From conifers to cognition: Microbes, brain and behavior

A diversity of bacteria, protozoans and viruses (“endozoites”) were recently uncovered within healthy tissues including the human brain. By contrast, it was already recognized a century ago that healthy plants tissues contain abundant endogenous microbes (“endophytes”). Taking endophytes as an informative precedent, we overview the nature, prevalence, and role of endozoites in mammalian tissues, centrally focusing on the brain, concluding that endozoites are ubiquitous in diverse tissues. These passengers often remain subclinical, but they are not silent. We address their routes of entry, mechanisms of persistence, tissue specificity, and potential to cause long‐term behavioral changes and/or immunosuppression in mammals, where rabies virus is the exemplar. We extend the discussion to Herpesviridae, Coronaviridae, and Toxoplasma, as well as to diverse bacteria and yeasts, and debate the advantages and disadvantages that endozoite infection might afford to the host and to the ecosystem. We provide a clinical perspective in which endozoites are implicated in neurodegenerative disease, anxiety/depression, and schizophrenia. We conclude that endozoites are instrumental in the delicate balance between health and disease, including age‐related brain disease, and that endozoites have played an important role in the evolution of brain function and human behavior.

(entomopathogens), including fungi and bacteria, because their presence can confer resistance to insect attack. 6 Other endophytes alter the metabolism of the host to generate defense molecules that in turn can fend off herbivore attack or reduce infection by other plant pathogens. 7 Paralleling the role of root microorganisms in plants, the microbiome in the gut of vertebrates promotes the metabolism and uptake of external nutrients from the diet, 8 can have beneficial effects on the immune system, 9 and can exert direct modulatory effects on brain and behavior (see below). Commonalities between root and gut, including a large surface area and a key role in nutrient assimilation, have been highlighted previously. 10,11 However, as in plants, other healthy tissues in vertebrates also contain diverse types of microbes.

| Endozoites are widely distributed in healthy tissues
In plants, endophytes are found in association with all tissues analyzed to date, and this could lead us to suspect that endozoites might also be present in multiple body tissues of vertebrates. In human, outside the intestine, attention has been paid to colonization of gastric mucosa by Helicobacter pylori that has also been associated with the initiation of gastric ulceration, 12 but most recent work has been carried out on epithelial surfaces that are a rich repository of bacterial species (e.g., the Human Microbiome Project 13 ). Bacterial species are widely reported in blood of healthy individuals (reviewed in Ref. 14). In another study an overt pathogen, Streptococcus pneumoniae, was found in the nasopharyngea of 4% of adults but in 53% of children, 15 blurring the distinction between endozoite and pathogen.
Diverse viruses are also present on external surfaces, including the naso-olfactory system and mouth, 16 and multiple viruses are found in cervical secretions. 17 Lymphoid and neuronal cells are additional repositories for lymphotropic/neurotropic viruses.
Indeed, the majority of the population is seropositive for multiple herpes viruses, including herpes simplex virus (HSV; Table 1).
HSV-1 and HSV-2 seropositivity increases with age in the United States, where >60% of the elderly population is HSV-1-positive, and >20% are positive for both HSV-1 and HSV-2. 18 Similar findings have been reported in Europe, with seropositivity rising with age to >80% in some countries, 19 pointing to progressive acquisition over a lifetime. Similar high seroprevalences have been reported for multiple herpes viruses ( Table 1 20 By contrast, the distribution of endozoites in peripheral tissues has not been systematically addressed. However, in addition to the gut, blood, and epithelial surfaces including the lung, 21 there is evidence that endozoites are present within other organs. PCR and deep sequencing of DNA from normal hamster liver revealed multiple bacterial species, confirmed by direct microbial culture from liver tissue. 22 There is evidence that the kidney also houses its own microbiota, 23 and the human urinary microbiome (a proxy for the kidney) contained multiple species of Firmicutes, Actinobacteria, Bacteriodetes, and Proteobacteria. 24 Studies on other key tissues are so far lacking, for example breast, heart, ovary, pancreas, prostate, skeletal muscle, testis, thymus, and thyroid; future work will be necessary to address this issue.

| Microbial diversity
It is important to recognize that our knowledge of the spectrum of endozoites is limited. Kowarsky et al. performed deep sequencing of blood cell-free DNA from 188 individuals and reported that, of new bacterial and viral (including bacteriophage) sequences, the majority were absent from current databases. 25 Notably, this study almost doubled the number of human anelloviruses sequenced to date. We are only seeing the tip of the iceberg.
In addition, although we tend to refer to viruses as unique entities, this is seldom if ever the case. For example, HSV-1 and HSV-2 each comprise multiple substrains; moreover, recombinants between HSV-1 and HSV-2 have been detected worldwide. 26

| GUT-BRAIN AXIS
Multiple studies report that the composition of the gut microbiota can influence brain function, 27,28 partly by release of metabolites including neurotransmitters that can enter the circulation, but also via direct neuronal communication between gut and brain (notably via the vagus nerve). For an excellent review see Ref. 29. However, our focus here is on microbes that directly enter the CNS. That endozoites are probably present in all human tissues is borne out by recent studies on the brain, discussed below.

| ENDOZOITES IN THE BRAIN
We saw above that the plant root contains a high diversity of endogenous microbes, thus resembling the vertebrate gut. However, ever since Darwin's time there have been suggestions that the plant rootthrough demonstrable adaptive behavior-in some ways resembles the vertebrate brain (the "root-brain hypothesis" 30 ) where endophytes play a pivotal role. The presence of microbes in the brain, and whether they modulate behavior, is the central focus of this review.
Box 2 overviews current knowledge of the presence of diverse endozoites in brain, not only including bacteria, protozoans, Archaea, and viruses, but also bacteriophages, higher eukaryotes, and plantderived agents (endogenous retroviruses/retroelements are not considered to be true endozoites; Box 3). One central conclusion emerging from this analysis is that there are parallels between the taxa found in plant tissues and those found in both gut and brain (Table 2, discussed in Box 2), suggesting that some species may be particularly adapted to a close association with multicellular organisms including both plants and animals.
Although these combined reports confirm the widespread presence of endozoites in the brain, a complexity of all these studies is that it is not always possible to distinguish between endozoites that endogenously inhabit brain tissue versus contamination with agents borne by the circulation (including migrating lymphoid cells). Nevertheless, HSV sequences within the brain parenchyma have been confirmed by in situ hybridization, 31,32 and in situ studies have directly demonstrated bacteria and protozoans within the brain (see below However, entry to the brain demands special mechanisms because the blood-brain barrier (BBB) prevents simple diffusion into brain tissue.
The BBB, generated by tight junctions between endothelial cells lining the cerebrovasculature, effectively prevents small molecules from entering the brain parenchyma. However, the BBB does not appear to constitute a barrier to many microbes. Porphyromonas gingivalis could be detected within the brain parenchyma following chronic oral administration to mice, 34,35 and replication-defective HSV is found in the brain within 3 days (see below). Intranasal or intratracheal administration of the fungus Cryptococcus neoformans in mice led to rapid dissemination into the brain as quickly as 3 h postinfection, although the titers were low (≤1% of the inoculum). 36 Similarly, intranasal administration of a filamentous bacteriophage to mice was followed by rapid appearance in the brain. 37 Microbes are thought to sidestep the BBB by exploiting at least four routes. First, many biologically active molecules such as polypeptide hormones are actively transported across the BBB into the brain, and many pathogens exploit host receptor-mediated transcytosis. Following receptor-ligand interactions at the endothelial cell surface, several microbes (bacteria and yeast) are internalized as vacuoles and thence pass into the brain without disrupting the integrity of the BBB (reviewed in Refs. [38][39][40]. For example, following intravenous inoculation in mice, the yeast Candida albicans is found within brain tissue as quickly as 4 days postinfection 41 by exploiting a specific receptor on endothelial cells. 42 Second, the BBB is not effective against migrating host cells, as exemplified by the common appearance of tumor cell metastases in brain tissue. Immune cells such as macrophages can efficiently enter the brain, and latent lymphotrophic viruses (e.g., HHV-6 and 7) borne by macrophages can thereby be delivered into the CNS. Viruses associated with mobile cells can therefore gain brain access (the "Trojan horse mechanism"), and similar pathways may apply to intracellular bacterial pathogens. 38,39 Passage parallels that in the gut, where intestinal dendritic, goblet, and M cells have been implicated in transporting bacteria from the gut lumen into the circulation. 14 BOX 2 Endozoites in the brain Bacteria For many years there have been reports of bacteria in brain, notably in Alzheimer's disease (AD), such as Borrelia 134,205 and Chlamydia/Chlamyophila spp. 101,139 ( Figure I), but also in healthy tissue. Systematic surveys in both health and disease using PCR and deep sequencing have revealed that the major phyla are α-Proteobacteria and Actinobacteria, with further Firmicutes and Bacteriodetes 136,138 ; similar findings were seen in macaque monkeys 136 (Table 2). Proteobacteria constitute a phylum of Gram-negative bacteria that include not only gut commensals (e.g., Escherichia) but also several human pathogens (e.g., Yersinia). The Actinobacteria also include human pathogens (e.g., Mycobacterium).
These are the same taxa that are seen in plants (Table 2), including the actinobacterium Frankia that can fix nitrogen, and several agriculturally important nitrogen-fixing α-Proteobacteria species that enter symbiotic relationships with leguminous plants (e.g., Rhizobium spp.).
These specific taxa may be predisposed to live in association with higher eukaryotes (noting that endosymbiosis with α-Proteobacteria is held to have been the primary driver for the emergence of Eukaryotes 206 ). In support, some human commensal bacteria of these phyla (e.g., Enterobacteriaceae)-gut organisms in humans-can colonize root tissues of plants such as maize, lettuce, tomato, and barley where they may persist as a reservoir for recolonization of humans, 207,208 and can even promote the growth of the new plant host. 207

Protozoans
Fungi are widespread in plant tissues, but there have been few studies on human, although Pisa et al. 101 report intermittent detection of several fungal species, principally Candida spp., in brain tissue of Alzheimer patients. Candida spp. are also prevalent plant endophytes, and species such as Candida metapsilosis are found both in plants and in human infections. The apicomplexan T. gondii, an obligate intracellular eukaryote, is widely present in healthy humans and animals, and 20-50% of the population is seropositive for T. gondii. 209,210 T. gondii can persist in a subclinical state in multiple tissues including the brain. 211 Toxoplasma sequences were detected in 16.5% of human brain samples. 104 A different apicomplexan, Neospora, is an endozoite of bovine species. Oomycetes such as Phytophthora spp. that are phylogenetically rooted alongside apicomplexans are important plant pathogens.

Archaea
These simple unicellular organisms are inferred to be the evolutionary precursor to all life on Earth, and are well represented in the plant root microbiome 212 (reviewed in Ref. 213). They are also present in human colonic, lung, nasal, pulmonary, and oral microbial flora, 214 However, no studies to date have systematically addressed whether they enter healthy tissues such as the brain, and so far no archaeal species has been demonstrated to cause disease in human. 215,216 However, methanogenic Archaea have been directly implicated in refractory sinusitis, 217 and one report described finding archaeal species in multiple brain abscesses, although only 1/27 control samples were positive 218 -possibly suggesting that archaea are not typically present in the normal CNS, perhaps because they are predominantly anaerobic. Archaeal viruses have been intermittently reported in blood. 20 Further studies on Archaea in healthy and diseased brain are certainly warranted.

Viruses
HSV in normal neuronal tissue (sensory ganglia) was reported in 1972, when virus was isolated by culture of brain tissue from 1 of 22 patients with no evidence of active HSV disease. 219 HSV in 18 of 39 normal trigeminal ganglia was demonstrated by passage and immunohistochemistry. 220 The presence of HSV sequences in normal brain samples has been confirmed by DNA-based analysis 221 and by PCR. 222 Multiple other virus sequences have also been detected. Infection with polyomaviruses is widespread in the human population, and Southern blot and PCR analysis confirmed the presence of BK and JC genomes in up to 20% of healthy brain samples. 223 Other viruses including HHV-6A/6B, EBV, CMV, VZV, and coronavirus have been reported in human brain, 136,221,224 and HHV-6A and -6B proteins were detected using specific antibodies in 22-32% of control brain samples. 145 EBV was detected by PCR in 24% of control brain samples. 225 Deep sequencing has now revealed an extraordinary diversity of viruses in normal human brain, ranging from several types of HHV and HSV to adenovirus, Duvenhage virus, hepatitis C virus, coronavirus, torque teno virus, and BK polymavirus, among others. 140 There was evidence of over-representation of HHV and HSV in AD brain. 140

Bacteriophages
There have reports of bacteriophages (or bacteriophage-like sequences) in multiple human tissues and that phages readily enter the brain (reviewed in Refs. [226][227][228]. For example, a sequence (dubbed Sphinx) with 70% homology to an Acinetobacter bacteriophage was reported in transmissible spongiform encephalopathy brain. 229 These may well be markers of bacterial coinfection rather than pathogens or endozoites in their own right because they are likely to lack the machinery for replication in higher eukaryotes. However, infection (e.g., of plastids of prokaryotic origin) may not be formally excluded. Arabidopsis chloroplasts contain a replication machinery similar to that of bacteriophage T7 230 and so-called mitochondrial viruses ("mitoviruses") have been reported in several plant species, 231 but not yet to the best of our knowledge in vertebrates.

Higher eukaryotes
Endozoites more rarely include multicellular species such as the tapeworm, Taenia solium. Neurocysticercosis caused by T. solium is the most common parasitic disease of the human CNS, and, although the parasite infects multiple body tissues, the larvae display a strong affinity for the CNS. 232 Common symptoms include headache, seizures, and meningitis, and in children include depression, social problems, and rule-breaking behavior. 233

Algae-and plant-derived agents
For completeness we include agents derived from photosynthetic species. Apicomplexans such as Toxoplasma have a secondary plastid-the apicoplast-whose distant ancestor was probably a photosynthetic plastid that originated from a red algal cell 234  Regarding plant-derived agents, there has long been speculation that some plant viruses may also interact with, or reside in, humans spreading to multiple brain regions including hippocampus and cortex. 48 Virus replication does not appear to be essential for dissemination. After delivery of a replication-defective marked HSV-1 either intranasally or intravenously, viral gene expression was detected in multiple brain regions as quickly as 3 days after inoculation. 49 Fourth, via the circumventricular organs. The circumventricular organs and choroid plexus lack a classical BBB, and trypanosomes (Trypanosoma spp.) appear to use this route for early invasion, whereas infiltration of the brain parenchyma occurs only later. 50,51 Thus, although the BBB may afford an obstacle to opportunistic pathogens, species-adapted endozoites appear to have evolved effective mechanisms to evade the BBB and gain brain entry. However, this remains to be clarified more extensively. To this list one must add biofilms, secreted layers of inert (nonimmunogenic) polymers that coat the local environment of the cellular endozoites, 53 as well as coronas of host proteins that can surround the endozoite, 54 and that may prevent recognition by innate immune receptors, antibodies, and cell-mediated immune mechanisms.

BOX 3 Retroviruses and retroelements
The genomes of humans and mice are (as in plants) also vastly punctuated by integrated elements such as endogenous retroviruses (e.g., HERVs) and retroelements (e.g., long and short interspersed nuclear elements: LINES and SINES) that are inferred to have had an earlier exogenous origin. Although these may have played a major evolutionary role, they are not generally regarded as endozoites, and the focus here is on agents acquired from the environment. However, in mouse, there is evidence for active retrotransposition of LINES in brain, with suggestions that these might be of benefit (so far uncharacterized) to the host (reviewed in Ref. 245). In human, LINE element mobilization in brain has been reported in schizophrenia, 246 and upregulation of HERV-W expression and increased copy number have been reported in multiple sclerosis and neurological disease (reviewed in Refs. 247,248). The potential benefits and adverse effects of endogenous retroviruses/elements warrant further study.
Other plant agents may potentially reside in mammals. Following antibiotic administration to mice, fecal DNA PCR amplicons were highly enriched in Streptophyta (a taxon comprising land plants and six main lineages of green algae) and Zea luxurians (a species of grass), perhaps reflecting residual plant-origin DNA introduced via feed. 242 Streptophyta have been widely detected in human mucosal microbiomes 243 and could derive from plant pollens. However, Streptophyta have been detected in mouse seminal fluid, 244 raising the possibility that green algae might potentially be true endozoites. Caution is urged, however, because the PCR apparatus might itself be contaminated with pollens.
Conversely, some human endozoites can infect plants (see earlier). Nonetheless, with the exception of apicomplexans, there have been no reports of algae-or plant-related agents in brain tissue, but future studies on the human microbiome (including the brain) should not limit themselves to known vertebrate endozoites.
As we will see later, several endozoites also cause local or systemic immunosuppression that further contributes to their persistence.

| LATENT/DORMANT ENDOZOITES ARE NOT SILENT
A longstanding view is that endozoites, after entering tissues such as the brain, remain in a hidden "silent" form that can persist for years. However, this view has been challenged by findings that latent HSV-1 infection is accompanied by persistent cytokine upregulation. Following low-dose infection of mice with HSV-1, brain cytokines including IFN-γ, IL-4, IL-6, and TNF-α were chronically upregulated for up to 120 days postinfection, despite evident viral clearance. 55,56 In an important study, Halford et al. 57 reported that ongoing treatment of latently infected mice with a potent inhibitor of HSV-1 DNA replication, aciclovir, led to an extensive decline in brain (trigeminal ganglia) expression of IFN-γ and TNF-α by 120 days postinfection (aciclovir was started at 15 days postinfection). This demonstrates that, instead of being inert, low-level HSV-1 gene expression and DNA replication continues to take place despite viral clearance, driving chronic cytokine production. 57 Low-level neuronal expression of multiple VZV proteins was also detected during latency. 58 In support, viral sequences are detected in blood, and saliva can be a source of HSV and VZV virions in otherwise healthy individuals. 59,60 The reported behavioral effects of T. gondii infection (discussed in more detail later) also argue that the parasite is not silent, and causes changes in the host despite persisting in a subclinical state.
In sum, far from being silent, the albeit limited evidence suggests that endozoites in subclinical infection (paralleling endophytes in plants) display low-level gene expression and turnover that may have an ongoing influence on their host. As we will see in the following sections, endozoites can both provide benefits to the host as well as manipulating local or systemic immunity and behavior to maximize their own persistence and/or onward transmission.

| THE ENEMY WITHIN: BEHAVIORAL CHANGES AND IMMUNOSUPPRESSION
Plant endophytes alter host physiology and metabolism, in some instances to promote host defense (see earlier), and in others to ensure their own propagation-for example by associating with seeds or pollen, 33,61 and also by suppressing host immunity. 62 Indeed, local immunosuppression is essential for maintenance of Rhizobium-legume symbiosis. 63 Similarly, specific endozoites cause behavioral and physiological changes that may facilitate their own proliferation. Although extensive work has been done in insects (not reviewed), the focus here is on vertebrates and the brain. Because most work has been done with acute replication, this is covered first before discussing more subtle changes taking place in subclinical infection.
Pathogens including viruses not only influence host behavior (perhaps to promote host-host transmission) but also can dampen the immune system (to prevent their elimination). A second route but less well studied route exploits the fact that limbic brain areas control both behavior and the immune system.

| Local immunosuppression
Damage to temporal brain including hippocampus, amygdala, and overlying cortex can cause Klüver-Bucy syndrome that is associated in monkeys and humans with hyperorality, hypersexuality, and decreased or increased aggression. 69  A case may therefore be made that hippocampus targeting by endozoites can contribute to immunosuppression.
As we will see, several classes of endozoites directly enter the brain and, through effects on key brain regions such as the hippocampus, may both dampen the immune system and cause behavioral changes to promote their own proliferation. Rabies virus affords the exemplar of how a virus can influence both behavior and immunity, and we consider this case first.

| Rabies virus: the exemplar
Rabies virus (RV) is a pathogen, and infection is predominantly lethal (unless rescued by vaccination 75 ), and it is therefore not a true endozoite (related endozootic lyssaviruses of humans and animals are reviewed in Box 4), but this neurotropic negative-stranded RNA virus is the paradigm for behavioral changes.
The name of the virus is reputed to derive from the old Indian word rabh, meaning "to make violent." 76 Following peripheral infection, the virus travels in a retrograde direction via the nervous BOX 4 Viral endozoites in the brain: effects on behavior and immunity Lyssaviruses: Duvenhage and Borna disease virus (BDV) In addition to rabies virus (main text), sequences for the closely related virus, Duvenhage, have been reported in normal human brain. 140 Further studies will be necessary to address whether endogenous Duvenhage-related viruses, like rabies, modulate the immune system and alter behavior. A third lyssavirus, BDV, is an endozoite in many vertebrate species. Infection with BDV in several non-human species is followed by long-term persistent infection, and the majority of the target animal population harbors this virus. 249 The virus exhibits tropism for the hippocampus. 250 In experimental animals BDV can cause anxiety and aggression without overt fever, 251 and disintegration of the hippocampal dentate gyrus was observed in late infection 197 (Figure 1(B)). Immunosuppression in BDV infection has also been reported. 252 A role for BDV as an endozoite of humans (vs animals) remains uncertain, 253 but the presence of integrated BDV-related sequences in primates including human 183,184 suggests a close association.

Herpes viruses
The majority of the human population harbors several types of herpes viruses (see earlier). The peculiar proclivity of herpes viruses for the brain (specifically HSV-1 and HSV-2-the HHV group of viruses tend to be lymphotropic) has been known for almost a century (Ref. 254 , reprinted from 1929). As noted earlier, the virus can persist lifelong in sensory ganglia, but the hippocampus displays the highest abundance of HSV-1 receptors 198 (Figure 1(D)) and latent virus is often found in hippocampus, amygdala, and olfactory system. 255 Regional apoptosis in hippocampus has been reported following HSV-1 infection. 256  For HSV-1 a subtle mechanism has been proposed. HSV-1 latency in the CNS is accompanied by persistent upregulation of cytokines (see earlier) and, as noted by Baker, 259 chronic production of specific cytokines may have detrimental effects on endocrine function and immunity, perhaps by targeting receptors that are expressed in the brain as well as by immune cells.
There is a further potential avenue for immunomodulation. Following ocular administration of HSV-1 in rats, a selective and intense focus of viral replication was seen in the hypothalamus, 264 the master regulator of body physiology including immunity. Thus, HSV-1 (and potentially other viruses) could directly target the apex of the HPA axis to cause immunosuppression.

Picornaviruses
The major cause of the common cold, rhinovirus, is among the most common viral infections in humans. However, rhinovirus infections of the brain appear to be uncommon. In mouse, a different picornavirus, Theiler's murine encephalomyelitis virus (TMEV), or murine poliovirus, is primarily an enteric endozoite that rarely causes overt disease, and reports of TMEV seropositivity in laboratory mice range from 0.1 to 48%. 265 However, acute TMEV infection of sensitive animals leads to neurological deficits and neuronal destruction in the hippocampus ( Figure 1). 199 As with other agents, infection can also lead to immunosuppression, including inhibition of innate immunity and lymphopenia, 266 suggesting that persistent rhinovirus infection in humans and animals might adversely affect the immune system.

Flaviviruses
These comprise a family of principally insect-borne viruses that are endemic in tropical and subtropical regions. We focus on two flaviviruses: Dengue and its recently emerged relative, Zika. Most individuals infected with Dengue recover, but, in a study on Brazilian Dengue-infected individuals, 6% had symptoms that persisted for more than 6 months, including memory loss, headache, and emotional lability, 267 consistent with hippocampal involvement. 268 Zika virus (that in neonates is associated with human microcephaly), also targets the hippocampal dentate gyrus even in the absence of microcephaly, 269 a finding replicated in adult mice. 270 system until it reaches the brain. Target brain regions in human and animals centrally include the limbic system and hippocampus, 77,78 a site that displays the highest density of RV receptors CHRNA1 (nicotinic acetylcholine receptor 79 ) and GRM2 (metabotropic glutamate receptor subtype 2 80 ; Figure 1

| Other lyssaviruses, herpes viruses, picornaviruses, flaviviruses
Box 4 overviews different types of viruses and their effects on the brain and immunity. One commonality emerging from this analysis is that many agents selectively target the hippocampus (Figure 1), where they can cause behavioral changes. They also induce immunosuppression, although in many cases the underlying mechanisms remain unknown. In addition, there is some evidence that HSV-1 may target the hypothalamus (Box 4), potentially affording a further mechanism for subverting host immunity.
As in rabies, saliva may be a major route of transmission, notably for herpesviruses. Indeed, some viruses (e.g., mumps rubulavirus) target the salivary gland.

| Coronaviruses: bats as an unusual source of human infection
Human coronaviruses such as the agents of severe acute respiratory syndrome (SARS), Middle East respiratory syndrome (MERS), and more recently COVID-19, have been in the headlines as a result of recent human epidemics, but their biology is not fully understood.
These pathogenic viruses principally cause respiratory tract disease; brain infection, although strongly suspected, 84 has not yet been studied in detail. By contrast, murine coronaviruses (rat JHM virus and a derivative of murine hepatitis virus, HMV) have been reported to cause selective cell destruction in the hippocampus. 85 90 This raises the intriguing question of why bats in particular are a rich source of human pathogens.

| Viruses in bats
There are over 1200 species of bats that comprise a quarter of mammalian species. Bats harbor more viruses per species than any other mammal. 91

| Bacteria and Archaea
Far less is known about the proclivity of bacteria for specific brain regions, but cognitive decline is commonplace in survivors of bacterial sepsis (Ref. 95 for review), and bacterial toxins including lipopolysaccharide (LPS) are known to exert negative effects on the hippocampus, 96,97 where they inhibit neurogenesis and neuronal replacement. 98 Hippocampal damage, particularly of the dentate gyrus, is a common feature of bacterial meningitis in human. 99 In addition, LPS administration to mice (a model for sepsis) can lead to immunosuppression. 100 Further studies on the relationship between enzootic bacteria, brain physiology, and immunosuppression are warranted.
To date there have been no systematic studies on Archaea, and their potential role as brain endozoites remains unknown.

| Yeast
Candida albicans is detected in the brain of AD patients, 101 including cortex and hippocampus, and Candida infection has been suggested to promote functional changes in the immune system and enhance immunosuppression. 102 However, this was not confirmed in another study, 103

| Toxoplasma and fatal attraction
Infection with the protozoan T. gondii is prevalent in the human population, 104 and some very high rates have been reported, 105 confirming that T. gondii is a true endozoite. In experimental rats, persistent infection remarkably blocks their innate aversion to the odor of cats ("fatal attraction"), the definitive host for T. gondii, thereby increasing the chance of transmission to that species. 106,107 Berenreiterova et al. 108 reported a significant association of T. gondii cysts with the limbic brain and cortex (Figure 1(C)), and more detailed studies highlight the amygdala, adjacent to the hippocampus, as the region responsible for the altered odor response. 109 There is also evidence for subtle behavioral changes in humans, including changes in the perception of animal pheromones 110 118 Tryptophol is reported to induce sleep in mice 119 and also causes immune system downregulation. 120 The lethargy and somnolence that name the disease may well maximize the likelihood that an infected individual is bitten by a further tsetse fly, thus completing the insect-human-insect life cycle.

| Neurodegeneration, bacteria, and herpesviruses
Host immunodeficiency inexorably leads to microbial proliferation and tissue damage in multiple organs. Thus, the persistence of endogenous microbes in healthy tissues such as the brain appears to reflect a delicate balance between microbial proliferation and elimination by the immune system. Therefore, aging, that is characterized by decline of the immune system, 127,128 is likely to be accompanied by reemergence of erstwhile clinically silent endozoites, as reported for HSV 129,130 and VZV. 131 Indeed, there is growing interest in the possibility that age-related reactivation of subclinical endozoites in brain might be causally linked to disorders such as AD. 132 This idea has a long history. At about the same that Lewis was studying endophytes in trees, Fischer and Alzheimer discovered deposits ("Drusen") in the brain of patients with AD that they suspected ("I emphasized the peculiar similarity of the Drusen with bacterial colonies") to be associated with microbes. 133 Several reports have recently appeared that address the potential relationship between AD and different types of infection including viruses, yeasts, and bacteria, 101,134-140 and these are not reviewed here.
Causal links are notoriously difficult to prove, but emerging population evidence argues that antiherpetic medication may reduce the incidence of AD, 141 and several studies are underway to reproduce or refute this finding. Antiviral treatment is also reported to reduce the incidence of Parkinson's disease (PD). 142 If confirmed, it would raise the prospect of eventual treatments not only in AD and PD 143,144 but also of conditions such as atherosclerosis and diabetes, among others, where an infectious trigger has long been suspected.
We raise a potential caveat regarding the brain microbiome in neurodegeneration (and other disorders) because most studies have been performed on postmortem samples from elderly patients, and it is difficult to distinguish between microbes that might play a role in brain disease such as AD versus those that invade the brain during terminal illness (e.g., the cause of death in AD is typically severe respiratory infection).

| Depression/anxiety and infection
There have long been suggestions that chronic infections may be associated with both depression and the associated condition, anxiety.
HHV infection has been associated with major depressive disorder, 145,146 and we saw earlier that subclinical infection with for example HSV leads to persistently elevated levels of circulating cytokines. These predominantly target limbic brain regions, centrally including the hippocampus, 147 and clinical administration of interleukins and interferons such as IL-1α, IL-2, IFN-α, IFN-β and TNF-α has been widely reported to cause malaise and sickness behavior that resemble anxiety/depression. [148][149][150][151][152][153] Indeed, subclinical infection of several types, perhaps not only in the brain, that lead to systemic inflammation may underlie depressive and anxiety disorders. 154

| Schizophrenia and Toxoplasma
Intense research is presently focused on endozoite involvement in AD and PD, as well as in depression, and we therefore draw attention to a neglected condition, SZ, where infection has long been implicated.
The cause of SZ, an enigmatic condition that is typically diagnosed in late teenage years, is unknown. A possible infective contribution to the etiology and pathogenesis of SZ has been investigated intermittently for over a century. In part this is because of the textbook example of the once common form of psychosis called general paralysis of the insane (GPI). Infection of the CNS by a single agent, Treponema pallidum, is responsible, and the disorder can be effectively treated with penicillin.
In support of a role of an infectious agent, genome-wide association studies of SZ have consistently reported by far the largest signal from the MHC region on chromosome six, 155 and part of the risk for SZ comes from allelic variations of the complement component 4 (C4) located in the MHC region. 156 This is of particular note because the complement receptor CR2 is a receptor for EBV 157 and complement C4 directly targets viruses for inactivation. 158 We overview in Box 5 some of the best-studied potential infectious organisms that have been associated with SZ, including influenza virus, HSV-2, Porphyromonas gingivalis, and Toxoplasma spp. All are associated with CNS invasion.
Of these, the case for an involvement of Toxoplasma spp. in SZ is supported by genetic and pharmacological findings. First, the DISC1 gene, that has long been recognized to be a key determinant of familial SZ, 159,160 is now reported to be a pivotal modulator of immune responses to T. gondii, 161 directly implicating Toxoplasma spp.
Second, SZ is widely treated with neuroleptics, but their mechanism of action is unknown. Intriguingly, these psychotropic drugs may inhibit the growth of T. gondii. Jones-Brando et al. examined the effect of a range of neuroleptic and mood stabilizing drugs on T. gondii cells.
Valproic acid together with haloperidol showed the strongest inhibitory effect on cell proliferation, but risperidone and trimethoprim also showed some effect. 162 In rats haloperidol or valproic acid can reverse behavioral changes induced by T. gondii infection, such as reduced fear of cats and attraction by cat odor. However, those drugs did not prevent acute infection nor decrease the number of tissue cysts in the animal brain, 163 and more recent studies have yielded less clearcut results. However, a recent study 164 confirms that antipsychotics, in particular, have antimicrobial effects. The available data suggest that some neuroleptic drugs may reduce psychosis not only through antidopaminergic action but also by direct inhibition of T. gondii 163,165 or other endozoites.

| HOST ADVANTAGES: AN EVOLUTIONARY ROLE FOR ENDOZOITES
Healthy tissues contain a multiplicity of endozoites, from bacteria to protozoans and viruses. These are not silent. In plants, select endophytes confer protection against pathogens and herbivores. Endozoites in vertebrates can also provide advantages to the host.

| Protection against superinfection
There are many examples. Ever since the time of Jenner it was observed that infection with one pathogen (e.g., poxvirus) could confer protection against a second unrelated pathogen (e.g., herpes), 166 and the 1927 Nobel Prize in Physiology or Medicine was awarded to Julius Wagner-Jauregg for the discovery that malaria infection is protective against GPI (i.e., neurosyphilis): inoculation of infectious malaria into patients remitted GPI in 83% of cases. 167

BOX 5 Endozoites and schizophrenia (SZ)
Influenza Influenza is one of the best-studied potential prenatal contributors to SZ. Mednick et al. reported an increased risk for SZ in people exposed prenatally to the 1957 influenza epidemic. 271 This was followed in rapid succession by papers from Scotland and Denmark essentially confirming the Mednick findings. National registry records were used and allowed examination of prenatal exposure to both the 1918-1919 and 1957 influenza epidemics. Unfortunately since then around 20 additional ecological studies have addressed the issue, with around half supporting the hypothesis and the other half failing to confirm. However, there are confounders that make interpretation of these studies difficult. For example, almost all studies of prenatal exposure to influenza are based solely on whether an individual was in utero at the time of an influenza epidemic. In these circumstances around 70% of individuals who were in utero during the 1957 type A2 influenza epidemic would have been misclassified as having been exposed. This increases the risk of false negative (type 2 error) associations. 272,273 In a nested case-control study, 274 Brown et al. demonstrated a threefold elevation in risk of SZ following influenza prenatal exposure during the first half of gestation. For first trimester exposure, the risk of SZ was increased sevenfold but there was no elevated risk following exposure during the second half of gestation. These results have been difficult to interpret especially because Seltzen et al. 275 pointed out that serological studies may have limited validity.

HSV-2
Neonatal exposure to HSV-2 is associated with congenital anomalies and neuropsychiatric disorders. 274 Three studies have examined the relationship between prenatal exposure to HSV-2 and risk of SZ in offspring. Two were derived from selected sites of the Collaborative Perinatal Project (CPP), a multisite study of population-based birth cohorts born from 1959 to 1967. In the first study 276 raised maternal IgG antibody levels to HSV-2 were associated with a significantly elevated risk of SZ and other psychoses in offspring with odds ratios of 3.4 to 4.4. In a much larger follow-up study, 277 which included 200 case subjects with psychotic disorders from three cohorts of the CPP, a 1.8-fold increased risk of SZ psychoses was observed among offspring of mothers who were seropositive for HSV-2, but only among seropositive mothers who has regular unprotected sexual intercourse. A third study based on the Child Health and Development Study cohort failed to replicate these positive associations. 278 Potential explanations for these discrepant findings are discussed by Brown,279 as are the limited and equivocal findings investigating measles, rubella, varicella zoster, rabies, and poliomyelitis.
There have been few and mostly negative subsequent studies of HSV-2 and SZ, 280 and the data so far argue that HSV-2 is not a major contributor to SZ.

Porphyromonas gingivalis
Multiple studies have demonstrated increased rates of periodontal disease in patients with SZ. 281-287 Indeed Fawzi et al. 287 demonstrated increased levels of P. gingivalis (the key pathogen in periodontal disease) in saliva from SZ patients compared to controls, and the severity of psychopathology was related to P. gingivalis levels. Although the most obvious explanation for these findings is that the periodontal changes are secondary to lifestyle, poor oral hygiene, and medications associated with SZ, a bidirectional link between the two conditions cannot be ruled out, especially given observations of the presence of P. gingivalis in postmortem AD and its presence in brain parenchyma following chronic oral administration in mice (see earlier).

Toxoplasma
This is the topic of excellent reviews, 288,289 much of which is paraphrased in the following outline. Toxoplasmosis is an infectious disease caused by the parasitic protozoan, T. gondii, that affects approximately one third of entire human population. T. gondii can be found in almost all warm-blooded animals, but cats are the only known natural hosts. T. gondii is highly neurotropic and, soon after the infestation, migrates within the brain tissue to localize in astrocytes, microglia, and neurons. 290,291 The dormant form or bradyzoite can persist in the host brain for many years. 292,293 The brain tissue cysts undergo continuous remodeling, but until recently were not thought to cause clinical symptoms in immunocompetent individuals. 294 However, given the high level of neurotropism and the fact that T. gondii is endemic in almost all cultures worldwide, it has long been postulated that there may be a link with SZ. 295,296 Increased rates of T. gondii infection are reported in SZ. Three meta-analyses of association between T. gondii exposure and SZ have been published. 295

| Ecosystem advantages
At the ecosystem level, plant ecologists argue that many endophytes within plant tissues remain latent until natural senescence, when they proliferate to promote recycling of biomaterial, to the benefit of seedlings and saplings, and thus to the ecosystem (see earlier). This is a perhaps a strange idea in the context of vertebrates, but this cannot be formally excluded over an evolutionary timescale, particularly for fungi and bacteria.

| Coevolution and horizontal gene transfer
There may be more direct routes. Herpes viruses and vertebrates have coevolved for at least 200 Ma. 175 In human, where the majority of the population harbors persistent infections with herpes viruses (Table 1), the divergence of HSV-1 and HSV-2 (and recombinants) accompanied human evolution from primates, and perhaps also migration out of Africa, 176,177 suggestive of functional effects.
In plants, integration of (non-retrovirus) viral sequences is commonplace, 178,179 raising speculation that "integrated viral sequences might reflect some functional advantage to the possession of the sequence." 178  However, we underline a major complicating factor-hygiene.
infection. 278,297 Similar results have been found assaying IgG and IgM anti-T. gondii antibody levels in neonatal blood spots from the Danish State Serum Institute, where increased IgG levels were present in neonates who later developed SZ (OR 1.79). 299 Because babies only start producing IgG antibodies around 3 months after birth, IgG antibodies assayed in the neonatal blood spots must be maternal in origin and suggest that earlier maternal exposure to T. gondii increases the risk of SZ in offspring. 299,300 Improved sanitation in developed countries has substantially cut infant mortality, but may have inadvertently increased other disorders.
Indeed, the "hygiene hypothesis" has been invoked to explain differential rates of autoimmune disease across the world, 192,193 including AD, 194 although recent data now argue against the hygiene hypothesis of AD. 310 One remarks that the situation of indigenous Amazonian villagers, who are chronically exposed over their entire lifetime to a myriad of endozoites, is a far cry from Western populations who are insulated from the vast majority of environmental microbes, and are only exposed to a restricted range of endozoites in later years. It is very possible that, despite benefits in terms of infant mortality, the lack of exposure to an "evolutionary" spectrum of microbes may predispose the human population to "modern" diseases including AD, PD and SZ.

| ENDOZOITES AND THE MISSING HERITABILITY
As a final note, we wonder if endozoites might explain a longstanding conundrum. Many CNS disorders, exemplified by SZ and autism, show high concordance between identical twins (a measure of "heritability"), as well as raised concordance between siblings. However, other than for specific single-gene defects (e.g., CFTR mutations in cystic fibrosis), multiple genomic analyses have failed to uncover gene variants (or groups of variants) that could explain this concordance, a phenomenon dubbed "missing heritability." 195 Indeed, for most disorders, genes explain no more than a small fraction of the heritability.
The widespread distribution of endozoites in the human population leads us to wonder whether endozoites, clusters of endozoites, and/or specific variants thereof could explain why twins and siblings display phenotypes that are closer to each other than to the general population. We undoubtedly inherit more from our parents (and from our prenatal and postnatal environments) that merely genes. Others have suggested that the gut microbiota might play a role, 196 but the broader spectrum of endozoites, specifically those reaching the brain, might have greater impact on diseases such as SZ that principally affect the nervous system. Comparative studies on twin/sibling microbiomes will be necessary to address this possibility.

| CONCLUSIONS
Given the precedent of plants, it comes as no surprise to discover that endozoites are widely present not only on superficial surfaces and in the gastrointestinal tract but also within healthy human tissues such as the brain. Indeed, for many host-adapted microbes it seems that there is no fundamental barrier to entering host tissues. The benefits of the close association are well documented in plants, but there is so far only limited evidence (except for the gut microbiome) that vertebrate endozoites benefit the host, and this is an area that demands further research. Instead, there is extensive evidence that endozoites manipulate host immunology and behavior to promote their own persistence and transmission.
Building on clear parallels between plants and animals, the key conclusions of this analysis are as follows.
1. Endozoites (like endophytes) are widely present not only in the circulation but in multiple body tissues including the brain.
2. Endozoites have accompanied the evolution of the lineage leading to Homo at least since the divergence of insects and vertebrates (0.5 Ga).
3. These passengers are not silent, and can influence both immunity and behavior.
4. As in plants, some of these passengers can be beneficial, and others harmful-a delicate balance. Endozoites are directly implicated in CNS disorders including AD, PD, and SZ, but in other cases endozoites may give their host a cognitive advantage.
In our view, endozoite modulation of behavior is the most intriguing of all the issues we have raised here. Is this mostly an incidental correlate of immunomodulation, given that the limbic brain governs both behavior/cognition and the immune system, or do endozoites deliberately manipulate our behavior?
We also wonder if there is an optimal (beneficial) brain microbiome that-paralleling GI tract microbiome transplantation in diabetes and colitis-we could perhaps resurrect (e.g., by simple measures such as intranasal inoculation) to prevent the adverse effects of key endozoites?
In sum, both plants and animals harbor a multiplicity of endogenous microbes that inhabit multiple tissues including solid tissues such as the brain. These are not silent, and harboring particular passengers may have both advantages and disadvantages. Understanding the mechanisms, roles, and ecology of endogenous microbes in different mammalian tissues including the brain will undoubtedly be a fertile field of investigation for the future.

ACKNOWLEDGMENTS
We would like to thank Hans Kuepper (Munich), Juergen Haas (Edinburgh), and Rob Moir † (Harvard) for constructive comments on the MS. This work was funded in part by the Benter Foundation.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.