The NBDs of all the D. melanogaster ABC transporters were used as queries for BLASTP search against the two silkworm genome databases, SilkDB and KAIKObase. Each potential silkworm ABC transporter was validated by searching its known orthologs from the protein database of NCBI and further searching its NBD and TMD with the Pfam program. As a result, a total of 51 putative ABC genes were identified in the silkworm genome. A previous study has identified 56 ABC genes in D. melanogaster 2 . For comparative analysis, we also identified 68 ABC genes from the flour beetle, Tribolium castaneum (Coleoptera), 52 from A. gambiae (8 more compared to 44 in 4 due to the updated mosquito genome), and 43 from the honeybee, Apis mellifera (Hymenoptera) (Table 1).

In SilkDB, 37 ABC genes have evidences of mRNA expression with EST sequences, which were collected from 36 cDNA libraries of multiple tissues during different developmental stages. Based on the EST information, BmABC010129 shows the highest transcript level with 26 hits. The largest ABC transporter (1794 amino acids) and the smallest one (280 amino acids) are encoded by BmABC012789 and BmABC010825, respectively (Table 2).

Each silkworm ABC transporter possesses one or two conserved NBDs (Figure 1). We performed a multiple sequence-alignment to construct a phylogenetic tree on the basis of NDBs of the 51 silkworm ABC transporters. Phylogenetic analysis revealed that silkworm ABC transporters can be divided into 8 subfamilies (A-H) with high bootstrap values. The ABCC subfamily can be separated into 2 groups, one of which shares high similarity to the ABCB subfamily (Figure 2). Among the 51 silkworm ABC transporters, there are 11 full transporters and 26 half transporters. The ABCA, ABCB and ABCC subfamilies contain full transporters, whereas half transporters exist in 6 subfamilies rather than ABCE and ABCF (Figure 1). The 4 ABCE and ABCF proteins lack of TMDs. One putative ABCB transporter contains one NDB and two TMDs, while one putative ABCC transporter contains two NDBs and one TMD. In addition, we found that 8 silkworm genes encode obvious NDBs show high similarity to ABC genes in the silkworm and other organisms. For convenience, we referred the 10 putative ABC proteins as incomplete ABC proteins. The incomplete sequences of those ABC genes might be because the genome of silkworm is not at completion and annotation of the entire genome is not complete. For comparative analysis, we also constructed phylogenetic trees for the ABC transporters from T. castaneum (Additional File 1: Fig. S1), A. gambiae (Additional File 2: Fig. S2), and A. mellifera (Additional File 3: Fig. S3). Similarly, the ABC transporters in each insect species can be divided into 8 subfamilies (A-H) (Table 1).

Of the 51 silkworm ABC genes, 50 are dispersed on 17 chromosomes and 1 on an unmapped scaffold. Gene structures of the 51 silkworm ABC transporters show high complexity with exon numbers ranging from 6 to 34 (Table 2). There are two major gene clusters located on chromosomes 12 and 15, which contain 9 and 10 ABC genes, respectively. In addition, chromosomes 10 and 22 each contain 4 ABC genes. Gene duplication is very evident in chromosomes 12 and 15, also occurs in chromosomes 10, 17 and 26 by tandem duplication (Additional File 4: Fig. S4).

Since the insect ABC transporters can be classified into 8 subfamilies (A-H), we characterized each subfamily in the silkworm and four other insect species for comparison.

In order to understand the possible physiological function and hormonal regulation of the silkworm ABC genes in the future, we here analyzed the expression patterns of all the silkworm ABC genes by microarray data. With the exception of BmABC010557, all other ABC genes have corresponding probes in the oligonucleotide chip 34 . We first investigated the spatial expression patterns of the silkworm ABC genes on day 3 of fifth larval instar using the microarray data of tissue-specific expression 34 . The analysis revealed that the expression of 24 ABC genes can be detected at least in one of the selected tissues, including testis, ovary, head, integument, Malpighian tubule, fat body, haemocyte, anterior and middle silk gland, and posterior silk gland (Figure 3). Significantly, a large number of ABC genes were highly expressed in the testis compared to other tissues. It is known that more than one thousand genes are expressed specifically in the testis 34 . However, whether all these testis-specific ABC transporters are functionally important in this male organ is questionable and worthy of further investigation.

Some tissue-specific ABC genes were verified by qRT-PCR in nine tissues, including fat body, midgut, testis and ovary, wing disc, trachea, brain, Malpighian tubule, haemocyte, prothoracic gland, and silk gland, on day 2 and 5 of fifth larval instar and day 1 of prepupae. Besides the testis, the Malpighian tubule is the second tissue enriched with ABC gene expressions. The Malpighian tubule serves as the excretory and osmoregulatory organs for insects, in which the urine is produced and transported to the hindgut for the selective absorption of water and ions. Thus, the Malpighian tubule plays an important role in excretion and it is also involved in xenobiotic detoxification 35 . BmABC002712, an ABCG gene, was exclusively and abundantly expressed in the Malpighian tubule, detected by both microarray (Figure 3) and qRT-PCR (Figure 4), suggesting that BmABC002712 might play an important role in the Malpighian tubule. BmABC002712 could serve as a Malpighian tubule-specific molecule marker for future studies of gene function and regulation.

In a microarray study published previously 13 14 , we detected multiple silkworm ABC genes that are highly expressed during molting and pupation, exhibiting possible regulation by 20E. We then analyzed the temporal expression patterns of the silkworm ABC genes from day 4 of fifth instar to the adult stage using the microarray data of developmental changes of mRNA 34 . The analysis revealed that the expression of 19 and 21 ABC genes can be detected in females and males, respectively (Figure 5). We found that BmABC005226, an ABCG gene, was highly expressed during pupation (12 hours after the wandering stage) and the early pupal stage (60 hours after the wandering stage), when the 20E level is high. As determined by qRT-PCR, BmABC005226 is exclusively expressed in the midgut and highly expressed during the wandering and prepupal stages, when the 20E level is high (Figure 6A). The microarray and qRT-PCR data suggest that the midgut-specific ABCG gene, BmABC005226, is regulated by 20E at the transcriptional level.

We are very interested in the ABCG subfamily, which is particular in both structure and function 30 . In an unpublished survey in D. melanogaster, we have found that several ABCG transporters are exclusively expressed in two important endocrine organs, the corpus allatum which produces juvenile hormone and the prothoracic gland which produces ecdysone (the 20E immediate precursor). To this end, we investigated the spatial expression patterns of all the 13 silkworm ABCG genes using qRT-PCR. Besides BmABC005226, four other genes, including BmABC005203, BmABC005202, BmABC010555, and BmABC010557, are also preferentially expressed in the midgut. Moreover, we investigated their temporal expression patterns in the midgut, which revealed that the 4 ABCG genes are highly expressed during the wandering and prepupal stages as well. The qRT-PCR data (Figure 6B-E) roughly matches the microarray data (Figure 5), implying that they are also 20E responsive.

In addition, BmABC002581 (brown) and BmABC012035 were highly expressed in brain, and the mRNA level of brown peaked during the feeding stage of the fifth instar (Additional File 12: Fig. S12). Unfortunately, we did not find any prothoracic gland-specific ABCG genes in the silkworm.

The expression patterns in Figure 6 imply that the 5 midgut-specific ABCG genes, including BmABC005226, BmABC005203, BmABC005202, BmABC010555, and BmABC010557, might be up-regulated by 20E during molting and pupation. To test this hypothesis, we injected 20E into day 2 of fifth instar larvae and USP dsRNA into larvae at the initiation of the early wandering stage, followed by measurements of their mRNA levels using qRT-PCR 13 14 . As expected, the mRNA level of the 20E-primary response gene E75 was significantly up-regulated after 6 hours of 20E treatment and down-regulated after 24 hours of USP RNAi. Similarly, all the 6 ABCG genes were up-regulated by 20E treatment and down-regulated by USP RNAi (Figure 7 and 8), demonstrating that the 5 midgut-specific ABCG genes are up-regulated by 20E during molting and pupation.

It has been reported that the TMD of ABCG2 binds steroids and the binding mediates modulation of ABCG2 activity 36 . However, little is known about whether and how ABCG genes are transcriptionally regulated by steroid hormones in mammals. Since gene duplication is very evident in the ABCG subfamily in insects but not in human, the similarity of ABCG transporters in both physiological function and hormonal regulation is of great interest. We suppose that transcriptional regulation of ABCG genes by the steroid hormone 20E in the silkworm holds promising values for further investigation.

Bombyx larvae were provided by the Sericultural Research Institute, Chinese Academy of Agricultural Sciences. They were reared with fresh mulberry leaves in the laboratory at 25°C under 14 hour light/10 hour dark cycles 13 14 .

We mainly used SilkDB 40 41 to search for potential ABC genes. First, we used the highly conserved NBDs of all 56 D. melanogaster ABC genes as queries to search against the updated GLEAN gene collection to identify silkworm ABC genes by local BLASTP 42 , with an E-value threshold of 10^-6. In addition, KAIKObase was also used, particularly for obtaining the full-length cDNA sequences. The identified putative ABC transporter genes were validated by search of the protein database of the NCBI with the putative ABC gene sequences as queries. Each potential ABC transporter was further analyzed by the program Pfam to identify its NBD and TMD domains. Finally, the identified silkworm ABC genes were used as queries to search SilkDB and KAIKObase again in order to avoid missing genes. The same methods were used for identification of ABC transporters in A. gambiae, A. mellifera and T. castaneum.

Protein sequence alignments were performed using ClustalX 43 . NBDs were then subjected to a phylogenetic analysis, using neighbor joining and bootstrapping with 1000 replicates in the program package MEGA4.0 44 .

We used SilkMap tool in silkworm genome database to map the loci of each ABC genes on the 28 chromosomes. Genes in clusters, for example, BmABC000724 and BmABC000725 on Chromosome 1, are indicated by a vertical line (Additional File 4: Fig. S4).

We used silkworm genome-wide microarray data to profile the expression patterns of ABC genes in multiple larval tissues and during the larval-pupal-adult metamorphosis. The expression patterns of ABC transporter genes were estimated from intensity values 33 . From SilkDB, we downloaded normalized microarray data (GSE17571) for genome-wide gene expression in the anterior/middle silk gland (A/MSG), posterior silk gland (PSG), testis, ovary, fat body, midgut, integument, hemocyte, Malpighian tubule, and head on day 3 of the fifth larval instar. Normalized microarray data for gene expression at 19 time points during silkworm metamorphosis was recently performed (unpublished data from Southwest University). The sequential time points include V4 (day 4 of the fifth larval instar), V5, V6, V7, W0 (0 h after wandering), W12, W24, W36, W48, W60 (just after pupation), W72, W96, W120, W144, W168, W192, W216, W240, and adult (moth). The expression pattern of the ABC genes was estimated from intensity values. An ABC gene was considered to be expressed if its normalized intensity value exceeded 0 34 .

Total RNA was extracted from the whole body or selected tissues of different developmental stages and used for qRT-PCR analysis as previously described 13 14 . Primers used here and somewhere else in this paper are listed in Additional File 13, Table S1.

As described previously 13 14 , day 2 of the fifth instar larvae (48 hrs after the fourth molting) was chosen for 20E injection (Sigma Aldrich, USA) (5 μg/larva) and the controls were injected with the same volume of control solvent. Six hours after 20E treatment, larvae were sacrificed to dissect tissues for qRT-PCR analysis. Ten animals were used for each group and 3 biological replicates were conducted.

dsRNA was generated using the T7 RiboMAX™ Express RNAi system (Promega, USA). At the initiation of the early wandering stage, each individual larva was injected with GFP dsRNA (10 μg) or USP dsRNA (10 μg). Twenty four hours after RNAi treatment, the larvae were sacrificed for qRT-PCR analysis 13 14 . Thirty animals were used for each group and 3 biological replicates were conducted.