The present study reports the first deep sequencing analysis of microbial populations within the normal appearing tissue in the human brain with confirmatory methodologies that highlighted the consistent presence of bacterial ribosomal RNA and associated bacterial products. The majority of the bacterial RNA sequences identified in all human and nonhuman primate brains were encoded by members of the α-proteobacteria class, regardless of the underlying disease process. This restricted bacterial diversity observed in normal brain tissues is in contrast to the findings of a wide variety of organisms from other groups studying brain abscesses 
; many of which display colonization by bacteria from other tissues. This dichotomy in findings suggests that the bacteria identified in normal appearing tissue in the present studies might be are outcompeted and/or supplanted by organisms from other body sites in the case of intracerebral abscesses. Peptidoglycan immunoreactivity and in situ
hybridization detection of bacterial rRNA were apparent within glial cells and in the extracellular space. In vivo
cerebral implantation of human brain homogenates into mice showed that the 16 s RNA sequences recovered from recipient animals’ brains were conserved and expressed at levels similar to the initial human brain homogenate. Transmission was interrupted by heat treatment of the initial homogenate, implying that viable bacteria were required for transmission. These findings indicated that bacteria were present in the primate brain and do not appear to be derived from the predominant populations at other human body sites. Indeed, the predominance of α-proteobacteria in the brain is unparalleled in other body sites where the microbiomes are dominated by Firmicutes, Bacteriodetes and Actinobacteria.
The bacterial rRNA V3–V4 region-based detection of α-proteobacteria in the present studies was supported by the unbiased (non-amplified) deep sequencing. The relatively greater abundance of β-proteobacteria identified using the V8 region amplicon might reflect a bias because of the poor discriminating power of the smaller V8 amplicon or the restricted diversity within this region relative to the analyzed V3–V4 region amplicon, leading to imprecise assignment of sequences to this closely related class. There might also be a bias introduced by the PCR conditions, despite the use of universal primers, resulting in an over-representation of β-proteobacteria sequences. This finding needs to be clarified through culturing of brain specimens as part of future analyses to identify definitively the components of the primate brain’s microbiome at the species level. Of interest, α-proteobacteria comprise one of the most diverse bacterial classes with wide spread biological niches and actions including detoxifying effects in the environment 
. The acquisition of this class of infectious agents by the brain might represent a beneficial organ-specific adaptation.
The potential for contamination of samples at any stage of the tissue preparation, together with blood contamination was considered throughout the present studies. Controls for all lots of reagents were used throughout this work at all steps of the RNA extraction, cDNA synthesis and conventional and quantitative RT-PCR. The current tissues varied in harvest times and sources; for example, autopsy times ranged from 12–24 hr, while surgical samples were collected under sterile conditions and immediately frozen on dry ice in the operating room. To confirm that 16 s rRNA amplicons identified were intrinsic to the tissue and not introduced during sample preparation RNA extraction, cDNA synthesis and PCR, extensive reagent and equipment controls were used throughout these studies up to and including having the entire process repeated by different personnel in another facility using different lots of consumables and reagents (Figure S3B
). White matter was intentionally selected to limit blood contamination of samples; additionally, bacterial genome and products were detected in brain parenchyma remote from blood vessels and within cells (glia) known to phagocytose foreign materials. Likewise, the current simian brain samples were harvested and processed by different investigators in a distant facility, yet the same phyla were predominant in the majority of macaque brains examined. Brain samples from experimental animals reared in SPF conditions with concurrent immunosuppression (RAG1−/−
mice and FIV-infected cats) were included as controls within these studies but did not show bacterial 16 s rRNA sequences in brain despite being processed in an identical manner to the human and macaque brains, even with prior amplification steps and identical preparatory methods. Despite the diversity of techniques and sources of the present brain tissues, α-proteobacteria represented the most prevalent bacterial class discovered within the human brain, which was in contrast to blood samples. The consistency of α-proteobacteria detection in primate brains but not in SPF experimental animal brains emphasized the reliability and specificity of the present observations.
Several of the bacterial classes observed in the present studies have been associated previously with human diseases. For example, an organism similar to many of the 16 s V8 region sequences, Delftia acidovorans
has been implicated in endocarditis 
, bacteremia 
, corneal keratitis 
and urinary tract infections 
. In addition to causing infections in compromised patients, D. acidovorans
and other members of the Comamonadaceae have been identified as part of the bacterial community in the arterial wall in patients who have had aortic aneurysms 
has also been isolated from cerebrospinal fluid, sputum, urine, pharynx and wounds without concurrent signs of disease 
. Taken together, this ubiquitous environmental organism likely represents a commensal organism with wide tissue distribution that can act as an opportunistic pathogen in vulnerable patients. As such, the prevalence of sequences showing similarity to Delftia sp.
in the majority of patients in all the present clinical groups bears close scrutiny as host specific factors might determine whether or not this organism contributes to brain disease. Two HIV/AIDS patients exhibited sequences similar to Alcaligenaceae; members of this family have been identified as components of the normal flora in Peyer’s patches 
, but members have also been implicated in endocarditis 
, bacteremia 
and meningitis with or without ventriculitis in neonates 
, HIV-infected persons 
, other immunocompromised adults 
or following invasive surgery 
. Thus, identifying members of this family may represent translocation of commensal gut organisms to the brain or a previously unidentified subclinical infection in these patients.
As the surgical cerebral resections were collected from patients as part of a procedure for epilepsy, they represent the samples with the lowest probability of ex vivo changes, such as RNA degradation or a possible artifact of bacterial growth. Differences in the bacterial populations between the surgically- and autopsy-derived specimens could represent post mortem–related differential levels of RNA degradation or changes in microbial replication. The >5 fold increase in the representation of Archea in the cerebral surgical specimens relative to the ODC group was interesting, as members of this kingdom were undetectable in all but one autopsy brain specimen.
In an organ widely assumed to be free of infectious agents in the absence of a specific disease process, autopsied and surgically-derived human brain specimens showed a restricted but distinct bacterial population in the present studies, which was composed of bacterial classes chiefly recognized in the physical environment, i.e., soil and water. The sources of these agents might include oral consumption or inhalation with eventual transport to the brain as intracellular agents in activated leukocytes trafficking into the brain. The brain is constantly surveyed by trafficking leukocytes (activated lymphocytes and macrophages), which provide a Trojan horse mechanism for microbial entry into the nervous system across the blood brain barrier. In fact, this mechanism is well recognized as a route by which viruses infect the brain and likely underlies the detection of herpes viruses in both the HIV and ODC brains. Corroborating this latter point is the report of peptidoglycan detection in brain lesions from multiple sclerosis patients, which are heavily infiltrated with blood-derived leukocytes 
. Since bacteria express multiple molecules that activate immune signaling cascades by engaging Toll- or NOD-like receptors, etc., their capacity for influencing brain function 
is immense. Hence, studies focused on delineating the brain’s microbiome at the species level together with their individual effects on host cell physiology might lead to a greater understanding of human neurobiology including cognitive, motor, sensory and behavioral functions.