Sex mosaics in the honeybee: how haplodiploidy makes possible the evolution of novel forms of reproduction in social Hymenoptera

Abstract

Hymenoptera are haplodiploid: females arise from fertilized, diploid eggs, while males arise from unfertilized, haploid eggs. The cytogenetic mechanisms underlying haplodiploidy enable remarkable phenomena including female cloning, male cloning and gynandromorphy (sex mosaics). We collected 11 newly emerged putative gynandromorph honeybees from a single colony, assessed the sex of various tissues morphologically and determined the genetic origin (maternal or paternal) of each tissue by genotyping. Ten bees were gynandromorphs with one to three distinct paternal origins. Remarkably, one bee carried no maternal alleles. This bee had female organs throughout, and arose from the fusion of two sperm nuclei. This is the first reported case in the Hymenoptera of sperm fusion resulting in a female, emphasizing the flexibility for social insect reproduction and potentially novel colony-level social structures.

1. Introduction

Hymenopteran (ants, bees, wasps) social insect colonies are characterized by a variety of genetic structures and reproductive modes. Examples include hybridogen ant species in which female and male reproductives are produced clonally, whereas workers are produced sexually [1,2]. Systems such as these are made possible by haplodiploidy (males haploid, females diploid), meaning that haploid eggs containing either a female or male gamete can develop into a male. Clonal females can be produced by fusion of two female gametes within an egg (thelytoky) [3] or clonal males by the development of a sperm in a gamete-free egg (androgenesis) [1]. We propose that haplodiploidy has been a key evolutionary innovation that has allowed novel social structures to evolve in social insects [1,2,4]. This is because almost any plausible combination of gametes within an egg, be they from a father or a mother, can result in viable offspring.

Gynandromorphy is a developmental abnormality in which individuals are chimeras containing both female and male tissue [5]. In honeybees, gynadromorphs generally develop from the combination of a diploid zygote and a supernumerary haploid male tissue originating from a second sperm [5,6]. This is made possible by the fact that honeybees are polyspermic where more than one sperm enters every egg [7].

Some honeybee colonies produce gynandromorphs at high frequency as a result of genetic mutations [8]. Such colonies provide the opportunity to discover novel developmental and reproductive modes in this model social insect. Here, we investigate the parental origins of 11 gynandromorph honeybees found in a single colony. Our study reveals previously unreported developmental pathways and complex parental origins.

2. Material and methods

(a). Biological material

During a previous study, 11 suspected honeybee gynandromorphs were identified among 265 bees emerging from worker cells of a brood frame that had been placed in an incubator overnight to provide day-old bees [9]. The mother of this brood was a naturally mated queen, of standard commercial stock, predominantly Apis mellifera ligustica. These 11 individuals, including a pink-eyed pupa, were stored in 70% ethanol at −20°C for later analysis.

(b). Phenotypic and genotypic analysis

For each bee, we dissected, imaged [10] and collected sex-specific tissues (gonads, legs, antenna, eyes, tongue, sting and associated gland, honey sack) and tissues with no clear sex-specific phenotype (brain, gut, fat body, flight muscle, tergite and malpighian tubules) for genotypic analysis.

We extracted DNA from each tissue independently using Qiagen DNA-easy kits (Hilden, Germany) and genotyped each sample using seven polymorphic microsatellites: A29, A79, A113, A8, A88, B124, The3 (electronic supplementary material, table S1). For some bees, additional loci (A107, Ap43, A35) were required to clarify paternity (electronic supplementary material, table S1). The genotypes of the samples were compared to the known queen and paternal genotypes of the colony of origin [9].

Honeybee queens are polyandrous, mating with at least 10 males [11]. Progeny of different males are known as patrilines. To allocate each tissue to a particular patriline, the maternal allele (if present) was identified, while the other allele(s) were assumed to be paternal in origin. To be conservative, every allele with fewer than three occurrences in different tissues of the bee was regarded as a non-specific peak and removed from further analysis. A paternal allele that was present in all diploid tissues was considered to have originated from the zygote. Other paternal genotypes were considered to have been derived from supernumerary sperm. Each paternal multilocus-genotype was then allocated to a known patriline [9]. Paternal genotypes that could not be unambiguously assigned to a patriline were assigned a new patriline. We considered all maternal and paternal alleles on a per-bee basis rather than on a per-tissue basis because of the possibility of cross-contamination between dissected tissues (electronic supplementary material, tables S2 and S3).

In one case, we further confirmed the maternity of one bee (in three tissues) and three putative sisters by sequencing part of the COI–COII intergenic region of the mitochondria using E2, H2 primers (electronic supplementary material, table S1) [12].

3. Results

(a). Phenotypes

Phenotypically, 10 of the 11 bees (bees A–J) were gynandromorphs, with a patchy distribution of male and female tissue. Bee K was phenotypically female, externally and internally (tables 1 and 2). Only an unusual mottled coloration on the abdomen distinguished bee K from normal workers (figure 1a). Bees A, B and E also had a mottled coloration on their abdomens (figure 1b). There was no apparent link between the external appearance of the bees and the internal sex organs (tables 1 and 2 and figure 1).

Table 1.

Presence and absence of sexually dimorphic external tissues based on phenotypic appearance. F, female; M, male; ?, undetermined.

bee	eye	tongue	antenna	abdomen	legs
A	F and M	M	F	M	F
B	M	M	M	M	F
C	M	M	M	F	F
D	M	M	M	F	F and M
E	M	M	F	M	F
F	F and M	M	F	F	F
G	M	M	F and M	F	F
H	M	F	M	F	F
I	M (patches of F)	F	F	F	F
J	M	M	?	M	F
K	F	F	F	F	F

Table 2.

Presence and absence of sexually dimorphic internal organs based on phenotypic appearance. √, present; ✗, absent; ?, undetermined; *, right side only; (Q) queen-like.

bee	sting (female)	ovary (female)	honey sack (female)	testis (male)	seminal vesicle (male)	mucus glands (male)
A	√	√	√	✗	✗	✗
B	√	✗	✗	√	√	✗
C	√	√ *	√	✗	✗	✗
D	√	√ (Q)	√	✗	✗	✗
E	√	√	✗	?	✗	✗
F	√	√	√	✗	✗	✗
G	√	√ (Q)	√	✗	✗	✗
H	√	✗	√	√ *	✗	✗
I	√	√ (Q)	√	✗	✗	✗
J	✗	✗	✗	√	√	√
K	√	√	√	✗	✗	✗

Figure 1.

External and internal morphology. (a) Female morphology but a mottled coloration on the abdomen. Bee K. (b) Predominantly male morphology and mottled coloration on the abdomen. Bee E. (c) Worker ovaries with approximately five ovarioles. Bee K. (d) Queen-like ovaries with approximately 40 ovarioles. Bee D. (e) Abnormal, underdeveloped testis (arrow). Bee B. (f) Normal male reproductive organs. Pupa J. (Online version in colour.)

Internally, the main differences between the bees related to their reproductive organs. Bees A, C, E, F and K had typical worker ovaries (figure 1c). Bees D, G and I had abnormal queen-like ovaries with 30–50 ovarioles (figure 1d). Bees B and H had partial male reproductive organs (figure 1e). The pupa (J) had normal male gonads (figure 1f).

(b). Genotypes

Microsatellites (table 3) revealed that individual tissues and organs of 10 gynandromorph bees had a maternal origin consistent with being offspring of the colony's queen [9]. Bee K lacked any maternal nuclear genetic material (table 3). To verify this finding, we genotyped it at three additional microsatellites (electronic supplementary material, table S5). Nonetheless, bee K's mitochondrial DNA matched that of the source colony (electronic supplementary material, table S4). As our bees emerged in an incubator in a sealed box, there was no possibility that this bee had drifted in from another colony.

Table 3.

Genotypes of 11 gynandromorph bees. ‘/’, the allele parental origin could not be unambiguously determined.

microsatellite	A29		A79		A8		A113		A88		B124		The3
queen allele	132,144		92, 123		179, 179		217, 223		142, 151		213, 215		200, 194
bee	paternal	maternal	paternal	maternal	paternal	maternal	paternal	maternal	paternal	maternal	paternal	maternal	paternal	maternal
A	156, 145, 132	144	123, 92, 111	92	179, 165	179	217, 223	217	151	151	213	215	194, 183	200
B	145, 132	132	92, 107	92	165, 179	179	223	217	151, 154	142	213	215	183, 200	194
C	145, 156	132	92, 123	92	165	179	223	217	151	142	213	215	183	194
D	132/144, 134	144/132	92	92	179, 165	179	217	223	151	151	213	215	194/200, 183	200/194
E	156, 154	132	105, 92	92	179, 165	179	223	223	151	142	213,221	213	200, 183	200
F	156, 144	144	105, 107	92	179	179	223	217	151	151	213	215	200	200
G	132, 149	132	92	92	165, 173	179	223	223	151, 142	151	217, 213	215	192	194
H	154, 149	132	92	92	165, 173	179	223	223	142	142	221, 213	215	183, 192	200
I	154, 156	144	92, 105	92	165, 179	179	223	223	142, 151	142	221, 213	215	183, 200	194
J	134	132	92	92	179	179	217	217	226	215	142	142	183	194
K	151	n.a.	107, 111	n.a.	165	n.a.	217, 233	n.a.	213, 215	n.a.	151	n.a.	192, 200	n.a.

In bees B–I, two separate paternal origins were present, one creating the diploid female tissue and another the haploid male tissue (table 3). Bee A contained three separate paternal origins, two giving rise to different haploid tissues (table 4). The remaining pupa (bee J), despite being an obvious sex mosaic, had one maternal and one paternal genome throughout all tissues tested, as would a normal diploid worker (electronic supplementary material, table S2) [13].

Table 4.

Paternal origin of zygotic and haploid tissues. Frequencies are calculated independently for colony workers sampled [9] and mosaic bees.

			frequency of patriline (%) in:
bee	origin	patriline	mosaics (N = 11)	colony workers (N = 254)
A	zygote	P18	9.0	4.7
	haploid tissue	P31	27.3	10.6
		P42	18.2	0
B	zygote	P31	27.3	10.6
	haploid tissue	P17	9.0	3.1
C	zygote	P31	27.3	10.6
	haploid tissue	P42	18.2	0
D	zygote	P46	9	0
	haploid tissue	P19	9	4.7
E	zygote	P41	18.2	0
	haploid tissue	P35	27.3	3.1
F	zygote	P41	18.2	0
	haploid tissue	P29	9	0.8
G	zygote	P2	9	6.7
	haploid tissue	P38	18.2	0.4
H	zygote	P35	27.3	3.1
	haploid tissue	P38	18.2	0.4
I	zygote	P35	27.3	3.1
	haploid tissue	P41	18.2	0
J	zygote	P43	9	0
K	zygote	P44	9	0
	zygote	P45	9	0

Fourteen patrilines were represented among the 11 individuals (table 4). Out of the colony's 40 known patrilines [9], six new patrilines were found in the gynandromorphs (electronic supplementary material, table S6). Patriline frequencies were significantly different between the normal worker population and the gynandromorph sample (p < 0.001, Fisher's exact test n = 46) (table 4). However, there was no evidence that particular patrilines were more likely to be present in female tissue (zygote) than male tissue (haploid tissue) (Fisher's exact test, p = 0.072, n = 14) (table 4).

4. Discussion

Our study reveals novel parental origins in honeybees. Nine of the gynandromorphs carried two (B–I), and even three (A), separate paternal genomes. These bees arose when supernumerary sperm underwent division to create androgenic, haploid, male tissue alongside the diploid tissue from the zygote. Some patrilines were more likely to sire gynandromorphs than others, suggesting that sperm can influence whether they develop androgenically. Interestingly, gynandromorph patrilines were rare in the general population, suggesting that the sperm of these fathers was only competitive in eggs where a zygote had already been formed.

Bee K was female and diploid throughout, but lacked any maternal nuclear genetic material. We propose that K was created by a fusion of two sperm nuclei, resulting in an androgenic diploid bee. This is the first report of an individual created by the fusion of two sperm in any Hymenopteran. K was able to develop to the adult stage, showing that a bi-paternal bee is viable. In mammals, parent-specific gene expression in key developmental genes means that either the paternally or maternally derived allele is expressed during offspring development. As a consequence, embryos with two paternal genomes are inviable in mammals [14]. The viability of K suggests that there are no impediments to the development of a bi-paternal female honeybee, as has previously been shown for bi-maternal [15].

The individuals studied here shared a common maternal origin. Therefore, it is likely that their gynandromorphy and androgenesis had a common cause. A mutation in a meiotic mechanism, such as disrupted microtubule formation, may cause a delay in pronuclei migration, resulting in either a paternal or maternal pronucleus undergoing cleavage before the formation of the zygote [16]. The timing of fusion can therefore result in different kinds of gynandromorphs and androgenesis, depending on whether paternal or maternal pronuclei have divided prior to formation of the zygote. If the maternal pronuclei do not migrate at all, then two sperm may fuse. Both gynandromorphy and sperm-fusion androgenesis may therefore occur as a result of the same cytological phenomenon.

Previous research [17,18], which used phenotype-based methods, reported no tendency for particular tissues to develop from paternal tissue or zygotic tissue. Our results confirm this, indicating that supernumerary sperm cells can start development anywhere within the embryo.

Of the eight bees with ovaries, three had approximately 40 ovarioles (figure 1d). We define these ovaries as ‘queen-like’ because honeybee workers typically have five ovarioles, whereas queens have 150 [10]. Workers with queen-like ovaries can occur when developmental genes in worker-destined larvae are silenced [19]. Thus, the development of queen-like ovaries in the gynandromorphs may be a result of a disruption to the gene network required to initiate caste-specific development. Such disruption could potentially give rise to novel mechanisms of caste development.

In conclusion, our genetic study emphasizes the range of developmental variants that can occur in haplodiploid organisms that nonetheless result in viable adults. Haplodiploidy allows almost any combination of gametes present in an egg (e.g. the four maternal pronuclei and one or more sperm pronuclei) to fuse and form a zygote or not fuse and develop independently as haploid tissue. Variants like those identified here can potentially lead to evolutionary novelty, including societies comprised solely of theltytokous clonal workers [20], hybridogens [1] and clonal social parasites [21]. Beyond these already-known examples, there are likely to be equally extraordinary social systems that have not yet been identified or even imagined.

Data accessibility

Supporting datasets are available as electronic supplementary material.

Authors' contributions

S.E.A. performed the genotyping, analysed data and wrote the paper; B.Y. identified gynandromorphs, analysed data and edited the paper; I.R. performed the dissections, identified, imaged tissue and edited the paper; B.P.O. performed dissections, analysed data and wrote the paper. All authors gave final approval for publication and agree to be held accountable for the work performed therein.

Competing interests

We have no competing interests.

Funding

This work was funded by the Australian Research Council, projects DP150100151 and DP180101696.