Table 1.
Clinical features, sex and neuropathological diagnoses.
Table 2.
Total sequence tag counts per patient for principal bacterial classes detected in brain1.
Figure 1.
Deep sequencing detection of bacterial and bacteriophage RNA sequences in human brain.
(A) Total sequence tags that were unambiguously identified as belonging to a bacterial phylum were grouped for each patient from which the percentages for each phylum were displayed. All patients showed a predominance of Proteobacteria-associated sequences. (B) Despite inter-individual variability the mean percentage of Proteobacteria sequences among the HIV, ODC and SURG groups was similar. (C) The majority of bacterial sequences identified in all patient samples belonged to the Proteobacteria phylum, which showed the greatest similarity to the α-proteobacteria class. (D) The majority of bacteriophage sequences identified matched Proteobacteria-tropic phage sequences although bacteriophage sequences were not detected in the SURG samples.
Table 3.
Detection of viral genomes in brain-derived cDNA or gDNA.1
Figure 2.
Human brain host gene responses.
Analysis of human transcript sequences implicated in (A) bacterial and (B) viral host responses identified by deep sequencing of brain-derived RNA showed variability within clinical groups with many genes showing a relative reduction or induction in expression in the HIV relative to ODC brain specimens. (C) Host neuroimmune responses were assessed as transcript abundance in terms of relative fold change (RFC) by real time RT-PCR, showing a general increase in pro-inflammatory gene expression among the HIV (n = 5) compared to ODC (n = 5) brain specimens with significant changes in cd3ε, IL-12 and egf expression.
Figure 3.
Bacterial detection in human brain.
Autopsy-derived ODC (A), and HIV (C) brain specimens were immunolabeled with anti-peptidoglycan antibody. Peptidoglycan (PGN)-positive bodies (arrows) were morphologically consistent with bacteria and smaller than CD45 immunopositive microglia from ODC (B) and HIV (D) patients, imaged at the same magnification. Double DIG-labeled EUB 338 probe in situ hybridization (ISH) against the 16 s rRNA gene was hybridized with slides from the same ODC (E) and HIV (G) patients and labeled with alkaline phosphatase-conjugated sheep anti-DIG FAB` fragments and stained with NBT/BCIP. ISH-positive bodies featured morphology resembling bacteria (arrow heads) and were smaller than GFAP immunopositive astrocytes in ODC (F) and HIV (H) patient sections. Peptidoglycan-labeled cells with both spherical and rod morphology were observed within the brain parenchyma and in a blood vessel (I) (White arrow) of an ODC patient. Peptidoglycan immunopositive bodies were observed within the cytoplasm of GFAP-immunolabeled astrocytes (I, inset) and Iba-1 immunolabeled microglia (J, inset). Spherical clusters of EUB 338 hybridized cells J) were evident (black arrowheads). Slides from the same ODC11 (K) and HIV11 (L) patients were processed under identical conditions except that the primary antibody was omitted. A scrambled DIG-labeled probe was hybridized to slides from the ODC11 (M) and HIV11 (N) under identical conditions used for the EUB 338 probe. In all cases specific signals were not detected. (Original magnification 200×). (O) A section from the forebrain of one of the FIV-infected cats was immunostained with the anti-PGN antibody and developed with DAB with no detectable signal. (Bar: A–H, 50 microns; I, 25 microns; J, 20 microns; K–N, 50 microns) (Magnification: A–H, 200×; I, 400×; J, 600×; I and J insets, 600×).
Figure 4.
Phylogenetic analyses of bacterial 16 s rRNA sequences from human brain and blood.
(A) Representative ethidium bromide gels show the amplification of a single band by a nested PCR protocol but in matched (nested) water controls, a product was not observed. (B) Real time RT-PCR showed similar levels of mean 16 s rRNA amplicon quantities in HIV and ODC brain samples, however SURG samples showed lower amplicon levels of 16 s rRNA products. (C) Phylogenetic analyses were determined among cloned amplicons generated from brain cDNA from ODC (n = 6), HIV (n = 6), SURG (n = 3) as well as cDNA from PBMCs (n = 2) and serum (n = 2) from healthy volunteers (PBMC 1,2; Serum 1) using the universal bacterial 16 s primers 514F and 806R and the equivalent region of various published 16 s rRNA genes. The alignments were used to generate a neighbour joining tree using 10,000 bootstrap trials. There was a predominance of α-proteobacteria in brain-derived 16 s rRNA clones although other classes, e.g., β- and γ-proteobacteria, and Bacilli, were amplified from blood-derived cDNA.
Figure 5.
16 s rRNA sequences from brain-derived cDNA of cynomolgus macaques.
(A) Ethidium bromide-stained agarose gel showed the amplicon generated by primers 16 s514F and 16 s 806R from SIV-infected macaques 1–10, SHIV-infected macaques 1–3 and FIV-challenged macaque 1 and 2. cDNA synthesized with ultrapure water, water carried through both rounds of nested PCR and water used as template only in the final round of PCR were all included as negative controls. (B) Phylogenetic analysis of 16 s rRNA region sequences amplified by the primers 514F and 806R derived from macaque brain specimens. Clustal alignments were generated comparing amplicon sequences with the equivalent position of published 16 s rRNA sequences identified by BLAST analysis. The Neighbor joining tree was generated based on 10,000 bootstrap trials and rooted on the macaque mitochondrial 12 s rRNA sequence. Again, these data showed a predominance of Proteobacteria in brain-derived 16 s rRNA clones although other bacteria, e.g., Actinobacter and Bacilli, were detected.
Figure 6.
Transmission of bacteria from human brain to RAG1−/− mice brains.
(A) Comparison of the relative quantity of 16 s rRNA in human and matched recipient mouse brains showed that mean bacterial rRNA levels in recipient mice were present at ∼40% of the mean ODC14 brain homogenate rRNA levels but less than 1% for the heated-treated ODC13 brain homogenate (prior to heat-treatment), measured by real time RT-PCR. (B) Mouse host inflammatory gene transcripts (ifn-α, il-1β and il-12) in mouse brains were not induced by implantation of untreated compared to heat-treated human brain homogenates. (C) The bacterial 16 s rRNA sequences from the ODC14 brain homogenates and the recipient mouse ODC14 brains were similar with substantial phylogenetic overlap between the bacterial genera, as evidenced by the 97% similarity for Delftia acidovorans sequences.