Introduction

The central nervous system (CNS) is one of the major anatomic HIV reservoirs in many HIV-infected patients, even those on highly suppressive combination antiretroviral therapy (cART), making it a major obstacle to HIV eradication and a cure. HIV reaches the CNS as early as 8 days post-infection (Valcour et al. 2012). Although cART can effectively reduce HIV loads in cerebrospinal fluid (CSF), decreases are slower and delayed compared to viral loads in plasma. Further, despite low to absent levels in blood of patients on cART, residual viral RNA can often be detected in the brain, and serve as a persistent reservoir for viral replication, resulting in neurocognitive dysfunction, despite the absence of direct neuronal infection (Heaton et al. 2010). Intensification of cART does not seem to have an impact on this low-level viral production in the CNS (Eden et al. 2010). Currently, no drugs are capable of eradicating HIV reservoirs from the CNS.

Many aspects of the HIV/SIV reservoirs in the CNS are still unclear (Brew et al. 2015; Churchill and Nath 2013). For instance, little is known regarding the cell types harboring persistent viruses, the distribution of virus and infected cells, viral diversity in tissues, or the possible genetic evolution in the brains of HIV-infected patients on cART. HIV diversity has been found in the brain of untreated patients showing neurological symptoms or opportunistic infections. Structural changes have also been observed in both the cerebral white and gray matters of HIV patients, even during suppressive cART in adults and adolescents (Fennema-Notestine et al. 2013). However, specifics of the distribution of HIV in white matter and gray matter are relatively unknown.

As different regions of the brain have different structure and function, the distribution and density of viral target cells is patchy, and the susceptibility of various regions of the brain to HIV infection differs. Few studies have extensively examined different regions of the CNS in HIV infection. A study of six brain tissues (meninges, frontal gray matter, frontal white matter, temporal subcortex, cerebellum, and basal ganglia) of HIV-infected patients demonstrated that all patients had HIV DNA in the brain, but with variable levels (Zhao et al. 2009). However, most studies have only examined one or two regions of the brain, which may not accurately reflect the whole extent of cerebral HIV infection. The challenge of studying multiple sections largely lies in the difficulties in accessing different, fresh brain tissues for sampling. To address these issues, animal models are essential.

Nonhuman primate (NHP) models have been instrumental in studying viral latency/persistence in the CNS (Clements et al. 2011) and continue to be used to address unanswered questions in the field (Garrido and Margolis 2015). In SIV-infected, treatment-naïve macaques, SIV DNA can be detected throughout SIV infection, but RNA is usually only detectable at the later stages of SIV encephalitis (Williams et al. 2001). In pigtailed macaques (Macaca nemestrina) infected with neurovirulent viruses SIV/17E-Fr or SIV/DeltaB670, cART reduces viral replication and inflammation, but viral DNA is persistently detectable (Zink et al. 2010). Further, viral compartmentalization occurs between the meninges and the brain parenchyma, and virus in CSF more closely matches that of virus in the meninges at least in one animal infected with neuropathogenic isolates SIVsmH631Br and SIVsmE804E (Matsuda et al. 2013). However, the density of SIV-infected cells can vary depending on the part of the brain sampled, which may confound interpretation of some studies, depending on what tissue sites and the number of sites biopsied (Brew et al. 2015). Thus far, it remains unclear as to whether cART can alter the distribution and density of SIV infection and addressing these issues is critical for new strategies targeting specific tissues for viral control and eradication.

Chinese rhesus macaques (ChRM) have been used in SIV pathogenesis and therapy studies, but rarely has the CNS been studied in this model. Recently, one study detected SIVmac239 RNA in the brain of SIV-infected ChRM resulting in neuropathological damage (Liu et al. 2016). In this study, we used SIVmac251, the most commonly used virus in the NHP SIV/AIDS model, as it more closely mimics the pathogenesis of HIV infection. Because myeloid cells such as perivascular macrophages and microglia are likely major reservoirs in the brain (Churchill et al. 2016; Dahl et al. 2014; Schnell et al. 2009; Williams et al. 2001), and because more microglia exist in white matter than gray matter in the human brain (Mittelbronn et al. 2001), we hypothesized that the white matter may be a more significant viral reservoirs in the brain than the gray matter which should be evident by higher DNA levels in the former. To address these issues, we examined SIV viral load in different regions of the CNS, compared viral distribution and genetic diversity in both SIVmac251-infected ChRM without cART and in animals on cART at the end of therapy. We also analyzed SIV gp120 envelope viral sequences isolated from the CNS. To our knowledge, this is the first time that SIV CNS reservoirs have been carefully examined in 20 regions of the CNS in a ChRM model, which closely mimics HIV infection and treatment in humans (Ling et al. 2010; Ling et al. 2014; Ling et al. 2013; Ling et al. 2007; Ling et al. 2002a; Ling et al. 2002b; Zhou et al. 2013).

Results

Basic information of SIV-infected Chinese rhesus macaques with or without antiretroviral therapy

Eleven adult ChRM (four males and seven females) 7 to 22 years of age (mean age 13.9 ± 4.6 s.d.) were studied in the cART drug-treated (DTs) group. These animals were chronically SIVmac infected for 3–87.6 months prior to initiation of cART with relatively low plasma viral loads (pVL). Animals received three-drug antiretroviral therapy for varying periods of time ranging from 7 to 24 weeks. By the end of cART termination, all animals reached undetectable levels of pVL (limit of detection was 30 copies/ml) (Table 1). Another group of 12 animals (six males and six females) 6.6–14.5 years (mean age 10.2 ± 2.7 s.d.) was used for studies of plasma and CSF viral loads (cVL). These animals received cART for a short period of time (8 weeks) starting at 4 weeks post-SIV infection, but they were off of cART for varying lengths of time when necropsied as shown in Table 2. All the viral loads on plasma and CSF shown were measured at the time of necropsy. No observable pathogenic lesions or evidence of SIV-associated encephalitis was detected by histologic examination.

Table 1 Basic information of SIV-infected Chinese rhesus macaques on antiretroviral therapy
Full size table
Table 2 Basic information of SIV-infected Chinese rhesus macaques
Full size table

CSF viral loads and their association with plasma viral loads in SIV-infected Chinese rhesus macaques

CSF viral load (cVL) levels were tested in 12 untreated animals. The levels ranged from undetectable (limitation of detection, 30 copies/ml) to 5.6 × 105 copies/ml. Compared to plasma viral loads, CSF had much lower levels (median 802 copies/ml, 25–75% percentile range 30–63,275 copies/ml) than pVL (median 77,650 copies/ml, 25–75% percentile range 366–282,750 copies/ml). However, cVL in untreated animals positively correlated with pVL levels (R = 0.7686, p = 0.0049) (Fig. 1). In the cART group, cVL were not detectable because these animals had pVL at ~ 103 copies/ml or less even prior to initiation of cART. In addition, although the ages of cART group were relatively older than that of the untreated group, analysis did not show significant correlation between age and pVL in each group; therefore, the impact of age discrepancy was considered insignificant in this study.

Fig. 1
figure 1

Correlations between SIV RNA loads in plasma and cerebrospinal fluid (CSF) in untreated SIV-infected Chinese rhesus macaques

Full size image

Frequency of positive SIV env DNA in different regions of the brain in animals with and without cART

To determine distribution and frequency of SIV infection in different regions of the brain and spinal cord in both untreated animals and DTs, we collected 16 brain regions that included corpus callosum, temporal lobe, frontal lobe, parietal lobe, occipital lobe etc., and four regions of the spinal cord (Table 3). We used nested PCR to amplify SIV envelope gp120 from the 16 brain regions and spinal cord in 11 cART animals and 2 typical progressors which had sustained pVL at 104–106 copies/ml off treatment (Table 3). Detectable viral DNA was spread throughout many brain regions from different individuals on cART, although 6 of 11 cART animals (55%) had no virus detected in the CNS. The highest frequency of detectable SIV DNA in cART group was in basal ganglia (4 of 11, 36.4%) followed by brain stem (3 of 11, 27.3%). Because the regions of gray and white matter are not distinctly separated in midbrain, detection of SIV was not separated into gray and white matter in this region (Table 3). Not surprisingly, no SIV RNA was detected in any of the drug-treated animals.

Table 3 Frequency of SIV env DNA detection in different regions of the CNS
Full size table

SIV DNA levels were significantly reduced in gray matter but not in white matter in cART macaques compared to untreated animals

To assess viral distribution in the gray and white matters, we used real-time PCR and quantified SIV gag DNA levels in the two untreated TPs and three DTs that showed SIV-positive signals in multiple regions of the brain. The detection limit of SIV gag DNA was 15 copies/106 cells. In DTs, virus DNA levels were significantly higher in white matter than in gray matter (p < 0.0001) (Fig. 2a). White matter had a median 455.7 copies/106 cells (25–75% percentile range 221–773 copies/106 cells), whereas gray matter had a median 50 copies/106 cells (25–75% percentile range 20–127.9 copies/106 cells). In contrast, gray matter had significantly higher viral levels than white matter in untreated SIV-infected animals (p < 0.0001) (Fig. 2b); the former with a median 4184 copies/106 cells (25–75% percentile range 1541–7697 copies/106 cells) and the latter with a median 472.5 copies/106 cells (25–75% percentile range 205.6–1190 copies/106 cells). Viral levels were significantly higher in the untreated group than treated group in gray matter (p < 0.0001) (Fig. 2c). However, strikingly, viral levels were not significantly different between the two groups in white matter, with median 567 copies/106 cells in untreated animals versus 530 copies/106 cells in treated animals (p = 0.3494) (Fig. 2d).

Fig. 2
figure 2

Comparison of levels of SIV gag DNA between gray matter and white matter in the group of cART-treated (a) and untreated Chinese rhesus macaques (ChRM) (b), and levels of SIV gag DNA in gray matter between groups of untreated and cART-treated ChRM (c) and in white matter (d). Solid diamond, SIV gag DNA viral load in gray matter of ChRM on ART; solid square, SIV gag DNA viral load in white matter of ChRM on ART; open diamond, SIV gag DNA viral load in gray matter of untreated ChRM; open square, SIV gag DNA viral load in white matter of untreated ChRM; ***p < 0.0001, N. S. no significance

Full size image

Viral diversity and presence of extensive G-to-A hypermutations in brains of cART animals

Next, we assessed the degree of viral mutation and diversity in TPs and cART controllers. We studied two TPs and two of the three DTs (DT1 and DT2). Since the third animal (DT3) in DTs group had very low DNA levels (range of 10 to 103 copies/106 cells) that were barely measurable by SIV gag real-time PCR, PCR quantification of the long fragment of SIV envelope gp120 and sequencing was not obtained from this animal. The limitation of SIV env detection is similar to those of single genome analysis (SGA), which can only reliably be successful on plasma samples with viral loads above certain thresholds (~ 4000 copies/ml). Therefore, only those samples that had SIV DNA levels above 1 × 103 copies/106 cells in the two TPs and two DTs were analyzed. Overall, the sequences could be easily distinguished between individual animals. Each animal had its own clades of virus (Fig. 3). The sequences from different regions of the brain in TP1 were highly homogeneous and close to the parental strain SIVmac251-PM series KC522165-KC522168 sequences (Del Prete et al. 2013) shown in Fig. 3. The viral sequences from TP2 were a little more variable and close to KC522142, and the HM045 series (Yeh et al. 2010) with a significant 21 nucleotide deletion in the V4 region in most variants.

Fig. 3
figure 3

Phylogenetic analysis of SIV envelope gp120 sequences of variants isolated from gray and white matter of different regions of the brain in 2 SIV-infected typical progressors without cART (TP1 and TP2) and 2 SIV-infected ChRM on cART (DT1 and DT2). Each sequence was named as animal name-label number of the brain followed with the left (L) or right (R) gray or white with more than one sequence if followed with a number, for instance, TP2-20L-G1 (the 2nd typical progressor animal-left temporal lobe-gray matter sequence 1) represents sequence 1 that was isolated from the gray matter of left temporal lobe in the 2nd typical progressor animal. The tree scale bar indicates the number of substitutions per site

Full size image

Sequences from cART animals DT1 and DT2 clustered closely, and were close to the SIVmac251 sequences of HQ187 series (Whitney et al. 2011) and the parental strain KC522142. Notably, DT2 had a high frequency of APOBEC3G (apolipoprotein B mRNA-editing, enzyme-catalytic, polypeptide-like 3G) hypermutations (G-to-A mutations), up to 75 (p value = 1.40E-11) in sequences derived from two regions: white matter of left parietal lobe and the brain stem (labeled as DT2-22L-W and DT2-24), when compared with sequences from other regions of the same animal and sequences from animal DT1 (Table 4).

Table 4 Significant hypermutation of G to A in parietal lobe and brain stem in DT2
Full size table

Changes of the envelope gp120 of SIV in the CNS in untreated typical progressors and animals on cART

To determine if the mutations of nucleotides resulted in amino acid changes in envelope gp120 protein, we further analyzed the amino acid alignment. The sites with most changes were in the V1 and V4 loops. The V1 region was rich in serine (Ser or S) and threonine (Thr or T) especially in cART animals but these were not very glycosylated, which differs from HIV-1M group that have more changes in glycosylation.

Sequences from different regions of the brain of TP2 showed some variations in the V1 and V4 regions. All but one of the TP2 variants had mutations from serine (Ser or S) to proline (Pro or P) at the V1-Loop position 155 (Fig. 4). Further, the V4 loop had a seven amino acid residue deletion at position 424-430 (QRPKERH) corresponding to a 21 nucleotide deletion in the majority of variants (Fig. 4), similar to findings previously reported in virus found in peripheral blood in chronic but not acute phase of SIVmac251 infection (Yeh et al. 2010).

Fig. 4
figure 4

Amino acid sequence comparison of SIV envelope V1 loop and V4 loop of SIVmac239, SIVmac251, variants from each sample group. TP1, test animal 1 (representative sequence); TP2, test animal 2; DT1, drug-treated animal 1; DT2, drug-treated animal 2. HQ series and HM series, GenBank sequences; and KC strains, parental strains. Dots represent amino acid identity with reference strain SIVmac239; dashes represent deletions

Full size image

Discussion

To our knowledge, this is the first study to evaluate viral reservoirs in up to 20 variant regions of the CNS in cART-treated SIV-infected ChRM model of HIV infection, although the SIV/ChRM model has been used to study SIV pathogenesis, as well as tissue reservoirs in lymphoid tissues and the gut (Ling et al. 2010; Ling et al. 2014; Ling et al. 2013; Ling et al. 2007; Ling et al. 2002a; Ling et al. 2002b; Marthas et al. 2001; Monceaux et al. 2007). As ChRM usually have lower viral loads and develop AIDS relatively slower than other SIV-susceptible nonhuman primates, it is not surprising that cVL were not detectable in SIV-infected LTNP of ChRM even prior to cART. However, undetectable cVL clearly did not mean that the virus did not enter the CNS of SIV-infected ChRM, as evidenced by the fact that SIV DNA could be detected in some of the LTNP even after cART. Similar to HIV-1-infected patients, cVL was positively correlated with pVL in typical progressors of ChRM without ART, indicating that ChRM with SIV infection mimics HIV-1 infection in the CNS.

In this study, we found that two sites, basal ganglia and brain stem, had the relatively highest frequencies of SIV detection in animals on ART from the 15 parts of the brain and spinal cord tested, indicating that these two sites might be major sites of viral persistence. However, it is possible that some of the viral DNA detected were from defective viruses. Our results are consistent with findings that despite the use of cART, cells carrying viral SIV DNA persist (Zink et al. 2010). In HIV-1-infected patients, abnormalities of the basal ganglia are associated with HIV dementia (Steinbrink et al. 2013). Although detection of viral DNA could not exclude defective virus, the results are still important because even isolated viral particles such as Tat can cause neurocognitive dysfunction, even in the absence of replication-competent virus. Determining which treated animals or treated patients still have substantial CNS reservoir would be very important, as identification and characterization of cellular CNS reservoirs is a critical step to understand the mechanism of viral persistence in the CNS for the development of eradication strategies.

We found significant differences in SIV DNA viral loads between gray and white matter in the same group of animals in untreated as well as cART animals as shown in Fig. 2. We also found that no significant differences in white matter viral loads between untreated and cART animals (Fig. 2d). Since the animals were not perfused when euthanized, it is possible that viruses examined in the gray matter could also contain vessel-associated blood-derived virus, as the human brain has higher blood flow in the gray matter than that of the white matter (Ballabh et al. 2004; Rengachary and Ellenbogen 2005). However, the > 2-fold log10 disparity of SIV load between untreated and treated was unlikely due solely to virus from blood contamination. The results suggested that antiretroviral therapy was effective in reducing viral DNA loads in the gray matter.

In contrast, the comparable viral levels in white matter between cART macaques and untreated group indicate that cART has little impact on this tissue, possibly due to poor penetration of antiretroviral drugs (Cory et al. 2013; Letendre et al. 2008). This result is in agreement with observations that HIV is more compartmentalized in the subcortical white matter in HIV-infected patients on cART (Nath 2015). A number of studies have demonstrated that gray matter and/or white matter undergo significant structural alterations during HIV infection, even in patients receiving cART (Becker et al. 2012; Fennema-Notestine et al. 2013; Sarma et al. 2014; Wang et al. 2016). Although cellular reservoirs in the CNS are not fully understood, studies have suggested that cellular HIV or SIV reservoirs in the CNS likely include perivascular macrophages, long-lived microglia cells (Dahl et al. 2014; Schnell et al. 2009; Williams et al. 2001) and astrocytes (Churchill et al. 2009). Further, there are more microglia in white matter than gray matter in the human brain (Mittelbronn et al. 2001). Based on the brain regions we studied, the microglial populations harboring SIV in these two regions are the most likely major factors contributing to the differences in virus loads observed between gray and white matter.

While viral diversity and CNS compartmentalization (in CSF) have been identified in multiple studies (Sturdevant et al. 2012; Sturdevant et al. 2015), especially in HIV-1-infected patients with HIV-associated dementia (HAD) (Schnell et al. 2011) and during the early years after HIV transmission (Schnell et al. 2010), here we mainly focused on viral variants isolated from different specific regions of the brain. We found that each animal had its own clade of viruses, which agrees with the vast majority of other HIV and SIV data sets that suggest even if a person or animal is exposed to viral quasi-species from a donor, or from an experimental stock of SIV, primary infections in the recipient are almost always essentially clonal. TP1 appears to have a bottleneck effect that indicates a homogenous population was formed in the brain tissues both in white and gray matter. This is probably due to the brain blood barrier (BBB) selection effect, and that particular virus may have had the best fit in these brain tissues. Although TP2 showed relatively more diversification, dominant variants had a specific 21 nucleotide deletion in V4 region which was not found in parental virus stock, similar to findings from chronically SIVmac251-infected macaques, which the HM045 series sequences reported elsewhere (Yeh et al. 2010) and a 15 nucleotide deletion at the same region in sequences of U18019-U18021 series (Zhu et al. 1995), suggesting the mutated virus is likely emerging in chronic infection. Whether the mutation occurred after virus entered the brain or was derived from peripheral blood in chronic infection or whether the mutated virus is the source of blood mutated virus remains to be determined, especially since other studies have demonstrated that meninges contain both CNS and peripheral blood-derived viruses (Matsuda et al. 2013), and there is a possibility of viral dissemination from the brain to the blood, or from the meninges to the brain parenchyma (Nath 2015). Since the function of V4 loop is not well understood, it is unclear whether there is selective pressure on the variants to shorten the V4 loop by 7 residues (QRPKERH). However, further studies are needed to elucidate the virological and immunological significance of this shortening of the V4 loop.

Interestingly, sequences from white matter of the left parietal lobe and brain stem (22L-W and 24-W) in animal DT2 had high frequencies of G to A mutations (calculate p value, Table 2). These findings are in agreement with results in viral sequences isolated from brain (Depboylu et al. 2007) and semen of SIV-infected rhesus macaques (Whitney et al. 2011) as well as in HIV-1-infected patients receiving antiretroviral drugs (Knoepfel et al. 2011; Mullins et al. 2011), possibly because of APOBEC3G editing under the selective pressure of drug treatment. While most of these viruses could be defective, DNA genome-like G-to-A mutated viruses have been detected in blood and rectal tissue (Fourati et al. 2012), and it is noteworthy that the G-to-A mutations were found only in the parietal lobe and brain stem, but no other sections of the brain in this animal, suggesting that viral hypermutation does not simultaneously occur in different sections of the brain.

Here, we show that SIV-infected ChRM on cART are an excellent model to study neuro-AIDS and CNS reservoirs. Moreover, given that the virus enters the CNS as early as days after infection (Valcour et al. 2012), and pVL are comparable during acute infection between Indian rhesus macaques and ChRM even those become spontaneous LTNP (Ling et al. 2002a), we show that the CNS reservoir is established and persistent after SIV infection in both progressors and LTNP of ChRM. Of note, not all animals have detectable virus in the CNS, just like HIV-1-infected patients, many of whom also do not have detectable virus in brain tissues (Zhao et al. 2009). We also show that basal ganglia and brain stem may be the major sites of viral persistence in the CNS after antiretroviral therapy. Moreover, cART group had significantly reduced viral levels in gray matter but had no significant changes in viral levels in white matter when compared with untreated macaques. Finally, G to A hypermutations occurred in certain parts of the brain in a cART animal. Therefore, sampling multiple regions of the CNS may be necessary when assessing effects of novel approaches on reduction of the size of viral reservoirs in the CNS. Of note, we only used PMPA and FTC in this study, effect of more potent regimen of cART on CNS reservoir, such as the clinically recommended combination of tenofovir disoproxil fumarate (TDF)/emtricitabine (FTC)/dolutegravir (DTG), can be thoroughly tested in this unique nonhuman primate model of HIV infection.

Materials and methods

Ethics statement

Experimental procedures performed on rhesus macaques used in this study were approved by the Tulane Institutional Animal Care and Use Committee (IACUC). All animals were housed indoors throughout the study period at the Tulane National Primate Research Center (TNPRC). TNPRC facilities are fully accredited by the Association of Assessment and Accreditation of Laboratory Animal Care International (AAALAC) in accordance with standard husbandry practices following the Guide for the Care and Use of Laboratory Animals (NIH). To avoid unnecessary discomfort and pain to animals, anesthesia and analgesic medications were used appropriately under the direction of a veterinarian. Rhesus macaques were anesthetized with 10 mg/kg ketamine or 3–8 mg/kg telazol whenever they were removed from their home cage. For surgical procedures, animals were anesthetized with ketamine and maintained in a surgical plane of anesthesia using isoflurane. For physical exams and blood sampling, 10 mg/kg ketamine was used. Buprenorphine (0.01 mg/kg) or buprenorphine extended release (0.2 mg/kg) was the standard analgesics administered for painful procedures.

Animals and virus inoculation

Rhesus macaques (Macaca mulatta) of Chinese origin (ChRM) were used in this study. A total of 23 animals were studied, which included 11 out of 12 SIV-infected DTs with exclusion of the one that had detectable viral load at necropsy, and also included 12 chronically SIV-infected untreated animals for CSF and blood collection to monitor viral loads in plasma and cerebrospinal fluid (CSF). The 11 DTs were further studied from 15 brain regions described in details below in brain tissue collection. Two representative untreated animals were also studied from 15 brain regions for comparison. These two animals were typical progressors (TPs), which were infected for 1–3 years with a set point plasma viral loads at 104~106 copies/ml. Basic animal information is shown in Table 1 for treated animals and Table 2 for untreated animals. Animals were intravaginally or intravenously infected with 500 TCID50 SIVmac251. The stock SIVmac251 was obtained by culturing and harvesting from CEMx174 cells and provided by the Virus Characterization, Isolation, and Production Core of the TNPRC, and the sequences of the stock are available in GenBank (KC522165-168, KC522142) (Del Prete et al. 2013).

Antiretroviral therapy

During chronic SIV infection, 11 ChRM were treated with the reverse transcriptase inhibitors (R)-9-(2-phosphonylmethoxyypropyl) adenine (PMPA, tenofovir; 20 mg/kg), and beta-2′, 3′ dideoxy-3′-thia-5-fluorocytindine (FTC, emtricitabine; 40 mg/kg) daily by subcutaneous injection for 8–24 weeks. Tenofovir and emtricitabine were generously provided by Gilead Sciences, Inc. (Foster City, CA) via Material Transfer Agreement.

Animal euthanasia and brain tissue collection

Animal euthanasia was consistent with the recommendations of the Panel on Euthanasia of the American Veterinary Medical Association. SIV-infected and/or drug-treated macaques were euthanized by first anesthetizing the animal with telazol and buprenorphine followed by a lethal i.v. injection of sodium pentobarbital following Tulane IACUC standards of operation (SOP). Tissues from brain and spinal cord (15 regions) were collected fresh and also snap frozen in liquid nitrogen during necropsies.

Quantification of viral RNA in plasma and cerebrospinal fluid in brain tissues

Plasma was separated from EDTA-treated whole blood by centrifugation at 700–800×g for 10 min, conservatively removed plasma to avoid cell pellet, transferred to a second centrifuge tube, and were centrifuged at 25,000×g for 1 h to concentrate virus particles especially in samples from animals on cART with expected low viral loads. CSF was obtained from the dorsal cervical spine using TNPRC’s standard operating procedures. CSF was centrifuged following the similar procedures as collection of plasma. SIV plasma viral loads (pVL) and CSF viral loads (cVL) were determined by real-time quantitative PCR analysis by the TNPRC Pathogen Detection and Quantification Core as described in detail elsewhere (Monjure et al. 2014). High Pure Viral RNA kit (Roche, Indianapolis, IN, USA) was used for viral RNA extraction with minor modification. Primers and probe were designed from the conserved gag region covering detection of SIVmac239, SIVmac251, and SHIV viruses.

DNA extraction and quantification of viral DNA in brain tissues

Total genomic DNA was extracted from each tissue using a QIAamp DNA mini kit following the manufacturer’s instructions. Determination of cell-associated SIV DNA viral loads for samples collected up to the time of necropsy followed methods for assays and analysis previously described elsewhere (Venneti et al. 2008). These standard methods have a 95% reliable threshold sensitivity of 30 total copies of SIV DNA or RNA per sample. Numbers of SIV copies were normalized to cell equivalents co-determined by quantitative PCR (qPCR) as described (Ling et al. 2014). For samples prepared from necropsy tissues, nested and qPCR/RT-PCR methods of analysis with reaction conditions and primer/probe sequences were performed as detailed in Hansen et al. (Hansen et al. 2011). DNA was prepared as above from multiple tissues to provide a greater amount of test material and a correspondingly lower limit of detection.

Viral DNA amplification by nested PCR, sequencing, and phylogenetic analysis

One to 0.5 μg DNA was used for nested PCR to amplify SIV envelope gp120. The PCR products were purified using Clonetech PCR purification kit for population-based sequencing. Expand High Fidelity PCR (Taqman enzyme) was used to minimize PCR-induced sequence errors. The primers were as below:

  • env-OF: 5′ CTATAATAGACATGGAGACACCCTTG 3′

  • env-OR: 5′ CTTCTTGCACTGTAATAAATCCCTTCC 3′

  • env-iF: 5′ GTAAAAAGTGTTGCTACCATTGCCAG 3′

  • env-iR: 5′ ACTGATACCCCTACCAAGTCATC 3′

PCR reaction conditions included denaturation at 94 °C for 2 min, followed by 35 cycles of denaturation at 94 °C for 50 s, annealing at 55 °C for 50 s, and extension at 72 °C for 2 min, with an additional extension of 72 °C for 7 min. Two microliters of the first-round PCR products was used for nested PCR under the same thermocycling parameters. The PCR products were purified using high pure PCR product purification kit from Roche Life Science. SIV gp120 env sequences from newly isolated brain SIV variants were aligned with SIVmac239, and SIVmac251 reference sequences from the Los Alamos National Laboratory HIV Sequence Database and other relevant sequences from GenBank using the CLUSTAL X program (Thompson et al. 1997). The alignment was manually adjusted, and poorly aligned regions were excluded. All phylogenetic trees were constructed by PhyML program for a maximum likelihood tree starting with BioNJ neighbor-joining method, and the reproducibility of the branching orders was estimated by 1000 bootstraps. Frequencies of G-to-A mutations were analyzed using the HIV Sequence Database Hypermut 2.0 program (http://hiv-web.lanl.gov). The amino acid sequences were aligned using Clustal W program of the BioEdit software. Highlighter for amino acid sequences v2.3.4 and analysis (Keele et al. 2008) was used to visualize the differences in amino acids between untreated and treated animals.

Statistical analysis

Nonparametric statistical analyses (Mann-Whitney test or Wilcoxon signed-rank test) were used to compare median viral loads in plasma and CSF, cell-associated SIV DNA in gray matter versus white matter, and untreated versus drug-treated groups. The Spearman correlation was used to assess the relationship of levels of viral loads between plasma and CSF. GraphPad Prism 5.0 statistical software (GraphPad Software, Inc. San Diego, CA, USA) was used to analyze data and statistical results were set two-sided at p < 0.05 as significant.

Nucleotide sequence accession numbers

Viral env sequences were deposited in GenBank under accession numbers MF284715-MF284792.