Bacteria influence eukaryotic biology as parasitic, commensal or beneficial symbionts. Aside from these organismal interactions, bacteria have also been important sources of new genetic sequences through horizontal gene transfer (HGT) for eukaryotes. In this Review, we focus on gene transfers from bacteria to eukaryotes, discuss how horizontally transferred genes become functional and explore what functions are endowed upon a broad diversity of eukaryotes by genes derived from bacteria. We classify HGT events into two broad types: those that maintain pre-existing functions and those that provide the recipient with new functionality, including altered host nutrition, protection and adaptation to extreme environments.
Genome fragments are sometimes transferred from bacteria to eukaryotes via horizontal gene transfer (HGT).
When these DNA fragments contain genes, these genes can retain their functionality in some cases.
If these bacterial sequences are maintained for long periods of time, they can acquire eukaryotic features such as introns.
If the eukaryotic recipient retains a stable bacterial endosymbiont, these HGT events can compensate for genome reduction in the endosymbiont.
These HGT events can also enable the eukaryotic recipient to protect itself from other organisms, survive in new environments or use new food sources.
Further study of neglected eukaryotic groups will help to clarify the frequency of bacteria–eukaryote HGT.
The genome of an organism is usually passed vertically from the parent to its offspring. As such, in the simplest case, the evolutionary history of a genome should reflect the evolutionary history of the organism. However, genomes are dynamic in content, size and rates of evolution. Genes can be lost, non-coding or selfish genomic regions can expand or contract over short timescales, different loci can evolve at different rates because of unequal selective pressures, and genes can be gained through both duplication within genomes and acquisition from foreign sources through horizontal gene transfer (HGT). HGT can in principle occur between any two organisms that contain DNA genomes, leading to different genes in the same genome exhibiting different evolutionary histories. However, HGT is not likely to occur equally among all branches of life.
HGT is now understood to be a major driver of genome evolution in bacteria and archaea1,2,3,4,5,6,7. In fact, it is so common that our ability to infer organismal relationships through phylogenetic tree analysis has been questioned for bacteria and archaea8,9. By contrast, the frequency and importance of HGT between bacteria and eukaryotes remain controversial and less clear7,10,11,12,13,14, in particular for animals15,16,17. This lack of clarity — at least partly due to the large size and complexity of animal genomes and the almost inevitable problem of microbial contamination in genome projects — has resulted in misattributions of HGT events in animal genomes (such as human genomes18,19,20,21,22) in the past, and more recently, similar misattributions have occurred in the genome of a tardigrade23,24 (Box 1). However, this history of sometimes over-ambitious claims of microbial–animal HGT events does not mean that HGT does not happen or that it is not important in eukaryotes.
Many cases of bona fide functional inter-kingdom HGT events have been documented, including several recent examples of bacteria-to-animal transfers25,26,27,28,29,30,31,32,33,34 (Fig. 1; Table 1). In this Review, we discuss these recent findings, aiming to incorporate examples from a wide diversity of eukaryotes. We focus on gene transfers from bacteria to eukaryotes because they seem to be the most common — or at least the most commonly reported — but we note that HGTs among nearly all branches of life have been described (Table 1; Supplementary information S1 (table)). Using this broad taxonomic sampling, we outline mechanistic similarities in HGT events that result in functional genes and organize these transfers into common functional categories. We classify these HGT events into 'maintenance' transfers and 'innovation' transfers and explain how HGT from bacteria has shaped, and continues to shape, eukaryotic genomes.
Mechanisms of DNA transfer
Most transferred DNA does not become functional. Factors that affect the frequency of gene transfer include the ability of an organism to take up foreign DNA, the accessibility of its germ line (specifically, whether it is sequestered away from somatic tissue), its recombinogenic tendencies and the frequency of the donor DNA in the environment. The presence of stably associated germline endosymbionts can dramatically increase the chances of HGT, as the DNA of any lysed endosymbiont cell has direct access to the germline nucleus. Upon acquisition, the foreign DNA must become subject to selection35, either by being transcribed into RNA or by providing a function at the genome level. For example, a transferred protein-coding gene must be transcribed and translated by the host for it to gain new functionality (Fig. 2). If the transfer does not become subject to selection, the newly acquired DNA will likely be deleted or otherwise erode in a manner dependent on the mutagenic tendencies of the host. Because of the differences in gene structure between bacteria and eukaryotes (for example, the presence or absence of introns, differences in GC content and differences in transcriptional promoters), gene inactivation and erosion are likely to be the most common outcomes of HGT between these organisms36,37,38,39. However, despite these structural hurdles, transferred DNA has been shown to become functional in many instances (see below).
Inferring the source of transferred DNA. Two issues hinder the precise identification of the donor organism for many HGT events. The first is that bacterial genomes can contain DNA fragments from many different organisms owing to the high frequency of HGT events among them, so the donor DNA may or may not have the same evolutionary history as most of the other genes in the donor bacterial genome40,41. The second is that phylogenetic problems such as long-branch attraction, differential gene loss and inadequate taxon sampling can often make it difficult to infer the origin of the donor DNA9,42,43.
If we assume that the more common the source organism is in the environment of the recipient organism, the more likely transfer will occur between these organisms, it is not surprising that bacterivorous or parasitic unicellular eukaryotes — in which every cell is a germline cell, and any gene transfer thus has the potential to be transmitted to the next generation — are among the most frequent recipients of genes from bacteria12,44,45. Many other examples also suggest that proximity matters: soil-dwelling nematodes tend to acquire genes from putative soil bacteria46; thermoacidophilic algae acquire genes from putative thermoacidophilic bacteria47,48; plant pathogenic oomycetes acquire genes from plant-associated bacteria49; stramenopile pathogens of the human digestive tract acquire genes from relatives of common members of the gut microbiome50; and many invertebrates acquire genes from common intracellular bacteria such as Wolbachia spp.25,32,34,51. By contrast, for some organisms that seem to be frequent recipients of foreign DNA via HGT, such as rotifers27,52 or fungi53, there is no clear environmental source of transferred DNA, and these organisms seem to acquire genes from multiple donor organisms from diverse environments.
Various models have been put forward to explain the patterns of frequency and taxonomy observed in bacteria–eukaryote HGT11,54,55. For example, the weak-link model56 expands previous proximity-based hypotheses11,54 by suggesting that genes primarily enter cells of their recipient organisms at particular stages of their life cycle in natural environments. As mentioned above, every cell of a unicellular eukaryote is both a germ cell and somatic cell and therefore has the potential to pass transferred DNA to its offspring56. In multicellular eukaryotes, early developmental stages that are exposed to their environment (for example, spores, zygotes or embryos) are likely to be recipient cells for foreign DNA that could be passed to offspring56. Bdelloid rotifers, which are microscopic freshwater animals that reproduce asexually, are the animals with the most reported HGT events from microorganisms, perhaps because of chromosomal instability formed by bouts of desiccation and rehydration39. In cases where early developmental stages are not exposed to the environment, the transfer, genome integration and maintenance of DNA is often promoted by parasites, symbionts or pathogens that infect germline cells28,56,57, as exemplified by the numerous transfers of Wolbachia spp. DNA into arthropod genomes37,58,59.
DNA acquisition and integration. Mechanisms of gene transfer4,11 include conjugation, transduction, transformation and gene transfer agents. Non-homologous end joining (NHEJ) seems to be the major mechanism of incorporation of foreign DNA into the eukaryotic genome60, although recombination with a homologue already present in the genome can also occur (for example, in plant mitochondria)61. As these mechanisms have been thoroughly discussed elsewhere4,11, we focus mainly on the steps that enable transferred genes to be expressed and thus provide the host with new functions.
Whether a newly arrived foreign sequence attains functionality probably depends on its features (for example, length of the acquired DNA, GC content, codon usage, genetic code and epigenetic marks) and the position of where the sequence is integrated in the genome (heterochromatin or euchromatin). The size of the transferred DNA seems to be a major factor in whether the transferred sequence becomes functional (Fig. 2A). Several examples from diverse eukaryotes suggest that HGT events involving large DNA sequences are more likely to result in junk DNA than those involving small fragments38,62,63. For example, large DNA fragments or even entire genomes of Wolbachia spp. that have been transferred to arthropod and nematode genomes are fairly common36,58,59,63,64. In most cases, such transferred fragments undergo rapid non-functionalization through pseudogene formation and DNA deletion and exhibit very low levels of transcription37,65. However, functional genes could still be hidden in these large genome regions. Many of the functional bacteria-to-animal HGT events discussed below are also from Wolbachia spp., but in these cases, the transferred sequences seem to be either much smaller in size or from larger fragments that were reduced in size after transfer25,32,34,51. A fascinating exception to this rule comes from the pillbug Armadillidium vulgare, in which incorporation of an ∼1.5 Mb fragment of Wolbachia spp. DNA into the genome of the pillbug (which subsequently duplicated to a final size of 3 Mb) seems to have been involved in creating a new female sex chromosome64.
Eukaryotic nuclei are often likely to be exposed to DNA fragments from lysed mitochondria and plastids that can become incorporated into the host chromosomes, but they often become non-functional or are eliminated. Transferred sequences from organelle genomes are called nuclear mitochondrial transfers (numts) and nuclear plastid transfers (nupts) or organellar gene transfers56. Several studies have shown that numts and nupts are inserted via NHEJ at double-strand breaks and that microhomology-mediated or blunt-end repair is involved in the incorporation of the DNA (reviewed in Refs 10,40). Incorporation is enriched at open chromatin regions and can occur in different genome regions in different lineages40. For example, numts are enriched at introns in the human genome66 but not in the intron-poor genome of Saccharomyces cerevisiae67. Genomic regions flanking numts are often rich in retrotransposons, and the insertion often occurs immediately adjacent to AT oligomers in mammals68. Transfers of DNA fragments ranging from only several bp to several hundred kb have been reported40, and numts and nupts can be further amplified after acquisition and give rise to tandem repeats40.
In principle, foreign DNA could be incorporated anywhere in a recipient genome that does not disrupt an existing functional genomic element. Several examples suggest that, similar to numts, horizontally transferred sequences are often integrated into genome regions that are enriched in DNA transposons and retrotransposons46,69,70,71,72 (Fig. 2A). In some genomes, sequences that were acquired long ago (in evolutionary terms) are found in gene-rich regions, whereas more recent horizontally transferred sequences are integrated in less conserved and less gene-rich locations, such as telomeric regions, or within or around transposable elements28,72. For example, horizontally transferred sequences that have been acquired by mealybugs tens of millions of years ago are found in gene-rich, possibly euchromatic, genomic regions (Fig. 2B), whereas sequences that were transferred more recently are found in less gene-rich and more poorly assembled genomic regions51. However, owing to a scarcity of data and the poor assembly quality of many eukaryotic genomes, it is difficult to draw firm conclusions with regards to the importance of the position at which the transferred gene is integrated.
Gaining functionality. Transfer events that become functional often seem to involve short foreign DNA fragments, which are mostly integrated in gene-poor but transposable element-rich and dynamic parts of the genome. Tandem duplications of horizontally transferred genes are frequently observed12,25,28,33,49,64,71,73 (Fig. 2A). For example, out of the 48 transferred sequences that are present in oomycete genomes, 38 exist in at least two copies49. Although these duplicates have not been studied in detail, it seems plausible that some copies might maintain the original function, some might gain new functions and some might become non-functional (Fig. 2A).
Intron gain might promote the increased expression of a transferred gene, as the presence of introns has been shown to increase gene expression in many eukaryotes74. Indeed, introns are commonly found in functional horizontally transferred genes in many eukaryotes26,28,29,46,53,72,75,76,77 (Fig. 2A), although these introns are sometimes found in the 5′-UTR or the 3′-UTR of the gene51. However, regardless of the location of the integrated transferred gene or whether it gains introns, eventually all functional transferred genes will evolve as does any eukaryotic gene: their GC content and codon usage will gradually adjust to the host, and they will be expressed in a cell-compartment-specific and/or tissue-specific manner.
Functions of transferred DNA
We classify horizontally transferred genes into two broad types: those that provide a new function in the recipient organism (innovative transfers) and those that replace or maintain a functional loss in the recipient organism (maintenance transfers) (Figs 3,4). Innovative transfers commonly enable recipient organisms to feed on nutritionally poor or toxic diets, parasitize other eukaryotes, survive in cold, hot, acidic, anaerobic or toxic environments or protect themselves from other organisms11,13,17,78 (Table 1). Maintenance transfers are most common in recipients that also house a bacterial endosymbiont that is required for normal host function and can often be explained as a mechanism to maintain the function initially encoded in a degenerating symbiont or organelle genome (Table 1).
Maintenance transfers: enabling the loss of endosymbiont function. All (or most; see below) eukaryotes rely on at least one bacterial endosymbiont, now called the mitochondrion, whereas other eukaryotes rely on an additional bacterial endosymbiont now called the plastid or chloroplast79. Extant mitochondrial and plastid genomes are small and encode only a few genes, and organelle function thus requires extensive participation from the host. Many of these host-encoded genes have been transferred from bacteria, the protein products of which are transported back into the organelle. As such, maintenance transfer of genes from bacteria has shaped the content of all eukaryotic genomes11,40,80,81.
However, the origins of these endosymbiont-related horizontally transferred genes are difficult to establish. Although it is not disputed that the endosymbiont that became the mitochondrion was an alphaproteobacterium (and the plastid a cyanobacterium82,83), controversy arises because the bacterial horizontally transferred genes that have been found in eukaryotic genomes affiliate with numerous other bacterial groups and not only the Alphaproteobacteria and Cyanobacteria in phylogenetic trees40,84. This taxonomic diversity has motivated several hypotheses to explain the data41,85,86,87,88,89,90, which fall roughly into two groups. The first group hypothesizes that the endosymbionts that eventually became organelles had highly mosaic genomes owing to HGT among bacteria40,41. The second group hypothesizes that the cell that became host to the organelle already had previously acquired genes by HGT before the endosymbiont became fixed91,92,93,94. These (not mutually exclusive) hypotheses will likely always be difficult to differentiate owing to the antiquity of organellogenesis. Might more recent endosymbioses provide some insight into the possible general mechanisms used to build mosaic metabolic pathways? The advantage of more recent endosymbioses, such as those found in sap-feeding insects25,32,34,51, is that inferring the origin of transferred genes is much more straightforward because the genes often come from bacteria, such as Wolbachia spp., that have extensive germ cell interactions with insects. If phylogenetic analyses of an HGT in an insect genome suggest that the gene came from Wolbachia spp., the most parsimonious explanation is that the transfer came from infections with the bacteria sometime in the history of the insect lineage. The disadvantage is that the cell biological and genetic contexts of these symbioses are different than those encountered at organellogenesis.
Paulinella chromatophora (Fig. 4a) is an amoeboid protist that acquired an organelle-like cyanobacterial endosymbiont ∼100 million years ago95. In this case, only ∼25% of its 229 nuclear genes of bacterial origin seem to result from transfer from the cyanobacterial endosymbiont44. The remaining 75% taxonomically affiliate with other bacterial groups in phylogenetic tress. Similarly, the incomplete nutritional pathways in some trypanosomatid bacterial endosymbionts are complemented by genes that were horizontally transferred from diverse bacteria to the protist genome96,97. Finally, analyses of insect nutritional endosymbioses in aphids, psyllids, whiteflies and mealybugs also show similar patterns of mosaic nutritional pathways that are constructed from multiple HGT events25,32,34,51. Many of these transferred genes seem to compensate for genes that have been lost in the genomes of the endosymbionts, but — as in the protist examples above— most of those genes have been transferred from bacteria other than the existing endosymbionts. Importantly, these transferred genes come from 'reproductive manipulators' such as Wolbachia spp. and Candidatus Cardinium spp., bacteria that are extremely common in insect germ cells. Thus, the transferred genes come not from the nutritional endosymbiont but from endosymbiotic bacteria with germline-cell tropism. Taken together, these results suggest that gene transfer from degenerate (or degenerating) extant endosymbionts to the host is not necessarily needed, or perhaps even common, and in some hosts, HGT events from other sources can compensate for gene loss in the endosymbiont28. We argue that these data support the idea that the taxonomic diversity of HGT events in eukaryotic genomes could have resulted from previously existing HGT from non-organelle sources44,54,84,98,99,100, probably in combination with the 'inherited chimaerism' of organelle progenitors resulting from bacteria–bacteria HGT events that predated endosymbiosis40,41.
A final example of a maintenance HGT event comes from the microbial eukaryote called Monocercomonoides sp. PS203 (Ref. 101). As a member of the eukaryotic supergroup Excavata, Monocercomonoides sp. PS203 is related to several microaerophilic or anaerobic parasites, such as Giardia intestinalis and Trichomonas vaginalis. Both G. intestinalis and T. vaginalis have highly degenerate mitochondria-related organelles that no longer produce ATP through oxidative respiration but still exist in the cytoplasm as double-membrane-bound organelles. These mitochondria-related organelles provide the only function conserved in all mitochondria and related organelles across eukaryotes: iron–sulfur (Fe–S) cluster biogenesis102. However, in Monocercomonoides sp. PS203, none of the proteins related to mitochondrial function, including Fe–S cluster biogenesis, could be identified in its genome101. Instead, the authors found that a related set of genes involved in Fe–S cluster biogenesis had been transferred from bacteria to the Monocercomonoides sp. PS203 genome101. These HGT events thus enabled the loss not only of all mitochondria-related proteins but also of the organelle itself101. This example highlights an important point about maintenance HGTs: the transfer of a new bacterial gene to the eukaryotic host likely needs to occur before the loss of the ancestral gene.
Innovative transfers: detoxification and protection. The genetic repertoire of multicellular eukaryotes is relatively small compared with that of bacteria. Thus, it is not surprising that eukaryotes have exploited the complexity of bacterial metabolism by acquiring genes to help detoxify novel environments (Fig. 4b). For example, herbivorous eukaryotes must cope with complex mixtures of toxic compounds that are produced by plants17. Horizontally transferred genes that are involved in detoxifying cyanide were found in phytophagous mites and various lepidopterans33. The beetle Hypothenemus hampei can feed exclusively on coffee beans owing to a mannanase that has been acquired through HGT70. The silkworm Bombyx mori has acquired the ability to circumvent alkaloids in the latex of mulberry plants103.
Fungi in the genus Epichloe are intercellular symbionts of grasses that protect their hosts from insect herbivores by use of a cocktail of fungal alkaloids. This anti-insect arsenal of Epichloe fungi seems to be supplemented by the mcf gene, which was acquired horizontally by the fungal genomes from a bacterium104. It was shown that mcf encodes a toxin produced by the endosymbiotic fungi to help kill insect larvae that feed on their host grass plants. Interestingly, the mcf gene is also found in the genome of the bacterial symbionts of entomopathogenic nematodes and was shown to be sufficient to kill insects105. Thus, HGT from one tripartite interaction (nematodes, bacteria and insects) has possibly altered the interactions in another tripartite interaction (grasses, fungi and insects)104.
Innovative transfers: nutrition. Plant material represents an enormous amount of the biomass of Earth, and many eukaryotes have evolved to feed on both living and dead plants. Herbivores must not only avoid plant defences but also degrade complex plant carbohydrates, and the genes required for degrading these complex carbohydrates have been acquired in numerous eukaryotic lineages. HGT of bacterial genes involved in carbohydrate metabolism has been found in herbivorous insects17, rumen ciliates106, oomycetes and fungi49,107, plant-parasitic nematodes46,77,108, necromenic nematodes from the Pristionchus genus109 and rotifers52. Conversely, tunicates can synthesize cellulose for their protective exoskeleton (the tunic) because of an ancient cellulose synthase gene that has been acquired horizontally110.
HGT of the genes encoding biosynthetic enzymes has enabled many eukaryotes to live in extremely nutrient-poor environments (Fig. 4c). Genes involved in amino acid, vitamin and carbohydrate metabolism are perhaps most often transferred via HGT (of note, in cases in which the host has an endosymbiont that provides these nutrients to the host, such transfer events are probably maintenance transfers28,32,34,73). Most of the transferred genes found in 13 genomes of unicellular protists involve amino acid and carbohydrate metabolism12. For example, the genomes of Cryptosporidium spp. contain genes that were transferred horizontally, such as genes encoding tryptophan synthase, aspartate–ammonia ligase and glutamine synthetase111, and trypanosomatids have acquired numerous horizontally transferred genes involved in arginine, tryptophan, threonine, methionine, cysteine, lysine and vitamin B5 metabolism from diverse bacterial donors96,97. Marine environments can be notoriously poor in nitrogen and iron, and HGT events have had a role in enabling several distinct eukaryotic lineages to scavenge these growth-limiting compounds from their environments. Iron-binding proteins of bacterial origin were found in diatoms112,113, an extremophilic red alga Galdieria sulphuraria47 and a soil-dwelling amoeba Dictyostelium discoideum114,115. Nitrogen metabolism was influenced by HGT events, at least in diatoms76 and green algae from the Micromonas genus116. In other environments, the limiting nutrient is not that obvious; for example, the acquisition of N-acetylneuraminate lyase probably enables T. vaginalis to scavenge sialic acids from its host for nutrition117. The transfer of genes involved in the biosynthesis of amino acids and vitamins from bacteria to animals includes genes involved in the diaminopimelic acid pathway in the choanoflagellate Monosiga brevicollis118, haeme biosynthesis pathway (aferrochelatase) in the nematode Brugia malayi29 and vitamin B1, vitamin B5 and vitamin B6 pathways in the plant nematode Heterodera glycines75,119.
Innovative transfers: adaptation to extreme environments. HGT events have enabled eukaryotes to live in extremely hot, cold and otherwise toxic environments. The transfer of bacterial genes that encode ice-binding proteins, which can enable recipients to survive in extremely cold environments, was found in the diatom Fragilariopsis cylindrus120, the haptophyte Phaeocystis antarctica121 and the green algae Pyramimonas gelidicola121 and Coccomyxa subellipsoidea122. HGT events that enable recipients to survive hot, metal-rich and acidic environments were detected in the genome of red alga G. sulphuraria47 (Fig. 4b), and transferred genes that enable survival in hypersaline conditions were detected in the genome of Halocafeteria seosinensis123. HGT events that protect hosts from other sources of environmental stress, such as desiccation and oxidative and osmotic stress, seem to be frequent in rotifer species from desiccating habitats39. Likewise, two HGT events seem to protect the microsporidium Antonospora locustae from oxidative and ultraviolet light-induced damage124,125.
Innovation transfers: bactericidal activities. Genes of bacterial origin that target the bacterial cell envelope are commonly found as functional transferred genes in eukaryotic genomes30,31. The phenomenon of eukaryotes using bacterial genes to defend themselves against other bacteria has been called 'the eukaryotes strike back' (Ref. 126). Bacterial cell envelopes can be disrupted by several mechanisms, but peptidoglycan is a common target (Fig. 3).
An example of eukaryotes acquiring bacterial genes to protect themselves from other bacteria involves a family of amide-bond-breaking genes called type IV secretion system amidase effectors (tae)30. Tae proteins cleave the amide linkages of peptidoglycan and were domesticated in at least three eukaryotic supergroups (Fig. 3; Table 1). Some of these transfers date back to more than 800 million years ago30. Importantly, the functional role of these HGT events was experimentally verified, and it was demonstrated that they limit the proliferation of Borrelia burgdorferi in the deer tick Ixodes scapularis30 (Fig. 4d). Other examples show that bacterial genes that encode lysozymes, which cleave peptidoglycan between N-acetylglucosamine (NAG) and N-acetyl muramic acid (NAM) moieties, have also been acquired by eukaryotes such as plants, fungi and insects multiple times independently25,31,127 (Fig. 3). The antibacterial effect of the horizontally transferred genes was also verified experimentally with a recombinant lysozyme, and the proteins encoded by the genes exhibited broad-spectrum antibacterial action31. Perhaps the most astounding toolkit for the disruption of the bacterial cell envelope acquired through HGT was reported for the marine protist Trepomonas sp. PC1 (Diplomonadida)45. This species clusters deeply within a clade of parasites but seems to have transitioned to a free-living and bacterivorous lifestyle. This nutritional transition is hypothesized to have been enabled by the acquisition of forty bactericidal, permeability-increasing proteins (which bind to lipopolysaccharides), twenty cell wall hydrolases, four N-acetylmuramoyl-L-alanine amidases and five lysozymes that now enable the species to feed on bacteria.
Innovation or maintenance: peptidoglycan production. Bacteria-to-eukaryote HGT events include not only genes for peptidoglycan degradation but also genes for peptidoglycan synthesis. Although the functional role of these transfers remains unclear (and thus their classification in the system we use in this Review remains unclear), putative horizontally transferred genes involved in peptidoglycan synthesis have been found to be expressed in bdelloid rotifers72, aphids25 and mealybugs28,51. Interestingly, in some Archaeplastida chloroplasts, the generation of peptidoglycan seems to also require numerous bacterial genes that were horizontally transferred to the host128,129. These genes have been shown to be essential for chloroplast division in both mosses130 and Arabidopsis thaliana131. Although a peptidoglycan layer has been found surrounding the chloroplasts of mosses120, no detectable peptidoglycan layer seems to exist in A. thaliana.
A similarly unclear situation exists for animals with obligate endosymbionts that have horizontally acquired amidases25,28,51. Amidases are related to peptidoglycan-recognition proteins (PGRPs), which are normally used by the innate immune system of animals as bacterial sensors132. Some PGRPs possess amidase activity that can cleave peptidoglycan at positions distinct from Tae proteins. Interestingly, PGRPs have been shown to be involved in endosymbiont maintenance rather than pathogen defense133. For example, native amidase-active PGRPs are used by tsetse flies to shield their nutritional endosymbionts from the host immune system by recycling peptidoglycan from lysed endosymbiont cells134. Some sap-feeding insects with obligate bacterial endosymbionts have acquired bacterial amidases by HGT25,28,51, raising the prospect that some of these horizontally acquired amidases are used not as antiacterial control measures by the host but rather for endosymbiont maintenance. In a potentially parallel process, bacterial β-lactamases have been found to have been transferred to the genomes of mealybugs and the slime mould D. discoideum28,114; β-lactamases are enzymes that provide resistance to β-lactam antibiotics, such as penicillin. Why would eukaryotes need genes targeting bacterial antibiotics? It seems plausible that, similar to PGRPs, these enzymes could be used for symbiont protection rather than defence from pathogens, as both slime moulds and mealybugs are tightly associated with beneficial bacteria51,135.
Both early and recent misattributions of HGT events in animal genomes have made it clear that care is required when hypothesizing that a gene has been transferred to a genome from an unrelated organism. However, numerous papers reporting HGT events from bacteria to eukaryotes stand on solid conceptual and methodological ground (Table 1). Although the field of HGT research is vibrant, there is clearly room for improvements in both the methods used to screen for contamination in genome projects and the methods used to verify putative HGT events (Box 1). As discussed in this Review, HGT events are used to maintain functionality — often in situations involving an endosymbiont that is subject to genomic erosion — but are also used to acquire new functions, such as altered nutrient acquisition and protection, and to explore new environments. To better understand the true phylogenetic and functional scope of bacterial genes in eukaryotes, a focus on phylogenetically important but neglected groups such as protists is warranted, as the majority of eukaryotic diversity resides in these understudied organisms.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
F.H. was supported by the Fulbright Commission and a fellowship from the European Molecular Biology Organization (EMBO; ALTF 1260–2016) while writing this Review. J.P.M. was supported by grants from the US National Science Foundation (IOS-1256680 and IOS-1553529), the US National Aeronautics and Space Administration (NASA) Astrobiology Institute (NNA15BB04A) and the Gordon and Betty Moore Foundation (GBMF5602).
HGT events from bacteria to eukaryotes.
- Horizontal gene transfer
(HGT). Movement of genetic material between organisms (also called lateral gene transfer) via non-vertical (not parent-to-offspring) transmission.
- Germ line
The specialized cellular lineage in multicellular sexual organisms that is used to pass on genetic material to the progeny.
Organisms living within the body or cells of another organism.
- Long-branch attraction
A phylogenetic artefact that causes distantly related lineages (often on long branches) to be incorrectly inferred as closely related in phylogenetic trees.
Gene transfer from a donor to a recipient by direct cell-to-cell contact, such as plasmid transfer between two bacterial cells.
Gene transfer carried out by a virus, such as a bacteriophage, transferring DNA from one bacterium to another.
Direct acquisition of DNA from the environment through the cell membrane.
- Gene transfer agents
Bacteriophage-like elements that package random DNA regions from a host cell and transfer them to a recipient cell.
- Non-homologous end joining
(NHEJ). A pathway for the direct repair of double-strand DNA breaks without a homologous template.
- Nuclear mitochondrial transfers
(numts). Transfers of mitochondrial DNA into the nuclear genome of a eukaryotic host; the encoded genes often become non-functional.
- Nuclear plastid transfers
(nupts). Transfers of plastid DNA into the nuclear genome of a eukaryotic host; the encoded genes often become non-functional.
- Transposable elements
DNA sequences (also known as jumping genes) that can move within a genome (and sometimes also between genomes) by a 'cut and paste' (DNA transposons) or a 'copy and paste' (retrotransposons) mechanism.
The process by which an endosymbiont becomes an organelle, in part by becoming (nearly) irreversibly integrated with its host cell at both a genetic and cell biological level.
- Reproductive manipulators
Bacteria, such as Wolbachia spp., that are transmitted in the egg cytoplasm of arthropods and nematodes and shift the sex ratio of the host population.
A structural matrix in bacterial cell walls formed by alternating N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM) residues, where peptide chains of up to five amino acids link NAM to other NAM-connected peptides.