Structure, function, and quantitative biology of the Drosophila gut microbiome

Abstract

Drosophila melanogaster is one of the most powerful animal models for developmental genetics, and it has a relatively simple microbiome. Flies can easily be reared germ-free, then reassociated with defined microbial strains to study colonization and bacterial effects on the fly, making a highly tractable model for microbiome studies. This review provides a biological background on the Drosophila microbiome for quantitative researchers interested in using the system. We discuss the current understanding of which microbes compose the fly gut microbiome, how they colonize, how they affect the fly, and how the overall microbiome structure influences its function. We also discuss existing large data sets and quantitative theory of the Drosophila microbiome.

Introduction

How is a microbiome constructed? What are its parts and what functions do they serve? How does the overall microbiome work in terms of its inputs, outputs, and dynamics? How do the individual parts of the microbiome affect its host? And is the whole simply the sum of its parts? These simple questions are difficult to address in model systems such as mice with 100s of bacterial species. Low complexity, small animal model systems can address some of these limitations [1].

The microbiome of Drosophila melanogaster has been studied as an experimental variable for nearly a century [2], and it has clear impacts on the fly’s physiology, including effects on lifespan and fecundity. However, there are relatively few quantitative studies exploring the relationship between individual members of the fly microbiome, how they colonize, and how they affect the fly. As with other host organisms, major questions about the relationship between the microbiome and its host concern: (i) whether changes in the microbiome are a cause or consequence of host physiological changes, (ii) whether causative effects of the microbiome are direct or indirect, and (iii) whether microbiome–host interactions are specifically co-evolved or more generic.

A major challenge in the microbiome field has been disentangling the complexities of microbial communities, which include the species diversity, the interactions with nutrients, the interactions with the host physiological systems such as immunity, and the various feedbacks in each of these relationships. Flies are already one of the most powerful models for animal developmental genetics, with the ability to turn individual genes on or off in specific cell types at defined time points in the animal’s lifespan [3–8]. Nutrition is also well studied in flies, and defined diets are in wide use with known consequences for fly physiology, allowing dietary factors to be disentangled from microbial ones [9–17]. Moreover, D. melanogaster from wild and laboratory environments all over the world are associated with a consistent community of roughly 5 to 20 species of bacteria from only two phyla [18], and it is straightforward to make germ-free flies and reassociate them with defined microbial communities [19]. These experimental advantages make flies an excellent choice for microbiome experiments.

In this review, we address the structure of the commensal Drosophila microbiome in terms of the species composition and the physical locations of these species. We address the function of the Drosophila microbiome in terms of how the microbiome species convert nutrients, how the species affect one another, and how all these factors affect the host. Where possible, we focus the review on quantitative studies that measure with precision how one factor changes as a function of another that is varied, for instance, the dose–response of colonization as a function of inoculum size [20]. We indicate data sets that can be utilized for theoretical work, and we emphasize current open areas for theory.

Structure of the Drosophila microbiome

The D. melanogaster commensal gut microbiome has been thoroughly inventoried both in the laboratory and the wild [21–24]. It is primarily composed of yeasts, acetic acid bacteria, lactic acid bacteria, and Enterobacteria [21] (Table 1). While the majority of studies have profiled only the taxonomic abundances of the microbes, some studies have spatially localized bacterial taxa within the gut.

Table 1.

Common D. melanogaster microbes.

Taxon	Location	Function	References
Yeasts: Hanseniaspora, Saccharomyces, Pichia, Candida others	Food	Nutrition, stimulate reproduction, necessary for development	[23,29,32,34,64,101]
Acetic acid bacteria: Acetobacter, Acidobacteria, Gluconacetobacter, Gluconobacter, others	Food, gut	Protein and fat production, sugar content reduction, vitamin production, acetate production, some pathogens	[22,24,38,61,102,103]
Lactic acid bacteria: Lactiplantibacillus, Levilactibacillus, Fructilactibacillus, Lactococcus, Leuconostoc, Enterococcus, others	Food, gut, cuticle	Lactate production, acidification, pathogen inhibition, growth factors	[17,22,38,42,46]
Gammaproteobacteria: Serratia, Morganella, Orbus, Gilliamela, others	Food, gut	Pathogens, commensals	[22,43–45,104,105]

Yeasts in the fly gut

Live yeasts are a major source of food and a stimulant for reproduction in adult flies; the fermentation products from live yeasts serve as attractants, with ethanol being the major metabolic end product [23,25–29]. Certain Saccharomyces cerevisiae strains from wild flies are the most attractive to flies, and these are best transmitted between flies [29]. The wild yeasts, Hanseniaspora uvarum, S. cerevisiae, Pichia kudriavzevii occur in the larval gut, aid development, and can survive pupariation [29–31]. However, yeast survive only transiently in the gut, suggesting that the gut environment is not a primary niche for yeasts [32]. Genetic screens for the factors required for yeast survival in the gut identified only sporulation genes, indicating that yeasts live in the gut as spores [33]. Hence, yeast have a strong and persistent relationship with Drosophila, but they appear not to be major inhabitants of the gut.

Bacteria in the fly gut

The total bacterial load in the fly gut has been quantified by both quantitative polymerase chain reaction (qPCR) and colony-forming unit (CFU) counts, with a wide range of 10² to 10⁶ microbial cells per fly across laboratory and wild flies [34]. The relative abundances of different bacterial taxa in the fly gut also vary widely across studies in wild flies, with the sample substrate, for example, apple versus peach, and environmental factors such as temperature potentially driving the major differences [35,36]. For example, lactic acid bacteria vary between ∼0 to 75% across different studies and sample sites, while acetic acid bacteria vary between ∼25 to 100% in the same samples [22,35–37]. Lactic acid bacteria and acetic acid bacteria are readily culturable, making them favored for laboratory experiments. In laboratory stocks of flies collected from different populations of wild flies from all over the world (e.g. [21,24,38–40]), roughly the same species of bacteria occur, particularly Lactiplantibacillus plantarum, Levilactibacillus brevis, and a handful of Acetobacter species, including A. pasteurianus and A. tropicalis. These species occur at roughly equal abundances and are stable in the fly, whereas some strains outcompete others on the food [41], suggesting that these are specific to the fly rather than the food, e.g. see [21,38,42]. In comparison with wild flies, the overall bacterial diversity is lower but composed of roughly the same microbial groups [22,24]. In addition to lactic acid bacteria and acetic acid bacteria, the Orbaceae are Enterobacteria that are often found as commensals in insects, and sequencing studies have identified them in Drosophila strains from wild and laboratory environments [22,43–46]. The Orbaceae were recently isolated in culture [44], and future studies will examine them under laboratory conditions.

Bacteriophage in the fly gut

Bacteriophage are important members of microbial communities. Currently, a single metagenomic study of bacteriophage in wild flies exists, which identified 167 phage genomes that were inferred to infect lactic acid bacteria, acetic acid bacteria, and Enterobacteria [45], in rough agreement with the known bacterial relative abundances in wild flies. The study of phage dynamics in the Drosophila gut microbiome is an open area for investigation.

Spatial localization of bacteria in the fly gut

There are several studies on the spatial localization of microbiota in the fly gut. Bacteria have been observed predominantly in the posterior midgut [47], but experiments to differentiate transiently passing cells from the fly food versus bacteria that stably associate with the gut found that most fly gut bacteria are transient [48,49]. However, specific strains of L. plantarum, A. thailandicus, and A. indonesiensis stably colonize a distinct physical niche in the Drosophila foregut, including the proventriculus, crop, and crop duct [20,49,50]. L. plantarum cells form a static adhesion with host tissue [50] using a specific glycan-binding adhesin that is expressed on the bacterial cell wall [51]. Colonization of the foregut niche by L. plantarum facilitates secondary colonization by A. indonesiensis and A. pasteurianus, which also colonize the foregut niche [49,50]. The same bacteria that live in the fly’s gut live on the external cuticle, and these microbes appear to be deposited via grooming and serve to limit pathogenic yeasts [46], including entomopathogenic species [52].

Drosophila viruses comprise another component of the microbiome that has been extensively reviewed, for example, Ref. [53], and is not covered here.

Function of the Drosophila microbiome

Microbes draw nutrients from host food, expel byproducts that affect other microbial species as well as the host, and use the host as both habitat and a vector to other food sources and other hosts. There are three distinct ways in which the microbiome contributes to the dietary composition: (1) consumption of compounds present in the food, (2) generation of secondary metabolites, and (3) serving as a direct nutrition source for the fly. The wild Drosophila diet largely consists of decaying fruit matter and microorganisms that produce ethanol, short-chain fatty acids, including lactate and acetate, lipids, and amino acids. Yeasts, Lactobacillus, and Acetobacter strains reduce carbohydrate content of laboratory fly food [10,54,55]. Bacterial colonization of Drosophila also appears to be protective against the negative physiological effects of ethanol, a common yeast fermentation product [56]. The microbiome produces a variety of secondary metabolites or metabolite precursors that are essential for the fly: B vitamins [57,58], amino acids [57], specifically the modified N-acetyl-amino acids [59], and nucleic acids [59]. Other non-essential metabolites affect physiology. For instance, acetate modifies host insulin signaling to promote weight gain through lipogenesis [17,60,61] and increases histone acetylation around antimicrobial peptide genes, increasing their expression, which inhibits bacteria [62]. In low-nutrient environments, bacteria also serve as a source of calories, lipids, and proteins [63,64]. The fly also benefits the bacteria. Specifically, Drosophila larvae actively secrete a substrate containing N-acetylglucosamine that facilitates Lactobacillus growth on the fly food and in vitro [65].

The role of the microbiome in lipid uptake

The role of the microbiome in host lipid metabolism requires further study [66]. The effects of the Drosophila microbiome on lipid uptake have been studied indirectly with interesting results. Treatment of wild Hawaiian flies with antifungals increased their total lipid content [67]. Antibacterial treatment slightly decreased total lipids and cuticular hydrocarbons in females, and significantly decreased testes lipid content in males [67]. Microbial colonization appears to decrease total lipid content when compared to axenic flies, although this effect is linked to the reduction of carbohydrate composition of the food [54,55]. Combining chemically defined diets and host genetics in gnotobiotic flies would be powerful for deciphering the mechanisms.

Influence on immunity

In controlled studies of antiviral immunity, it was found that Gram-negative bacteria, including Escherichia coli and A. pomorum, induce resistance to Drosophila C virus through bacterial priming of the antiviral ERK pathway via NF-kB stimulation of Pvf2 ligand production [68]. Particular commensal gut species can also influence the host survival of bacterial pathogens, such as Vibrio cholera, which is more virulent when given to flies with Acetobacter than to flies with lactobacilli due to the Gram-negative Acetobacter activating the V. cholera type VI secretion system, which injects toxins into fly epithelial cells in the gut [69]. In addition to anti-pathogen immunity, the onset of intestinal cancer and cellular structure of the gut are also influenced by microbiome composition [70–72]. There are numerous additional knowledge gaps with regard to the influence of the microbiome on distinct developmental processes of which the immune system is but one component.

Influence on lifespan

The microbiome composition is both correlated with and causative of aging. Elderly Drosophila reared germ-free express reduced stress response and innate immunity genes compared to equally aged flies with commensal bacteria [73]. Bacteria can provision nutrients and change the nutritional composition of the fly food through metabolic processes. For instance bacterial genes for consumption of cysteine and methionine were causally linked to extended fly lifespan [74,75]. Increased microbial load is a consistent factor associated with aging — higher numbers of gut bacteria are correlated with shorter host lifespan [76], and the proportion of lactobacilli decreases while Acetobacters increase in older flies [43]. Age-related shifts in microbiome compositions are associated with age-related intestinal dysfunction [71,77,78], evidenced by loss of stem cell compartmentalization in the posterior midgut [43,70,71], loss of intestinal barrier integrity [43,79], systemic inflammation [77], and neurodegeneration in Drosophila Alzheimer’s models [80]. While L. plantarum can inhibit pathogens, L. brevis and Gluconobacter switch from commensal to pathogenic [38,81]. An open question is why and how these bacterial strains switch from a commensal state to a pathogenic one. Detailed studies assessing the role of each combination of individual species show that microbial interactions have large effects on fly lifespan [41,76,82] through unknown mechanisms. A key finding from these experiments is that lactobacilli regulate the interactions between other taxa in the microbiome that influence host lifespan [82]. However, the effects of the microbiome on lifespan are diet dependent. On low-nutrient food, bacteria can increase fly lifespan [40], whereas on high-nutrient food, bacteria decrease fly lifespan [14,63]. Strain-dependent effects further complicate the issue. For instance, while L. plantarum often extends lifespan [41,83], one strain shortens lifespan by disrupting intestinal homeostasis through impaired epithelial renewal [84].

Theory for the Drosophila microbiome

What determines the microbiome composition of an individual animal? Why do two individual animals of the same species have differences in their microbiome composition? Does the order in which species arrive influence the final composition? To what extent do deterministic versus stochastic processes play a role?

Colonization theory

To explain whether an individual exposure to a defined number of bacterial cells will result in stable colonization of the host, a stochastic framework based on the dose–response was introduced to account for the observation that two identical germ-free flies inoculated with the same dose of bacteria can end up with different outcomes: either colonized or remaining germ free [20]. This model calculates the odds of colonization from a binomial distribution where each bacterial cell inoculated into the fly has an equal probability of establishing stable colonization [20]. Obtaining reliable data requires precise measurement of the input dose, the resulting colonization outcome (colonized or not), and a large number of experimental replicates in individual germ-free flies, which can be accomplished by high-throughput methods, for example, [85].

Integration of these findings with pulse-chase experiments, where flies are first colonized by a strain labeled with one fluorescent protein and then subsequently fed the same strain labeled with a different color, allows one to calculate the rate of population turnover and use microscopy to spatially localize the site of turnover [50]. These measurements in L. plantarum and A. indonesiensis allowed refinement of the theory to account for spatial turnover of the population, which differentiates two important modes of bacterial colonization [50] (Figure 1). The first mode we call population-level colonization, where colonization is maintained by the bacteria growing faster than they are lost due to death inside the host and flushing by peristaltic flow (Figure 1). This mode has ecological features like Allee effects, where a population cannot be maintained below a certain threshold size. Population-level colonization is stochastic and can be lost due to perturbations [20]. Bacterial growth rates and population dynamics can be tracked by measuring the probabilistic loss of plasmid from labeled strains with neutral markers [20].

Figure 1.

Distinction between population-level bacterial colonization and cell-level bacterial colonization. A bacterial population can stably colonize with continuous turnover of cells. Bacterial cells can be gained by consumption in the diet and cell division. They can be lost by cell death and washout by peristaltic flow. Cell-level colonization occurs when individual bacterial cells stably colonize host tissue. These two modes of colonization can be distinguished experimentally by video tracking via Bellymount [92] or pulse-chase experiments [50].

The second mode we call cell-level colonization, where colonization is stable at the single cell level due to adhesive contacts that cause individual bacterial cells to bind the host and therefore resist peristaltic flow (Figure 1). These adhesive contacts make colonization deterministic, if they are specific for a particular gut region where the bacteria have the correct physiological conditions for growth. Both types of colonization occur in Drosophila. Among the colonizers, L. plantarum, A. thailandicus, and A. indonesiensis strains from wild-caught flies can stably colonize at the single cell level in the foregut, while most bacteria are population-level colonizers [20,49,50].

Interactions between colonizers result in priority effects where the order of introduction modifies the probability of colonization [86]. The more different the two species, the more positive their interactions [86]. As the diversity of a community increases, the number of possible interactions increases as 2ⁿ – 1, giving need for theory of complex interaction networks.

Higher-order interaction theory

The theory of higher-order interactions has been extended and applied to microbial communities, including the fly gut microbiome. Interactions between bacterial species that affect the function of the community can be viewed as a microbial equivalent of genetic epistasis, where the interacting species are equivalent to genes and the function can be either a host trait, like lifespan [41,82], or a community function, like usage of a complex carbon source [87,88]. This framework applies to higher-order interactions, including context dependence. A challenge with applying traditional epistasis frameworks to higher-dimensional interactions is that while there are many interactions that are not calculated by standard frameworks, there are far too many interactions to calculate with exhaustive approaches [89,90]. A new mathematical framework called ‘epistatic filtration’ addresses this challenge by analyzing the complete interaction landscape by an unbiased method that reveals the important interactions without exhaustively testing each one. In particular, this approach identifies master regulators of complex microbial communities — species that regulate the interactions of the rest of the community in their effects on the host [82].

Stronger interactions occurred between bacterial strains isolated from a single wild D. melanogaster than between the same bacterial species isolated from laboratory flies [82]. This approach identifies important members of microbial communities for follow-up mechanistic work. For instance, the strong colonizer, L. plantarum, was found to be a master regulator. Thus, the fly can control its microbiome and its own physiology by proxy through L. plantarum [83]. A convenient, web-based interface is provided so that users can load data and receive the analysis [82].

Theory for the evolution of host–microbe associations

Deterministic interactions are typically driven by molecular mechanisms, which are encoded by genes that favor fitness. The evolutionary problem of how animal-microbe associations might evolve through initially nonspecific interactions has been approached from a disease ecology perspective on transmission dynamics [91]. Overlapping habitat preferences between microbes and host — perhaps around a shared, transient food source such as fallen fruit — and transmission dynamics in a patchy landscape are sufficient to drive co-occurrence between microbes and hosts and could serve as a starting point for the evolution of host–microbe specificity [91]. An open area for future theory could examine how these associations develop from nonspecific initial associations into complex communities. The relatively simple fly microbiome, with a stable community of as few as five bacterial species, allows these questions of complexity to be examined within a tractable framework both in terms of experiments and theory.

Quantitative data sets for microbiome theory in Drosophila

Given the naturally low microbiome diversity and experimental tractability of the Drosophila system, experimentalists have generated data sets that can be useful for theorists. Different aspects of colonization can be measured using (1) frequent transfer to sterile food to push out unstable colonizers [48], (2) pulse-chase using labeled bacteria to measure population turnover [50], and (3) single-cell microscopy in live fly guts to measure the adhesion of individual bacterial cells [92]. (4) High-throughput colonization quantification enables the study of stochastic processes [20,85,93]. (5) Fecal time series can infer population dynamics within the gut [94]. Advances in fecal sampling have increased the throughput and precision [95] to allow a large number of flies to be sampled with a minimally invasive technique. There are limitations to each technique. For instance, the fluorescent proteins labeling strains can change the fitness of the strain they label; single-cell microscopy is low throughput; and high-throughput experiments miss the spatial localization of bacteria in the gut. However, the techniques cover a wide range of precision and effort, letting experimenters select the technique most suited to their application.

Quantification of colonization across a wide range of bacterial strains with sequenced genomes was used to detect bacterial genes associated with colonization in a whole metagenome-wide association study (MGWAS) [74,96,97]. In the MGWAS framework, a microbial gene is correlated with a trait of interest using a large panel of microbial strains for which the complete genome sequences are available. Drosophila also provides a rapid system for genome-wide association studies where fly genes are associated with traits conferred by bacteria [98]. These experimental frameworks allow researchers to screen for genes of interest, and the approach can be applied to study the genetic basis of many different microbial interactions.

Dose–response curves relate the odds of colonization for a single inoculation of an individual fly to the number of bacterial cells in that inoculation. These curves clearly differentiate strong colonizers from weak ones, with weak colonizing stochastically and strong colonizing deterministically. The Obadia et al data set contains dose–response curves constructed from thousands of individual flies that either began the experiment germ-free or colonized with a conventional microbiome [20]. The Dodge et al data set contains dose–response curves for flies that began the experiment either (1) germ free, (2) colonized by the same strain as the inoculum but carrying a different label, or (3) colonized by a different species [50]. Priority effects, where an early colonizing strain influences the colonization of a later arrival, were observed both where the first strain inhibited the second and where the first strain facilitated the second. Thus, these data sets provide a resource for theory on ecological priority effects.

To study higher-order interactions, combinatorial data sets are needed, where all 2ⁿ – 1 possible combinations of bacterial strains are constructed and measured for a focal trait. Combinatorial approaches, where each possible combination, for example, of one, two, three, four, and five species was inoculated into germ-free flies, were used to generate datasets to model both the effects of microbes on one another for colonization studies [41,86] as well as to model the effects of differing microbial compositions on the host physiology [41,76,82]. This combinatorial experimental design allows the interactions between the different bacteria to be calculated. The data include measurements of CFU counts for each bacterial strain in the combinations as well as measurements of fly development time, fecundity, and lifespan, which can be combined to calculate fitness (Table 2). The combinatorial data sets for colonization can be reanalyzed to further develop theories of priority effects, and the data sets for fly physiology can be reanalyzed to further develop the theory of higher-order interactions. Notably, the experiments to generate these data sets were performed on acidic fly food, where the bacteria cannot grow, which means the measured phenotypes are indicative of interactions inside the fly rather than on the food.

Table 2.

Data sets.

Reference	Traits measured	Bacterial species	Relevance for future research
Obadia et al, 2017[20]	Colonization dose–response curves	Three L. plantarum strains as focal species; 17 total strains from genera including Acetobacter, Enterococcus, Lactococcus, and Leuconostoc.	Theory of colonization including priority effects
Dodge et al, 2023 [50]	Colonization dose–response and priority effects	Focal species are L. plantarum and A. indonesiensis. Limited data in supplement for A. pasteurianus.	Theory of colonization including priority effects
Gould et al, 2018 [41]	Colonization, development, fecundity, lifespan	L. plantarum, L. brevis, A. pasteurianus, A. tropicalis, and A. orientalis.	Theory of higher-order interactions
Jones et al, 2022[86]	Colonization	L. plantarum, L. brevis, A. pasteurianus, A. tropicalis, and A. orientalis.	Theory of microbial interactions in colonization
Eble et al, 2023[82]	Lifespan	L. plantarum, L. brevis, A. pasteurianus, A. tropicalis, A. orientalis, A. cerevisiae, and A. malorum	Theory of higher-order interactions

These microbial data sets help elucidate bacterial communities that are important for fly physiology. It should be noted that a major advantage of Drosophila as a model system is the availability of genetics tools to determine the cell physiological mechanisms of host phenotypes. The recent discovery of the host cell types where colonization occurs and development of new genetic tools to control gene expression in these cell types opens up new opportunities to study the cellular and molecular mechanisms of the host–microbe interactions that were discovered through these quantitative data sets [99].

Conclusions

From an ecological and evolutionary perspective, it is important to consider how host–microbe interactions evolved. In what context is there a mutual fitness advantage in host–microbe specificity? Which interactions can persist without a co-evolved mechanism? While some host–microbe traits are nonspecific due to ecological filtering [100], others are highly specific [20,49,50]. Quantitative studies in flies and bacteria provide precise traits, which can be mapped to genes in the respective genomes of bacteria and flies.

An open and understudied area for investigation is what are the evolved molecular traits that confer specificity to host–microbe interactions. A related and understudied area, for which the fly model system is well suited, is the quantitative study of host-to-host transmission and how it has shaped the molecular repertoire of colonization factors in the bacteria and their fly hosts. Flies provide an excellent host model for these studies because of their extremely tractable genetics, high-throughput microbiome techniques, and staggering host genetic resources available.

By taking quantitative approaches integrated with theory, the field has gained critical insights into the process of gut colonization. These types of critical insights provide a model for understanding more complex host-microbiome systems such as those of humans.

Data Availability

No data were used for the research described in the article.