Caste-specic gut symbionts contribute to the different adult longevity in the honeybee

Honeybees are important pollinators, and their health is important to agricultural production and ecosystem. Queen-bees contain same genome as worker-bees, but live longer and healthier than worker-bees; thus, queen and worker pairs are natural biological models for studying longevity. Concerns are increasing regarding the relationship between gut microbes and honeybee health. We compared the hindgut microora of queen and worker (Apis mellifera carpatica) by sequencing the bacterial 16S DNA, then salvaging the caste-specic microbes using LEfSe analysis and predicting the microbial functions using Tax4Fun, hoping to nd potential gut symbionts associated with longevity. The hindgut microora of queens differed from those of worker. Queens had higher abundances of Commensalibacter, Lactobacillus and Bidobacterium than workers. The dominant microora in the worker hindguts were Gilliamella (29.37%), Lactobacillus (15.28%), Commensalibacter (13.65%), Snodgrassella (11.56%), Bidobacterium (6.07%) and Frischella (3.51%). The dominant microora in the queen hindguts were Commensalibacter (44.89%), Lactobacillus (38.42%), Bidobacterium (6.74%), Gilliamella (2.44%) and Bombella (2.41%). Queen-specic microbes was Bombella genus, and worker-specic microbes included Snodgrassella alvi, Frischella perrara and Gilliamella apicola. Queen and worker hindgut microbes exhibited diverse functions in lipid metabolism, neurodegenerative diseases, endocrine system, nervous system and immune system; those functions were linked to honeybee tness and longevity. The queen-specic symbiont, Bombella, was predicted to be involved in host endocrine and immune regulation, which may contribute to queens living longer and healthier than worker bees. the queen-hindgut help the queen bees live longer and healthier than the worker bees. This study may help determine the mechanisms of queen longevity and enable further understanding the positive roles of gut symbionts in honeybee tness. (R compare the microora communities of the workers and queens, the diversity and distances of the samples were analyzed. PCoA based on Bray-Curtis showed differences in gut bacterial compositions between the worker and queen bees with a profound difference along the PCO1 axis (reaching 59.47% of overall variation, P = 0.003684). The aggregation of samples within groups and dispersion of samples between groups indicated higher similarity within groups with a clear difference between groups (Fig. 1A). ANOSIM results (Fig. 1B) showed that the intergroup distance between the worker and queen bees was greater than the intragroup distance between the worker and queen bees (R = 0.926, P = 0.028). The distance index between the worker and queen samples was larger than that among samples in either the worker or queen bees (Fig. 1C). Larger distance indexes indicated greater distances between samples. These results conrmed that the worker and queen gut microora communities differed. gut microbiotas 26]. However, studies honeybee guts differences among hosts 25– 28]. Previously characterized results on Apis mellifera ligustica showed that the main core phylotypes in worker guts included Gammaproteobacteria (G. apicola), Betaproteobacteria (S. alvi), Firmicutes (Lactobacillus Firm-4 and Lactobacillus Firm-5) and Bidobacterium spp. (listed in order of highest to lowest abundance). Other studies analyzed the gut microbiomes of other species and Apis subspecies, including Apis dorsata, Apis andreniformis, Apis mellifera carnica, Apis mellifera capensis, and Apis mellifera scutellata [9, 25, 27]. The results showed that the gut microbes differed only slightly among the various honeybee species and subspecies. Most core bacterial taxa in the Apis guts included the nine dominant phylotypes mentioned above; however, some subtle differences were noted: Alpha-2.1, Firm-4 and Firm-5, which were common in Apis mellifera ligustica guts, were undetected in Apis dorsata, while A. andreniformis was universally colonized with high abundances of Alpha-1, Beta, Firm-4, and Gamma-1 [25]. Simonsiella was detected in Apis mellifera scutellata, and Gluconacetobacter, Serratia, and Simonsiella were detected in Apis mellifera capensis [9]. Gluconacetobacter, Curvibacter, Comamonas, Ralstonia, Simonsiella, Salmonella and Serratia occurred sporadically in the Apis mellifera carnica guts[28], but these bacteria were rare in the Apis mellifera ligustica guts. In the present study, the dominant microora in the worker guts (Apis mellifera carpatica, Kaqian black ring strain) were Gilliamella (Gamma-1; 29.37%), Lactobacillus (Firm-5; 15.28%), Commensalibacter (Alpha-2.1; 13.65%), Snodgrassella (Beta; 11.56%), Bidobacterium (6.07%) and Frischella (Gamma-2; 3.51%); no Lactobacillus Firm-4 was identied. Apis species or subspecies taxa, in addition to methodological and environmental factors, may have contributed to these subtle differences.

Introduction level. The higher functional abundances of worker bees (22 items) mainly involved nutrient synthesis and metabolism (11 items, 50%), of which, three items were associated with vitamins (nicotinate, folate, vitamin B6), 2 were associated with lipopolysaccharide, 2 were associated with amino acids (valine, leucine, isoleucine and taurine), 1 was associated with carbohydrates, 1 was associated with fatty acids, 1 was associated with dibasic acid, and 1 was associated with ubiquinone. The higher functional abundances in worker bees were secondly involved in genetic information processing (chromosomes, ribosome biogenesis, DNA replication proteins, chaperones and folding catalysts), with higher relative abundances and higher con dence levels. Additionally, a nutritionsensitive signaling/insulin signaling pathway differed between workers and queens, although it was less abundant.

Correlation between caste-speci c biomarker taxa and functions
Seven caste-speci c species (Lactobacillus sp SF6D, Lactobacillus kullabergenisis, Lactobacillus apis, Lactobacillus sp Adhmto19, B. intestini, F. perrara, and S. alvi; Figs. 3C and 4) were targeted to predict their associated functions at level 2 ( Fig. 6). Five functions were signi cantly associated with S. alvi (r > 0.8 and P < 0.05): folding, sorting and degradation; glycan biosynthesis and metabolism; signaling molecules and interaction; endocrine system; and excretory system, of which, folding, sorting and degradation; glycan biosynthesis and metabolism; and endocrine system were positively correlated, while the others were negatively correlated. The endocrine system and excretory system functions were signi cantly associated with B. intestini (r > 0.8 and P < 0.05). The r between immune system and B. intestini was 0.8015, but P was > 0.05. The most relevant function for L. sp. Adhmto19 was immune system (r = 0.9196 and P = 0.009432). The functional pro les of L. sp. SF6D, L. kullabergenisis and L. apis were similar, and their more relevant functions (0.5 < r < 0.8) included amino acid metabolism, metabolism of cofactors and vitamins, environmental adaptation, circulatory system and immune system diseases.
To further study the potential functions of the caste-speci c taxa, correlations between the caste-speci c taxa and the level 3 functions from the signi cantly related functions at level 2 were analyzed using Pearson correlation analysis. Figure 6B shows the level 3 functions with r > 0.5. Eight tertiary functions were signi cantly related to B. intestini, 9 were signi cantly related to L. sp. Adhmto19, 6 were signi cantly related to F. perrara, and 11 were signi cantly related to S. alvi (r > 0.8; P < 0.05). The tertiary functions closely related to B intestini and L. sp. Adhmto19 primarily comprised antigen recognition and innate immunity (RIG-I-like receptor signaling pathway, cell antigens, antigen processing and presentation, and NOD-like receptor signaling pathway), biosynthesis and degradation of polysaccharides (N-glycan biosynthesis, glycosaminoglycan degradation and glycosphingolipid biosynthesis) and endocrine signaling pathways (PPAR signaling and adipocytokine signaling pathways). The functions closely related to S. alvi included ubiquinone and other terpenoid-quinone biosynthesis, proximal tubule bicarbonate reclamation, bacterial toxins, alpha-linolenic acid metabolism and the insulin signaling pathway. F. perrara functions mainly included ubiquinone and other terpenoid-quinone biosynthesis, alpha-linolenic acid metabolism, insulin signaling pathway, sulfur relay system, ubiquitin system and chaperones, and folding catalysts. These predictive functions of the caste-speci c taxa indirectly indicated that the host bees (workers and queens) may differ in physiological functions, substance metabolism, and immune recognition and helped clarify the differences between workers and queens.
Queen bee guts contained more Acetobacteraceae, Lactobacillus and Bi dobacterium, differing from the worker bee guts In eusocial insects, there were many different phenotypes between queen and worker, including morphology, genitalia and longevity. Studies of the differences in longevity between castes are valuable because individual longevity is often closely associated with physical tness. Recent studies regarding how microbial communities in bee guts are involved in pathogen protection and nutrition metabolism have drawn attention to the impact of the microbiota on bee tness. Thus, we examined which differences in the gut microbial communities of the queen and worker bees would contribute to differences in their lifespans.
Previous studies of the gut microbial communities of queen and worker bees showed that queens lack the stable core microbiotas that are associated with the workers, although these studies used queens of different ages and reproductive stages (i.e., 4-6-month-old, 16-18-month-old, 4-day-old virgin, 14-day-old spawning and 7-day-old virgin queens) ( Anderson et al. (2018) found that the queen-speci c micro ora of Apis mellifera ligustica included Parasaccharibacter apium (Alpha-2.2) and Lactobacillus kunkeei (Firm-5); worker-speci c micro ora included Bartonella apis, F. perrara, S. alvi and G. apicola, and shared core micro ora were Lactobacillus Firm-4, Lactobacillus Firm-5, Bi dobacterium asteroids and Acetobacteraceae Alpha 2.1. Our results showed that the queenspeci c taxon was B. intestini (Alpha 2.2; Fig. 3C), the worker-speci c taxa were S. alvi, F. perrara and G. apicola (Figs. 4 and 3C), and queens had higher abundances of Commensalibacter, Lactobacillus and Bi dobacterium than worker bees (Fig. 3B). However, B. intestini was not the most abundant bacterium in the queens, but the fth most abundant, with Commensalibacter (44.89%; Alpha 2.1), Lactobacillus (38.42%), Bi dobacterium (6.74%), and Gilliamella (2.44%) ranking before it (2.41%). Powell et al. (2018) showed that Acetobacteraceae (Alpha-2.1 and Alpha-2.2) and Lactobacillus Firm-5 dominated queen Apis mellifera guts. Kapheim (2015) reported similar results in that the top four most abundant bacteria were P. apium, Alpha-2.1, Lactobacillus Firm-4 and Lactobacillus Firm-5. These studies collectively suggest that queen guts harbor more abundant Acetobacteraceae, especially Alpha-2.2, than do worker guts, likely related to the honeybee diets. Previous research concluded that Acetobacteraceae Alpha 2.2 were proli c in the crops, hypopharyngeal glands of nurse bee, royal jelly and larva fed on royal jelly, but were negligible in the nurse and forager midguts and hindguts (Anderson et al. 2013; Vojvodic, Rehan & Anderson 2013; Corby-Harris et al. 2014). Some niches were characterized by the availability of royal jelly, which is the main food of queen bees and differs from the bee bread and honey eaten by the workers. Thus, queen bee guts are a niche accessible to royal jelly, suggesting that royal jelly may promote Acetobacteraceae Alpha 2.2 proliferation. The widespread distribution of Alpha 2.2 in many niches of honeybee indicates some specialized biological functions in the hosts. Our latest study found that overwintering honeybee colonies with higher abundances of Acetobacteraceae exhibited a lower rate of overwintering loss (Wang et al. 2020a), suggesting that Acetobacteraceae plays a positive role in honeybee tness. One Alpha 2.2 isolate was shown to increase honeybee larval survival in vitro (Corby-Harris et al. 2014). Acetobacteraceae are symbionts of various insects. Studies on Anopheles and Drosophila indicated that Acetobacteraceae provided nutrition to the host (Crotti et al. 2010), bene tted host growth and development [29-31], and modulated host immunity (Ryu et al. 2008). So it was inferred that the abundant Acetobacteraceae, including Commensalibacter and Bombella, in queen bee guts might be another important bene cial bacteria enabling queen bees to live healthier and longer lives than worker bees, except for the well-known bene cial symbiotic bacteria Lactobacillus and Bi dobacterium.
Gut microbial functions in queen bees differed from those of worker bees Regarding the integral functions of gut microbes, the microbiota gene functions that differed between workers and queens at level 2 included lipid metabolism, neurodegenerative diseases, and the endocrine, nervous and immune systems (Fig. 5). Of these, lipid metabolism and the endocrine and immune systems are correlated with honeybee longevity. Regarding lipid metabolism, the fatty acid composition of the phospholipid membrane affects bee longevity [32]. The membrane phospholipids of adult worker bees include richer polyunsaturated fatty acids (PUFAs) and increase with age. However, the membrane phospholipids in adult queens remain highly monounsaturated throughout the bee's adult life [32,33]. PUFAs are 1,000 times more likely to oxidize than are monounsaturated fatty acids [34]. Accumulation of lipid oxidative damage over a lifetime is one main cause of aging [35, 36]. Studies in mice have shown that gut bacteria can alter the saturation and length of host fatty acids [37]. Therefore, gut microbes likely in uence the lifespans of queens and workers by regulating lipid metabolism. Regarding the endocrine system, the insulin pathway was shown to be associated with caste differentiation in female honeybees [38]. In the present study, the honeybee gut microbes were predicted to be functionally associated with host insulin signaling, consistent with previous studies [31,39]. Reducing insulin/insulin-like growth factor signaling (IIS) activity inhibits juvenile hormone secretion, thereby increasing vitellogenin expression [40,41]. Upregulating vitellogenin expression can prolong bee longevity [42]. Studies have shown that older queen bees have lower IIS activity than do older worker bees [40], and the insulin signaling abundance associated with gut microbes was predicted to be lower in queens than in worker bees. Thus, queen bees live longer than do worker bees. Studies of model organisms have shown that reducing IIS activity increases longevity [43][44][45]. Thus, bee gut microbes likely affect host longevity via endocrine pathways, particularly insulin signaling. Regarding the immune system, a common characteristic among aging animals is reduced immunity [46], which also occurs in aging bees [3,47,48]. That is, factors that affect bee immune performance tend to affect their longevity. Studies on insects have shown that gut microbes play key roles in establishing and regulating the host immune system [49][50][51][52], suggesting that gut microbes may affect host longevity via immunoregulation. Differences in lipid metabolism and endocrine and immune function abundance in gut microbes between queen and worker bees indicate that gut microbes contribute to the differences in longevity between queen and worker bees.
Another gut microbial function predicted to differ between female castes is folate biosynthesis. Folic acid functions as a one-carbon unit carrier involved in substance metabolism and synthesis. The nitrogen 5-trimethyl-tetrahydrofolic acid (N5-CH3-FH4) of the one-carbon units provides methyl for homocysteine to produce methionine. Methionine is activated into S-adenosylmethionine, which is the methyl donor in DNA methylation [53,54]. DNA methylation is one of the main molecular mechanisms for female bee castes differentiation [55,56]. Methionine as a methyl donor also plays a regulatory role in differentiating female honeybee castes [57]. Thus, gut microbes may be involved in regulating the ontogenetic trajectory of female bees.

Bombella positively affected honeybee tness
The functions of the caste-speci c taxa, L. sp. SF6D, L. kullabergenisis, L. apis, L. sp. Adhmto19, B. intestini, F. perrara, and S. alvi, were predicted using Tax4Fun. As worker bee-speci c taxa, the functions of F. perrara and S. alvi were previously investigated [58,59]; these microbes are involved in immunity[60], defense[61] and maintaining the anaerobic fermentation environment [39]. However, less is known about the microbial functions associated with queen bees. In this study, ve queen-speci c taxa were found, including four species of Lactobacillus and B. intestini. The bene ts of Lactobacillus on animal tness have been well documented. However, little is known about B. intestini originating from honeybees. In the present study, B. intestini appeared as a queen-speci c gut microbe; thus, it received our attention. B. intestini, which was rst isolated from bumblebee crops [62], is part of a clade of acetic acid bacteria (a group within the family Acetobacteraceae). To date, B. intestini from honeybees is unreported. However, Bombella apis, another member of the genus Bombella with 98% sequence similarity to B. intestini, was detected from honeybee midguts [63]. However, little is known about the role of genus Bombella in honeybee tness. Whole-genome sequencing of B. intestini and B. apis disclosed the general genomic features and functional annotations of the coding gene [64,65]. Fusaric acid resistance (FUSC) genes were found in the B. apis genome. Several fungal species produce fusaric acid [66]. FUSC genes in the B. apis genome indicate that B. apis in honeybees may protect honeybees from fungal infection [67]. In this study, B. intestini was predicted to be associated with immune and endocrine functions, including the RIG-I-like receptor signaling pathway, cell antigens, antigen processing and presentation, the NOD-like receptor signaling pathway, N-glycan biosynthesis, glycosaminoglycan degradation and glycosphingolipid biosynthesis, the PPAR signaling pathway and the adipocytokine signaling pathway. Of these, RIG-I-like and NOD-like receptors are members of pattern-recognition receptor families that sense nucleic acids derived from viruses and trigger antiviral innate immune responses [68,69]. Therefore, honeybee symbiotic bacteria belonging to Bombella may function by resisting pathogen infection and play important roles in maintaining honeybee health. Further studies and more direct evidence are required.     Diversity differences in the gut micro ora between workers and queens. A. Principal coordinate analysis (PCoA) plot based on Bray-Curtis similarity for the worker and queen samples. B. Analysis of similarities (ANOSIM) testing based on unweighted UniFrac was used to evaluate the intergroup (workers and queens) and intragroup (workers or queens) distances. R, interpretation degree of the different groups in the sample differences. R=(the between-group variance)/(total variance) . C. Sample distance analysis based on the Bray-Curtis distance index. Larger indices indicate greater distances between samples.

Figure 1
Diversity differences in the gut micro ora between workers and queens. A. Principal coordinate analysis (PCoA) plot based on Bray-Curtis similarity for the worker and queen samples. B. Analysis of similarities (ANOSIM) testing based on unweighted UniFrac was used to evaluate the intergroup (workers and queens) and intragroup (workers or queens) distances. R, interpretation degree of the different groups in the sample differences. R=(the between-group variance)/(total variance) . C. Sample distance analysis based on the Bray-Curtis distance index. Larger indices indicate greater distances between samples.

Figure 2
Numbers of OTUs in the worker and queen guts. Numbers of OTUs in the worker and queen guts. Gut microbial abundance and taxonomic distribution of the workers and queens. A. phylum level; B. genus level; C. species level. Column diagram, relative abundances of micro ora. Heatmap, relative abundances normalized by z-score.

Figure 3
Gut microbial abundance and taxonomic distribution of the workers and queens. A. phylum level; B. genus level; C. species level. Column diagram, relative abundances of micro ora. Heatmap, relative abundances normalized by z-score.      Correlation between caste-speci c taxa and functions using Pearson correlation analysis. A. Correlation heatmap of caste-speci c taxa and functions at level 2. B. Correlation network of caste-speci c taxa and functions at level 3 from the functions at level 2, which are signi cantly correlated with the caste-speci c taxa (A). Violet node size indicates the average relative abundances of the microorganisms. Blue node size indicates the average relative abundances of the functions. Lines linked to nodes indicate signi cant correlations between the nodes (r>0.5), with red dotted and black solid lines showing negative and positive correlations, respectively. * represents a signi cant correlation (P<0.05), ** represents a signi cant correlation (P<0.01), *** represents a signi cant correlation (P<0.001).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.