Recognition of polymorphic Csd proteins determines sex in the honeybee

Abstract

Sex in honeybees, Apis mellifera, is genetically determined by heterozygous versus homo/hemizygous genotypes involving numerous alleles at the single complementary sex determination locus. The molecular mechanism of sex determination is however unknown because there are more than 4950 known possible allele combinations, but only two sexes in the species. We show how protein variants expressed from complementary sex determiner (csd) gene determine sex. In females, the amino acid differences between Csd variants at the potential-specifying domain (PSD) direct the selection of a conserved coiled-coil domain for binding and protein complexation. This recognition mechanism activates Csd proteins and, thus, the female pathway. In males, the absence of polymorphisms establishes other binding elements at PSD for binding and complexation of identical Csd proteins. This second recognition mechanism inactivates Csd proteins and commits male development via default pathway. Our results demonstrate that the recognition of different versus identical variants of a single protein is a mechanism to determine sex.

Fig. 1.. Complementary sex determination in the honeybee.

(A) Multiple alleles of the complementary sex locus in the population. Heterozygosity and hemi/homozygosity determine sex. i and j symbolize different alleles. (B) The sex-determining pathway. The sex locus signal controls female-specific splicing of fem transcripts, which produce Fem proteins. Fem protein controls female development via the female splicing of downstream transcripts, including those of the dsx gene. Males result from default regulation. The genomic splicing scheme and proteins are shown. ^F and red color indicate female-specific regulated products; ^M and blue indicate male-specific ones. (C) Highly polymorphic Csd protein variants. The range of amino acid (aa) differences between Csd protein variants is presented. HV, hypervariable region domain; PR, proline-rich domain. i and j: different protein variants are shown by different green colors.

Fig. 2.. The combination of two different csd coding sequences is essential for female determination.

(A) The generation of csd ^{i/j stop} genetic (diploid) females. i/j and light/dark green color symbolize different alleles in the testing. Blue triangle, induced stop codon in exon 2 using CRISPR-Cas9. Polymorphisms in the i allele restricted mutations to the j allele as shown by deep sequencing of the target site using amplicons for each bee. (B) Sex-specific splicing of genetic female with the csd ^{i/j stop} genotype at late larval/ early pupal stage. The mutations were independently induced for each bee. Resolved amplicons are presented as black/white negatives. Reverse transcription polymerase chain reactions (RT-PCRs) were semiquantitatively adjusted across individuals using ef-1α (elongation factor 1α) transcripts as a control. fem^F, female fem transcript; dsx^F/dsx^M, female/male dsx transcripts; S, marker in 100–base pair (bp) steps; NC, negative PCR control. (C) The frequencies of female/male transcripts and reproductive organs in csd ^{i/j stop} genetic females. (D) Reproductive organ of csd ^{i/j stop} genetic females at the early pupal stage (reared on worker nutrition). Yellow arrowhead, horizontally packed testioles; white arrowhead, few vertically organized ovarioles. Scale bars, 2 mm.

Fig. 3.. The combination of two different csd coding sequences is sufficient to produce female-determining activity.

(A) Schematic of the transgenic experiments to examine the female-determining activity of two different versus two identical csd alleles in haploid males. Combinations of endogenous allele and transgenic allele were tested. i/j and light/dark green color symbolize coding sequences from different alleles; tg, transgene; long gray box, actin-5c promoter. (B) Female- and male-specific splicing of the fem and dsx transcripts in the transgenic males at larval stage. Size-resolved amplicons from RT-PCRs, which were semiquantitatively adjusted across individuals using ef-1α transcripts as a control. The numbers above the lanes indicate RT-PCRs from single bees. The pictures are black/white negatives. ^F, female; ^M, male transcript. (C) Summary table of the allele combinations examined. The feminizing activity in transgenic males with two different (csd ^{i, tg actin5C csd j}) or two identical (csd ^{i, tg actin5C csd i}) csd coding sequences. Males (csd ⁱ) and females (csd ^i/j) were WT controls. Black, ****P < 0.0001; purple, comparison of the identical alleles tested under both different and identical allele combinations: **P < 0.01 and ***P < 0.001.

Fig. 4.. Trimeric complex formation between different and identical Csd protein variants.

(A) Schematic of the fluorescence lifetime (LT) reduction upon protein binding via FRET. LT was measured using fluorescence lifetime imaging microscopy (FLIM). (B) LT values of the Csd protein variants that were fused with either cerulean protein or YFP. Proteins were expressed in Sf21 insect cells using a baculovirus system. Values from n ≥ 6 per condition were compared using a t test. (C) Example of a Western blot using oligomer-forming conditions. Three different protein variant combinations and three identical Csd protein variants were expressed. Myc-tagged Csd proteins were resolved in 8% acrylamide gels and stained with anti-myc antibody. The complexes equal in size to trimers and monomers are marked. (D) Highly resolved trimeric complexes of different and single-protein variant expression using 6% acrylamide gels. All lanes are from the same Western blot. Bottom: Detected intensities along the position in each lane (lane profile) of the Western blot shown in (D). Arrows indicate the peak intensities. AU, arbitrary units. RU, relative units. (E) The relative amounts of the detected trimeric complexes. The number of replicates is indicated. The SEM is displayed. z test, *P < 0.05.

Fig. 5.. The presence or absence of polymorphisms controls nonbinding and binding at the PSD.

(A) Schematic of the fragments used in the Western blot experiments. The amino acid polymorphisms at PSD of the protein variants used are shown. (B to E) The relative abundance of trimeric complexes using oligomer-forming conditions in SDS-PAGE. Proteins were detected in Western blots. Three different protein variant combinations and three identical variants were examined. The SEM with n ≥ 6 is shown. (B) Homotrimers are formed between elements of PSD. (C) Heterotrimers are not formed between elements of PSD. (D) PSDs form homotrimers together with full-length proteins. (E) The N-terminal/CC domain is a minimal segment for heterotrimer formation together with full-length proteins. The two values shown in (D and E) represent the measures of testing the two possible variant combinations of full-length proteins with fragments.

Fig. 6.. The CC motif encodes the shared heterotrimeric binding ability and is essential for the binding of different Csd protein variants.

(A) The position and sequences of the CC motifs in the different Csd protein variants. Light pink, heptad repeats; dark pink, hydrophobic/amphipathic amino acid residues at positions one and four. Further below, the examined amino acid replacements are shown for only one allele. (B) Schematic presentation of the proposed trimeric binding. Usually, heptad repeats form α helices with hydrophobic surfaces (displayed in dark pink). Helices from different Csd proteins (shown for helix 1 only) interact to form a trimeric complex (the CC structure). (C) The quantitative difference in homotrimer formations of mutated versus WT sequences using Western blots from SDS-PAGE using oligomer-forming (low denaturation) conditions. The SEM and the number of replicates are indicated. (D) Fluorescence lifetime values derived from combining two different Csd-CC^DQVEHR protein variants. Values from n ≥ 6 cells per condition were compared using a t test. (E) To the left: Highly resolved trimeric complexes (6% acrylamide gels) derived from combining two Csd-CC^DQVEHR or two WT Csd protein variant expressions. Lanes are from the same Western blot. Western blots were from SDS-PAGE using oligomer-forming conditions. To the right: Detected intensities along the position in each lane (lane profile) of the Western shown to the left. Arrows indicate the peak intensities.

Fig. 7.. The CC domain binding of different protein variants is essential for female determination.

(A) Schematic of the transgenic experiments to examine the function of the CC domain for the female-determining activity using CC^DQVEHR replacements. Combinations of endogenous allele and transgenic allele were tested in haploid genetic males. i/j and light/dark green color symbolize the different csd coding sequences in transgenic males. Blue arrowhead, the introduced replacements compromising the CC domain. long gray box, actin-5c promoter. (B) Female- and male-specific splicing of the fem and dsx transcripts in the transgenic males at larval stage. Resolved amplicons as black/white negatives. RT-PCRs were semiquantitatively adjusted across individuals using ef-1α (elongation factor 1α) transcripts as a control. (C) Summary table of the allele combinations tested. The feminizing activity in transgenic genetic males with two different coding sequence and compromised CC domain (csd ^{i, tg csd j CC DQVEHR}) and with two different WT alleles (csd ^{i, tg actin5C csd j}). Males (csd ⁱ) and females (csd ^i/j) were WT controls. (D) Detection of CC^DQVEHR peptide in csd ^{i, tg csd j CC DQVEHR} males. MALDI-TOF fragment spectra of the peptide (blue box), which matches the spectra of a synthetic, identical, but isotope labeled peptide (fig. S12).

Fig. 8.. The molecular regulation and evolution of complementary sex determination.

(A) The molecular basis of complementary sex determination in the honeybee. The sex locus expresses either different or identical Csd protein variants, which determine the sex. The different protein variants in females and identical variants in males are recognized by distinct bindings. In females, the amino acid differences at PSD of different Csd variants triggers the selection of CC domain binding for heteromeric complexation. In males, the absence of the amino acid differences of identical variants mediates binding at other PSD elements for homomeric Csd protein complexation. These distinct recognitions produce two possible binding states of Csd proteins (flip/flop mechanism of binding) that switch the activity state of Csd proteins either ON or OFF. The active Csd proteins (ON state) direct the female-specific splicing of the fem gene transcripts, a component of the conserved sex determination pathway. The inactive Csd proteins (OFF state) ensure male-specific splicing, which results by default regulation. i/j different csd alleles. Light versus dark green colors, different csd alleles/Csd proteins. (B) The molecular evolution of complementary sex determination. Two key components of the recognition mechanism newly and adaptively evolved after the csd gene evolutionary originated by gene duplication from the fem/tra progenitor gene. The evolution of the CC domain by amino acid replacements established a shared binding site among the Csd protein variants. The amino acid and sequence length difference establish between diverging PSDs a nonbinding ability, which is used to direct selective CC domain binding. The amino acid (aa) values are the changes for the formation of the CC domain (22) and the adaptive changes for the functional divergence of PSDs (18).