DISCUSSION
Our results reveal that the manipulation of social structure affects the development of the gut microbiome and associated physiology. Worker midgut and ileum microbiotas differed in composition and abundance by both age and behavioral task (Figure 2 and Table 1), as did immune and oxidative stress gene expression (Figure 5). Based on our chosen genes, host physiology was best explained by an interaction of age and task, with the greatest variation explained by behavioral task. This result is likely a reflection of differential nutrition and Vg titers associated with both age and task performance (Fig. 5). Beyond its role in nutrition, Vg is a potent antioxidant. We found that precocious foragers (PFs) with low Vg expression incur oxidative damage via the accumulation of carbonyl contents in the hemocoel at a significantly greater rate than both age-matched nurses and age-right (27-day) foragers (Figure 6). Young bees transition to foraging faster when there is limited social contact with older bees (Z.-Y. Huang & Robinson, 1996; Pankiw, 2004). This can occur in response to various biotic (predators, pathogens) or abiotic (pesticides, anthropogenic factors) pressures. The nutrient deficient physiology of PFs may be poorly suited for tasks outside the hive (Vance et al., 2009) and evidence suggest that PF individual risk of death increases relative to older foragers (Prado et al., 2020). Precocious foraging is also less productive (Chang, Barron, & Cheng, 2015), factors that conspire to accelerate colony loss (Perry, Søvik, Myerscough, & Barron, 2015).
Associated with the control of complex social behaviors in honey bees, vitellogenin (Vg) is a major reproductive protein that functions as a regulatory gene in workers (Gro V. Amdam, Norberg, Fondrk, & Page, 2004). Consistent with this result, we found that Vg expression was differentially expressed between 7-day old nurses and PFs of the same age (Figure 5A). High titers of circulating Vg provide several benefits to young bees; immunity, oxidative stress resistance, increased longevity, and suppression of foraging behaviors (Gro V. Amdam, 2011; Gro V. Amdam et al., 2005; Marco Antonio, Guidugli-Lazzarini, Do Nascimento, Simões, & Hartfelder, 2008; Nelson, Ihle, Fondrk, Page, & Amdam, 2007; S.-C. Seehuus et al., 2006). Because Vg expression is tightly coupled with division of labor, we can confirm that our experimental design was successful in distinguishing the general physiology associated with nursing and foraging.
The midgut microbiota varied in composition based on age and behavioral task. Previous studies have suggested that the nurse midgut is inhospitable to microbial colonization because of the continual shedding of the peritrophic membrane (Harwood & Amdam, 2021). Here we report the aging midgut as a potential niche for microbial invasion based on aging gut physiology and forager gene expression (Figure 3A, 5). Recently, it was shown that the peritrophic membrane is absent or greatly reduced in foragers (Harwood & Amdam, 2021), which may leave the tissue vulnerable to microbial opportunism. We saw a massive increase in midgut microbiome size in 27-day-old foragers (Figure S1). Gilliamella apicolaabundance in the midgut was greatest in age-matched nurses vs. foragers and comprised nearly half of bacterial cells in 27-day-old foragers.G. apicola often dominates the midgut (Ludvigsen et al., 2015), but is often lacking in the ileums of young nurse workers, becoming better established at middle age (Anderson & Ricigliano, 2017).G. apicola strains have varying capabilities to degrade pollen cell wall components (Engel & Moran, 2013a), metabolize toxic monosaccharides (Zheng et al., 2016), and encode battlefield genes such as type VI secretion systems (Steele, Kwong, Whiteley, & Moran, 2017). Because we clustered OTU’s at 97% sequence similarity, G. apicola strain taxonomic refinement is lost and more research is required to unlock their individual genomic potential (Anderson & Ricigliano, 2017). G. apicola abundance in the ileum showed the strongest relationship with task, establishing more efficiently in nurses than precocious foragers. Tasks within the hive may increase the chance of compatible G. apicola establishment, or fortify its establishment via other mechanisms.
The succession of gut bacteria in honey bee workers is poorly known, and is typically considered in terms of both task and age. The G. apicola / S. alvi relationship is one primary metric of a healthy gut microbiome. For the adult worker, S. alvi is considered a keystone species in the ileum/pylorus, interfacing with host epithelium and creating a biofilm with G. apicola and Lactobacillusfirm5 (Martinson et al., 2012). S. alvi protects the host from opportunism (Maes et al., 2016), while producing nutrients to support other gut bacteria (Kwong & Moran, 2016). Both G. apicola andS. alvi increase with age and stabilize in ratio abundance in the midgut and ileum, succession apparently accelerated and reinforced by an extended nursing role in early life (Tables S5, S6). With natural ontogeny, variation of the three major ileum bacteria was minimized at 27 days of age relative to younger bees, suggesting a long-term adult refinement of ileum microbiome structure. In stark contrast, the midgut microbiota of the oldest workers was an order of magnitude larger than younger bees including significantly more (non-core) bacterial diversity, suggesting opportunism (Fig. 3). The different successional patterns documented between midgut and ileum might suggest an important function of the gut microbiome in mitigating gut opportunism and dysbiosis. Following the decreased production of midgut peritrophic membrane, bacteria populate the midgut, a process that may rely on early bacterial succession and strong establishment of core ileum species. This hypothesis is supported by the significant negative association ofG. apicola abundance with the abundance of non-core ”other” diversity in the midgut of 27-day-old bees (Rsq = 0.42, F = 10.3, p < 0.006). Both G. apicola and Enterobacteriales increase with age in the hindgut and are positively associated with fungal abundance suggesting opportunism associated with senescence (Maes et al. 2020).
Although not considered part of the core gut microbiome, Enterobacteriales have been repeatedly discovered in the honey bee gut (Corby-Harris et al., 2014; Dalenberg, Maes, Mott, Anderson, & Spivak, 2020; Kešnerová et al., 2020; Maes et al., 2021), and here we found that they increase with age in both the midgut and ileum regardless of behavioral task. Many genera belonging to Enterobacteriales have significant impacts on human health and recently, a strain ofSerratia marcescens was shown to be an underreported pathogen of honey bees (Burritt et al., 2016; Raymann, Coon, Shaffer, Salisbury, & Moran, 2019). Enterobacteriales and other lesser understood genera are omnipresent in the guts and hives of honey bees and require more investigation in relation to the aging midgut.
Apilactobacillus kunkeei and Fructobacillus fructosus were highly intercorrelated and more abundant in nurses than foragers.A. kunkeei is often considered a hive and foregut-inhabiting bacteria that favors oxygen and high fructose niches but is also found in the hindgut in low numbers. F. fructosus is often ignored as a transient member of the microbiota, but recently it was discovered that it can utilize lignin, another component of pollen (Rokop, Horton, & Newton, 2015). Coculture assays demonstrate that A. kunkeei andF. fructosus support the growth of other honey bee symbionts considered “core hindgut bacteria” (Rokop et al., 2015). Together, this suggests that so-called transient microbes may have a co-evolved functional roles within the honey bee gut as more ubiquitous microbial members. Core gut microbiota Firm4, Firm5-2, and Bifidobacterium sp . were found together and explained by age and behavioral task because they were more abundant in foragers relative to nurses. Our sequences clustered Firm5 as two distinct clusters with Firm5-1 most abundant in the ileum. This is supported by previous research indicating there are many adapted strains of Firm5 throughout the honey bee gut, some more proximately available to takeover niches during age-based succession (Anderson, Rodrigues, Mott, Maes, & Corby-Harris, 2016).
Overall we detected gene expression patterns explained primarily by the interaction of age and task (Table S7). Antimicrobial peptides (AMPs) are expressed as part of the innate immune system of the honey bee (reviewed in Alberoni et al. , 2016). Explained primarily by task, we found high levels of hymenoptaecin in all ages of foragers and strong upregulation of apidaecin in 27-day foragers. This pattern is supported by studies that show foragers express genes encoding AMPs greater than nurses (Vannette, Mohamed, & Johnson, 2015). Similar to our findings, honey bees inoculated with gut microbiota or mono-colonized withS. alvi , upregulated apidaecin and hymenoptaecin constitutively in the fat body (Kwong, Mancenido, & Moran, 2017). The core microbiota tends to have increased tolerance for host AMPs compared to allochthonous microorganisms (Kwong et al., 2017), therefore it’s advantageous to constitutively express AMPs as a prophylactic measure given that foragers are exposed to more pathogen pressure outside the hive. Younger foragers also expressed DSCAM higher than age-matched nurses, with an age-associated decline by 19-days (Figure 5D). InDrosophila , DSCAM is a highly diverse Ig-superfamily receptor that may affect phagocytic uptake of bacteria by host hemocytes (Watson et al., 2005). The number of honey bee hemocytes decrease in relation to age and behavioral task (Gro V. Amdam et al., 2005; Gro V. Amdam, Simões, et al., 2004; Schmid et al., 2008). Specifically, foragers had decreased hemocyte counts in the hemolymph and a higher number of pycnotic cells than nurse bees. The honey bee DSCAM gene has the potential to generate as many as 12,000 splice variants which may allow them to target specific pathogens (Graveley et al., 2004). The expression of DSCAM shows a strong negative association with Vg expression (Table S8), suggesting it is not governed by nutritional state. Taken together, higher DSCAM expression in nutrient depleted foragers may serve to increase efficiency of the decreased number of hemocytes.
We also considered the effects of oxidative stress in relation to aging and behavioral task. Oxidative stress produced by intensive foraging flights is likely mitigated by host enzyme expression within the limits of host physiology. We found that CuZnSOD increased with age, MnSOD was higher in nurses and decreased with age, and catalase was highest in precocious foragers. CuZnSOD and MnSOD detoxify the free radical superoxide (O2·-) into the less reactive hydrogen peroxide (H2O2), which is then processed by catalase into water and oxygen (Lei et al., 2016). Mitochondrial activity during aerobic respiration (flight) is the main cause of ROS generation, which tends to generate more H2O2. Fat body respiration is greater in nurses due to the continuous production of substances (e.g. Vitellogenin) needed to sustain brood rearing. Foragers experience ROS production in the flight muscles of the thorax that would circulate throughout the hemolymph. A previous study found that abdominal H2O2 levels were elevated in forager flight muscles as a possible result of increased mitochondria density (Cervoni et al., 2017). Foragers also have decreased abdominal lipid stores (Toth & Robinson, 2005), less developed fat body (Ament et al., 2008; Wilson-Rich, Dres, & Starks, 2008), and a decrease in Vg expression compared to nurses (S.-C. Seehuus et al., 2006). While young bees have generally more resistance to oxidative stress, foragers incur a gradual accumulation of tissue damage reflecting age-associated declines in the efficiency and degradation of ROS (Williams et al., 2008). We showed that 19-day foragers that had been foraging since at least 6-days old, had a high level of fat body protein carbonylation relative to age-matched nurses and age-right 27-day foragers (Figure 6). The age-right 27-day foragers had the highest oxidative stress gene expression, which could explain their low levels of protein carbonylation. These were the bees that remained nurses the longest and most recently transitioned to foraging (see Vg expression relative to 7-13-19-day foragers Figure 5A). The 19-day PFs antioxidant capacity may have reached its physiological limit. There were some significant correlations between CuZnSOD expression levels and bacterial cell abundance in the midgut, including G. apicola . Paradoxically, ROS can have positive effects such as acting in redox signaling pathways (Lei et al., 2016) or modulating the microbiota (Engel & Moran, 2013b). Taken together, these findings warrant further investigation into host-gut microbe interactions focused on the honey bee midgut.