nevertheless, chemoreception must be most important in the various interactions between mites and their honey bee hosts. For example, V. destructor prefers to parasitize nurse bees rather than foragers during its phoretic phase (Kraus, 1993;Xie et al., 2016). For its reproductive stage, it locates fifth instar honey bee larva and enters the brood cell prior to capping (Aumeier et al., 2002). These behaviors are considered to be mediated by chemical cues derived from the adult bee, larva, and larval food. Since T. mercedesae has a very similar life cycle to V. destructor, both honey bee mites should be equipped with thermo-, hygro-, and chemosensation, as observed in other mite/tick (Acari) species. Accordingly, V. destructor was found to have a sensillarich sensory organ on the foreleg tarsus (Dillier et al., 2006;Häußermann et al., 2015), which corresponds to Haller's organ in ticks (Belozerov et al., 1997). Proteomic and transcriptomic characterization were conducted for the forelegs of V. destructor, which identified potential semiochemical carriers and sensory proteins (Iovinella et al., 2018).
Ionotropic receptors (IRs) represent a subfamily of ionotropic glutamate receptors (iGluRs), which are conserved ligand-gated ion channels. IRs have specifically evolved in protostomes (Croset et al., 2010) and are best characterized in the fruit fly, Drosophila melanogaster. Most IRs are expressed in sensory neurons and function as chemoreceptors to detect various odorants and tastants (Benton et al., 2009;Rytz et al., 2013). Recent studies have also demonstrated that IR21a, IR40a, IR68a, IR93a, and IR25a are critical for thermo-and hygrosensation, suggesting that IRs have diverse physiological roles as well as gating mechanisms (Enjin et al., 2016;Knecht et al., 2016;Ni et al., 2016;Frank et al., 2017). IR25a has the same protein domains as iGluRs, is expressed broadly in various sensory neurons, and is deeply conserved in protostomes. These findings suggest that IR25a is likely to function as a co-receptor with other IRs, similar to the Orco pairing with other olfactory receptors (ORs).
In this study, we aimed to identify and characterize a sensilla-rich sensory organ in T. mercedesae using scanning electron microscopy (SEM). By comparing the transcriptomes of forelegs and hindlegs (the second to fourth legs), we identified potential genes that may be highly expressed in the sensory organ. Identification of this major sensory organ and its associated proteins in T. mercedesae inform our understanding of the mechanisms of sensory perception in honey bee parasitic mites.

Mite Sampling
Tropilaelaps mercedesae infested honey bee colonies were obtained from a local beekeeper in Suzhou, China. Adult females of T. mercedesae were collected from the capped brood cells and dissected under a light microscope using fine forceps. The collected mites were directly used for all experiments and kept together with honey bee pupae in 33 • C incubator when necessary.

SEM
A cold field emission gun SEM (Hitachi S-4700, Hitachi Company) was used for characterizing sensory organs of T. mercedesae. The whole mites and dissected legs were sprayed with gold alloy first, and then mounted on a conductive adhesive tape. During the observation, each sensillum was assigned with a number to classify the types of sensilla.

RNA-Seq
Total RNA was extracted from the forelegs, hindlegs, and main bodies of 50 adult females of T. mercedesae using TRI Reagent (Sigma). High-quality RNA samples in duplicate were then sequenced at BGI (Shenzhen, China) using Illumina HiSeq 4000 platform. After sequencing, the raw data were filtered to remove the adaptor sequences, contamination, and low-quality reads by BGI. The quality control (QC) was further analyzed using FastQC. All RNA-seq data are available in SRA database with the accession #: PRJNA510306.

Bioinformatics
The reference genome and annotated genes of T. mercedesae were first acquired from NCBI 1 , and then used for building the index by Hisat2-build indexer (Kim et al., 2015). The generated index files were used to align the clean reads of six RNA-seq samples to the reference genome. Subsequently, SAM file outputs from the previous step were sorted using SAMtools (Li et al., 2009). HTSeq-count (Anders et al., 2015) was further applied to obtain the raw read counts for downstream analysis of identifying the DEGs in R (V3.4.3) based Bioconductor edgeR package (V3.20.9) (Robinson et al., 2010). DEGs were cut-off by a false discovery rate (FDR) at 0.05, and then they were subjected to gene ontology (GO) term enrichment analysis using Blast2GO (Conesa et al., 2005). The results of GO enrichment analysis between the forelegs, hindlegs as well as main bodies were cutoff by FDR at 0.05.

Western Blot
HEK293 cells in 12-well plate were transfected with 1 µg of above expression construct (the V5-epitope tagged version) using 2 µL of Lipofectamine 2000 under OPTI-MEM medium (Thermo Fisher Scientific) for 24 h. The transfected cells were washed once with PBS, and then lysed with 200 µL of SDS-PAGE sample buffer. The cell lysates were sonicated and heated at 60 • C for 5 min. The proteins were separated by 8% SDS-PAGE and transferred to a NC membrane (PALL, 66485) by Pierce 2 Fast Blotter (Thermo Fisher Scientific, B103602038). The membrane was first blocked with 5% BSA/TBST (10 mM Tris-HCl, 150 mM NaCl, and 0.1% Tween 20) for 40 min, and then incubated with rabbit anti-V5-epitope antibody (SIGMA) (1:1,000) for 2 h at room temperature. The membrane was washed with TBST for five times (5 min each), and then incubated with IRDye800conjugated secondary antibody (LI-COR) (1:10,000) for 1.5 h at room temperature under dark and washed as above. The fluorescent signal was detected by Odyssey (LI-COR).

Thermotaxis Test
To assay the temperature preference of fruit flies, a temperature gradient of 10-40 • C with a slope of 1.07 • C/cm was produced in an aluminum block (27 long × 15 wide × 2.5 cm high) as previously reported (Sayeed and Benzer, 1996). The temperature gradient was established using a cold circulating water chamber and a hot probe at each end. The aluminum block was covered with moist paper to maintain a uniform relative humidity (RH) along the gradient. This paper was divided into 20 observation fields with a black pencil for recording the distribution of fruit flies. A glass plate with three separate lanes was placed 5 mm above the block, creating suitable corridors for fruit flies to migrate. Approximately 30 adult flies (4-5 days old) per lane were placed in the middle of testing arena around 25 • C between the aluminum block and the glass plate, allowed to migrate for 3 h, and photographed every 10 min with a digital camera. When the positions of fruit flies in the apparatus were stabilized between 1.5 and 2.5 h (This time period did not differ between the experimental groups), the number of fruit flies located at the area <24 • C was counted. Preference index was calculated by the number of flies at <24 • C/the total number of flies. The preference indexes of all tested groups were statistically analyzed by one-way ANOVA with multiple comparisons followed by Dunnett test. In each test, wild type, mutant, and the rescued fruit flies were examined simultaneously. All experiments were performed in a room where the temperature was kept constant at 25 • C.
Thermotaxis of T. mercedesae was tested as above except with 15-40 • C gradient or without temperature gradient. The testing arena was photographed every 10 min for 1h and we counted the number of mites located at the area >32 • C followed by calculating the preference index as above. The mites distributed evenly in the testing arena without temperature gradient and we counted the number of mites in the same area as we did with 15-40 • C gradient.

Humidity Preference Test
Hygrosensory behavior was assayed as previously reported (Enjin et al., 2016). A 12-well cell culture plate was modified to make a well-defined chamber with two spaces. A half of wells was filled with saturated solution of LiCl (20% humidity at 25 • C) while another half was filled with saturated NaCl (70% humidity at 25 • C) to maintain stable RH on the liquid surface in an enclosed space. The plate was then covered by a nylon mesh and closed with a lid matching the plate. In each test, approximately 80 adult flies (4-5 days old) were briefly ice anesthetized and placed at the center of apparatus. The lid was sealed to stabilize RH inside the apparatus. Humidity preference of the fruit flies with different genotypes was recorded using a digital camera and the number of flies on each side was recorded manually at 3-5 h after the start of recording. Humidity preference index was calculated by (the number of flies on NaCl side -the number of flies on LiCl side)/total number of flies. The preference indexes of all tested groups were statistically analyzed by one-way ANOVA with multiple comparisons followed by Dunnett test.
Humidity preference of T. mercedesae was tested as above except saturated NaCl was replaced with H 2 O (96% humidity at 25 • C). The control experiment using 12-well cell culture plate filled with H 2 O was also carried out. The number of mites on each side was counted every 10 min for 1 h and we calculated the preference index as above.

Identification of a Sensilla-Rich Sensory Organ on the Foreleg Tarsus of T. mercedesae
We observed the forelegs and hindlegs of T. mercedesae using SEM and found that only the foreleg tarsus contained a putative sensory organ on the dorsal side, with more than 20 sensilla of various shapes and sizes (Figures 1A-D). Most of the sensilla were equipped with well-defined sockets ( Figure 1A). We characterized the shape of each sensillum at high magnification and found that they could be classified into four different types based on the shape: type 1 had a rough surface with many indentations, sensillum #3 (Figure 1F), type 2 had a terminal pore, sensillum #18 (Figure 1G), type 3 included 18 sensilla with smooth surface, e.g., sensillum #8 (Figure 1H), and type 4 had surface pores at various densities-sensilla #2, #7/#12, #10 had pores at high, medium, and low density, respectively (Figures 1I-K). Several long sensilla were found on all legs and these are likely to be mechanosensory bristles. The pedipalp of T. mercedesae also contains a cluster of sensilla at the distal end ( Figure 1E); however, all of them have smooth surface (Figure 1L).

Identification of Potential mRNAs Enriched With the Sensory Organ
To identify potential mRNAs highly expressed in the sensory organ, we obtained RNA-seq reads from the forelegs, hindlegs, and main bodies (without legs) and then identified the differentially expressed genes (DEGs) between the forelegs and hindlegs. Since only foreleg tarsi were equipped with the sensory organs, we expected the DEGs to represent the sensory organ-associated mRNAs. We found that 46.1-83.9% of the sequence reads were aligned with the T. mercedesae genome (Supplementary Table S1) and we used these to identify DEGs between the forelegs and hindlegs, the forelegs and main bodies, and the hindlegs and main bodies. The lists of DEGs are shown in Supplementary Tables S2-S4. Table 1 and  Supplementary Tables S5, S6 indicate GO terms enriched for the genes highly expressed in the forelegs compared to the hindlegs and main bodies, and the ones highly expressed in the hindlegs relative to the main bodies, respectively. For the genes highly expressed in the forelegs, many of the GO terms were associated with ion channel activity, particularly iGluR activity, as well as microtubule motor activity in the "molecular function" category. In the "biological process" category, GO terms related to cilium assembly, microtubule-based processes, and detection of chemical stimulus involved in sensory perception were most prevalent. All GO terms in the "cellular component" category were related to cilium, intraciliary transport particle, and BBSome ( Table 1). Several GO terms related to mitochondrial activity were also enriched in the forelegs, compared with the main bodies, and this was similar for the genes highly expressed in the hindlegs relative to the main bodies (Supplementary  Tables S5, S6). These results are consistent with the finding that the sensory organ on the foreleg tarsus had many sensilla (Figure 1) and with the ciliated sensory neurons and the expression of abundant iGluR mRNAs. It is likely that higher expression of mRNA of genes involved in mitochondrial activity in the legs relative to the main bodies would be necessary to supply energy for leg movement.
In addition to iGluRs, the forelegs expressed high levels of transient receptor potential channel A1 as previously reported , anoctamin-7 (TMEM16 family), and two gustatory receptors (GRs): Tm03548 and Tm05586   (Supplementary Table S2 and Figure 2). Orthologs of these GRs were also present in Ixodes scapularis, but were not found in D. melanogaster, indicating that they are specifically expanded in the Acari lineage. Thus, the T. mercedesae sensory organ appears to be equipped with various sensory proteins with ion channel activity.

Conserved Sensory Functions Between D. melanogaster and T. mercedesae IR25a and IR93a
We previously annotated eight IR and 33 iGluR genes in the T. mercedesae genome and showed that two IR mRNAs, Tm15229 and Tm15231, are abundantly expressed in the forelegs, using qRT-PCR . These two genes are included in the above DEGs and we also found that mRNAs for two non-NMDA iGluRs (Tm15234 and Tm15241), as well as two other IRs (Tm15230 and Tm15243), were also highly expressed in the forelegs (Figure 2). Thus, a small fraction of iGluRs and half of IRs appear to play roles in mite sensory perception.
Based on the phylogenetic tree of T. mercedesae IR and iGluR genes, together with those of D. melanogaster and Ixodes  scapularis , we found only two (out of eight) IRs (Tm15229 and Tm15231) were conserved, having the D. melanogaster orthologs, IR93a and IR25a, respectively. DmIR93a and DmIR25a have been shown to play roles in temperature and humidity preferences (Enjin et al., 2016;Knecht et al., 2016;Ni et al., 2016). To test whether the sensory functions of IR93a and IR25a are deeply conserved between fruit flies and mites, we first obtained the full length cDNAs of Tm15229 and Tm15231 by determining both the 5 and 3 end sequences using RACE methods. Tm15229 (TmIR93a) and Tm15231 (TmIR25a) share the same protein domains with DmIR93a and DmIR25a, respectively (Figure 3). Tm/DmIR25a contains the N-terminal leucine/isoleucine/valine-binding protein (LIVBP)-like domain and PBP2_iGluR domain. Meanwhile, Tm/DmIR93a contained only the PBP2_iGluR domain. The protein expression was confirmed by ectopic expression in HEK293 cells, followed by western blot (Supplementary Figure S1). We then compared the thermotactic behavior of D. melanogaster IR93a and IR93a mutants expressing TmIR93a under DmIR25a-Gal4 with the wild type. Expression of DmIR93a and DmIR25a overlapped in the antennae (Knecht et al., 2016). We also analyzed D. melanogaster IR25a and IR25a mutants expressing DmIR25a or TmIR25a under DmIR25a-Gal4. From our assay to test thermotactic behavior, the fraction of animals in the area with temperatures <24 • C significantly increased in both IR93a and IR25a mutants compared with the wild type; however, expression of TmIR93a, TmIR25a, or DmIR25a rescued this behavioral defect ( Figure 4A). We then tested the humidity preferences of the fruit fly stocks described above. Wild type flies preferred high (saturated NaCl, 70%) over low (saturated LiCl, 20%) humidity but this preference was significantly impaired in IR25a and IR93a mutants (Figure 4B), as previously reported (Enjin et al., 2016;Knecht et al., 2016Knecht et al., , 2017Ni et al., 2016). Expression of DmIR25a FIGURE 2 | Expression of ionotropic receptors (IRs), NMDA iGluRs, non-NMDA iGluRs, and gustatory receptors (GRs) mRNAs in the forelegs, hindlegs, and main body of T. mercedesae. The level of expression of each mRNA in the forelegs, hindlegs, and main body is shown by a graded color (blue to red) based on the counts per million mapped reads (CPM).
or TmIR25a was able to rescue the humidity preference defect of the mutant fly. Expression of TmIR93a partially rescued the defective phenotype of IR93a mutant and preference for 70% humidity of the rescued flies is significantly lower than that of wild type flies ( Figure 4B).
We also examined whether T. mercedesae is capable of detecting ambient temperature and humidity and has the specific preferences. As shown in Figure 5A, the mites accumulated at the area of 32-40 • C in a gradient of 15-40 • C but randomly positioned on the testing arena without temperature gradient. Similarly, the mites were equally present on each side of the testing apparatus filled with H 2 O (no humidity gradient); however, they accumulated on the side with H 2 O (96% RH) over saturated LiCl (Figure 5B). These results demonstrate that the mites detect temperature as well as humidity and prefer 32-40 • C and high humidity, which are equivalent to the inside environment of honey bee hive.
characterization of each sensillum is, of course, necessary to support this hypothesis.

T. mercedesae Sensory Organ Enriched With mRNAs for Sensory Proteins and Proteins Necessary for Ciliary Biogenesis/Transport
We sought to identify mRNAs differentially expressed in the forelegs of mites as candidates for those expressed in the sensory organs. Although we have no direct evidence to show that these mRNAs are indeed expressed in the sensory organ, their specific existence in the forelegs, as well as the identified DEGs, support this approach. The same method was used with two tick species, Dermacentor variabilis and Ixodes scapularis, to identify mRNAs associated with the Haller's organ Josek et al., 2018). Our results to show the enrichment of TmIR25a and TmIR93a mRNAs in the forelegs of T. mercedesae are consistent with the results for I. scapularis (Josek et al., 2018). Eliash et al. (2017) also reported that the V. destructor homolog of IR25a (this may not be the ortholog since it does not have the N-terminal LIVBP domain) was highly expressed in the forelegs. These results suggest that IR25a and IR93a may represent the major thermo-and hygroreceptors in acarids, based on their physiological roles in fruit flies. This hypothesis was further supported by our finding that TmIR25a and TmIR93a rescued the defective thermo-and hygrosensation in D. melanogaster IR25a and IR93a mutants (Figure 4) and T. mercedesae is capable of detecting ambient temperature and humidity (Figure 5). It is notable that not only the structure, but also the physiological roles, have been deeply conserved during Arthropod evolution (Croset et al., 2010). IR21a, IR40a, and IR68a are necessary for thermo-and hygrosensation together with IR25a and IR93a in D. melanogaster (Enjin et al., 2016;Knecht et al., 2016Knecht et al., , 2017Ni et al., 2016;Frank et al., 2017). However, as we previously reported, there are no orthologs of IR21a, IR40a, and IR68a in T. mercedesae genome. Two other IRs enriched in the mite foreleg are Tm15230 and Tm15243 which appear to be Acarispecific and the orthologs are absent in fruit fly. As the exclusive parasite of honey bee larva and pupa, T. mercedesae must adapt to the colony environment where the temperature and humidity are different and relatively constant compared to the outside environment. In fact, the mite prefers temperature between 32 and 40 • C in contrast to D. melanogaster at 25 • C. This may have resulted in evolving the novel IRs, Tm15230 and Tm15243, which are different from D. melanogaster IR21a, IR40a, and IR68a. However, the possibility that Tm15230 and Tm15243 are involved in chemosensation cannot be ruled out at this stage. Two (Tm03548 and Tm05586) and eight GR mRNAs were highly expressed in the sensory organs of T. mercedesae and I. scapularis, respectively (Josek et al., 2018). However, these GRs do not appear to be orthologs and Josek et al. (2018) reported that the expression of other I. scapularis GRs was too low to make a comparison between the forelegs and hindlegs. Furthermore, most of the GRs have expanded in Acari in a lineage-specific manner (Eyun et al., 2017;Josek et al., 2018). Except for Gr28b in D. melanogaster, which has an important role in thermosensation (Ni et al., 2013), GRs are generally considered to function as chemoreceptors. Thus, the above two GRs of T. mercedesae may detect, for example, a few odorants/tastants derived from honey bee adults, larva, and larval food. This is consistent with the finding that the numbers of IR and GR genes in parasitic T. mercedesae were dramatically reduced compared to those in "free-living" mites/ticks . The TRPA1 channel was also enriched in the forelegs and may function as a sensor to detect nociceptive stimuli (temperature and chemicals) for avoidance, as previously reported (Peng et al., 2015Dong et al., 2016). In summary, T. mercedesae may depend on IR25a, IR93a, and TRPA1 for thermosensation, IR25a and IR93a for hygrosensation, and two acarid-specific GRs (Tm03548 and Tm05586) as well as TRPA1 for chemosensation. It is difficult to extrapolate the precise physiological roles of two acarid-specific IRs (Tm15230 and Tm15243) at this stage.
Another group of proteins enriched in the forelegs is associated with cilium assembly and intraciliary transport processes and includes kinesin, dynein, and intraflagellar transport proteins. Cilia are organelles present on the cell surface that concentrate signaling molecules to organize sensory, developmental, and homeostatic function. Movement of the signaling receptor from the basal body into the cilia requires IFT-A and its exit depends on IFT-B and BBSome (Nachury, 2018). Many sensilla are present in the sensory organ of T. mercedesae (Figure 1) and sensory neurons associated with the sensilla have a ciliated dendrite, which requires the protein complexes described above to control traffic of, for example, sensory proteins. GPCRs are considered to be the major target for intraciliary transport (Schou et al., 2015); however, the four IRs of T. mercedesae may also depend on IFT-A, IFT-B, BBSome, and other proteins for transport. Consistent with the presence of few sensilla in the Haller's organs of two tick species, enrichment of these mRNAs was not observed . In contrast to Carr et al. (2017), we did not observe high expression of mRNAs for the downstream signaling pathway components of sensory proteins in the forelegs of T. mercedesae.
Our study uncovers the ancient roles of IR25a and IR93a in thermo-and hygrosensation of arthropods. We also found the potential roles of evolutionarily conserved intraciliary transport proteins for the entry and exit of sensory proteins in the ciliated dendrites of sensory neurons. The functional disruption of these proteins could be considered as an effective method to control honey bee parasitic mites as well as other mites/ticks that represent major pests for plants and animals.

AUTHOR CONTRIBUTIONS
JL and QL conducted the experiments. TK supervised the research project and wrote the manuscript with JL.

FUNDING
This work was supported by Jinji Lake Double Hundred Talents Programme to TK.

ACKNOWLEDGMENTS
We thank Bloomington Stock Center for providing us fruit fly stocks for the experiments. We are grateful to Xinyi Li for sampling the mites and Ziwen Xie for preparing the fruit fly food. This manuscript has been released as a pre-print at bioRxiv.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys.2019. 00556/full#supplementary-material FIGURE S1 | Expression of TmIR25a and TmIR93a proteins. The IR proteins (V5-epitope) and β-actin expressed in HEK293 cells transfected with empty vector (Mock), TmIR25a-, and TmIR93a-expressing constructs were analyzed by western blot. The size (kDa) of protein molecular weight marker (MW) is at the left.