Non-human primates were used in strict accordance with UK Home Office guidelines, under a license granted by the Secretary of State for the Home Office. NIBSC is governed by the Animals (Scientific Procedures) Act 1986, which complies with the EC Directive 86/609 and performs under license (PPL 80/1952) granted only after review of all license procedures by the NIBSC Ethical Review Process. All macaques were purpose bred and group housed for the study duration, with daily feeding and access to water ad libitum. Regular modifications to the housing area were made by husbandry staff to further enrich the study environment. Animals were acclimatized to their environment and deemed healthy by the named veterinary surgeon prior to study inclusion.

All animals were sedated prior to bleeding or virus inoculation by venipuncture. Frequent checks were made and unexpected changes in behavior followed up, including seeking of veterinary advice where necessary. Regular blood samples were obtained to assess hematological parameters that might provide evidence of incipient disease and veterinary advice sought when persisting abnormalities detected. The study was terminated and all animals killed humanely by administering an overdose of ketamine anesthetic prior to development of overt symptomatic disease. All efforts were made to minimize animal suffering, including the absence of procedures not essential to the study.

Twelve purpose bred juvenile, simian retrovirus negative, and simian T lymphotropic virus negative cynomolgus macaques were used in these studies. Macaques K4, K6, K7, K9, K10, and K11 were vaccinated against full length SIVmac239 gag before being subjected to viral challenges. Specifically, they received three 100 µg DNA intradermal inoculations 23, 19, and 15 weeks before viral challenge, and a boost with replication-defective adenoviruses 5 vector 6 weeks before challenge. Animals K20, K21, K23, K24, K25, and K26 were the unvaccinated control group. All animals were challenged intrarectally with 150TCID₅₀ SIVmac251 starting at week 1 and for up to 10 weeks. Blood samples were taken weekly and tested for plasma SIV RNA by PCR, as described below. Following a positive PCR result, animals were no longer challenged. Animals were euthanized 19–20 weeks after study.

The implications that the vaccination protocol may have had for the data on neuroinflammatory responses reported in this study are considered in both the Sections “Results” and “Discussion.”

Simian immunodeficiency virus viral RNA loads were determined in plasma samples collected at various times throughout the course of the study by quantitative real-time reverse transcriptase PCR, as previously described (37).

Whole brains were harvested no longer than 1 h post-termination. Tissue explants of approximately 50–100 mg were collected from the frontal lobe, midbrain, and cerebellum, immersed in RNAlater (Qiagen, Manchester, UK) and stored at −80°C until RNA extraction. Whole brains were then fixed in 10% (v/v) formal saline for 4 weeks at 4°C, dissected into defined brain regions and processed for paraffin embedding (33). Following dewaxing, 4 µm formalin-fixed paraffin embedded brain sections were stained with hematoxylin in accordance with standard histochemical techniques.

Because the animals were originally included in a study aimed at assessing the efficacy of a vaccination protocol, fresh tissue, and blood samples had to be collected at termination. The multiple breaks introduced during removal of tissues meant that the vascular system could not support the flushing of saline and then formalin. Thus, none of the animals had undergone perfusion prior to brain collection. The implications that residual leukocytes or viral particles within the blood vessels may have had on our findings are considered in the Sections “Results” and “Discussion.”

Total RNA was extracted from RNAlater-preserved explants using the RNeasy Midi Kit (Qiagen), according to the manufacturer’s instructions. Samples were thawed at room temperature, removed from RNAlater, and placed in the lysis buffer provided with the RNeasy Midi Kit. Tissue homogenization was performed using a Precellys Minilys bead beater homogenizer (Bertin Technologies, France) in tubes of 2 ml capacity with 1.4 mm ceramic (zirconium oxide) beads (Precellys soft tissue homogenizing kit, CK14). Total RNA was eluted in molecular grade RNase/DNase-free water (Qiagen), yield and quality were tested by absorbance measurement at 260, 280, and 230 nm using a NanoDrop™ Lite Spectrophotometer (Thermo Scientific, UK), and samples were stored at −20°C until use. Reverse transcription was carried out using the RT² First Strand Kit (Qiagen), according to manufacturer’s instructions. cDNA samples were stored at −20°C until use in the real-time PCR arrays. This same procedure was used for both the multigene arrays and for the additional single gene real-time PCR described below.

Two targeted real-time PCR gene arrays were performed on cDNA from SIV-macaques brain tissue samples: (1) an array measuring the expression of 84 genes coding for inflammatory cytokines, chemokines, and their receptors (RT² profiler gene array PAQQ-011Z, Qiagen) was used on cDNA samples from frontal lobe, midbrain, and cerebellum; and (2) an array measuring the expression of 84 genes that are either upregulated or downregulated during processes associated with neurotoxicity (RT² profiler gene array PAQQ-069Z, Qiagen) was used on cDNA samples from frontal lobe and midbrain. Full lists of the genes included in each array are given in Tables S1 and S2 in Supplementary Material. Arrays were performed according to the manufacturer’s instructions on a LightCycler-480 (Roche Diagnostics Ltd., Burgess Hill, UK) with 96-well plate block. Briefly, the reaction plates are already seeded with appropriate primers set, including five housekeeping genes, one genomic DNA control, triplicate RT, and PCR positive controls. A reaction mix containing sample cDNA and all reagents required for a SYBR green-based detection system, with the exception of the primers, was prepared using the RT² qPCR Master Mixes provided with the kit and dispensed in each well. Reactions were performed according to the following thermal profile: initial denaturation at 95°C for 10 min; 45 cycles of 95°C for 10 s and 60°C for 1 min, fluorescence data collection performed at the end of the 60°C step.

Initial data analysis was performed using the LightCycler-480 operating software for automatic estimation of background noise and fluorescence threshold. Data were normalized on the average of at least three housekeeping genes, which showed a coefficient of variability (CV = SD/mean) lower than 10% across all tissues and among all animals within the same array. Specifically, data from the PAQQ-011Z array were normalized on the average of three housekeeping genes (ACTB, CV = 9.76%; B2M, CV = 8.06%; GAPDH, CV = 9.03%; combined CV = 8.02%); data from the PAQQ-069Z were normalized on the average of all five housekeeping genes (ACTB, CV = 4.77%; B2M, CV = 6.35%; GAPDH, CV = 3.23%; HPRT1, CV = 4.51%; RPL13A, CV = 5.39%; combined CV = 4.41%). RT and PCR positive controls were used to calculate inter-plate variability (PAQQ-011Z: RT CV = 0.79%, PCR CV = 1.06%; PAQQ-069Z: RT CV = 0.29%, PCR CV = 0.29%).

The expression of genes encoding for interleukin (IL)-10, cytotoxic T lymphocyte antigen 4 (CTLA-4), the forkhead family transcriptional regulator FOXP3, transforming growth factor (TGF)-β, and the SIV capsid protein gag was tested by SYBR green-based real-time PCR on cDNA from brain explants collected as described above. All reactions were performed in a LightCycler-480 (Roche) using SYBR green PCR mastermix (Qiagen); a list of primers used for each target and the housekeeping gene GAPDH is provided in Table S3 in Supplementary Material. All reactions were performed according to the following thermal profile: initial denaturation at 95°C for 10 min; 40 cycles of 95°C for 20 s, 60°C for 20 s, and 72°C for 20 s (fluorescence data collection performed at the end of the 72°C step).

Cerebrospinal fluid (CSF) was collected at necropsy from all animals with the exception of K10. CSF concentrations of TRP and KYN were measured by high performance liquid chromatography as previously described (38).

Statistical analysis was performed using R version x64 3.2.2 (The R Foundation for Statistical Computing) and SPSS version 22 (IBM, Armonk, NY, USA). Microsoft Excel 2013 (Microsoft Corporation, Redmond, WA, USA) and GraphPad Prism version 5 (GraphPad Software, La Jolla, CA, USA) were used to generate some of the plots in the figures. Comparison between uninfected and SIV-infected animal groups was performed using non-parametric Mann–Whitney test. Due to the small sample size, the results obtained in this study did not survive stringent false discovery rate controlling procedures. Therefore, only genes showing at least 1.5-fold difference between the two groups, calculated as the ratio between median values, and uncorrected P < 0.05 were considered to be upregulated or downregulated. All correlations were assessed using non-parametric Spearman test.

Twelve cynomolgus macaques were included in the study. All animals were challenged intrarectally with 150 TCID₅₀ SIVmac251 at weekly intervals starting at week 1 for up to 10 weeks or until they tested positive for plasma SIV-RNA and were euthanized at week 19 or 20 (Figure 1A). All animals were originally included in a prophylactic vaccination study, schedule of vaccination and viral challenge is shown in Figure 1A. Eight of the twelve macaques showed positive plasma SIV RNA measurement by week 5 and one (K24) tested positive at week 11, after the last challenge (Table 1). Two of the nine infected animals (K4 and K7) had undetectable viremia at the time of necropsy and will be referred to as aviraemic (Table 1). Animals showing detectable plasma SIV RNA at necropsy (median 8.2 × 10³ SIV RNA copies/ml, range 1.5 × 0²–5.1 × 10⁵ SIV RNA copies/ml) will be referred to as viremic (Table 1). Peak of viremia generally occurred on the second week in which plasma tested positive for SIV RNA, with the exception of animals K10 (peak on the first positive sample) and K26 (peak on the third positive sample). A direct correlation was observed between peak of viremia and viremia measured at termination (Figure 1B). Three animals never showed detectable viremia throughout the study (K6, K9, and K11) and will be referred as uninfected (Table 1).

All uninfected (K6, K9, and K11) and SIV+ aviraemic animals (K4 and K7) belonged to the vaccinated group (Figure 1; Table 1), whereas only animal K10 among the viremic group had previously received prophylactic vaccination. These observations suggest that the vaccination protocol employed in the study may have exerted a protective effect on infection or favored control of viral load in the periphery.

SIVgag RNA was detected only in the midbrain of animal K24 and frontal lobe of animals K20 and K23 and was undetectable in all other tissue explants. Because the animals were not perfused before collection of the brains, we cannot exclude the possibility that part of the detected viral RNA derived from viral particles trapped in blood vessels, rather than from virus produced in situ. Nonetheless, the vast majority of the explants tested negative for viral RNA. It is noteworthy that macaque K23, which tested positive for SIVgag RNA in the frontal lobe, displayed the lowest detectable plasma viral load at termination among viremic animals (Figure 1C).

The expression of 84 genes associated with inflammatory responses was measured by multigene real-time PCR array in homogenized brain explants collected postmortem from midbrain, frontal lobe, and cerebellum from N = 12, N = 11, and N = 11 macaques, respectively (frontal lobe and cerebellum explants were not collected from animal K4), and levels of gene expression were compared between uninfected (SIV−) and SIV+ animals.

We found three genes upregulated by at least 1.5-fold with P < 0.05 in the midbrain of SIV-infected macaques (Figure 2A). Surprisingly, the same analysis in the frontal lobe showed 12 genes to be downregulated by at least 1.5-fold (P < 0.05) in SIV-infected compared to -uninfected animals (Figure 2B). No differences in any of the genes analyzed were observed in the cerebellum (Figure 2C). The expression profiles in both midbrain and frontal lobe of SIV-infected aviraemic animals had greater similarity to that of viremic rather than uninfected animals (Figure 3), suggesting that inflammatory responses in the CNS ensue early during infection, and are not affected by the level of ongoing viral replication. An alternative, non-mutually exclusive explanation is that plasma viremia is not a reliable indicator of viral replication in other anatomical compartments, including the CNS.

We considered whether the protective effect of the prophylactic vaccine had any influence on the expression of inflammatory genes in the brain. Focusing on the 15 genes that were differentially regulated between SIV− and SIV+ animals, we compared the expression levels in aviraemic animals that had received the vaccine (K4 and K7) to that observed in SIV+ viremic animals (SIV+_<50 vs. SIV+_>50 in all plots in Figure 3). Although stringent statistical analyses could not be performed due to the limited number of aviraemic animals, the levels of gene expression observed in animals K4 and K7 closely resemble those observed in viremic SIV+ animals. In addition, when we repeated the analyses shown in Figure 3 on a dataset that excluded animals K4 and K7, 10/15 of the parameters tested remained significantly different (P < 0.05) between SIV− and SIV+ animals, and the P values for the other 5/10 genes remained <0.1. These observations suggest that even if prophylactic vaccination may have allowed control of viral replication in the periphery, it did not prevent the immune alterations detected in the CNS.

We tested whether the downregulation of inflammatory genes in the frontal lobe was associated with differences in the expression of immunoregulatory genes encoding for interleukin (IL)-10, CTLA-4, the forkhead family transcriptional regulator FOXP3 or TGF-β. Expression of all immunoregulatory genes was below detection level in at least 80% of the samples tested, with the exception of TGF-β, which, however, did not differ between SIV-infected and -uninfected animals (data not shown).

Increased tryptophan (TRP) catabolism by indoleamine 2,3-dioxygenase 1 (IDO1) has been suggested to contribute to inflammation-mediated neuronal damage by leading to the production of neurotoxic metabolites such as quinolinic acid (QA) (39). We measured the levels of TRP and its key metabolite kynurenine (KYN) in CSF collected at necropsy from the macaques in the study. The KYN/TRP ratio, a well-accepted measure for IDO1 activity, correlated directly with plasma viremia at termination in the SIV-infected group (Figure 4A). This correlation was mainly driven by alterations of KYN levels, rather than TRP (Figure 4B), suggesting an increase in catabolism at either peripheral or central level, rather than systemic depletion of TRP.

We used multigene real-time PCR array in the same samples prepared from midbrain and frontal lobe to analyze the expression of 84 genes, which are either upregulated or downregulated during biological events associated with neurotoxicity (Figures 5 and 6). We did not observe a clear and consistent pattern of gene expression associated with neurotoxicity in either midbrain or frontal lobe. However, subsets of genes associated with neuron development (EREG, PLP1, YWHAE) and nitric oxide response (DYNLL1, GUCY1A3), as well as the ion transporters TRPM1 and TRPM4 were downregulated in the midbrain of SIV-infected compared to -uninfected macaques (Figure 6). The frontal lobe of SIV-infected animals showed upregulation of the apoptotic mediator CASP7, and downregulation of EIF2AK3 and NOTCH4, which have all been associated with neurotoxicity (Figure 6).

Histologic sections from frontal lobe and midbrain were analyzed for perivascular leukocyte accumulation by standard hematoxylin staining. Evidence of perivascular cuffing was found in midbrain, but not frontal lobe sections of SIV-infected animals, independent of viremia at the time of necropsy (Figure 7). No perivascular lymphocyte accumulation was observed in either midbrain or frontal lobe sections from uninfected macaques (Figure 7). Sections from one representative non-vaccinated, uninfected, non-SIV-challenged macaque (#910) with no signs of perivascular cuffing are shown for comparison.