Introduction

Insects show a wide range of anatomical diversity in appendage morphology. The larvae of many holometabolous insects possess abdominal appendages called prolegs (Nagy and Grbic, 1999). There is considerable diversity in proleg distribution on the insect body, including wide variation in segmental arrangement and number. For example, hymenopteran species have prolegs on each abdominal segment, whereas the prolegs of some dipteran species may be a useful adaptation for certain larval habits or predatory needs (Suzuki and Palopoli, 2001). Lepidopteran larvae have prolegs in segments A3–A6. The mechanism underlying proleg development is unclear. The body segments of many insects are specified by Hox, or homeotic, genes (Ouweneel, 1976). Changes in the functions of these genes and/or in the downstream genes under their control may result in these morphological differences.

In the silkworm Bombyx mori, a homeotic gene complex known as the E complex specifies the identities of the body segments. The E complex genes are part of the silkworm Hox gene cluster, which is located on chromosome 6 (Yasukochi et al., 2004). These genes include Bm Ultrabithorax (Ubx), Bm abdominal-A (abd-A) and Bm Abdominal-B (Abd-B; Ueno et al., 1992), which are homologs of the Drosophila bithorax complex (BX-C; Bender et al., 1983). In Drosophila, these Hox genes, especially abd-A, are important for determinative decisions in the anterior abdomen (Karch et al., 1985). The Hox protein BX-C represses the development of ventral limbs by preventing the transcription of Distal-less (Dll), a limb-promoting gene. In Lepidoptera, it is thought that E complex genes also repress Dll since in the abdominal segments, circular holes lacking Hox protein expression showed Dll expression and proleg development (Warren et al., 1994). Dll seemed to escape from repression by the Hox genes in these special holes. What produces this pattern of Hox gene expression in the abdominal segments of Lepidoptera? Do changes in the regulation of the Hox genes or their downstream targets play a role in the occurrence of prolegs? The answers to these questions are largely unknown (Averof, 2002; Hughes and Kaufman, 2002).

In the silkworm, more than 30 types of homeotic mutants, called E mutants, have been identified by phenotype and analyzed for pseudoallelism. Most of the mutations in these mutants are dominant and induce the expression of extra markings or supernumerary legs/prolegs (that is, rudimentary abdominal legs; Itikawa, 1943; Tazima, 1964). Until recently, the molecular nature of the E mutants was unknown, except for the mutants EN and ECa. Ueno et al. (1992) reported that deletions within Bm Ubx may be responsible for the phenotypes of these two mutants and that the function of Bm Ubx might be similar to that of Dm Ubx in repressing leg and proleg development. Although the mechanism of proleg development remains unknown, changes in Hox gene function (Warren et al., 1994) or in the expression of genes downstream of the Hox genes (Panganiban et al., 1994), including novel target genes (Ueno and Suzuki, 1997), have been implicated. The analysis of mutations will help to explain how segmental developmental mechanisms may have been modified during evolution.

The Ekp mutant is an ideal subject for such studies. Larvae homozygous for Ekp (Ekp/Ekp) exhibit various shifts on their dorsal and ventral sides (Figure 1), including extra crescents on the A3 segment (anterior transformation) and small supernumerary prolegs on the A2 segment (posterior transformation). In contrast, larvae heterozygous for Ekp (Ekp/+) show only the A2 segment phenotype. Regardless of the crescent phenotype of the Ekp mutant, its supernumerary prolegs can be considered a typical phenotype for studying the mechanism of proleg development. We have named this phenotype Ekp-1. The various phenotypes of the Ekp mutant are thought to be associated with mutations in E complex loci (Goldsmith, 1995), but experimental evidence is still lacking.

Figure 1
figure 1

Phenotype and appendage morphology of p50 and the Ekp mutant. (a1) E+/E+ (p50 strain). There is one crescent on the A2 segment and another on the A5 segment. The A2 segment lacks ventral appendages. (a2) E+/Ekp (F1 progeny of p50 and Ekp). The dorsal morphology is the same as that of the E+/E+, except the E+/Ekp individual has small supernumerary prolegs on the dorsal side of its A2 segment (that is, the Ekp/Ekp phenotype). The arrow shows the supernumerary prolegs. (a3) Ekp/Ekp (Ekp strain). The mutant has extra crescents on its A3 segment and small supernumerary prolegs on its A2 segment; on the dorsal side, it is missing a crescent on the A5 segment. The arrows show the extra crescents and supernumerary prolegs. (b1) Thoracic leg. (b2) Ventral side of a normal A2 segment. (b3) Abdominal leg (that is, proleg). (b4) Supernumerary proleg, which is similar to a proleg but distinct from a thoracic leg.

Full size image

In this study, we used simple sequence repeat (SSR) and cleaved amplified polymorphic sequence (CAPS) markers to restrict Ekp-1 to within a 220-kb region in the Hox gene cluster, thereby providing the starting point for molecular analysis of the locus, which may play a role in silkworm proleg development. This is the first study to finely map a phenotypic locus in the silkworm. Our results demonstrate that it may be feasible to locate new genes in the silkworm by map-based cloning.

Materials and methods

Silkworm material and mapping populations

Ekp (Ekp/Ekp) is maintained by single-pair mating in a stable silkworm strain at the Chinese Academy of Agriculture Sciences’ Sericulture Research Institute. Ekp and the wild-type strain p50 are the parental insects. Ekp-1 was previously placed within the sixth chromosomal linkage group based on SSR linkage analysis (Miao et al., 2005). To map the Ekp-1 locus on a fine scale, we used nine male F1 individuals from a single-pair cross of an Ekp female with a p50 male and mated them with a p50 female to generate backcrossed F1 (BC1♂) progeny (designated p50 × (Ekp × p50)); thus, a total of 2396 BC1♂ offspring were constructed. The larvae were reared at 25 °C until the fifth instar.

Genetic mapping of Ekp-1

Total DNA was extracted from the hind parts of 2396 BC1♂ larvae according to the method of Li et al. (2006) with the modification that we added SDS to the extraction buffer at a final concentration of 2%. We used the p50 and Ekp parents to test for polymorphisms at each of the SSR markers developed by Miao et al. (2005). In total, 192 BC1♂ backcross progeny were then genotyped using the polymorphic SSR markers. SSR polymorphism determination and genotyping were performed using an Applied Biosystems 377 DNA sequencer (ABI PRISM) for 2 h with 96 lanes run per gel. Next, published sequence-tagged site (STS) markers that map to the Hox cluster on chromosome 6 (Yasukochi et al., 2004) were amplified using p50 DNA as the template and sequenced to develop CAPS markers for the BC1 population. BLAST searches using the sequences of the STS markers and silkworm SSR sequences were performed to identify linkages between the two sets of markers. Using this approach, we developed a CAPS marker, B422251, within Scaffold000028 that bears both an SSR marker (S0608) and a published STS marker (B4C5Ga; Yasukochi et al., 2004). We also developed three polymorphic STS markers, B3l10Gb, B9B9Ba (Yasukochi et al., 2004) and ab17577 (an Abd-B-derived marker in Scaffold003414) on either side of B422251. Genotyping of the same BC1♂ population was conducted using these markers.

To determine the relationship between Ekp-1 and the molecular markers linked to it, MAPMAKER (version 3.0) was used to calculate genetic map distances (Li and Garvin, 2003). Two-point linkage was determined at log of odds (LOD)=5.0.

On the basis of our low-resolution mapping results (see Results), a total of 2396 BC1♂ larvae were genotyped using two closely linked markers, ab17577 and B422251, to produce a fine map of the Ekp-1 locus.

Candidate gene cloning

The epidermis and fat body were dissected from fifth instar p50 and Ekp larvae in sterile phosphate-buffered saline (50 mM sodium phosphate, pH 7.5 and 150 mM NaCl), frozen rapidly in liquid nitrogen and stored at −70 °C until use. Total RNA was extracted using TRIzol (Invitrogen, Salt Lake City, Utah, USA) and digested with DNase I (Takara, Tokyo, Japan) to remove any remaining DNA. cDNA was synthesized using a first-strand cDNA synthesis kit (Toyobo, Osaka, Japan).

To clone the complete coding region of Ubx, we designed the primer Ubx1f (5′-ATGAACTCTTACTTCGAC-3′) based on a conserved motif in the translation initiation region. Similarly, the primer Ubx3′ORFr (5′-TTAATGTTCGGGGTGTCCCTG-3′) was designed based on a motif close to the terminal coding region that included the predicted ending code. cDNA from p50 and Ekp was used as the template for PCR using the primers Ubx1f and Ubx3′r.

To clone full-length abd-A, we screened a silkworm gonad cDNA library constructed in our laboratory (Liu WB et al., unpublished) by PCR using the primers abd-Aboxf (5′-GCGAGGACGACAAACTTA-3′) and abd-Aboxr (5′-TGCTGCTGCTTCATTCTG-3′). Positive clone pools were subsequently amplified using either universal primer M13 forward with abd-Aboxf or universal primer M13 reverse with abd-Aboxr.

The PCR products were cloned into pMD18-T (Takara) and sequenced using an ABI 3700 DNA Analyzer (ABI PRISM). The sequences were then BLAST-searched against the silkworm genome database. The matching scaffolds were then used to develop several CAPS markers.

Marker development

For high-resolution mapping, seven CAPS markers were developed based on published data (Yasukochi et al., 2004), and sequence information was inferred from the candidate genes. These CAPS markers include Ubx3′A, which is located about 29 kb downstream of the Ubx 3′ORF, designed from Scaffold000017; B12J2Aa, which is in the Hox cluster (Yasukochi et al., 2004); B910325 and B923511, which are in Scaffold005660 (B910325 is about 12 kb from the published STS marker B9B9Ba (Yasukochi et al., 2004), while B923511 is actually a B9B9Ba-derived CAPS marker); abd-A5′, which is about 9 kb upstream of the abd-A 5′ open reading frame (ORF), designed from Scaffold000405 and abd-A3′A and abd-A3′B, which are about 9 and 19 kb downstream of the abd-A 3′ ORF, respectively, designed from Scaffold000075.

Construction of BAC pools

The bacterial artificial chromosome (BAC) library was kindly provided by Kazuei Mita and Chengcang Wu (Wu et al., 1999). The library has 36 864 clones, which are in 96 384-well microtiter plates. For each plate, we constructed four pools, labeled A, B, C and D, each with 96 BAC clones. In total, 384 pools were constructed. The BAC DNA from each pool was isolated using a conventional alkali method (Sambrook et al., 1989), diluted 200-fold and then put into the corresponding well in one of four 96-well microtiter plates marked pool-A, pool-B, pool-C and pool-D.

PCR-based BAC library screening and construction of a physical map

The markers used for fine mapping plus additional monomorphic STS markers, including the Ubx-derived markers Ubx3′B and Ubx5′, were used to screen the BAC library. We discovered that the predicted non-coding locus iab-4 (Tong et al., 2006) was located in Scaffold005660 (where B9B9Ba is located); thus, iab-4-derived markers were designed and used in the screening. All markers developed in this study are listed in Table 1.

Table 1 Markers used in this study
Full size table

The first round of PCR was performed using the CAPS markers known to be closely linked to Ekp-1 with the four pools containing all BAC clones in the library. For each positive pool, we constructed a row pool and a column pool. The BAC DNA of each pool was isolated using the method described above. A second round of PCR was performed on the row and column pools using the corresponding markers to confirm the positive BAC clones. To prepare the DNA of the positive clones for further analysis, single colonies were cultured for 24 h in 20 ml of Luria–Bertani containing 12.5 μg ml−1 chloramphenicol. The BAC DNA was then prepared as described by Yin et al. (2006). The positive BAC DNA was digested with NotI (NEB, Ipswich, MA, USA) at 37 °C for 3 h and analyzed by pulsed-field gel electrophoresis in 1.0% agarose gel with 0.5 × tris-borate-EDTA buffer under the following conditions: 6.0 V cm−1, 14 °C, pump setting 80, switch time 1–12 s, 120° for 16 h. The end sequences of the positive BAC clones were determined using the primers T7 and Sp6. One BAC-end derived STS marker was designed based on the sequence (Table 1).

Results

Low-resolution mapping of the Ekp-1 locus on chromosome 6

Among the 2396 BC1 individuals, 1230 had supernumerary prolegs and 1166 were normal, giving a segregation ratio of 1:1. The χ2 analysis confirmed segregation from a single gene (χ2=0.45), proving that the Ekp-1 mutation affects a single locus. Ekp-1 was previously mapped to the sixth chromosome using BC1♀ offspring (Miao et al., 2005). Four SSR markers on chromosome 6 (S0601, S0603, S0606 and S0609) were polymorphic between p50 and Ekp. We found that S0608 colocalized with the B4C5Ga-derived polymorphic CAPS marker B422251; thus, S0608 was also mapped. The recombination data are shown in Figure 2a. The order of the SSR markers in the linkage group was S0601, S0603, S0606, S0608 and S0609, which is consistent with previously published data (Miao et al., 2005). Four STS markers, Abd-B (represented by its derived marker ab17577), B9B9Ba, B4C5Ga (represented by its derived marker B422251) and B3l10Gb were also mapped. The linkage map was 39.5 cM in length (Figure 2a), which is shorter than that determined for the C108 × (p50 × C108) population (46.5 cM, Figure 2b; Miao et al., 2005). Ekp-1 mapped at 29.2 cM, between the SSR markers S0606 and S0609. The genetic distance between these two markers was 25.9 cM, which is shorter than the distance found using p50- and C108-derived BC1♂ progeny (32.2 cM; Miao et al., 2005). Abd-B (ab17577) mapped at a distance of 2.7 cM from Ekp-1. B4C5Ga (B422251) and B3l10Gb were clustered on the other side, 2.7 cM away. B9B9Ba was clustered with the Ekp-1 locus.

Figure 2
figure 2

Low-resolution mapping of the Ekp-1 locus. The genetic distance between the markers S0601 and S0609 in the p50 × (Ekp × p50) population was 39.5 cM (a), which is shorter than but comparable to that in the C108 × (p50 × C108) population (46.5 cM; (b)) (Miao et al., 2005).

Full size image

High-resolution genetic mapping of the Ekp-1 locus

In the low-resolution genetic map, B422251 and B3l10Gb were clustered (Figure 2a). As the corresponding physical map indicated that B422251 might be 1 Mb nearer to the Ekp-1 locus than B3l10Gb (Yasukochi et al., 2004), we chose B422251 and ab17577 for fine mapping of the Ekp-1 locus. We identified 21 recombinants for ab17577 and 30 for B422251, placing Ekp-1 between the two markers (Figure 3a). These recombinants were used to map our previous and newly developed CAPS markers (B3l10Gb, Ubx3′A, abd-A3′A, abd-A3′B, abd-A5′, B12JaAa, B910325 and B923511). The segregation data were then used to generate a high-resolution genetic map. Markers B3l10Gb, B422251, Ubx3′A, abd-A3′A and abd-A3′B were mapped to one side of the Ekp-1 locus. B3l10Gb and B422251 were clustered, as were abd-A3′A and abd-A3′B. abd-A3′A and abd-A3′B were the closest to Ekp-1, while B3l10Gb and B422251 were the farthest from Ekp-1. The other two markers, ab17577 and B923511, flanked Ekp-1 on the other side (Figure 3a). B9B9Ba (represented by its derived marker B923511), which was initially clustered with Ekp-1, was spaced farther away in the high-resolution genetic map. The markers B910325, abd-A5′ and B12J2Aa, which were previously mapped to the Hox cluster (Yasukochi et al., 2004), were clustered with Ekp-1. These markers spanned a total distance of 2.28 cM, with Ekp-1 in the middle, 0.88 cM from ab17577 and 1.39 cM from G422251; in comparison, a 5.4-cM interval was previously estimated for the same markers in the low-resolution genetic map. The order and distance of some of the markers in the high-resolution genetic map were different from those of the physical map of the Hox gene cluster (Yasukochi et al., 2004); this was especially true for B3l10Gb, B4C5Ga, Ubx and abd-A (abd-A3′A and abd-A3′B).

Figure 3
figure 3

Genetic map and BAC contig of the Ekp-1 locus. (a) High-resolution genetic map of the Ekp-1 region. Genetic distances are given in centiMorgan, and the number of recombinants/number of segregants tested is indicated. Vertical lines denote the position of each marker or gene. The black dot indicates the genetic position of the Ekp-1 locus. (b) BAC contig spanning the Ekp-1 locus. Dotted vertical lines mark the relative positions of the corresponding markers. The Ekp-1-containing BAC insert (BAC083DE9) is highlighted in black.

Full size image

PCR-based BAC assembly to narrow the Ekp-1 region to a single BAC clone

To develop a BAC contig encompassing the Ekp-1 locus, we screened the BAC library using B9B9Ba and B4C5Ga. In total, 11 BAC pools were positive for B9B9Ba, and 20 were positive for B4C5Ga. No pools included both B9B9Ba and B4C5Ga. These positive pools were confirmed using all other markers that were used for fine genetic mapping (Figure 3b). Among the 20 BAC pools that were positive for B4C5Ga, 16 were positive for B3l10Gb, among which two were also positive for Ubx3′A; none was positive for abd-A5′, abd-A3′A or abd-A3′B.

Two B9B9Ba-positive BAC pools and their corresponding clones, 083DE9 and 045CC12, were also positive for the markers closely linked to and clustered with the Ekp-1 locus (Figure 3b). In particular, 083DE9 was positive for abd-A3′A and abd-A3′B. The end sequence of this clone, determined using the primer Sp6, indicated that the left end of the BAC is located 26 kb downstream of the abd-A 3′ ORF within Scaffold000075; thus, 083DE9 harbors the region corresponding to Ekp-1. The insert size, which was determined by CHEF electrophoresis after digestion with NotI, was about 220 kb. The BAC-end-derived marker 083DE9end, which was developed from the BAC clone BAC083DE9, was used to confirm the overlap between 083DE9 and 045CC12 (Figure 3b).

Candidate gene analysis

Among the silkworm Hox genes, only Bm abd-A is located in the interval between the markers flanking the Ekp-1 locus (Figure 4). We detected T-C substitutions at four sites within the abd-A cDNA from Ekp compared to that from p50, but they did not cause amino-acid changes and thus were considered to be single nucleotide polymorphisms. The Bm abd-A ORF contains three exons (Figure 4). The four single nucleotide polymorphisms were all in exon 3. Two recombination events were detected about 9 kb downstream, indicating that the Ekp-1 locus is in the upstream direction. No recombinants were detected in intron 2 or the 5′ flanking region. The coding sequences of Ubx and abd-A from the p50 and Ekp strains have been deposited in GenBank (accession numbers EU365396–365399).

Figure 4
figure 4

Structure of the Bm abd-A gene. Open boxes indicate the exons. The polymorphic markers are indicated on top of the vertical lines. Scaffold000405 contains exons 1 and 2, and scaffold000075 contains exon 3. There are four single nucleotide polymorphisms in exon 3. The gray bar indicates BAC083DE9, which contains the Bm abd-A gene.

Full size image

Bm iab-4 is clustered with the Ekp-1 locus

In the Ekp-1 region, we also found a predicted non-coding locus, Bm iab-4. Since no polymorphic markers were developed from iab-4, we were unable to map this locus to the genetic map. On the basis of the physical map, however, the locus is between the markers abd-A5′ and B910325, which are clustered with the Ekp-1 locus. The iab-4 locus was also found to be clustered with the Ekp-1 locus. The sequence of the fragment containing putative iab-4 was deposited in GenBank under accession number EU365400.

Discussion

The location of a gene on a genetic linkage map provides a starting point for map-based cloning. Advances in the silkworm DNA molecular map and genomic sequencing (Xia et al., 2004; Miao et al., 2005) have made it possible to identify the chromosomal location of a gene and to quickly screen a large number of molecular markers and identify those closely linked to the target gene for further fine mapping.

The Ekp-1 locus was previously placed on the sixth chromosome (Miao et al., 2005). In the silkworm, owing to a lack of crossing over in females, BC1 progeny are usually used for linkage analysis (Miao et al., 2005; Li et al., 2006). The genetic distance of the linkage map constructed here was smaller than that constructed by Miao et al. (2005), which affected the genetic distances of the SSR markers flanking the Ekp-1 locus. This indicates a difference in the number of recombination events between the two pairs of parents (Ganal and Tanksley, 1996; Chetelat et al., 2000). The colocalization of S0608 and the STS marker B4C5Ga allowed us to link the map of Ekp-1 with the physical map of the Hox gene cluster (Yasukochi et al., 2004); thus, we were able to finely map the Ekp-1 locus in a relatively short time. With regard to the physical map produced by Yasukochi et al. (2004), Ubx and abd-A were the only Hox genes in the region, so they were the first candidate genes to be considered. Cloning and analysis of these genes allowed us to create new closely linked markers for fine mapping.

A large-insert silkworm BAC library, derived from the p50 and C108 strains, has been established that can be used for physical mapping of a silkworm gene or locus of interest (Wu et al., 1999). PCR analysis confirmed that the marker B4C5Ga, located downstream of the Ubx 3′ UTR, was close to the marker B3l10Gb. This is quite different from the previously published position of B4C5Ga, which was mapped between B12J2Aa and B9B9Ba in the Hox gene cluster (Yasukochi et al., 2004).

The physical and genetic mapping data for Ekp-1 suggest that the region shows reduced recombination. The genetic distance between abd-A3′A and B923511, the innermost loci that flank the target, was 0.13 cM; however, the physical size of this region is estimated to be less than 220 kb. Therefore, the average physical/genetic (P/G) distance ratio of the target interval is about 1.24 Mb cM−1. This is much higher than the estimated average ratio for the entire silkworm genome (about 0.25 Mb cM−1; Goldsmith et al., 2005). This phenomenon has also been reported in plant studies. For example, the rice blast resistance locus Pib (Wang et al., 1999) showed a ratio of 1.34 Mb cM−1, while in the rice lesion mimic gene Spl11, the ratio was even higher (2.46 Mb cM−1; Zeng et al., 2002). In plants, the offspring used for mapping, which may include the F2, F6 or even F7 generation, might be one reason for the apparent reduction in recombination (Zeng et al., 2002); however, that does not apply in this study. Recombination suppression can also be caused by a high degree of sequence divergence between the two parents (Ganal and Tanksley, 1996; Chetelat et al., 2000). p50 and Ekp are two strains of the species B. mori and thus can mate normally, so there is no high degree of sequence divergence between them. Another reason for recombination suppression is the proximity of the locus to the centromere (van Daelen et al., 1993; Haupt et al., 2001). Some insects, such as the Hemiptera and Lepidoptera, which include the silkworm, are believed to have diffuse centromeres (Murakami and Imai, 1974; Pimpinelli and Goday, 1989), resulting in holocentric chromosomes. Chromosomes with full-length kinetochores are not sensitive to breakage (Okazaki et al., 1993); thus, recombination is not easy to occur. Whether Ekp-1 is near a kinetochore region remains unknown. Fluorescence in situ hybridization using marker-specific probes as well as heterochromatic centromere probes may be used to test this possibility. Moreover, fine mapping of the genomic sequence in the Ekp-1 region will provide a powerful tool with which to predict the possible heterochromatic zones therein.

Our physical mapping data pinpointed Ekp-1 to a region less than 220 kb in length and containing an unknown number of genes. However, there is at least one gene, abd-A, and a non-coding locus, iab-4, for microRNA (Tong et al., 2006) in this region. By comparing the ORF of the abd-A gene from Ekp to that from p50, we found mutations in exon 3, but they were synonymous. Thus, these mutations cannot be the cause of the mutant phenotype. Moreover, because we did not explore the UTR region, the promoter, or other upstream regulatory elements, abd-A cannot be excluded as a candidate. In Drosophila, abd-A represses limb development in the abdomen (Vachon et al., 1992) by preventing expression of the limb-promoting gene Dll. However, in Lepidoptera the effect of abd-A on Dll is unclear. In the larvae of the butterfly Junonia coenia, abd-A is absent from areas of Dll accumulation in abdominal segments A3–A6 (that is, where larval prolegs form). The otherwise uniform domain of Hox gene expression in this species shows ‘holes’ where Dll accumulates (Warren et al., 1994). These holes imply that the Ubx/abd-A proteins in Lepidoptera have repressive effects on Dll (Warren et al., 1994; Suzuki and Palopoli, 2001). Zheng et al. (1999) have addressed the possible difference between Ubx and abd-A in regulating Dll. In the silkworm, the ECa/ECa mutant embryo, which has a deletion near the homeobox region of abd-A, does not develop prolegs (Ueno et al., 1992). One possibility is that Ubx represses proleg development, whereas abd-A may have a distinct regulatory role and might have associations to promote proleg development. Using RNAi to prevent Hox gene expression, Xiang (2008) also found that Ubx but not abd-A repressed ventral limb development. Here, we localized Ekp-1 to the abd-A region. Detailed analysis of this gene will hopefully allow the identification of the causal mutation.

Studies in Drosophila have shown that in this homeotic gene complex, many different classes of cis-regulatory elements, such as insulators, silencers, enhancers and promoters, work together to control gene expression during development (Sanchez-Herrero, 1991; Drewell et al., 2002), and they have begun to unravel the intricate nature of these regulatory interactions, including the suggested activation of cis-sequences at the complex under the control of non-genic transcription, providing an additional level of regulatory complexity to regions of non-protein coding DNA (Akbari et al., 2006; Chopra and Mishra, 2006). In particular, a recent study showed that in Drosophila, microRNA transcribed from the iab-4 locus regulates Ubx and can result in transformation of the haltere into a wing (Ronshaugen et al., 2005). In the present study, the putative Bm iab-4 locus was in the Ekp-1 region, although no difference in sequence was found between p50 and Ekp at the iab-4 locus. This might be due to the very limited number of sequences compared between the two strains in this region. As it is only a predicted locus, further research is needed to isolate Bm iab-4 and to characterize its functions. In addition, the abd-A region in silkworm is larger than that in Drosophila, so this region, which is about 220 kb in length, may have diverse regulatory elements such as cis-regulatory elements and non-coding loci. In the near future, the sequencing of the silkworm genome will be completed, allowing for more informative annotations. This may help identify the Ekp-1 locus and further our understanding of the mechanism of proleg development in Lepidoptera. Further, cloning of Ekp-1 may reveal key elements that respond to proleg development in Lepidoptera, such as non-coding loci.