Introduction

In honey bees, sex is controlled by a mechanism termed complementary sex determination (csd) [1, 2]. Haploids containing a single csd allele will develop into normal males (drones), while diploids, heterozygous for csd, become females. When csd genes are homozygous, a bee will develop into a diploid male, which is consumed at the larval stage by workers [3]. Therefore, nucleotide mutations leading to heterozygotes at csd gene are selectively favoured. The complementary sex determination mechanism of honey bee remained a hypothesis for a long time [4]. In 2003, Beye et al. cloned the csd gene from Apis mellifera by positional cloning [5]. They found that no transcription differences existed between the two sexes but suppression of csd in females with double stranded RNA for csd resulted in male phenotypes. Csd encodes an arginine serine-rich (SR) type protein, which contains an RS domain in the middle and a proline-rich region at its C terminus, between these two domains is a hypervariable region that differs highly between alleles, and has a variable number of asparagine/tyrosine repeats. The honey bee csd protein is homologous to the Drosophila Tra protein, which is involved in Drosophila sex determination [5].

Previous research has demonstrated that the csd genes of three honey bee species (A. mellifera, A. cerana and A. dorsata) have evolved under balancing selection, and several parts of the coding region are possible targets of selection [6–10]. Moreover, the polymorphic level is approximately seven times higher in the csd coding region than in the neutral regions [7].

Although much evolutionary research on the csd gene was conducted in A. mellifera, A. cerana and A. dorsata, no research on csd has yet been conducted in different subspecies of A. mellifera. A. mellifera includes several different geographic subspecies, which arose through long-term natural selection in different geographic environments, and it is worth investigating whether the csd gene is polymorphic among these subspecies.

In this study, we analyzed the polymorphism of csd gene in six A. mellifera subspecies, A. m. anatolica, A. m. caucasica, A. m. carnica, A. m. carpatica, A. mellifera ssp., A. m. ligustica. In A. mellifera, the csd gene contains nine exons, which form three clusters separated by two large introns [5]. The genomic region of the third cluster (from exon six to nine) has the highest polymorphism compared with the other two regions. Therefore, we chose region 3 to study its polymorphism.

Materials and methods

Sample collection

All samples were collected from the bee breeding apiary of the Honeybee Research Institute, Jilin province, China, with 50 workers for each subspecies. Every subspecies is reproductively isolated from other subspecies through artificial breeding. The samples were first collected into 95% ethanol and then stored frozen at −70°C until further use.

DNA extraction

Total genomic DNA was extracted from the cephalothorax of each sampled bee according to the protocol of the Animal Genomic DNA Extraction Kit (BEST ALL-HEAL LLC, NY, USA).

PCR and sequencing

The primers used for amplifying region 3 of the csd gene in this study were designed by consulting those reported by Cho et al. [9] with some modifications. The primers are 5′ AATTGGATTTATTAATATAATTTATTATTCAGG 3′ (forward) and 5′ ATYTCATTATTCAATATGTTNGCATCA 3′ (reverse). The high fidelity LA Tag DNA polymerase (BEST ALL-HEAL LLC, NY, USA) were used for all PCR reactions. PCR conditions were: denaturation at 94°C for 3 min, followed by 30 cycles of 94°C for 30 s, annealing at 54°C for 30 s and extension at 72°C for 2 min, with a final extension at 72°C for 10 min. PCR products were purified using DNA GEL EXTRACTION kits (Sangon, Shanghai, China) and cloned into the pEASY-T3 vector (Transgene, Beijing, China). To obtain as many csd alleles as possible, the genomic region 3 of the csd gene was cloned from cephalothorax of each sampled worker bee, and 1–3 clones of each cloned fragment were subjected to double-strand sequencing. Single-sequencing reads were assembled using the Seqman program in the DNAstar software [11].

Sequence analysis

The exons, introns and coding regions of our sequences were determined by consulting those sequences of the genomic region 3 of A. mellifera csd gene reported by Cho et al. [9] and cDNA sequences of A. mellifera csd gene reported by Hasselmann and Beye [10]. Nucleotide sequence alignments were performed with Clustal X version 1.8 [12], and alignment results were adjusted manually for obvious alignment errors.

DAMBE 4.1.19 [13] was used to identify haplotypes. Phylogenetic trees were constructed using MEGA version 4.0 program [14]. The minimal evolution (ME) method and Kimura’s 2-parameter distances were adopted to obtain an unrooted tree with 2,000 bootstrap replications. Nucleotide diversity (π) was calculated by using DNAsp5.0 program [15]. Two tailed Z test was adopted to detected significant difference between two π values. Fst distances were computed using ARLEQUIN 3.0 software [16]. Kimura’s 2-parameter genetic distance was calculated by MEGA 4.0.

Results

Polymorphism of the csd haplotypes in six A. mellifera subspecies

We cloned the genomic region 3 of the csd gene from six A. mellifera subspecies, A. m. anatolica, A. m. caucasica, A. m. carnica, A. m. carpatica, A. m. ssp., A. m. ligustica. After sequencing, we obtained 6, 10, 19, 14, 28 and 7 haplotypes from these subspecies (table 1), respectively. There are a total of 79 haplotypes, and five haplotypes exist in two subspecies. We compared the nucleotide diversity (π) of the csd gene of the six A. mellifera subspecies. As shown in Table 1, the csd gene has a high level of polymorphism in all the six subspecies. Of them, the π value of the A. m. anatolica subspecies is the highest among the six subspecies. It is significantly higher than that of the A. m. ssp. subspecies (two tailed Z test, P < 0.05), but failed to be significant compared with π values of the other four subspecies (two tailed Z test, P < 0.1). Except for the A. m. anatolica subspecies, there is no significant difference in π values among the other five subspecies (two tailed Z test, P > 0.05) Table 2.

Table 1 Nucleotide diversity (π) (mean ± SD) of csd haplotypes in six A. mellifera subspecies
Full size table
Table 2 Pairwise Fst values (lower-left matrix) and Kimura’s 2-parameter genetic distance (upper-right matrix) between all the subspecies pairs
Full size table

Phylogenetic tree of all the csd haplotypes

A genealogy tree was constructed based on all the haplotypes from the six subspecies. As shown in Fig. 1, all the csd haplotypes from different A. mellifera subspecies mainly form two clades in the tree, and they are mixed in the genealogy tree while not forming different clades according to subspecies.

Fig. 1
figure 1

The gene genealogy of csd haplotypes in region 3 of six A. mellifera subspecies. The minimum evolution method and Kimura’s two parameter distances are used to construct the tree. Bootstrap percentages are shown on internal branches. The scale bar represents the number of nucleotide changes per site. Amcau A. m. caucasica, Amssp A. m. ssp., Amcan A. m. carnica, Amcap A. m. carpatica, Amlig A. m. ligustica, Amana A. m. anatolica

Full size image

Hypervariable region of CSD proteins in six A. mellifera subspecies

We analyzed the amino acid sequence of the csd sequences in these subspecies, since this region contains a hypervariable region critical for determining the specificity of csd alleles. The exons, introns and coding sequences on the obtained csd haplotypes were determined by consulting sequences of A. mellifera csd gene reported by Cho et al. [9] and Hasselmann et al. [10]. Similar to that in A. mellifera, A. cerana and A. dorsata, the coding region of this part also contains an RS domain at the N terminal, a P-rich domain at the C terminal and a hypervariable region between these two domains. The hypervariable region is rich in asparagine (N) and tyrosine (Y), and they form a basic (N) 1–4Y repeats terminated mainly with KK, KQ in each haplotype (supplementary Fig. S1).

Genetic differentiation among six A. mellifera subspecies

The pairwise Fst values between different subspecies range from 0.02570 to 0.23848, and of them, Fst value between A. m. anatolica and A. m. carnica is the highest, while that between A. m. carpatica and A. m. caucasica is the lowest. From the Fst values, it indicated that the genetic differentiation levels between A. m. caucasica and all other five subspecies is very low. The genetic differentiation between A. m. carpatica and both A. m. ligustica and A. m. anatolica is also not significant (P > 0.05). Except for these subspecies pairs, the remaining pairs showed significant genetic differentiation (P < 0.001).

Genetic distance and phylogenesis of six A. mellifera subspecies

The kimura’s 2-parameter genetic distances between different subspecies range from 0.04216 to 0.06415. Of them, the genetic distance between A. m. anatolica and A. m. carnica is the largest, while that between A. m. ssp. and A. m. caucasica is the nearest.

When a NJ tree was constructed based on Kimura’s 2-parameter genetic distance, A. m. caucasica and A. m. carnica formed a clade first, followed by A. m. ssp, A. m. ligustica, A. m. carpatica and A. m. anatolica that were gathered in the tree in turn (Fig. 2).

Fig. 2
figure 2

NJ tree of six A. mellifera subspecies based on Kimura’s 2-parameter genetic distances

Full size image

Discussion

Previous studies have shown that the csd genes in A. mellifera, A. cerana and A. dorsata have a very high level of polymorphism [9, 10]. In this study we found that the csd gene in the six A. mellifera subspecies also shows a high level of polymorphism. This result further confirms that the complementary sex determination mechanism is common for all the honey bee species [1].

Cho and Hasselmann found that the (N)1–4Y repeats and (KHYN) 1–4 motif are two types of important repeat sequences in the hypervariable region of CSD protein, but the (KHYN)1–4 motif exists in A. cerana and A. dorsata while not in A. mellifera [8, 9]. In this study, we also did not find (KHYN)1–4 motif in any of the six A. mellifera subspecies—it may have existed in the ancient csd alleles, but was lost in A. mellifera during evolution, since the A. dorsata and A. cerana speciated prior to A. mellifera in their evolutionary history.

Some researchers have divided A. mellifera into four evolutionary clades based on the analysis of mitochondrial genes [17–22]. They are the African type, western and northern Europe type, eastern Mediterranean type, and the near Eastern type. According to these researchers, A. m. carnica, A. m. carpatica and A. m. ligustica belong to the Eastern Mediterranean type, A. m. caucasica, A. m. ssp. and A. m. anatolica belong to the near East type. In this study, the structure of the NJ tree of six A. mellifera subspecies based on the Kimura’s 2-parameter genetic distances did not match the above geographical division. For example, the A. m. anatolica subspecies did not group with the A. m. caucasica and A. m. ssp. subspecies to form a clade, although they are meant to belong to the same geographical type. One reason may be that other different strains were once introduced into this subspecies during artificial breeding, causing an increase in the polymorphism of csd gene in this subspecies, as well as a relatively distant relationship with the A. m. caucasica and A. m. ssp. subspecies.

Apis mellifera carpatica was once considered to be a strain belonging to A. m. carnica subspecies [23]. But our pairwise Fst analysis indicated that there is significant genetic differentiation between these two subspecies. Meanwhile, Liu also obtained significant genetic differentiation between them by analyzing the polymorphism difference of malate dehydrogenase between these two subspecies [24]. Therefore, the A. m. carpatica and A. m. carnica subspecies maybe really have a large genetic difference, despite their close geographical distribution.

Our analysis showed that the genetic differentiation level between A. m. caucasica and all other five subspecies is very low; this may suggest that A. m. caucasica is a transitional or intermediate subspecies among the six subspecies.

In conclusion, we found a high level of polymorphism of the csd gene in six A. mellifera subspecies by molecular analysis, and detected significant genetic differences between some of these subspecies pairs. The present study has further verified that gender in bees is determined by the csd gene, and has expanded our understanding of sex determination mechanism in bee subspecies.