Disentangling host–microbiota complexity through hologenomics - Nature.com

Abstract

Research on animal–microbiota interactions has become a central topic in biological sciences because of its relevance to basic eco-evolutionary processes and applied questions in agriculture and health. However, animal hosts and their associated microbial communities are still seldom studied in a systemic fashion. Hologenomics, the integrated study of the genetic features of a eukaryotic host alongside that of its associated microbes, is becoming a feasible — yet still underexploited — approach that overcomes this limitation. Acknowledging the biological and genetic properties of both hosts and microbes, along with the advantages and disadvantages of implemented techniques, is essential for designing optimal studies that enable some of the major questions in biology to be addressed.

Download PDF

Introduction

Biological systems composed of a eukaryotic organism and its microbiota are widespread in nature^1,2,3. Also known as holobionts⁴, host–microbiota systems range in complexity from one-to-one symbiotic associations between a host and a single microorganism (such as the bioluminiscent Aliivibrio bacteria in light organs of bobtail squids⁵) to intricate arrangements between a host and a dynamic community of microorganisms (such as mammals and their gut microbiota⁶, or plants and their root microbiota⁷) (Fig. 1a). Many such systems play core functional roles in ecosystems or are fundamental to agricultural production (Fig. 1b). Hence, their study and manipulation have become a global strategic priority^8,9.

Fig. 1: Diversity and ecological relevance of host–microbiota systems.

a | Complexity gradient of microbial communities associated with specific body parts of various animals. Many Lepidoptera caterpillars have no detectable resident gut microbiota¹⁶³. Bobtail squids maintain a symbiotic relationship with a single type of bacteria that colonizes their light organs, which provides them with skin luminescence⁵. Pea aphid bacteriocytes harbour one or two dominant bacteria along with a few secondary taxa with considerably lower representation¹⁶⁴. The gut microbiota of bees is composed of a few types of bacteria¹¹⁴. Birds harbour a complex microbiota with hundreds of bacterial species¹⁶⁵. b | Examples of different host–microbiota systems and their relevance for ecosystem or applied processes. Photosynthetic dinoflagellates provide corals with a source of carbon in nutrient-poor tropical waters, forming the basis of coral reefs¹⁶⁶. Lignocellulose-degrading bacteria in the gut of termites transform complex polysaccharides into short-chain fatty acids (SCFAs) and modify lignin from wood substrates¹⁶⁷. The root microbiota of plants plays a central role in transforming complex organic compounds into simple nutrients that plant roots can absorb, thus boosting productivity⁷. The rumen microbiota is important for producing SCFAs and other essential compounds through the fermentation of complex plant polysaccharides, which act both as energy sources and signalling elements in animal hosts¹⁶⁸. The gut microbiota of mice plays a crucial role in laboratory disease models because of their effects on many systemic conditions related to energy metabolism and immune response¹⁶⁹.

Full size image

The study of functional associations of eukaryotic hosts with microbial communities dates to at least the late nineteenth century, when researchers noticed, based on microscopy, that termite guts contained microbial species on which the termites appeared to rely¹⁰ and the intestines of infants hosted bacteria that aided in digestion¹¹. Subsequent research on insects¹², farm animals¹³ and humans¹⁴ relied on a combination of culture-based approaches, functional assays and microscopy to characterize the microbial taxa associated with these hosts. The more recent development of culture-free molecular technologies has provided new ways to study and understand the complex details of such systems. Today, we are able to generate high-quality eukaryotic genome sequences (at the level of near-complete chromosomes)¹⁵ and recover virtually complete bacterial genomes from samples that contain complex mixtures of DNA (such as faeces or plant roots)¹⁶ relatively quickly and in ways that are replicable within and among host species. Taken together, these advances enable the simultaneous characterization of the so-called hologenome¹⁷, that is, the entirety of the genetic information contained by a eukaryotic host and the microbial community associated with that host.

The terms hologenome and hologenomics used in this Review are primarily instrumental^18,19. They reflect an approach to the study of microbes and hosts that is holistic rather than atomized and that transcends disciplinary and taxonomic boundaries. The simultaneous analysis of host and microbial genes and genomes naturally allows the study of fundamental evolutionary aspects of host–microorganism associations²⁰ and, as a result, will contribute to discussions about general evolutionary models for the role of selection on hosts, microbes and their amalgams^21,22,23,24. However, the utility of hologenomics is much broader as it has also proven valuable for the study of diseases²⁵, the improvement of farming practices^19,26 and in understanding molecular interactions between eukaryotic and prokaryotic cells²⁷, to cite but a few examples.

Hologenomics leverages the latest advances in genomics and metagenomics to disentangle the complexity of host–microbiota systems by using a systemic approach. This enables not only detailed profiling of the host genotype but also characterization of the microbial metagenotype and, by extension, the definition of the hologenotype. High-resolution characterization of all these units of analysis is only possible through the combination of high-throughput DNA, RNA, protein and metabolite profiling²⁸, novel mathematical tools to integrate the different omic layers^29,30, and the generation of high-quality reference genomes of both eukaryotic and prokaryotic organisms^15,31. This hologenomic approach is unlocking domains of knowledge that have remained inaccessible until recently because of their excessive complexity and the costs of data generation.

In this Review, we first address five fundamental criteria that should ideally be considered in the design of any hologenomic study. In doing so, we explain the relevant methods and model systems available to address a variety of biological questions using a hologenomic approach. Subsequently, we outline how the information extracted using this methodology can be used to address major outstanding questions in life sciences and to design novel interdisciplinary approaches to expand our knowledge horizon.

Designing hologenomic studies

Hologenomics can be used to understand how the combined features of hosts and microorganisms shape biological processes relevant for hosts (such as adaptation), for microorganisms (such as meta-community dynamics) or both (Box 1). Depending on the aims and features of the study system, hologenomics can be implemented using different study designs, model systems and techniques^18,19,32. We propose that this landscape of possibilities is shaped around five essential questions that need to be considered when designing and interpreting hologenomic studies (Fig. 2), which relate to five core topics: hologenomic complexity, control of hologenomic variables, hologenomic resolution, spatiotemporal resolution and explanatory versus response variables.

Fig. 2: Overview of the five essential criteria for designing and interpreting hologenomic studies.

1. Complexity. The complexity of a hologenomic system can be decomposed into three elements: host genomic complexity, microbial metagenomic complexity and environmental complexity. 2. Control of variables. What information can be recovered from host–microbiota systems will depend on whether and to what degree host genomic, microbial metagenomic and environmental complexity can be controlled, minimized or removed by the researcher. 3. Genomic resolution. The resolution at which the host genome and the microbial metagenome are characterized and, in consequence, how genotypes, metagenotypes and combined hologenotypes are defined, determine the genomic level of complexity at which to study a host–microbiota system. 4. Spatiotemporal factors. Which questions can be addressed also depends on the intrinsic spatial and temporal features of hosts and microbial communities considered as well as on the spatial and temporal features of the study design. 5. Explanatory and response variables. Due to the multi-layered nature and bi-directional interactions in host–microbiota systems, the definition of explanatory and response variables and, thus, the question researchers want to answer will often be determined by the study design. ASV, amplicon sequence variant; COI, cytochrome oxidase subunit 1 gene; GBS, reduced representation genome sequencing through genotyping-by-sequencing; MAG, metagenome-assembled genome; mtDNA, complete mitochondrial genome; OTU, operational taxonomic unit; WGS, whole-genome sequencing.

Full size image

Box 1 Hologenomics from different perspectives

While host–microbiota systems cover a range of biological attributes (Fig. 1), the asymmetry between the host genome and the microbial metagenome is a fundamental feature that characterizes them all. The host genome typically belongs to a single multicellular macroorganism, whose generation time is usually much longer than that of its associated microbes. By contrast, the microbial metagenome represents the composite of the genomes of millions of unicellular organisms that may belong to hundreds or thousands of different taxa. In addition to their typically shorter generation times, they often exhibit a capacity for horizontal gene transfer that increases their adaptive potential¹⁷⁰. As a result, whereas the host genotype is mostly fixed throughout the host's life (although see discussion below), the microbial metagenotype and the resulting hologenotype change continuously^150,171. Because of this and other differences addressed elsewhere²¹, the routes available for the host to affect the microbiota and vice versa differ, as does the information hologenomics provides when applied to describe a system from the microbial versus the host perspective.

The microbial perspective

The host provides the physical basis for the microbiota and defines many properties of their ecosystem¹⁷². Hence, the genomic features of the host contribute to configuring ecological niches that microorganisms can occupy¹⁷³. For instance, gut morphology shapes oxygen gradients⁴³, the production of lipids determines the structure of skin microbiota¹⁷⁴ and host macrophages can exert direct selection on the microbial community¹⁷⁵. In animals, the host genome can also indirectly affect the transmission, acquisition and community dynamics of the microbiota through moulding host behaviour — sociality, nursing behaviour and diet are three of the main factors that shape microbial communities associated with animals^176,177,178.

Hologenomics is not only useful for studying how host genomic variation changes the conditions that the microbial communities experience but also for understanding how the microbiota alters host processes that shape such a landscape. Microorganisms can modulate the expression of specific genes in the host. In mouse models, for instance, intestinal spore-forming bacteria release metabolites that modify the expression of various host genes involved in the biosynthesis of serotonin¹³². Microorganisms can also trigger more complex systemic changes that induce developmental¹⁷⁹ or behavioural^180,181 modifications in the host, which can indirectly affect the environment in which microorganisms live.

The host perspective

Genomic features of microorganisms can influence host biology by causing disease¹⁸² or enhancing biological capabilities¹²⁵. Bacteria can significantly shape a myriad of host phenotypic traits due to their capacity to metabolize complex foods¹⁸³, produce essential biomolecules¹⁸⁴, modulate host gene expression¹³², promote epigenomic changes¹⁸⁵ or trigger hormonal cascades¹⁸⁶. Hologenomics can be used to understand how host genomic features condition the metabolic functions provided by its microbiota and how the metagenomic features of these microorganisms, along with the genotype of the hosts¹²⁸, modulate essential biological processes for the host¹⁸.

Hologenomic complexity

Hologenomic complexity can be broadly defined as the amount of information relevant to the study that the biological system under analysis contains and it can be decomposed into three major elements: host genomic, microbial metagenomic and environmental complexity. Within each of these elements, two sources of complexity can be defined: the intrinsic complexity of the system under study, including host genome size and number of bacterial genomes, and the complexity introduced by the degree of difference between the organisms under comparison such as gene expression differences versus distinct genomes (Fig. 3). The combination of host genomic, microbial metagenomic and environmental complexity will determine the relevance of each factor in the study system and the questions that can be answered (Fig. 4). This information could be used either for selecting the most appropriate system to address a specific biological question or, when researchers are bound to a given system, for adjusting the study design to the properties of the system.

Fig. 3: Decomposition of hologenomic complexity.

The design and interpretation of hologenomic studies depend on the host genomic (part a), microbial metagenomic (part b) and environmental (part c) complexity of the system under study. Within each axis of complexity, two types of gradients can be defined based on whether the features are intrinsic to the system or introduced by the researcher through the selection of groups under comparison.

Full size image

Fig. 4: Hologenomic complexity of study systems.

Different study systems hold different levels of complexity, which limit the range of scientific questions that can be addressed. a | Six examples of study systems with different levels of genomic, metagenomic and environmental complexity. b | Three-dimensional representation of the complexity of the examples. The area of the plain represents the combined host genomic and microbial metagenomic complexity of the system, while the height represents the environmental complexity. The combined three-dimensional volume represents the overall hologenomic complexity of the system. HMP, Human Microbiome Project.

Full size image

Host genomic complexity

The intrinsic complexity of host genomes is primarily determined by features such as genome size, number of chromosomes, ploidy, amount of functional genetic elements and number of repetitive sequences³³. These features vary across eukaryotes^34,35 and not only determine the degree of complexity under analysis but also how data are generated (for example, by considering complications related to reconstructing high-quality reference genomes). Thus, effective implementation of a hologenomic approach in a host organism whose genome is relatively simple and well characterized (such as Caenorhabditis elegans or Mus musculus) is more feasible than those with extremely large, duplicate-rich genomes (such as salamanders³⁶ or polyploid plants³⁷) (Fig. 3a).

Host complexity is also determined by the experimental setup itself. Hosts under study may range from individuals with practically identical genomic features (such as monozygotic twins³⁸ or inbred mice³⁹) to intraspecific variants (such as different landraces of plants⁴⁰ or reproductively isolated populations of animals⁴¹) to different species or evolutionary lineages (for example, monocotyledonous versus dicotyledonous plants grown in a common garden experiment⁴²) (Fig. 3a). In this gradient of host genomic complexity, the lower extreme considers interindividual genomic variation mainly introduced by SNPs, which normally introduce subtle differences in the morphological (for example, intestinal crypt properties⁴³) and biomolecular (for example, antimicrobial peptide and immunoglobulin production⁴⁴) features of the host. The upper extreme, meanwhile, typically addresses major genomic differences that yield important morphological and physicochemical changes that create conspicuous phenotypic differences across the hosts studied⁶. One interesting exception relates to the consideration of how genetically determined sex can introduce large-scale changes to microbial communities within conspecifics⁴⁵ through the major structural differences sexual dimorphism can induce on the body⁴⁶. Both types of genomic variation are known to shape the microbial communities associated with the hosts through multiple underlying mechanisms¹⁸.

Microbial metagenomic complexity

One fundamental aspect that determines intrinsic metagenomic complexity is the breadth of taxa included in the definition of the microbial metagenome. Whereas early metabarcoding-based studies of hologenomes often focused on single taxa, such as bacteria (for which both marker genes and databases were available early), shotgun sequencing-based approaches allow the consideration of the entirety of the microbial metagenome. In truth, most microbial metagenomes can also include viruses⁴⁷, other prokaryotes (such as archaea⁴⁸ and Candidate Phyla Radiation nanobacteria⁴⁹), and eukaryotes such as fungi⁵⁰, protozoa⁵¹ and helminths⁵². Even when studied in its entirety, the microbial community of interest can range from a single microbial symbiont to a complex microbial community composed of thousands of taxa (Fig. 1a).

Metagenomic complexity is also shaped by the taxonomic and functional differences among the microbial communities being compared, which can differ in terms of relative abundances of taxa or be compositionally distinct (Fig. 3b). However, gene expression differences caused by the highly dynamic nature of microbial communities do not always lead to significant structural changes and functional redundancy means that microbial turnover is not always translated into an effective functional change in the microbial ecosystem⁵³.

Environmental complexity

The complexity of environmental factors, defined as any feature not directly determined by host genomic and microbial metagenomic properties, also plays an essential role when designing and interpreting hologenomic studies. Here, the environment might include abiotic conditions (such as cold versus warm) but also habitat (such as swamp versus forest or zoo versus wild) or diet (such as plant based versus animal based). Gene expression patterns of both hosts and microorganisms as well as other features of hosts, such as behaviour or physiological state, are largely driven by environmental cues^54,55,56. A gradient of intrinsic complexity of the environment can be defined, spanning from laboratory experiments, in which all environmental features are kept constant, to natural ecosystems, in which both abiotic and biotic conditions differ from place to place, time to time, and between individual hosts (Fig. 3c). Similar to host and microbial complexity, environmental complexity is also determined by the experimental setup. The degree to which environmental conditions differ across the biological systems being compared will determine the overall environmental complexity of the study system.

Controlling hologenomic variables

Controlling the complexity of hologenomic variables is essential for addressing specific research questions. Broadly speaking, the more detailed and mechanistic the question under study, the greater the required control. For instance, research on specific biomolecular processes using laboratory models will require a higher level of control than studying biogeographical patterns of host–microbiota interactions in wild organisms. The control of hologenomic variables can be achieved through a number of strategies.

Controlling host genomes

The control over host genomic complexity largely depends on the model organisms studied and the technical approaches employed. In laboratory organisms that can reproduce asexually, such as water fleas (Daphnia, Crustacea) and Lamiaceae plants, absolute control over host genotypes can be achieved by using clonal organisms^57,58. When clones cannot be used, inbred laboratory animals can provide a high level of genomic homogeneity. The use of groups of genetically homogeneous hosts allows the effects of contrasting environmental conditions or specific microbial communities to be compared^59,60. Clonal and inbred models also enable the effects of a specific host genetic factor to be studied in a controlled genomic background through the application of targeted techniques for modulating gene expression (such as RNA-mediated interference) or for genomic engineering (such as CRISPR–Cas9)^60,61. Working with humans and wild organisms does not enable such a degree of control over the genotypes studied unless in vitro models, such as organ-on-a-chip co-cultures of animal tissues and microbial communities, are generated⁶². When this level of control is not possible, coarse control over host genotypes can be achieved through contrasting animals from different populations or from closely related species⁶³, while greater control can be achieved through comparing individuals across different degrees of kinship, such as monozygotic versus dizygotic twins³⁸, and family members to other individuals⁶⁴.

Controlling microbial metagenomes

Control over microbial metagenomic complexity is usually achieved through the modulation of microbial communities. Some strategies, such as modification of dietary regimes or the administration of microbiota-targeted additives or prebiotics, aim to modify microbial ecosystems by changing nutrient availability. However, unless compounds that match the unique enzymatic capabilities of specific microorganisms are used⁶⁵_, it is difficult to accurately modulate the microbiota owing to the complexity of ecological relationships among microorganisms. Alternative approaches to modify microbial communities include the inoculation of target bacteria (such as probiotics)⁶⁶ and faecal microbiota transplantation⁶⁷. The efficacy and accuracy of these methods is also variable; there is no guarantee that inoculated bacteria will establish or modulate the microbiota, while transplantation does not enable accurate control over the microbial community introduced or the secondary elements that are transplanted along with bacteria⁶⁸. These issues complicate the interpretation of results; for example, bacteriophages transferred alongside bacteria may severely impact the gut microbiota composition⁶⁹. A higher level of control could potentially be achieved through transplanting synthetic microbial communities^70,71. While this approach has been successfully implemented in diverse in vitro setups^72,73, the complexity of microbial communities still hinders its efficient use as a routine scientific procedure in live animals⁷⁴. The metagenotypes of these synthetic communities could be further controlled by genetically engineering bacteria with the desired genetic properties^75,76, which could eventually facilitate the assembly and control of interactive microbial consortia⁷⁷.

Controlling the environment

In most laboratory studies, environmental complexity is reduced so that no or very few environmental parameters (usually only experimental treatments) vary among groups and subjects. Climate chambers and aquaria provide absolute control of abiotic conditions such as light/dark cycles, humidity and temperature variations. Outdoor common garden experiments⁴² do not provide full control over environmental factors but they ensure that the effect on the systems being compared is identical. Some natural systems can also provide special conditions that enable environmental features to be controlled, such as cuckoo nestlings that are bred by other birds⁷⁸, or sympatrically occurring but allochronically isolated populations such as salmon populations that breed in the same rivers in alternating years⁷⁹. Research on wild organisms usually incorporates more complex and dynamic environmental conditions; when controlling them is not possible, it is useful to collect relevant environmental metadata, which can be incorporated as covariates in the statistical analyses. A century of ecological research has revealed the advantages of each of these approaches. On the one extreme, laboratory microcosms allow the most reductive control; on the other, studies in the macrocosm of the real world provide perspective on emergent properties of natural ecosystems that cannot be anticipated based solely on microcosms.

Hologenomic resolution

The complexity of a study system is not only determined by its inherent properties and study design but also the techniques and procedures employed to analyse it. Researchers can decide how much a system is simplified by altering the resolution of the hologenomic features under study⁸⁰; in essence, zooming in or zooming out. The biological resolution of hologenomic studies can be increased by not limiting the study of host and microbial community features to the genomic level but rather by incorporating additional molecular data layers associated with both host genotypes and microbial metagenotypes⁸¹ (Table 1). As long as appropriate sample preservation and laboratory processing procedures are adopted, high-throughput DNA sequencing platforms and mass spectrometry instruments can be used to generate multiple layers of biological information that enhance the resolution when trying to ascertain biological interactions between hosts and microbial communities. Multi-omic approaches that incorporate data from hosts and associated microorganisms, defined as holo-omics¹⁸, generate hypercomplex datasets that may require the dimensionality of the data to be reduced to gain power. However, it is crucial to acknowledge that reducing resolution comes with an increased risk of overlooking essential parts of the total variability that lie under a certain grouping threshold (for example, strain-level variation of microbiota or interindividual variation in hosts). The use of automatic dimensionality reduction techniques, such as principal components analysis or partial least squares regression, can also complicate the biological interpretation of the results⁸².

Table 1 Examples of techniques used in hologenomics and the level of information they generate

Full size table

Host genotypes

In host–microbiota studies, host genotypes can be defined at different levels, including species, breeds, populations^83,84, strains⁸⁵, sex or individuals⁸⁶. Genotypes can be defined as categorical variables, without analysing the differences between them, or can be studied in more detail by considering their actual genetic content and establishing correlations among them⁸⁶. When using an evolutionary perspective, phylogenetic relationships between genotypes are established based on phylogenomic markers⁸⁷, which usually vary above population and species level but not among individuals. This implies that genomic variability among the individuals included within each genotype is overlooked. Studying the effect of interindividual genomic variability on host–microbiota systems, such as identifying candidate host genomic variants associated with microbial features, requires a higher level of resolution. This is achieved by defining genotypes at the individual level and using techniques based on whole-genome resequencing that enable the complexity of host genomes to be screened at a much finer level such that differences between the individuals contrasted are not only defined based on their kinship but also on the functional properties of their genomic variants. Currently, this approach requires high-quality reference genomes from which high-density SNP profiles of individuals can be generated, for example, through SNPchip or resequencing studies¹⁵. The genomic resolution could be further refined by incorporating structural variants⁸⁸, methylation patterns^89,90 or, even, we hypothesize, chromosome 3D folding structure as revealed through techniques such as Hi-C⁹¹. In doing so, researchers can identify associations between SNPs or gene variants and specific microbiota traits, such as the relative abundance of certain taxa or the enrichment of a given function, and thus identify mechanisms by which a host exerts control over the composition and function of its associated microbiota⁹².

Microbial metagenotypes

The structure and resolution at which microbial metagenotypes are defined also affects the complexity of the metagenome under analysis. Metagenotypes can be defined as arrays of microbial taxa, microbial genes or a combination of both. The most common approach to define them is to rely on short marker sequences targeted for metabarcoding purposes such as the 16S ribosomal RNA (rRNA) or the internal transcribed spacer⁹³. Based on sequence similarity, the microbial sequences detected can be clustered into operational taxonomic units that approximate the species-level identity of microorganisms (usually using a 97% similarity threshold) or can be analysed at a finer level as amplicon sequence variants with the aim of approaching strain-level resolution⁹³. However, these procedures often do not enable reliable taxonomic assignment at genus or species level⁹⁴, do not capture sub-amplicon sequence variant strain-level community dynamics (as shown by a recent non-peer-reviewed preprint⁹⁵) and are prone to generating biased functional inferences as bacteria with identical marker genes (particularly those associated with wild taxa) might carry very different catalogues of genes⁹⁶. Thus, while useful for estimating microbial diversity and obtaining preliminary insights into functionality^97,98,99, targeted sequencing approaches do not provide conclusive evidence about the metabolic capabilities of the microbiota, particularly when working with non-human systems¹⁰⁰.

By contrast, if appropriate strategies and adequate sequencing depths are employed, shotgun metagenomics enables bacterial genome sequences to be recovered from which genes can be predicted and annotated to create a gene catalogue that can define a metagenotype^101,102. However, these genes are not randomly distributed but enclosed within genomes of specific bacteria or other microorganisms, with a particular combination of genes that shape their expression and the specific biological features (such as oxygen affinity, reproduction time, metabolic capacity) that determine their ecology. Hence, a more refined characterization of microbial metagenotypes can be achieved through binning algorithms that enable bacterial genome reconstruction from metagenomic mixtures¹⁶, yielding metagenome-assembled genomes (MAGs). Nevertheless, unless short-read sequencing is combined with long-read approaches, it is challenging to capture multi-copy genes such as the 16S rRNA marker gene¹⁰³, which is often employed in metabarcoding studies and therefore represents a useful link to a large number of existing studies. However, machine learning-based solutions to link 16S rRNA marker gene sequences with MAGs are being developed¹⁰⁴. Finally, regardless of the approach used to define the microbial metagenotype, the complexity of microbial communities will often require a dimensionality reduction to increase statistical power^105,106. This can be achieved by defining co-abundance clusters, ecological guilds or more complex strategies that also consider temporal features of microbiota variation such as compositional tensor factorization¹⁰⁷.

Envirotypes

The characterization of environmental factors that affect the host–microbiota system under study enables the definition of envirotypes, a term drawn from crop sciences¹⁰⁸ that is useful in accounting for environmental factors in the hologenomic context. Any different physical place or a place sampled at different time points will be exposed to a different environment as conditions will seldom be identical between two spatial and temporal points. Hence, the resolution at which the composite of environmental factors is considered will define whether these two environments will be thought of different envirotypes or not. For example, if only considering water temperature, killer whales sampled in the Arctic and the Antarctic seas experience the same envirotype. However, if the biotic composition is also considered in the definition of the environment, the Arctic and the Antarctic will need to be split into two distinct envirotypes as some killer whales will have access to penguins as a food source while others will not¹⁰⁹. The same principle applies to laboratory setups or mesocosm experiments: a temperature shift of 2–3 °C might not be considered relevant under some experimental setups, while it can define different envirotypes under other study designs. Finally, the failure to recognize environmental factors that affect host–microbiota interactions and thus define relevant envirotypes can lead to increased noise and a decreased capacity to achieve statistical significance.

Spatiotemporal resolution

Spatial resolution

Microbial communities associated with animal and plant hosts vary not only across coarse body parts¹¹⁰ but also at the microscale^111,112 such as between the lumen and the intestinal crypts¹¹³. Thus, the resolution at which a body site is defined will also determine how a hologenomic system is characterized. For example, the animal gastrointestinal tract can be considered a single sampling unit¹¹⁴, 4–5 units^65,115 or hundreds of micro-units^111,112 depending on the sampling and data processing strategies employed. Naturally, each level of resolution will allow different questions to be addressed and will require the use of different technologies and analytical approaches.

Temporal resolution

Temporal features to be considered include when, how often and for how long host–microbiota systems are to be analysed. Researchers must consider when a host is first exposed to microbes with regard to temporal benchmarks (number of days or years), as well as the order in which it is exposed to them. Priority effects relate to how the order of species arrivals in an ecosystem shape the potential for subsequently arriving taxa to establish themselves¹¹⁶. Although originally discussed at the macroorganismal level in the context of plant communities¹¹⁷, the phenomenon is also relevant for building host-associated microorganism communities¹¹⁸, for example, as documented in the human gut^119,120. In addition, microbial communities are known to vary daily¹²¹, seasonally¹²² and relative to life-stage patterns¹²³. Hence, the extent and frequency of sampling determine which of these dynamics will be observed or, conversely, missed. Finally, it is important to consider that the consequences of changes at one time period or life stage may appear only later in time and thus detection of such effects obviously requires that the subsequent period is also studied. For example, interventional animal experiments show that, when the immune system develops early in life, there is a window of opportunity where the gut microbiota composition shapes the risk of developing diseases in the future^123,124.

Explanatory and response variables

Host genomic and microbial metagenomic data generated under hologenomic setups can take on different roles when generating statistical models. While the environment is most often considered as an explanatory variable (though one can also study how the hologenome affects the environment), the host genome and the microbial metagenome are sometimes viewed as explanatory and sometimes as response variables, depending on the aim of the research. In many cases, directionality is set by the researcher rather than the biological system itself, as host–microbiota systems contain many bi-directional interactions and circular processes, which complicate the establishment of causal relationships. Here, we define three basic models in which the three main variables (genome, metagenome and environment) are assigned different roles to address different types of fundamental questions (Fig. 5).

Fig. 5: Examples of biological processes addressed by the different models of host–microbiota interactions.

a | How does the hologenome shape animal phenotypes? Only the combination of specific host genotypes (G) and microbial metagenotypes (MG), probably resulting from a selective force exerted by the presence of predators (that is, a specific envirotype (E)), enables rough-skinned newts to have skin toxicity¹²⁸, an ecologically relevant phenotypic trait. b | How do the microbial metagenome and environment shape host genomic features? Short-chain fatty acid (SCFA)-producing bacteria along with a fibre-rich diet enhance chromatin accessibility and thus activate immune gene expression¹³⁵. c | How do the host genome and the environment shape microbial genomic features? Only the combination of a lactase non-persister genotype combined with the milk-drinking envirotype generates a microbial metagenotype characterized by the enrichment of Bifidobacterium. HDAC, histone deacetylase; Na_V, voltage-gated sodium; P, phenotype.

Full size image

Phenotype as a product of genotype, metagenotype and envirotype

This is the main model used when hologenomics is conducted to ascertain how genome–metagenome–environment interactions affect the biological properties of a host such as disease susceptibility, performance or fitness. It is an especially common and relevant model for health, agricultural, ecological and evolutionary research^{19,125,126,127}. One clear example of a phenotype shaped by host genomic, microbial metagenomic and environmental factors was recently reported for rough-skinned newts¹²⁸. The study showed that bacteria on the skin of the newts produce a deadly neurotoxin from which th...