Animals and tissues

European honeybee (Apis mellifera L.) colonies were purchased from the Kumagaya bee farm (Saitama, Japan) and maintained at the University of Tokyo (Tokyo, Japan). Nurse bees were collected when they were feeding brood, and foragers were collected when they returned to the colony after foraging pollen and honey (Kubo et al., 1996). After the workers were anesthetized on ice, the heads were removed and the HPGs were dissected from them with fine tweezers and a surgical knife under a binocular microscope. Nurse bees with well-developed HPGs and foragers with shrunken HPGs were used in experiments, as described previously (Ohashi et al. 1999). Tissues to be used for extraction of RNA were stored frozen at -80°C until use.

Differential display

Total RNA was extracted from HPGs using TRIZol Reagent (Invitrogen, Carlsbad, CA), treated with DNase I, and reverse-transcribed by using SuperScript III (Invitrogen) with rhodamine-labeled downstream primer 1 (TaKaRa, Tokyo, Japan). Differential display was performed using a Fluorescence Differential Display Kit, Rhodamine Version (TaKaRa) and LA Taq (TaKaRa) with a combination of 24 primers, as described previously (Ito et al., 1994; Kamikouchi et al., 1998). PCR conditions were (94°C × 2 min + 40°C × 5 min + 72°C × 5 min) × 1 cycle + (94°C × 30 sec + 40°C × 2 min + 72°C × 1 min) × 34 cycles) + 72°C × 5 min. To ensure the reproducibility of the differential display profiles, duplicate reactions with two lots of RNA obtained from two honeybee colonies were performed.

Bands whose signal intensities differed between nurse bees and foragers in the two lots of RNA samples were selected as candidate bands. Bands whose intensities differed between nurse bees and foragers were excised from gels, and the DNA in the bands was reamplified by PCR. The resulting PCR products were cloned into pGEM-T easy vector (Promega, Madison, WI) and transfected into Escherichia coli DH5α competent cells (TaKaRa). The cDNA sequences were determined using an ABI PRISM 3100 Genetic Analyzer with the BigDye Terminator v3.1 Cycle Sequencing Kit. Similarity searches of cDNA sequences were performed by using NCBI Honey Bee Genome Resources.

Cloning of full-length open reading frame sequences

RT-PCR was performed using Ex Taq (TaKaRa), with cDNA derived from the HPG as the template. Primers (Ambuffy, 5′-GAAATGAATCGTTCGTTGTAG-3′ and 5′-CAAGATGAAATATTTA-ATGAATCATTTATGGAGG-3′; AmMMP1, 5′-GACGATCTACGG-GAACAC-3′ and 5′-GATGACACAATATTGCATGCGAC-3′) were designed on the basis of sequences in the 5′ and 3′ untranslated regions, obtained from the NCBI Honey Bee Genome Resources. Cloning and sequencing were performed as described above. Domain and motifs in deduced amino acid sequences were predicted by using the Pfam database (GenomeNet Database Resource:  http://motif.genome.jp/), SignalP (The Center for Biological Sequence Analysis at the Technical University of Denmark:  http://www.cbs.dtu.dk/services/SignalP/), and information previously reported by Llano et al. (2000) and lgaki and Miura (2004). The Cterminal hydrophobic membrane anchor (MA) was predicted by using TMHMM (The Center for Biological Sequence Analysis at the Technical University of Denmark:  http://www.cbs.dtu.dk/services/TMHMM/). The amino acid sequences of Ambuffy and AmMMP1 were aligned by using ClustalX (Thompson et al., 1997).

Quantitative RT-PCR

To investigate tissue-preferential gene expression, we simply dissected worker bodies into HPGs, the head except for the HPGs, thorax, midgut, and abdomen except for the midgut. We dissected the midgut from the abdomen, because the midgut contained much food pollen, which might have disturbed the RNA extraction procedure. We washed the pollen from the midgut and used the remaining midgut also for RNA extraction. Total RNA was extracted from these tissues with TRIZoI Reagent (Invitrogen), treated with DNase I, and reverse-transcribed by using SuperScriptlll (Invitrogen) with the oligo dT primer. Real-time PCR was performed with SYBR premix Ex Taq II (TaKaRa) and LightCycler (Roche, Basel, Switzerland) according to manufacturers' protocols, using genespecific primers. The amount of transcript was normalized with that of elongation factor 1α-F2 (EF1α-F2; Danforth and Ji, 1998) or ribosomal protein 49 (rp49) (Ben-Shahar et al., 2003). We also examined the expression of mrjp2 and α-glucosidase as reference genes that are expressed preferentially in nurse bee and forager HPGs, respectively, which we previously showed by Northern blotting analysis (Ohashi et al., 1997).

Gene-specific primers (Ambuffy, 5′-CATGGCACTTCTCATC-CTTTTC-3′ and 5′-GAGAACGGTTTCAGCATCAATC-3′; AmMMP1, 5′-GCTTCCCGATAATCTTGATG-3′ and 5′-CATCCGAACCACCAG-TAAG-3′; MRJP2, 5′-AAATGGTCGCTCAAAATGACAGA-3′, and 5′-ATTCATCCTTTACAGGTTTGTTGC-3′; α-glucosidase, 5′-TACCTG-GCTTCGTGTCAAC-3′ and 5′-ATCTTCGGTTTCCCTAGAGAATG-3′; EF1 α-F2, 5′-CATCAAAAACATGATTACTGGTACCTC-3′ and 5′-CAGAATACGGTGGTTCAGTGG-3′; and rp49, 5′-AGAAACTGGCG-TAAACCTAAAG-3′ and 5′-GTTCCTTGACATTATGTACCAAAAC-3′) were derived from cDNA sequences for each gene and information from Honey Bee Genome Resources. PCR conditions were: (95°C × 30 s) + (95°C × 5 sec + 60°C × 15 s + 72°C × 20 sec) 45 cycles.

Screening of nurse bee and forager HPG-preferentlal genes

To search for genes whose expression in the HPGs differs between nurse bees and foragers, we used the differential display method. By screening approximately 900 bands, 48 candidate bands were obtained (23 nurse beeselective and 25 forager-selective bands). Among these, 14 bands were selected based on the extent of differences in intensity between nurse bees and foragers in the differential display images. Sequence analysis assigned these bands to seven genes: MRJP1, 3 (Ohashi et al, 1997), and 7 (Stefan and Jaroslav, 2004), α-glucosidase (Ohashi et al, 1996), and three other clones (1 to 3). In the present study, we focused on clones 1 and 2, as we could not detect significant expression of clone 3 (data not shown).

A database search suggested that clone 1 corresponded to a part of a predicted gene, GB18455, which is located in linkage group 3 of the honeybee genome and encodes a Bcl-2-like protein (Fig. 1A). To identify full-length open reading frame sequences, we performed RT-PCR using primers binding within the 5′ and 3′ untranslated regions of each gene, designed based on information obtained from the NCBI Honey Bee Genome Resources, and obtained a 1233-bp sequence for clone 1 that contained initiation and stop codons. The cDNA identified for clone 1 encoded a protein consisting of 283 amino acid residues. The deduced amino acid sequence contained putative Bcl-2 homology domains (BH domains), including the BH3, BH1, and BH2, and a putative C-terminal hydrophobic membrane anchor (MA), which are typical of Bcl-2 proteins (Fig. 1B, C). A database search using protein BLAST indicated that the BH3, BH1, and BH2 domains of the identified protein had 50%, 57%, and 50% sequence identity with buffy, which is a Bcl-2 protein in Drosophila melanogaster (Fig. 1B). In Drosophila, there are two Bcl-2 family genes, Drab-1/Debcl/dBorg-1/dBok and Buffy/dBorg-2 (lgaki et al., 2000; Cloussi et al., 2000; Brachmann et al., 2000; Zhang et al., 2000). In contrast, no other bcl-2 family genes apart from clone 1 were found in the honeybee genome, indicating that the gene corresponding to clone 1 is the honeybee buffy homolog (Ambuffy).

Clone 2 corresponded to part of a predicted gene, GB19151, located in linkage group 10 and encoding a matrix metalloproteinase (MMP) homolog (Fig. 2A). We obtained a 2531-bp sequence for clone 2 that contained initiation and stop codons. The cDNA identified for clone 2 encoded a protein consisting of 608 amino acid residues. The deduced amino acid sequence showed all the structural features typical of the MMP family (Fig. 2B). The open reading frame sequence contained the sequence PRCGVXD, which is a conserved motif in the prodomain of MMP, as well as a putative catalytic domain including the consensus motif HEXGHXXGXXHS, and a putative hemopexin domain (Fig. 2C). The hemopexin domain has sequence similarities to the serum protein hemopexin, a hemebinding protein that transports heme to the liver (Gomis-Ruth et al., 1996). The hemopexin domain binds to several proteins such as cell surface proteins and tissue inhibitor of matrix metalloproteinases (TIMPs). Thus this domain is thought to be involved in cellular protein-protein interactions and the inhibition of MMP (Page-McCaw et al., 2003). A protein BLAST search indicated that these three domains had 57%, 75%, and 72% sequence identity with MMP1 of Drosophila melanogaster (Fig. 2B). The database search revealed another predicted MMP gene, GB16274, in the honeybee genome. Drosophila melanogaster MMP1 (Dm1-MMP) was more similar in amino acid sequence to clone 2 than to GB16274 (data not shown), indicating that the gene corresponding to clone 2 is the honeybee MMP1 homolog (AmMMP1). AmMMP1 contained at the N terminus a putative signal peptide for secretion, suggesting the product could work in the extracellular matrix (Fig. 2B, C).

Expression analysis of Ambuffy and AmMMP1 in HPGs

We performed quantitative RT-PCR to confirm the differential expression of Ambuffy and AmMMP1 between nurse bees and foragers. For this, 9 to 14 nurse bees and foragers were collected at the same time from a single colony, and gene expression was compared using four batches of samples derived from four different colonies. The nurse bees and foragers were collected based on behavior as well as on HPG development (Ohashi et al., 1999). Quantitative RT-PCR showed that the expression of Ambuffy in the HPGs was approximately 12-fold higher in nurse bees than in foragers (Fig. 3A). We also analyzed the expression of MRJP2 and α-glucosidase as reference genes, because our previous Northern blotting analysis showed that differential expression depended on the role change (Ohashi et al., 1997). Expression of MRJP2, which showed the most prominent differential expression between nurse bee and forager HPGs among MRJPs 1 to 3 (Ohashi et al., 1997), was approximately 10-fold higher in nurse bees than in foragers (Fig. 3C), indicating that the extent of role-dependent change in Ambuffy expression in the HPGs was almost comparable to that in MRJP2 expression.

Fig. 1.

Identification of clone 1, obtained by differential display, as Ambuffy. (A) Genomic organization of the gene for clone 1. Exons (filled boxes) and introns (lines) of the newly identified gene, the gene predicted by NCBI Honey Bee Genome Resources, and the cDNA structure of clone 1 obtained by differential display are indicated below the corresponding linkage group. (B) Comparison of the domain structures of Ambuffy and Dmbuffy. Homologous domains (the Bcl-2 homology domains [BH3, BH1, BH2] and the C-terminal hydrophobic membrane anchor [MA]) are labeled. The number above each domain in Dmbuffy indicates the amino acid sequence identity of that domain between Ambuffy and Dmbuffy. (C) Alignment of the predicted protein sequences of Ambuffy and Dmbuffy. The predicted BH3, BH1, BH2, and MA domains are boxed. The BH domains were predicted with the Pfam database (Bateman et al., 2002) and information reported by lgaki and Miura (2004). The MA was predicted by using TMHMM (Sonnhammer et al., 1998).

Fig. 2.

Identification of clone 2, obtained by differential display, as AmMMP1 (A) Genomic organization of the gene for clone 2. Exons (filled boxes) and introns (lines) of the newly identified gene, the gene predicted by NCBI Honey Bee Genome Resources, and the cDNA structure of clone 2 obtained by differential display are indicated below the corresponding linkage group. (B) Comparison of the domain structures of AmMMP1 and Dm1-MMP. Homologous domains (the pro-, catalytic, and hemopexin domains) are labeled. The number above each domain in Dm1-MMP indicates the amino acid sequence identity of that domain between AmMMP1 and Dm1-MMP. (C) Alignment of the predicted protein sequences of AmMMP1 and Dm1-MMP. The predicted pro-, catalytic, and hemopexin domains are boxed. Amino acid residues corresponding to the conserved PRCGVXD motif in the prodomain and the consensus motif HEXGHXXGXXHS in the catalytic domain are shaded. Domains and motifs were predicted by the Pfam database, SignalP (Bendtsen et al., 2004), and information reported by Llano et al. (2000).

In contrast, the expression of AmMMP1 in the HPGs was approximately 4.5-fold higher in foragers than in nurse bees (Fig. 3B). Quantitative analysis of the differential expression of α-glucosidase between nurse bee and forager HPGs indicated that its expression in the HPGs was approximately 360-fold higher in foragers than in nurse bees (Fig. 3D). Although the extent of differential expression of AmMMP1 in the HPGs between nurse bees and foragers was smaller than that of α-glucosidase, there was a reproducible and significant difference in AmMMP1 expression level between nurse bee and forager HPGs.

Expression analysis of Ambuffy and AmMMP1 in various tissues

We investigated whether tissues other than HPGs also express buffy and MMP1, and whether expression in those tissues also changes in association with the role change of workers. The amounts of transcripts of these genes in various body parts (the HPGs, the head without the HPGs, thorax, midgut, and abdomen without the midgut) were examined by quantitative RT-PCR. Each sample was prepared by using two or three nurse bees or foragers per batch, and eight batches of samples were prepared from three colonies and analyzed. In the above experiments to analyze Ambuffy and AmMMP1 expression in the nurse bee and forager HPGs (Fig. 3), the amount of Ambuffy transcript was normalized to that of elongation factor 1α-F2 (EF1α.-F2) (Danforth and Ji, 1998), which was expressed almost equally in both nurse bee and forager HPGs (data not shown). However, we used rp49 (Ben-Shahar et al., 2003) as a reference to normalize Ambuffy and AmMMP1 expression in various tissues, because we could not detect any significant EF1α-F2 expression in the other tissues, and quantitative RT-PCR revealed that there were almost equivalent levels of rp49 expression in all these tissues (data not shown).

Fig. 3.

Quantitative analysis of Ambuffy and AmMMP1 transcripts in the HPGs. Nurse bees or foragers (n=9–14/group) were collected as one batch, and four batches prepared from four different colonies were subjected to real-time RT-PCR. Bars indicate mean relative mRNA levels with the standard deviation for (A) Ambuffy, (B) AmMMP1, (C) MRJP2, and (D) α-glucosidase, with the amount of mRNA in nurse bee HPGs defined as 1. Asterisks indicate significant differences between nurse bees and foragers (^*p<0.01; Welch's t-test).

Ambuffy expression was high not only in the HPGs but also in the other body parts and tissues examined. However, interestingly, although the differential Ambuffy expression in the HPGs between nurse bees and foragers was again confirmed in this experiment, the expression level of Ambuffy in nurse bee HPGs was largely comparable to that in the other body parts and tissues of nurse bees, and the Ambuffy expression in other body parts or tissues other than the HPGs did not differ significantly between nurse bees and foragers (Fig. 4A).

Fig. 4.

Quantitative analysis of Ambuffy and AmMMP1 transcripts in various body parts and tissues. Samples were collected from two or three nurse bees or foragers for each batch, and eight batches of samples were prepared from three colonies and subjected to realtime RT-PCR. HPGs of foragers were collected from only seven batches due to a failure in sample preparation. Bars indicate the mean relative mRNA levels with the standard deviation for (A) Ambuffy and (B) AmMMP1 in various body parts and tissues, with the amount of mRNA in nurse bee HPGs defined as 1. Asterisks indicate significant differences between nurse bees and foragers (^* p< 0.05; Welch's t-test).

Similarly, all body parts or tissues examined showed high AmMMP1 expression, and the differential AmMMP1 expression in the HPGs between nurse bees and foragers was again confirmed in this experiment (Fig. 4B). AmMMP1 expression varied to a larger extent among body parts and tissues than Ambuffy expression. AmMMP1 expression was distinct from Ambuffy expression in that it was significantly higher in foragers than in nurse bees, not only in the HPGs but also in the midgut, indicating that Ambuffy and AmMMP1 expression differed between nurse bees and foragers in a tissue-preferential manner.