The gene expression network regulating queen brain remodeling after insemination and its parallel use in ants with reproductive workers

Abstract

Caste differentiation happens early in development to produce gynes as future colony germlines and workers as present colony soma. However, gynes need insemination to become functional queens, a transition that initiates reproductive role differentiation relative to unmated gynes. Here, we analyze the anatomy and transcriptomes of brains during this differentiation process within the reproductive caste of Monomorium pharaonis Insemination terminated brain growth, whereas unmated control gynes continued to increase brain volume. Transcriptomes revealed a specific gene regulatory network (GRN) mediating both brain anatomy changes and behavioral modifications. This reproductive role differentiation GRN hardly overlapped with the gyne-worker caste differentiation GRN, but appears to be also used by distantly related ants where workers became germline individuals after the queen caste was entirely or partially lost. The genes corazonin and neuroparsin A in the anterior neurosecretory cells were overexpressed in individuals with reduced or nonreproductive roles across all four ant species investigated.

Fig. 1. Experimental groups and time points.

(A) Conceptual design showing ants have two fundamental developmental bifurcations: caste differentiation (virgin gynes versus lifetime unmated workers established before pupation) and adult role differentiation (virgin gynes versus inseminated queens). Left: The lifecycle and associated differentiation processes of M. pharaonis are unusual because caste is decided already in the egg stage and gynes failing insemination assume worker roles and survive alongside inseminated queens. Right: Several other ants secondarily lost the queen caste, partly or entirely (dashed arrows), and evolved role differentiation between reproductive and sterile workers via gamergate replacement queens or parthenogenesis. Queenless Dinoponera colonies always have a single gamergate, but Harpegnathos colonies have multiple gamergates forming a dominance hierarchy (fig. S1). (B) Cohorts of M. pharaonis gynes were divided, such that one group was allowed to mate on the third day to become inseminated queens, while the other group continued maturation without insemination. On day 4, all individuals were CO₂-anesthetized. Gynes (blue) were collected at the same time points as queens (red) for neuroanatomical and transcriptomic investigations. (C) Behavioral observation experiments (n = 10, 3 hours per subcolony experiment) showed that the motility (average active time) and foraging activity (average number of foraging runs) of day 30 queens were significantly lower than that of same age gynes (t₁₈ = 11.635, P < 0.001 and t₁₈ = 3.311, P < 0.05, respectively). *P < 0.05, ***P < 0.001.

Fig. 2. Comparison of neuropil volumes of M. pharaonis gynes and queens over time.

(A) Representative brain reconstruction of a day 5 gyne obtained from image stack of transparent head. In total, 64 brains were reconstructed in parallel series with AMIRA software 6.2.0 and enabled volume quantification based as sums of paired (ALs, OLs, and MBs) or triplet (OCs) parts. (B) Volumes of whole brains showed significant age or role differentiation effects in single comparisons (*P < 0.05) on days 15 and 30 and between days 5 and 15 gynes (horizontal *). An overall two-way ANOVA gave a stronger brain volume effect of gyne/queen status (F_1,44 = 8.12, P = 0.007), adjusted for age and combining all sample sizes given above the x axis. (C) Volumes of separate brain parts also showed significances (*P < 0.05) in single t test comparisons for gyne/queen status (vertical *) or age (horizontal *). Capturing all data (days 5, 15, and 30) in a single three-way ANOVA again offered a more powerful test of overall differences (table S1). In particular, the age × gyne/queen status interaction term (F_2,220 = 4.96, P = 0.0078) illustrated the generality of bifurcation across brain parts, while the brain-part × gyne/queen status interaction (F_4,220 = 5.05, P = 6.54 × 10⁻⁴) illustrated that bifurcation also varied across brain parts (see table S1 for details and fig. S2 for relative volumes).

Fig. 3. Development of differential gene expression as gyne-queen role differentiation in M. pharaonis proceeds.

(A) PCA of the brain transcriptomes of gynes and queens of different ages, with each point representing the transcriptome from a single brain, colored according to binary reproductive role and with symbols representing different ages. The mean PC values for gynes and queens at days 1, 5, 15, and 30 were connected chronologically by solid lines (arrowed, young to old), showing that transcriptome profiles were similar on day 5 but had diverged on days 15 and 30. The larger scatter among the gyne dots may have contributed to the number of significant DEGs between gynes and queens on day 30 being less than on day 15. (B) The number of DEGs between gynes and queens of the same age increased drastically between days 5 and 15, which is the time period during which gynes started laying unfertilized eggs and queens fertilized (and possibly also some unfertilized) eggs, after which no further increase at day 30 was observed. (C) Multidimensional scaling clustering according to semantic similarities (gene membership) of Gene Ontology (GO) terms (62). Representative gene members were written next to GO clusters, showing that genes with queen-biased expression were involved in hormonal responses on day 15 and in biogenesis and metabolic processes on day 30, whereas genes with gyne-biased expression on day 30 were mainly involved in ion transport functions.

Fig. 4. Network analysis documenting effects of insemination in M. pharaonis queens.

(A) Coexpression profiling of DEG connectivity associated with gyne-queen role differentiation, calculated as the proportion of coexpressed genes among all 648 DEGs, ranging from 0 (no coexpression with other genes) to 1 (coexpression with all other genes). The dashed rectangle highlights 191 genes exclusively and highly activated in queens 5 days after eclosion (2 days after insemination). (B) Transitional coexpression networks for brain transcriptomes of gynes (day 1) and gynes and inseminated queens (day 5). Connected gene pairs had significantly correlated expression levels (P < 1 × 10⁻³). The 191 high-connectivity genes in young queens (red dots) clearly cluster relative to the remaining DEGs (gray dots), but the gyne coexpression network has hardly changed. Three genes (blue, yellow, and green for tachykinin, ChAT, and nAChRα2, respectively) are highlighted, showing that they were all closely connected (coexpressed) in inseminated queens, while tachykinin was isolated in gynes (see fig. S4 for coexpression networks at other time points). The coexpression networks are visualized with a force-directed graph drawing algorithm where edges are assumed to be of equal length (63), so that genes with higher interconnections (coexpression coefficient) cluster together.

Fig. 5. Expression comparisons for genes involved in gyne-queen role differentiation for M. pharaonis and for three species from other ant subfamilies that secondarily evolved reproductive workers.

(A) Gene expression profiles (projected scores along the first PC axis) in the three ant species with reproductive role differentiation in the worker caste (left) and in the gyne caste for M. pharaonis differentiated by age (right) as illustrated in Fig. 1A. The reference PC axis for M. pharaonis was extracted from the combined data on days 15 and 30, using 170 DEGs with consistent differential expression on these 2 days and with one-to-one orthologs in all four ant species. The plotted values were the PC scores calculated from the expression profiles of these 170 genes, weighted by the gene-specific relative contributions to the overall gyne-queen DEG segregation in M. pharaonis. (B) Expression levels for two genes that were significantly differentially expressed between gynes and queens of M. pharaonis and exhibited the same direction of expression bias in nonreproductive (NR) versus reproductive (R) workers in the three other ant species. Dots represent the gene expression levels in each brain sample and solid lines connect the averages. Even shallow negative slopes represent considerable differences because expression abundances are plotted on a logarithmic scale. TPM, Transcripts Per Million.

Fig. 6. Localization of Crz- and NPA-positive cell clusters in the brain of a 5-day-old gyne.

(A to D) Confocal microscopy image stacks for a gyne brain of M. pharaonis with (A) only nuclei staining [4′,6-diamidino-2-phenylindole (DAPI); blue], (B) mRNA in situ HCR probes of Crz-positive cells (yellow), (C) mRNA in situ HCR probes of NPA-positive cells (green), and (D) combined view after reducing all signals to a two-dimensional projection view. (E) Volume reconstructions of neuropils and crz-positive tissues enabled the localization of two mirrored crz-positive cell clusters between the medial and lateral calices and anterior to the respective hemispheres (LNSCs). The NPA-positive cells are part of the MNSCs surrounding the medial nerve fiber that connects the medial OC to the brain. No DAPI signal was detected in the LNSC or the MNSC, likely because the nuclei in these cell clusters are highly active. Scale bar, 50 μm.