A highly divergent gene cluster in honey bees encodes a novel silk family

Abstract

The pupal cocoon of the domesticated silk moth Bombyx mori is the best known and most extensively studied insect silk. It is not widely known that Apis mellifera larvae also produce silk. We have used a combination of genomic and proteomic techniques to identify four honey bee fiber genes (AmelFibroin1-4) and two silk-associated genes (AmelSA1 and 2). The four fiber genes are small, comprise a single exon each, and are clustered on a short genomic region where the open reading frames are GC-rich amid low GC intergenic regions. The genes encode similar proteins that are highly helical and predicted to form unusually tight coiled coils. Despite the similarity in size, structure, and composition of the encoded proteins, the genes have low primary sequence identity. We propose that the four fiber genes have arisen from gene duplication events but have subsequently diverged significantly. The silk-associated genes encode proteins likely to act as a glue (AmelSA1) and involved in silk processing (AmelSA2). Although the silks of honey bees and silkmoths both originate in larval labial glands, the silk proteins are completely different in their primary, secondary, and tertiary structures as well as the genomic arrangement of the genes encoding them. This implies independent evolutionary origins for these functionally related proteins.

Figure 1.

Analysis of proteins in the labial gland of late final instar Apis mellifera by SDS-PAGE showing that major protein bands correspond to identified silk proteins. Molecular weight markers (kDa) are shown alongside the gel. AF1, AmelFibroin1; AF2, AmelFibroin2; AF3, AmelFibroin3; AF4, AmelFibroin4. As Coomassie brilliant blue dye binds to basic and aromatic residues (Compton and Jones 1985), the relative intensity of the staining seen in this gel may not be linearly proportional to protein abundance. AmelSA1 and AmelSA2, the two most prominent bands, contain approximately twice the basic plus aromatic content of the AmelFibroin proteins.

Figure 2.

A schematic showing the genetic organization of AmelFibroin1–4 and AmelFibroin-rel genes in the Apis mellifera genome and the GC content and CpG islands in the corresponding genomic region. The open reading frames of the AmelFibroin genes are shown as block arrows with the regions encoding heptads shaded. Intergenic regions are shown to scale as lines. The GC content is redrawn from BeeBase (http://racerx00.tamu.edu/bee_resources.html). Regions corresponding to CpG islands (predicted by newcpgseek) are indicated by stars.

Figure 3.

Dot matrix similarity analysis of honey bee silk and related proteins. The horizontal and vertical rows of dots seen as hatched patterns demonstrate repeats of the same sequence character. (A) AFrel; (B) AmelFibroin1; (C) AmelFibroin2; (D) AmelFibroin3; (E) AmelFibroin4; (F) AmelSA1.

Figure 4.

Protein sequence alignment translated from the nucleotide alignment of the AmelFibroin proteins. The bulk of the nucleotide alignment (Supplemental Fig. II) was derived from aligning primary sequence and conserving heptad periodicity as described in the Methods. The alanine residues populating the hydrophobic heptad positions are marked (position a shaded and position d in bold). The predicted signal peptide is underlined. Regions shown in italic were aligned using MUSCLE.

Figure 5.

Maximum parsimony tree showing the relatedness of the four AmelFibroin proteins with branch length indicating the number of amino acid changes between the proteins. The bootstrap value separating the two gene clusters was calculated by carrying out 1000 replicates with PAUP* (Swofford 2002).