Group size influences maternal provisioning and compensatory larval growth in honeybees

Summary Environmental variation selects for the adaptive plasticity of maternal provisioning. Even though developing honeybees find themselves in a protected colony environment, their reproductively specialized queens actively adjust their maternal investment, even among worker-destined eggs. However, the potentially adaptive consequences of this flexible provisioning strategy and their mechanistic basis are unknown. Under natural conditions, we find that the body size of larvae hatching from small eggs in large colonies converges with that of initially larger larvae hatching from large eggs typically produced in small colonies. However, large eggs confer a persistent body size advantage when small and large eggs are cross-fostered in small and large colonies, respectively. We substantiate the increased maternal investment by identifying growth-promoting metabolomes and proteomes in large eggs compared to small eggs, which are primarily enriched in amino acid metabolism and cell maturation. Thus, our study provides a comprehensive adaptive explanation for the worker egg size plasticity of honeybees.


INTRODUCTION
All parents face the fundamental life-history decision of how much to invest in individual offspring produced.Often this decision results from a concurrent optimization of offspring number and size, 1,2 but trade-offs with survival or future reproduction are also prominent. 3The exact outcome of such trade-offs is influenced by the external environment that affects the costs and benefits to parent and offspring. 4Low-quality environments can increase the benefit of increased parental investment per offspring due to disproportional gains in offspring survival. 5This observation has led to the adaptive investment hypothesis, which explains that larger offspring or heightened parental care frequently occur in poor compared to favorable environments. 6,7[10] Parental investment per offspring is readily quantified in females of oviparous species and many studies have characterized egg size plasticity in plants and animals, particularly birds and insects. 1,11Social insects, including the highly social honeybees of the genus Apis, are a notable exception to this abundance of studies, presumably because their colony organization and social protection mechanisms 12 are thought to generate a highly predictable environment, which does not promote selection for egg size plasticity. 13However, honeybee queens, females that are solely specialized on reproduction, adjust the size of their eggs in multiple contexts: Queens produce larger eggs when ovipositing into specialized wax cells that predetermine the resulting offspring to be raised into new queens than when ovipositing into regular worker cells. 14Concomitantly, larvae hatching from larger eggs have a higher probability to be reared as new queens than larvae from smaller eggs when nurse bees are given the choice, 15 and larger eggs also result in higher-quality queens, which might even have effects on the quality of offspring that these queens produce. 16,17oneybee queens also increase the size of worker-destined eggs in response to small colony size and low food availability. 18,19This correspondence to the adaptive investment hypothesis 2,6 is surprising because even in small colonies hundreds of workers are available for nursing the brood based on stored food resources 20 and food shortages lead to demographic responses at the colony level that can buffer individuals. 21Thus, we set out to investigate the consequences of worker egg size plasticity for juvenile development in the context of small and large colonies.We identified growth advantages of larvae from large eggs that were dependent on colony size because the growth of larvae that hatched from small eggs in large colonies show compensatory growth.Our subsequent comparison of the small-versus large-egg ll OPEN ACCESS maternal provisioning tactics with a combined metabolomics and proteomics approach demonstrated that large eggs contain not only more material than small eggs but are also better primed for growth and development.

RESULTS AND DISCUSSION
Facilitated by their social organization, intense brood care, and protected hive environment, honeybee workers grow 1000-fold during five days from a 0.1 mg egg to a full-sized fifth larval instar. 22The highly social, homeostatically controlled hive environment thus may seem to make maternal plasticity in egg provisioning superfluous, in contrast to most other species. 6Yet, we discovered previously that honeybee queens reversibly adjust egg size according to perceived colony size; eggs were more than 10% smaller in colonies with 16,000-20,000 workers than in colonies with 500-700 workers. 19

Compensatory late larval growth in large colonies
To determine the developmental consequences of this difference, larvae from small and large eggs that developed in their own colony were weighed at each larval instar at 24-h intervals and as emerging adults (Figure 1A; Table S1).Larvae hatching from large eggs were significantly heavier as first instar (t (5,5) = 7.3, p = 0.00008) and second instar (t (5,5) = 10.8, p = 0.000005), and to a lesser extent also as third (t (5,5) = 2.3, p = 0.048) and fourth (t (5,5) = 2.5, p = 0.035) instar.However, after five days, the two groups reached similar in weights (t (5,5) = À0.9, p = 0.42) and continued to display similar weights one day later (t (5,5) = À0.5, p = 0.62) and as adults (t (5,5) = 0.7, p = 0.50).These results were confirmed in a second replicate (Figure 1A; Table S1) and demonstrate that larger eggs convey an advantage in the developmental growth trajectory of young honeybees, 23 which is common in insects 1 and other organisms. 11The results also show that honeybee larvae hatching from small eggs in large colonies exhibit compensatory growth, 24,25 facilitated by nursing workers that also stabilize the phenotype of developing workers. 26,27However, this growth might come at a survival cost, which could potentially explain the survival advantage of large over small eggs when reared in a common, large-colony environment.

Social environment limits compensatory growth in small colonies
In a reciprocal cross-fostering experiment of small eggs in small colonies and large eggs in large colonies, the initial weight differences persisted into adulthood (Figure 1B; Table S1).Large eggs developed into significantly heavier individuals than small eggs across all measurement time points (L1: t (5,5) = 8.0, p = 0.00004, L2: t (5,5) = 12.6, p = 0.000002, L3: t (5,5) = 6.0, p = 0.0003, L4: t (5,5) = 10.2, p = 0.00008, L5: t (5,5) = 10.9, p = 0.000004, L6: t (5,5) = 17.1, p < 0.000001, Adult: t (5,5) = 10.8, p = 0.000005).These findings were also confirmed by matching results of a second replicate (Figure 1B; Table S1).The sharp contrast between the first and second experiments highlights the importance of colony size for brood care and the social environment in general. 28,29Nurse bees play a key role in honeybee development and the phenotype of the offspring, 26,30 and the observed enhanced larval development in large colonies thus may be due to enhanced nursing.However, it remains to be tested whether the quality of honeybee nursing behavior changes with colony size or other potential benefits of large colony size, such as better homeostatic control, 12,31 can explain our results.Although it is common for mothers to reduce maternal investment in many social species in response to alloparental helpers, 9 our finding is remarkable because honeybees are swarm-founding species that always live in large groups with thousands of workers throughout their life cycle. 32By demonstrating individual-level consequences of the queens' adjustment of egg size, we resolve the open question of the adaptive value of the observed egg size plasticity, 19 although it remains to be studied how these individual effects translate to colony-level effects.

Large eggs contain more metabolites and proteins related to growth and development
To better distinguish adaptive from non-adaptive explanations of the observed egg size plasticity in honeybees, we compared small and large eggs in terms of metabolome and proteome content.Capitalizing on eggs collected during a reciprocal queen transfer experiment that was performed in July/August 2021 and indicated the reversibility of egg size adjustment in honeybees, 19 we compared pools of 100 eggs per queen when they produced large eggs (in small colonies) and small eggs (in large colonies).Data from ultra-high performance liquid chromatography coupled with high-resolution mass spectrometry input to various databases yielded 390 compounds with an average peak area > 1E5.The four groups separated according to date (July vs. August 2021) and egg size (Figure S1).In July, 120 metabolites were significantly more abundant and 24 less abundant in large compared to small eggs.In the August samples, 98 more-and 17 less-abundant metabolites were found correspondingly.Core sets of 78 more-and 7 less-abundant metabolites were derived from the overlap of all four possible comparisons between large and small eggs across time points (Figure 2A).The overlap among more-abundant metabolites in large eggs (78/143) was significantly (Fisher's Exact p < 0.0001) higher than the overlap among the more-abundant metabolites in small eggs (7/42).
Based on these core sets, egg size divided the samples into two exclusive clusters (Figure 2B; Table S2).KEGG enrichment analysis identified 17 significantly enriched pathways in large eggs with the following top nine: ''Alanine, aspartate and glutamate metabolism,'' ''ABC transporters,'' ''beta-Alanine metabolism,'' ''Arginine and proline metabolism,'' ''Cysteine and methionine metabolism,'' ''Pantothenate and CoA biosynthesis,'' ''Glycine, serine and threonine metabolism,'' ''Butanoate metabolism,'' and ''Aminoacyl-tRNA biosynthesis'' (Figure 2C; Table S3).The higher relative quantity of a number of metabolites in large eggs than in small eggs indicates that the flexible reproductive strategy of queens involves quantitative and qualitative differences.Large eggs not only provide more mass but also better nutrients, particularly in terms of amino acid metabolism, in contrast to houseflies 33 and butterflies. 34Resource transfers from workers to queens in eusocial species can prevent resource-driven reproductive trade-offs 35,36 and thus may facilitate the positive association between size and quality of eggs.Early embryonic development in honeybees is characterized by protein degradation and synthesis with amino acid intermediates, 37 and the metabolome profiles suggest that these processes are up-regulated in large eggs.A higher abundance of metabolites related to ABC transporters also suggest that transport may be upregulated in larger eggs, but two ABC transporter gene sub-families in insects are also related to protein biosynthesis. 38ikewise, the LC-MS/MS-based proteome characterization indicated that more proteins were up-regulated in large relative to small eggs.Our four experimental groups clustered without overlap according to date and egg size based on 2040 identified protein groups, of which 1829 could be reliably quantitated (Figure S2).In July, 275 proteins were significantly more abundant and 72 less abundant in large compared to small eggs.In August, 260 more-and 64 less-abundant proteins were detected correspondingly.Based on overlap among all four possible contrasts between small and large eggs across both sampling dates, a core set of 243 more-and 56 less-abundant proteins was identified (Figure 3A; Tables S4, S5, S6, S7, and S8).The shared core represented a significantly (Fisher's Exact p = 0.0002) higher proportion of the proteins that were more abundant in large eggs (243/318) than of proteins up-regulated in small eggs (56/99).The core proteome profiles were sufficient to separate all samples into two principal clusters of small and large egg samples, with secondary clustering mostly according to sampling date (Figure 3B).
Thus, the proteome content of the differently sized eggs corroborates our interpretation of the metabolomic differences; in addition to their size advantage, large eggs are also qualitatively superior to smaller eggs, enriched with more proteins that function in cellular growth, maturation, and development.For example, the up-regulated core proteins contained insulin-like growth factor I (IGF-1) and eight translation initiation factors.IGF-1 controls metabolism 39 and plays a key role in honeybee caste differences 40 and division of labor. 41In contrast to adults, 41 honeybee eggs display a positive association between IGF-1 and nutrient levels that is common in other species. 42These proteome differences presumably provide developmental benefits, 43 enhancing embryogenesis and subsequent growth, which can thus explain the observed early developmental advantage of large eggs. 44Cellular transport processes, including cytoskeleton organization, are upregulated in ovaries that produce large eggs, 19 providing a mechanistic explanation for the better provisioning of large compared to small eggs.
Overlap analysis of the core proteome with the proteome profiles of honeybee eggs aged 24, 48, or 72 h 37 indicated no significant relationship (Fisher's exact p = 0.73) between egg size and temporal differences, refuting the hypothesis that larger eggs simply undergo accelerated development.This result is corroborated by a similar lack of significant directional overlap between our core protein set and Drosophila proteins that change during the egg maturation process upon fertilization 43 (Fisher's exact p = 0.31, n = 16).The enrichment of proteins related to ''Pole plasm oskar mRNA localization'' is intriguing because oskar is not present in honeybees. 45Thus, the finding should be interpreted to indicate more generally an up-regulation of mRNA-binding proteins, 46 which is also corroborated by our KEGG pathway and PPI results.The PPI analysis further indicated ''Innate Immunity'' as one major cluster containing five proteins up-regulated in small eggs and 16 up-regulated in large eggs.This mix suggests varying the regulation of different aspects of immunity and related processes, and it explains why GO and KEGG analysis did not identify this biological function as an overall difference between small and large eggs.9][50] Many proteins related to energy generation and homeostasis were up-regulated in larger eggs, indicating a more intensive

Metabolomic Differences between Small and Large Eggs
Overlap analysis among different comparisons between small and large eggs indicated that more metabolites are over-abundant in large eggs compared to small eggs and that most of the differences are shared among comparisons (left Venn diagram).In contrast, few metabolites were more abundant in small compared to large eggs and the majority was not shared among all comparisons (right Venn diagram) (A).Clustering based on the abundance of core differential metabolites (indicated by the heatmap) unambiguously separated small and large eggs but did not completely resolve the temporal differences between July and August samples of large eggs (B).KEGG pathways enrichment analysis indicated 17 pathways (p < 0.05) with dot size symbolizing the number of metabolites and color representing the ratio (number of identified metabolites/total number of metabolites) for each pathway (C).LJ: large colonies in July; SJ: small colonies in July; LA: large colonies in August; and SA: small colonies in August.metabolism, 51 and our results also suggest that proteins involved in cell maturation and development are upregulated correspondingly.However, summarized in the GO terms ''Proton transport,'' ''centrosome duplication,'' and ''ribosomal proteins,'' some growth-related proteins were also found to be more abundant in small eggs.Potentially, these processes lay the foundation for compensatory larval growth after egg hatching under adequate conditions, which was observed when small eggs developed in large colonies.

Conclusions
Overall, our study characterizes the qualitative differences that accompany egg size plasticity in honeybees, a species that should not be strongly selected for such plasticity because its advanced sociality provides a homeostatically regulated colony environment in which the offspring develop comparatively well-buffered from environmental fluctuations.The identified metabolomic and proteomic differences between small and large eggs suggest that there are no size-quality trade-offs in honeybee egg production.A well-defined and large set of metabolites and proteins are upregulated in large eggs, which are therefore larger and higher quality than small eggs.We further SJ vs LJ Similar to our metabolomic results, a large number of proteins is more abundant in large relative to small eggs (A).The core of upregulated proteins in large eggs that are shared among all four possible contrasts between small and large eggs contained 243 proteins (left Venn diagram), more than 4x the number of proteins in the corresponding core set for proteins upregulated in small eggs (56, right Venn diagram).Based on these core proteins, perfect clusters according to egg size and sampling date form, except for one large egg sample from July that clustered with the large egg samples from August (B).GO terms enriched in the proteins upregulated in large eggs eight GO terms, while the corresponding list in small eggs contained three (C).LJ: large colonies in July; SJ: small colonies in July; LA: large colonies in August; and SA: small colonies in August.
demonstrate the consequences of the different provisioning strategies; under normal circumstances, early developmental advantages of large eggs in small colonies are compensated by small-egg offspring that develop in large colonies to achieve similar adult body size.However, in offspring that lack their typical social environment, the initial developmental differences translate into different sized adults.The consequences of different body size in honeybee workers are unexplored, and it remains to be studied whether egg size plasticity of honeybee queens is a passive consequence of their tremendously plastic egg-laying rate or an evolved reaction norm with either current or past adaptive value.

Limitations of the study
Our study was performed with unselected honeybees of the species Apis mellifera L., which is a diverse, cosmopolitan species and therefore we cannot exclude the possibility that our results are specific to the populations studied here.In addition, the proteomic and metabolomic work was done with samples from North America, while the growth experiments were performed in colonies from China.We cannot exclude the possibility that this could lead to a disconnect between these two parts of our study even though our previous study 19 indicates that the observed effects are comparable in both populations.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:  The protein-protein interaction network analysis indicated six major clusters and numerous solitary proteins.One of the clusters (structural constituent of ribosome) was dominated by proteins upregulated in small eggs (colored blue), while the other five clusters and solitary proteins mostly contained proteins upregulated in large eggs (colored pink).The clustering thus reflects the overall results that large eggs are enriched in numerous proteins and also suggests that this enrichment is related to many biological functions because it includes several clusters of proteins and various solitary proteins.
(M, +15.99); maximum allowed variable PTM per peptide = 3.A fusion target-decoy approach was used for the estimation of FDR and controlled at % 0.01 both at peptide and protein levels.Proteins were identified based on at least one unique peptide.
Quantitative analysis of egg proteome from different experiment groups was performed using the PEAKS Q module.Feature detection was performed separately on each sample by using the expectation-maximization algorithm.The features of the same peptide from different samples were reliably aligned together using a high-performance retention time alignment algorithm.The label-free quantitation was applied based on extracted ion chromatograms and normalized using total ion chromatograms.These analyses were not performed blind.
Venn diagrams were again generated using Venny v2.1.0,which reflects the number of shared proteins in different sample groups.All the quantified proteins were used for PCA by SIMCA v14.1 software (Umetrics).
A heatmap of the overlapping differentially expressed proteins was created and clustered using TBtools v1.1043 software. 54The clustering was performed with Euclidean distance and complete method.

Bioinformatics analyses
In order to provide a better interpretation of the experimental results, honey bee proteins were mapped to Drosophila melanogaster for further bioinformatics analyses using KOBAS v3.0. 55Proteins of interest were uploaded in fasta format, Drosophila melanogaster was selected as target species, and sequence similarity mapping was conducted with default cutoffs (BLAST E-value < 1E-5 and rank % 5).
To perform functional enrichment analysis of differentially expressed proteins, the ClueGO + CluePedia v2.5.7, 56 a Cytoscape v3.8.2 57 plugin, was employed to enrich functional categories based on biological processes of Gene Ontology (GO).Two-sided hypergeometric test (enrichment/depletion) with p-value % 0.05 was used followed by a Bonferroni correction.The GO tree interval was set between 3 to 8 with a minimum of five genes and 1% of all genes.Kappa score R 0.4 was applied to generate term-term interrelations and functional groups based on shared genes between the terms.KEGG pathway enrichment analysis was done by Metascape 58 with default settings: min overlap = 3; p-value cutoff = 0.01; min enrichment = 1.5.
For the exploration of functional protein connections involved in biological activities, proteinÀprotein interaction (PPI) networks were constructed by STRING v12.0. 59A full STRING network was built with medium confidence (0.4) and medium FDR (0.05).The PPI networks were visualized using Cytoscape v3.8.2.

QUANTIFICATION AND STATISTICAL ANALYSIS
For the analysis of larval growth, the weight data were analyzed by two-tailed unpaired student t tests for each larval stage because the larvae were weighed destructively and thus a repeated measures approach was inappropriate.Significantly different metabolites between pairs of experimental groups were determined by variable influence on projection (VIP R 1) values derived from the OPLS-DA result, Benjamini-Hochberg corrected p-value (FDR % 0.05).The label-free quantitation was applied for protein quantification based on extracted ion chromatograms and normalized using total ion chromatograms.These data were analyzed with pairwise ANOVAs to specifically target the four pairwise comparisons between small and large eggs (LJ vs. SJ, LJ vs. SA, LA vs. SJ, and LA vs. SA), adjusting for multiple comparisons with the Benjamini-Hochberg correction (FDR % 0.05).Sample sizes were determined in advance based on the results of previous results, 18 and all available data were included in the analyses.

Figure 2 .
Figure 2. Metabolomic Differences between Small and Large EggsOverlap analysis among different comparisons between small and large eggs indicated that more metabolites are over-abundant in large eggs compared to small eggs and that most of the differences are shared among comparisons (left Venn diagram).In contrast, few metabolites were more abundant in small compared to large eggs and the majority was not shared among all comparisons (right Venn diagram) (A).Clustering based on the abundance of core differential metabolites (indicated by the heatmap) unambiguously separated small and large eggs but did not completely resolve the temporal differences between July and August samples of large eggs (B).KEGG pathways enrichment analysis indicated 17 pathways (p < 0.05) with dot size symbolizing the number of metabolites and color representing the ratio (number of identified metabolites/total number of metabolites) for each pathway (C).LJ: large colonies in July; SJ: small colonies in July; LA: large colonies in August; and SA: small colonies in August.

CFigure 3 .
Figure 3. Proteome Differences between Small and Large EggsSimilar to our metabolomic results, a large number of proteins is more abundant in large relative to small eggs (A).The core of upregulated proteins in large eggs that are shared among all four possible contrasts between small and large eggs contained 243 proteins (left Venn diagram), more than 4x the number of proteins in the corresponding core set for proteins upregulated in small eggs (56, right Venn diagram).Based on these core proteins, perfect clusters according to egg size and sampling date form, except for one large egg sample from July that clustered with the large egg samples from August (B).GO terms enriched in the proteins upregulated in large eggs eight GO terms, while the corresponding list in small eggs contained three (C).LJ: large colonies in July; SJ: small colonies in July; LA: large colonies in August; and SA: small colonies in August.

Figure 4 .
Figure 4. Interaction Analysis of Proteome Differences Between Small and Large EggsThe protein-protein interaction network analysis indicated six major clusters and numerous solitary proteins.One of the clusters (structural constituent of ribosome) was dominated by proteins upregulated in small eggs (colored blue), while the other five clusters and solitary proteins mostly contained proteins upregulated in large eggs (colored pink).The clustering thus reflects the overall results that large eggs are enriched in numerous proteins and also suggests that this enrichment is related to many biological functions because it includes several clusters of proteins and various solitary proteins.

TABLE
d RESOURCE AVAILABILITY B Lead contact B Materials availability B Data and code availability d EXPERIMENTAL MODEL AND SUBJECT DETAILS d METHOD DETAILS B Larval development rate measurement B Egg metabolome analysis B Egg proteome analysis B Bioinformatics analyses d QUANTIFICATION AND STATISTICAL ANALYSIS