Organization of the Hox gene cluster of the silkworm, Bombyx mori: a split of the Hox cluster in a non-Drosophila insect (original) (raw)

Introduction

The Hox genes, which encode homeodomain transcription factors, have been analyzed for a wide variety of animal species. They determine developmental fate along the anterior-posterior (A-P) axis of the embryo and are located along the chromosome in the same order as their functional domains along the A-P axis. Understanding the organization, function, and expression profiles of the Hox genes is important for elucidating the evolution of the Arthropoda (for recent review, see Hughes and Kaufman 2002a).

Insect Hox genes were first analyzed in Drosophila melanogaster, where they are split into two clusters: the Antennapedia complex [(ANT-C) Kaufman et al. 1990], which includes labial (lab), proboscipedia (pb), Deformed (Dfd), Sex comb reduced (Scr), and Antennapedia (Antp), and the Bithorax complex (Bender et al. 1983), which includes Ultrabithorax (Ubx), abdominal-A (abd-A) and Abdominal-B (Abd-B). Additional Hox genes, zerknült (zen), zerknült2 (z2), bicoid, and fushi tarazu (ftz), are present in the Drosophila ANT-C. These four genes are thought to have lost their original roles and gained novel functions in the developmental process, in contrast with the more typical eight Hox genes described above, whose original roles have been conserved among different animal phyla. Ubx, abd-A, and ftz have not yet been identified outside the onychophora/arthropod clade, suggesting these Hox genes have evolved only in the clade (Grenier et al. 1997).

In the red flour beetle, Tribolium castaneum, the sequence of 280 kb of the Hox cluster revealed that this portion contains Tclab, maxillopedia [(mxp) the pb orthologue], Tczen2, Tczen, TcDfd, Cephalothorax [(_Cx_] the Scr orthologue), Tcftz, and prothoraxless [(ptl) the Antp orthologue] (Brown et al. 2002). In the grasshopper, Schistocerca gregaria, Sgzen, SgDax (the ftz orthologue), Scr, abd-A, and Abd-B genes were shown to be present in a single cluster by fluorescence in situ hybridization (FISH) and PFGE analysis (Ferrier and Akam 1996). The Hox genes in the malaria vector mosquito, Anopheles gambiae, were also reported to be organized in a single cluster (Powers et al. 2000; Devenport et al. 2000). Splits at different locations in the Hox cluster were reported in other Drosophila species (Von Allmen et al. 1996; Lewis et al. 2003; Negre et al. 2003). Therefore, splits in the Hox cluster are thought to be limited to the Drosophila lineage.

In the lepidopteran insects, orthologues were previously reported in the silkworm, Bombyx mori, for Dfd, Scr (Kokubo et al. 1997), Antp (Nagata et al. 1996), Ubx, abd-A, and Abd-B (Ueno et al. 1992). Orthologues of Antp, Ubx (Nagy et al. 1991), and abd-A (Zhen et al. 1999) were also reported for the tobacco hornworm, Manduca sexta. However, there have been no reports on lab, pb, zen, or ftz, and the organization of the Hox cluster in a lepidopteran is not known. Here, using data obtained from linkage analysis, construction of bacterial artificial chromosome (BAC) contigs, and FISH analysis, we provide evidence that the Hox gene cluster in B. mori is divided not between Antp and Ubx, but between lab and the remaining Hox genes. All Bombyx Hox cluster genes are located on chromosome 6; however, the lab gene is far from the rest of the cluster.

Materials and methods

Amplification of genomic fragments coding for the homeodomain

Genomic fragments coding for the homeobox of the Bombyx homologues of the lab and pb genes were amplified from genomic DNA by nested PCR experiments with the use of degenerate primer pairs GGNMGNACNAAYTTYACNAA/ACNCYTTYTTYTGYTTCAT for lab for the first reaction, ACNGCNTAYACNAAYAAYACNCA/GTYTGNCKYTTRTGYTTCAT for pb for the first reaction), and GARYTNGARAARGARTTYCA/CATNCKNCKRTTYTGRAACCA for the second reaction. Amplified fragments were cloned, and nucleotide sequences were determined by standard methods with an ABI-377 automated sequencer (Applied Biosystems).

Chromosome walking

A BAC library (Wu et al. 1999) constructed from strain p50, with an average insert size of 134.5 kb, was used for the PCR screening. The detailed method of PCR screening is described elsewhere (Yasukochi 2002). First, PCR screening was performed using sequence-tagged sites (STSs) designed from already known Hox genes and the lab gene newly identified in these experiments. BAC DNA was prepared from the positive clones with an automatic plasmid isolator (Kurabo PI 50), digested with _Hin_dIII and inserted into pUC19. The resultant subclones were sequenced to design STSs. Direct sequencing of BAC clones was also performed for the same purpose. STSs used in chromosome walking are shown in Table 1.

Table 1 Sequence-tagged site (STS) primers used in this experiment

Full size table

Linkage analysis

STSs polymorphic between silkworm strains p50 and C108 (Table 1) were used for linkage analysis on the same population of 166 F2 individuals in the same manner as described previously (Yasukochi 1998, 1999).

Fingerprinting assay of BAC clones

The BAC DNA was digested with _Hin_dIII and fractionated in 1% agarose gels at 16°C. The agarose gels were stained with SYBR Green I, and banding patterns of _Hin_dIII fragments were scanned with an image analyzer (Bio-Rad Molecular Imager FX). Estimation of the size of BAC contigs was performed with FPC, version 4.7, physical mapping software (http://www.sanger.ac.uk/Software/fpc/, Soderlund et al. 2000).

FISH analysis

Chromosomes were prepared from p50 females according to the methods described elsewhere (Sahara et al. 1999, 2003b), with slight modifications. We used the p50 ovaries of 12- to 36-h-old fifth-instar larvae and dissected in _Ephestia_’s saline (see Lockwood 1961). The ovaries were pretreated in hypotonic solution (85 mM KCl and 15 mM NaCl) before fixation with Carnoy’s fluid (6:3:1 ethanol:chloroform:acetic acid). Cells were dissociated in 60% acetic acid and spread on glass microscope slides placed on a heating plate at 50–52°C according to Traut (1976). After passage through an ethanol series (70%, 80%, 98%), the preparations were stored at −30°C until use.

The BAC FISH was carried out essentially according to the methods described in Sahara et al. (2003b). The DNA from the BACs covering the chromosome regions of interest was extracted with a Plasmid Midi Kit (Qiagen) after overnight culture. For probes, DNAs were labeled with Cy3-dCTP or FluorX-dCTP (Amersham), using a Nick Translation System (Invitrogen). Air-dried chromosome preparations, passed through the ethanol series after removal from the freezer, were treated with 5× Denhardt’s solution at 37°C for 30 min. The preparations were denatured at 72°C for 3.5 min in 70% formamide, 2× SSC. The probe cocktail for one BAC FISH contained either 250 ng or 500 ng labeled BAC DNA, 25 μg sonicated salmon sperm DNA (Sigma–Aldrich), and 20 μg sonicated Bombyx male genomic DNA as competitors in 10 μl hybridization solution (50% formamide, 10% dextran sulfate, 2× SSC).

Hybridization was performed in a moist chamber at 37°C for 3 days. After post-hybridization washing at 62°C in 0.1× SSC, 1% Triton X-100, slides were counterstained and mounted with antifade [0.233 g 1,4-diazabicyclo(2.2.2)-otane, 1 ml Tris-HCl, pH 8.0, 9 ml glycerol] containing 0.5 μg/ml of DAPI (4′6-diamidino-2-phenylindole, Sigma–Aldrich).

Black-and-white digital photos were taken with a Leica HC fluorescence microscope with a Photometrics CoolSNAP CCD camera through the A, L5, and N2.1 filters, and image processing was carried out using Adobe Photoshop, version 6. Routinely, red color was used for Cy3, green for FluorX, and light blue for DAPI images.

Results

PCR amplification and sequence determination of genomic fragments coding for homeoboxes of lab and pb

In the absence of reports on lab and pb, isolation of these two genes was attempted first. As a result of nested PCR with degenerate primer pairs designed from conserved amino acid sequences, small PCR fragments of the expected size were amplified for both genes. Nucleotide sequence determination confirmed that these fragments were highly similar to previously reported lab and pb genes (Fig. 1).

Fig. 1

figure 1

Sequence comparison of homeodomains of Bmlab (AB120761) and Hox 2-like genes (AB120762) to known insect Hox class 1 and 2 genes. Tclab and TcMxp, Tclab (AF231104) and mxp (AF187068) of Tribolium castaneum; Dmlab and Dmpb, lab (X13103) and pb (X13103) of Drosophila melanogaster; and Aglab, lab of Anopheles gambiae (AF269153). Accession numbers are in parentheses

Full size image

PCR screening of a BAC library was performed to isolate clones containing these two fragments. A number of positive clones were isolated for the Bombyx homologue of lab, but none were isolated for that of pb because of the difficulty of designing adequate primers from the short sequence that was determined. Sequences adjacent to the first cloned fragment of the putative lab gene were determined, which also showed significant similarity to the lab genes of D. melanogaster, T. castaneum, and A. gambiae (Fig. 1).

Linkage analysis

Isolation of BAC clones containing the Antp, Ubx, abd-A, and Abd-B genes was previously reported (Wu et al. 1999). In addition, clones containing the Dfd and Scr genes were isolated from the BAC library and utilized for linkage analysis.

Linkage analysis was performed by PCR, using STS primers designed from sequences of _Hin_dIII fragments of BAC clones (Table 1). Hox genes Dfd, Scr, Ubx, abd-A, and Abd-B were found to be closely linked to the previously reported Antp locus (Yasukochi 1999) on linkage group (LG) 6 (Fig. 2a), as expected. Although the lab gene was also mapped on LG 6, the locus was far from the other Hox genes (Fig. 2a).

Fig. 2

figure 2

a Linkage map of linkage group (LG) 6 of Bombys mori. Twin bars represent LGs of C108-dominant markers (left) and p50-dominant markers (right) (Yasukochi 1998). b Overlapping of bacterial artificial chromosome (BAC) clones covering the homeobox gene cluster of B. mori. c Schematic representation of overlapping of BAC clones around the Bmlab gene. Horizontal bars represent BAC clones. Vertical lines represent DNA markers listed in Table 1. Underlined sequence-tagged sites were polymorphic and used for linkage analysis. Open circles indicate that a BAC is positive for each DNA marker. Arrows indicate the range of BAC subgroups used for estimation of physical distance by fingerprint analysis, (1G6G, 2A5C, 3A5B, 3E12A, 4C5G, 4D1A, 5E8G, 6J9F, 7B7F, 8A12C, 9B9B, 9D12D, 10B2F, 12D4D, 12G3C, 13K12B, 13L7H, 14F4A, 16E8A, 17C6E, 20B1E), (11J4A, 14C7G), (2L6A, 3I10G, 3L10G, 4B6C, 4C8H, 5F7H, 8A6G, 8H1G, 8I9A, 9K9E, 12J4H, 13C7F, 13D3D, 14G3G, 14G12F, 15D6G, 15K2C, 16E4E, 18A7G, 18A9E, 18L5F, 19D4H, 19E2H), and 1C10F

Full size image

Identification of the Bombyx Hox 3/zen gene

Since amino acid sequences of the homeobox of insect Hox 3/zen genes are not as well conserved as other Hox genes, we did not try to isolate Bombyx Hox 3/zen genes by nested PCR with degenerate primer pairs. Instead, chromosome walking was performed from the 3′ region of the Dfd gene. The end sequence of BAC clone 19H9H showed significant similarity to Hox 3/zen genes of insects and other arthropods (Fig. 3). Although the sequence also showed similarity to the even-skipped genes of other arthropods (Fig. 3), it was not identical to the reported Bombyx even-skipped gene (Xu et al. 1997). In addition, the Bm even-skipped gene was mapped to LG 2 (unpublished results), which made less likely the possibility that it was a derivative of the Bm even-skipped gene generated by gene duplication. Judging from its similarity and localization downstream of the Dfd gene, we designated the sequence as Bmzen.

Fig. 3

figure 3

An alignment of homeodomains of Bmzen (AB120763) and other arthropod Hox 3 and even-skipped genes. Deduced amino acid sequence of the partial Bmzen homeodomain (42 residues) were compared. Sequence identity indicates the similarity between each of the Hox 3 and even-skipped genes and Bmzen. Alhox3, Hox3 of a mite, Archegozetes longisetosus (AF085352); Bmeve, even-skipped (D38486) of B. mori; Cazen, Elzen and Hpzen, zen of Clogmia albipunctata (AJ419659), Empis livida (AJ419661) and Haematopota pulvialis (AJ419660); Cshox3, Hox3 of a spider, Cupiennius salei; Dmbcd, Dmeve, Dmzen, and Dmzen2, bicoid (P09081), even-skipped (M14767), zen (P09089), and zen2 (P09090) of D. melanogaster; Mabcd and Mazen, bicoid (AJ133024) and zen (AJ133025) of Megaselia abdita; Sgzen, zen of S. gregaria (X92654); Tczen1 and Tczen2, Tczen1 (X97819) and Tczen2 (AF452568) of T. castaneum; Lahox3, Hox3 of a centipede, Lithobius atokinsoni (AF435001). Accession numbers are in parentheses

Full size image

Construction of a BAC contig of the homeobox gene cluster

Using BAC clones containing Hox genes as starting points, chromosome walking was performed until gaps among Hox genes were closed. As a result, a completely overlapping BAC contig was constructed showing that the order of the Hox genes from the Bmzen to Abd-B genes was conserved, as seen in other animals (Fig. 2b). Duplicated primer pairs designed from 5′ and 3′ portions of the Scr and Antp genes showed that the transcriptional orientation of the genes was from Abd-B to Bmzen (Fig. 2b). Sequence determination also revealed that the Bmzen gene was transcribed in the same orientation because the 3′ end of the homeodomain was consistent with the 3′ end of clone 19H9H.

The size of the BAC contig was estimated by fingerprint analysis of _Hin_dIII fragments of BAC clones. Since the redundancy of BAC clones was not high enough to detect all the overlaps, we could not estimate the full, uninterrupted length of the BAC contig. Alternatively, the sizes of non-overlapping four subgroups were estimated (Fig. 2b). The total of the estimates was 2.32 Mb; however, the actual size of the contig was larger than 2.32 Mb because of considerable gaps between subgroups.

A BAC contig covering the chromosomal region around Bmlab was also constructed (Fig. 2c). During the process of chromosome walking, two already known genes, the attacin gene coding for an antibacterial protein (accession No. D76418) and the VAP-peptide gene coding for a kind of peptide hormone (accession No. AB001053), were found in this contig. However, no other detectable Hox genes were found.

FISH analysis confirmed the separation of lab from other Hox genes

To confirm the results obtained from linkage analysis and BAC contig construction, FISH analysis was carried out using BAC clones covering the chromosome regions of interest as probes. Two-color FISH analysis revealed that the lab gene was located on one end of putative chromosome 6, whereas the 3′-most BAC clone from the Hox contig was located two thirds of the way across the same chromosome, as expected (Fig. 4). Continuous blocks brightly stained by DAPI, probably heterochromatinization of the chromosome, were observed between the two loci (Fig. 4a′–e′). These heterochromatin blocks are continuously present on the sixth chromosome bivalents from the early (longer chromosomes) to late (shorter) pachytene stages. To determine the orientation of the Hox gene cluster, 5′ end BAC clones of the Hox contig were labeled with the same dye (Cy3, red color) as that of the lab gene. As a result, an additional red hybridization signal from the 5′ end of the Hox contig appeared between the red signal from the lab clone and the green signal from 3′ end of the Hox contig, in good agreement with the results obtained from linkage analysis (Fig. 5). The order of the three hybridization signals, red-red-green, never changed irrespective of the degree of condensation of the sixth chromosome bivalents.

Fig. 4

figure 4

BAC fluorescence in situ hybridization (FISH) images on an oocyte pachytene complement showing location of Hox genes. A pachytene complement (a), four pachytene bivalents in various stages (b–e), and gray-colored DAPI images of the bivalents (a′_–_e′) from the p50 strain of B. mori. Chromosomes were counter-stained with DAPI (blue). Red and green represent hybridization signals from 10L5C (lab) and 17H8F (3′-Hox) probes, respectively. Note that there are brightly DAPI-stained regions, considered heterochromatin blocks, between the signals (a′_–_e′). #6 Sixth chromosome bivalent, N nucleolus, bar represents 10 μm

Full size image

Fig. 5

figure 5

BAC FISH images on the sixth chromosome in oocyte pachytene spreads showing relative organization of lab and Hox gene clusters. A pachytene complement (a) and seven bivalents in various stages (b–h) from the p50 strain of B. mori. Chromosomes were counter-stained with DAPI (blue). Red end signals were from 10L5C probes (lab), and internal signals were from 3A5B probes (5′-Hox). Green signals represent hybridization from 17H8F probes (3′-Hox). Note that the signal order (red-red-green) never changes from the earlier (longer chromosomes) to later (shorter) pachytene stages

Full size image

The distance between the two ends of the contig seemed to be about one tenth of the chromosome length. The haploid genome size of B. mori (_n_=28) is estimated to be 530 Mb (Gage 1974), indicating that the average size of a chromosome is approximately 18.9 Mb. Since chromosome 6 is medium size (Figs. 4a, 5a), these observations were consistent with the estimation that the size of the contig obtained by BAC fingerprinting is 2–3 Mb.

Discussion

The E locus in B. mori represents a number of dominant homeotic mutations that cause distinctive phenotypes, such as addition of dorsal markings and extra legs in the abdominal segments. The E locus is closely linked to the Nc locus, another homeotic mutation mapped on LG 6 (Hasimoto 1941; Itikawa 1952). Earlier studies have attempted to characterize the organization of the Hox gene cluster using such mutants. The 3′ portion of the Bombyx Antp gene was shown to be deleted in the Nc mutant (Nagata et al. 1996), and homeodomains of the Bm Ubx and abd-A genes were not detected in homozygous E N/E N embryos (Ueno et al. 1992). Therefore, Bm Antp, Ubx, and abd-A genes were thought to be clustered (Nagata et al. 1996). Although Bm Dfd, Scr (Kokubo et al. 1997), and Abd-B (Ueno et al. 1992) were also cloned, there was no direct evidence that Bombyx Hox genes are clustered in the same order as in other organisms.

In this report, a BAC contig was constructed containing Bmzen, Dfd, Scr, Antp, Ubx, abd-A, and Abd-B genes. The order of Hox genes was consistent with other animal species. The distance between Bmzen and Abd-B genes was estimated to be less than 2.82 Mb, calculated by BAC fingerprinting analysis from BAC clones 20H9G to 3A5B (Fig. 2b). The size of insect Hox gene clusters is generally larger than that of vertebrate, which ranges from 100 kb to 120 kb (Ferrier and Akam 1996). The total size of Drosophila ANT-C and BX-C is approximately 700 kb (Adams et al. 2000), and the ANT-C portion of the homeotic complex in Tribolium is almost as large as the fruit fly ANT-C (Brown et al. 2002). The distance between Sgzen and Abd-B in the Hox gene cluster in the grasshopper, S. gregaria, was reported to be more than 700 kb, but no longer than 2 Mb (Ferrier and Akam 1996). Thus, the size of Bombyx Hox cluster is about four times as large as that of Drosophila and probably larger than that of S. gregaria. The haploid genome size of B. mori is 3.1 times as large as that of Drosophila, suggesting that the difference in size might simply reflect the extent of expansion around the whole genome.

Larger intervals among Hox genes are disadvantageous for sequence determination, but advantageous for linkage analysis. Genetic recombination has been observed between several alleles of the E locus, and alleles causing mutations in the anterior segments are more closely linked to the Nc locus than alleles causing mutations in the posterior segments (M. Hirokawa, unpublished results), suggesting that the E locus is defined by a heterogeneous set of mutations in Ubx, abd-A, Abd-B, and the intergenic regions between them. We are now executing a detailed linkage analysis of several alleles of the E locus, which will facilitate the understanding of the molecular mechanisms in one of the oldest known groups of homeotic mutations (Hasimoto 1930).

The finding that the lab gene was a long way from the Hox gene cluster is intriguing. It is likely that an intra-chromosomal inversion or translocation occurred in the Bombyx lineage. Relocation of the lab gene was recently reported in the species of the Drosophila repleta group (Von Allmen et al. 1996; Negre et al. 2003). Splitting of the Hox gene cluster might not be such a rare event, since chromosomal locations of all Hox genes have not been thoroughly identified in most insect species. In the most extensively analyzed Drosophila species, three types of splits (between lab and pb, between Antp and Ubx, and between Ubx and abd-A) were already known (Negre et al. 2003). This question might be addressed by performing comparative FISH analysis using large-insert DNA clones containing Hox genes as probes to reveal the organization of the Hox gene clusters in various insect lineages.

FISH analysis also revealed that the region upstream of the major Hox gene cluster in B. mori was brightly stained by DAPI (Fig. 4) for several megabases, indicating that the G+C content in this region is low, which is a typical characteristic of heterochromatin. This finding is in good agreement with the large interval between polymorphic markers in this region (i.e., between 5′ end of the Hox gene cluster and RAPD anchor markers, R0411X and R105132X, Fig. 2a), since the large heterochromatin block probably consists of repetitive sequences (Sahara et al. 2003a). If the large heterochromatin block really exists in the upstream region of the Hox gene cluster, it might function to minimize effects of transcription from other genes.

Hox 3 genes in the higher insects, called zen or bicoid, have evolved rapidly from an ancestral gene, accompanied by gene duplication and changes in function (Falciani et al. 1996; Stauber et al. 1999), whereas canonical Hox 3 function is maintained in basal arthropods (Telford and Thomas 1998; Hughes and Kaufman 2002b). Bmzen seems to be relatively similar to zen of Clogmia albipunctata, a lower dipteran in which no _bicoid_-type gene has been observed (Stauber et al. 2002), and not similar to bicoid which is a recent duplication of zen in the Drosophila lineage. Bmzen has evolved faster compared with basal arthropods, Schistocerca and Tribolium, and not so fast as bicoid genes of Diptera (Fig. 3). Since there are some differences in the expression pattern of Hox 3 genes among Drosophila, Schistocerca, and Tribolium, further analysis is needed to characterize its expression pattern in B. mori.

In this report, we could not identify the chromosomal location of the Hox 2 gene, which was critical to confirm whether the separation of the Hox gene cluster occurs between the lab and Hox2 gene, the Hox2 and Bmzen gene, or both. Nested PCR was also attempted to amplify a Bombyx orthologue of the ftz gene on genomic DNA or BAC DNA pools, but all of amplified fragments were found to have been amplified from the BmAntp gene (data not shown). Since ftz has been identified in a wide variety of arthropods and Onychophora (Grenier et al. 1997), it is unlikely that there was no ftz orthologue in the Bombyx ancestor. It is possible to speculate that the ftz gene was lost in the Bombyx lineage. Alternatively, it may have quickly evolved because of changes in function, and sequence similarity is no longer high enough to amplify fragments by nested PCR under the conditions used in this report.

Further analysis is needed to characterize the complete organization of the Bombyx Hox cluster. Sequence determination of the cluster is now underway, which will reveal the presence or absence of ftz and additional copies of the Bombyx Hox 3 gene, and conserved sequence motives among dipteran, coleopteran and lepidopteran insects, including _cis_-acting elements necessary for proper expression of individual Hox genes.

References

Download references