Cerebrospinal fluid CD4+ T cell infection in humans and macaques during acute HIV-1 and SHIV infection

HIV-1 replication within the central nervous system (CNS) impairs neurocognitive function and has the potential to establish persistent, compartmentalized viral reservoirs. The origins of HIV-1 detected in the CNS compartment are unknown, including whether cells within the cerebrospinal fluid (CSF) produce virus. We measured viral RNA+ cells in CSF from acutely infected macaques longitudinally and people living with early stages of acute HIV-1. Active viral transcription (spliced viral RNA) was present in CSF CD4+ T cells as early as four weeks post-SHIV infection, and among all acute HIV-1 specimens (N = 6; Fiebig III/IV). Replication-inactive CD4+ T cell infection, indicated by unspliced viral RNA in the absence of spliced viral RNA, was even more prevalent, present in CSF of >50% macaques and human CSF at ~10-fold higher frequency than productive infection. Infection levels were similar between CSF and peripheral blood (and lymph nodes in macaques), indicating comparable T cell infection across these compartments. In addition, surface markers of activation were increased on CSF T cells and monocytes and correlated with CSF soluble markers of inflammation. These studies provide direct evidence of HIV-1 replication in CD4+ T cells and broad immune activation in peripheral blood and the CNS during acute infection, likely contributing to early neuroinflammation and reservoir seeding. Thus, early initiation of antiretroviral therapy may not be able to prevent establishment of CNS viral reservoirs and sources of long-term inflammation, important targets for HIV-1 cure and therapeutic strategies.


Introduction
Successful viremia suppression by anti-retroviral therapy (ART) has significantly reduced neurological pathologies associated with HIV-1 infection. However, increased life expectancies and co-morbidities in the ART era has led to higher prevalence of milder forms of HIV-related neurocognitive dysfunction [1,2], which remain a challenge for people living with HIV-1 (PLWH). The mechanisms underlying the neurocognitive dysfunction associated with HIV-1 are complex and currently unclear [3,4], though it is widely believed that HIV-1 replication within the central nervous system (CNS) prior to therapy initiation triggers neuropathogenesis. Overall neurological dysfunction in treated infection is thought to mainly result from widespread inflammation in the CNS in response to the virus rather than direct viral infection and replication [5], though a contribution of viral infection/persistence to neurologic injury during ART is an area of intense study. The very early appearance of HIV-1 RNA in the CSF of acutely infected PLWH suggests that these processes may begin within the first weeks of infection [6,7]. The cellular origin of this virus and the extent to which it represents independent CNS viral replication is unknown. Importantly, early viral replication in the CNS may also establish viral reservoirs that persist during ART and pose a barrier to HIV-1 cure.
There remains debate in the field as to how HIV-1 enters the CNS compartment and the relative role of myeloid versus CD4+ T cells in this process. Infected microglia and macrophages in the brain along with aberrant inflammatory responses among these resident cell populations are established drivers of neuropathogenesis in chronic infection [8][9][10][11][12]. However, evidence of CSF inflammation and viral detection within two weeks of HIV-1 infection along with very early compartmentalization of HIV-1 sequences in the CNS indicate an early phase of viral entry when CD4+ T cells are more likely to be involved [7,[13][14][15]. This is supported by CSF transmitted viral variants early in human HIV-1 infection that are T cell-tropic and inflammatory infiltrates in the brain parenchyma of acute CCR5-tropic SHIV-infected macaques comprised predominantly of T cells [15,16]. Gradual emergence of viral lineages with greater macrophage tropism in human CSF months post-infection suggests that selection for non-T cell entry occurs following CNS invasion [15]. Recent studies characterizing CSF virus during ART-suppressed chronic infection also implicate T cells as the origin of HIV-1 RNA detected in the CNS at this later stage [17,18]. Direct evidence of CNS cells harboring virus and supporting viral replication is necessary to delineate early events in CNS seeding.
To define the early cellular targets of HIV-1 replication in the CNS and the association with CNS cellular inflammation and activation, we assessed CSF cell populations for viral RNA and phenotypic changes during acute infection using a novel combined flow cytometric and PCR approach developed for samples with low cellular input. CSF samples from humans and SHIV-infected rhesus macaques were interrogated to study these events in two complementary models, at peak viremia in humans and longitudinal time points in macaques. The SHIV strain was selected for its prior characterization in macaque CNS during acute infection and its nonneurovirulent phenotype, distinct from accelerated neurovirulent SIV models that result in severe CNS disease more rapidly than seen in humans [19][20][21]. Assays specific for spliced and genomic viral RNA were used to distinguish different stages of cellular infection, including cells actively replicating virus. We compared the infected cell burden in CSF to that in peripheral blood (PB) and lymph nodes to discern compartmentalized replication across anatomic sites. Lastly, we measured the immunologic activation state of CSF cell populations and soluble indices during acute infection. Our findings clarify early host cell infection events and cellular inflammatory responses occurring in the CNS during acute HIV-1 infection.

Ethics statement
All animal in vivo procedures were carried out in accordance with institutional, local, state, and national guidelines and laws governing research in animals including the Animal Welfare Act. Animal protocols and procedures were reviewed and approved by the Animal Care and Use Committee of both the US Army Medical Research and Development Command (USAMRDC, Maryland, USA) Animal Care and Use Review Office as well as the Institutional Animal Care and Use Committee (IACUC) of Bioqual, Inc. (Maryland, USA, protocol number 14-B077) or the IACUC of the Armed Forces Research Institute of Medical Science (AFRIMS; Bangkok, Thailand, protocol number PN 13-07) where the non-human primates were housed. Bioqual, Inc., AFRIMS, and the USAMRDC are accredited by the Association for Assessment and Accreditation of Laboratory Animal Care and are in full compliance with the Animal Welfare Act and Public Health Service Policy on Humane Care and Use of Laboratory Animals. The clinical study protocols were approved by the institutional review boards of Chulalongkorn University (Bangkok, Thailand) and the Walter Reed Army Institute of Research (Silver Spring, MD). All participants provided written informed consent to participate in the study.

Soluble protein biomarker concentration
Levels of soluble biomarkers were measured in CSF and plasma via a combination of Luminex-based custom assays and ELISAs, as per manufacturers' protocol and as described previously [31]. IP-10/CXCL10 and IL-10 were measured using the Luminex based Milliplex Map Human Cytokine / Chemokine Panel (Millipore Sigma, St. Louis MO). Soluble CD163 was measured via the Bio-Plex Pro Human Inflammation Panel 1 (Bio-Rad Laboratories, Hercules CA). Neopterin was measured by chemiluminescent detection ELISA (Genway Biotech, San Diego CA). Data was collected on a FlexMap 3D reader (Bio-Rad Laboratories, Hercules CA) or VersaMax plate reader (Molecular Devices, Sunnyvale CA) and analyzed in Prism v.9 for Mac OS X (GraphPad, La Jolla CA) using 4-parameter fit standard curves.

Statistics
Poisson distribution statistics were used to estimate the frequency of infected cells harboring gag and/or env mRNA in replicate limiting diluted PBMC and LNMC as described previously [26]. For CSF, the infected cell frequency was estimated by conservative assignment of one positive cell to each replicate in which viral mRNA was detected. Wilcoxon matched-pairs rank test was used to compare infected cell frequency between tissues. Flow cytometry data were analyzed in FlowJo v9.9.6. Mann-Whitney test (P<0.05) was used to assess differential surface protein expression. Spearman correlation was used to measure associations between parameters.

Viral and CSF cellular dynamics in acute SHIV infection
To study the dynamics of cellular inflammatory processes and cell-associated viral replication within the nervous system during acute infection, 18 rhesus macaques were infected intrarectally or intravaginally with the well-characterized SHIV-1157ipd3N4, which encodes an R5-tropic subtype C HIV-1 env also able to infect monocyte-derived macrophages [16,22]. Infection with this virus recapitulates many aspects of acute HIV-1 infection, including early CSF viral detection and inflammation [16]. Peak plasma viremia occurred two weeks postinfection (PI) and viremia was sustained in most animals for three months (Fig 1A). CSF SHIV RNA was detected in cell-free CSF supernatant in 4 of 12 animals assessed 12 weeks PI and ranged from 6-722 copies/mL as previously reported [16]. Given dynamic cellular processes that occur during acute infection, including CSF pleocytosis in HIV-1 infection [15,32], we characterized CSF mononuclear cell composition throughout early SHIV infection. Cryopreserved CSF cells collected 4, 8, and 12 weeks PI from a subset of the animals (N = 7-12) and six uninfected controls were thawed and stained with lymphocyte and myeloid cell lineage markers and analyzed by flow cytometry (S1A Fig). PBMC were analyzed in a similar manner for comparison (S1B Fig). The relative abundance of monocytes in the CSF was similar between uninfected animals and following SHIV infection, as was the frequency of CD4+ and CD8+ T cells (Fig 1B). The CD4/CD8+ T cell ratio in both CSF and PB also did not change significantly (Fig 1C). CSF CD4+ T cells were predominantly of a central memory phenotype (median~65%) with limited effector memory cells (median <10%), and the memory subset composition was unaltered throughout acute infection (Fig 1D). In contrast, CD4+ T cells circulating in PB comprised a more balanced ratio of memory and naïve subsets and suffered transient effector memory cell loss 8 weeks PI, consistent with depletion of this population by CCR5-tropic SIV and HIV-1 [33,34]. These data

PLOS PATHOGENS
Cerebrospinal fluid CD4+ T cell infection during acute HIV-1 infection indicate that the CSF cellular makeup is distinct from PB and may undergo modest changes during acute SHIV infection.

SHIV replication in CSF CD4+ T cells of acutely infected macaques
Though detection of SHIV RNA in the cell-free component of CSF was limited in this SHIV model, we hypothesized that cells within the CNS harbored virus based on prior observations of viral RNA+ cells in the brain parenchyma of SHIV-infected macaques [16]. To probe for cell-associated virus in the CSF and identify cell targets, CD4+ T cells, monocytes, and CD8+ T cells were FACS sorted in replicate and analyzed by viral RNA (vRNA) RT-qPCR directly ex vivo (Fig 2A). Cellular vRNA was selected for analysis in an effort to capture active viral transcription, which is not reflected by integrated DNA or intact proviral DNA measurements. SIV gag was used as a marker of unspliced vRNA, which can be derived from virions due to its presence in the viral genome [26,28]. To detect active replication, we developed and validated an env/vpu/nef mRNA assay ("env") specific for these spliced transcripts of SHIV-1157ipd3N4 (S2 Fig). Previous analysis of SIV-infected CD4+ T cells using gag and a related SIV-specific env assay demonstrated a~100-fold increase in genomic vRNA in env+ cells relative to envnegative gag+ cells, consistent with viral replication in cells identified by the spliced env assay [26]. vRNA was present in the majority of the CSF CD4+ T cell specimens, with 75% of animals positive at week 4 PI (N = 12) (Fig 2B). gag vRNA in the absence of env predominated, indicating widespread non-productive infection (e.g. early, latent, abortive). Transcriptionally active or productive CD4+ T cell infection (env+gag+) was more limited and detected in CSF of two of twelve animals, those with the highest plasma viremia 4 weeks PI. The median percentage of infected CD4+ T cells was 0.4% for both gag+ and env+gag+ cells among animals with values above the limit of detection. Analysis of week 12 CSF, which was prioritized only for animals with detectable week 12 PBMC CD4+ T cell infection (described below), demonstrated gag vRNA+ infection in all animals (N = 8; median 0.2% gag+) and productive infection in one animal (1.3% env+). Productive infection was not observed in CSF monocytes, while gag was present infrequently (Fig 2B, bottom), though limited monocyte yields (~20% that of CD4+ T cells) constrained sampling depth. No vRNA was detected in any samples from uninfected controls and vRNA was exceedingly rare among CD8+ T cells from SHIVinfected animals (S3A and S3B Fig).

PLOS PATHOGENS
Cerebrospinal fluid CD4+ T cell infection during acute HIV-1 infection

Lymphocyte activation in CSF and PB during acute SHIV infection
Since cellular activation can mediate CNS inflammation and may increase susceptibility to viral replication [35], we examined the activation status of CSF cells during acute SHIV infection by surface marker staining and flow cytometry. CD38 expression increased on both CD4+ and CD8+ T cells in CSF 4 weeks PI (P<0.05) (Fig 3A). T cell activation was transient and largely resolved by week 12. CSF monocytes did not show evidence of activation based on surface expression of the cell adhesion molecule and activation marker CD169 (Fig 3B), though expression of CXCL10, which encodes the chemoattractant IP-10, was elevated at week 8. In a parallel analysis of PBMC, similar results were observed for CD4+ and CD8+ T cells, with activation apparent as early as 2-3 weeks PI, sustained at week 4, and diminished by week 12 (Fig 3C). Blood monocyte activation was also observed 2-3 weeks PI, but mostly resolved by weeks 4 and 12, consistent with lack of CSF monocyte activation at these times (Fig 3D).

HIV-1 replication in CD4+ T cells in CSF during acute infection in humans
To extend these findings to human HIV-1 infection, which is generally characterized by more robust HIV-1 RNA levels in both plasma and CSF than that observed in this macaque SHIV model, we performed a similar analysis of acute HIV-1 infection in humans. We selected six participants from the RV254 early acute HIV-1 infection cohort in Bangkok, Thailand, with well-characterized viremia. Individuals identified in Fiebig stages III and IV, which correspond to peak HIV-1 RNA levels in plasma and CSF [6], respectively, were prioritized for optimal detection of cell-associated virus ( Table 1). These Fiebig stages represent an estimated 14 and 19 days post-infection, respectively [24]. Five additional participants were included in additional analyses described below. As previously reported in this cohort, CSF HIV-1 RNA ranged from 10 3 −10 5 copies per mL (N = 11), which was~2-3 logs lower than that in plasma.
To measure HIV-1-infected cells in the CNS, CSF CD4+ and CD8+ T cells and monocytes were FACS sorted as described above followed by RT-qPCR for HIV-1 gag and env, using assays optimized for CRF01_AE detection. Assay validation for sensitivity and specificity to detect active viral replication was performed on CD4+ T cells infected in vivo from participants of the RV254 cohort. CD4+ T cells positive for spliced env vRNA contained 10 3 −10 4 copies of unspliced (gag) HIV-1 RNA,~10-100-fold more than vRNA+ cells lacking spliced transcripts, and downregulated surface CD4 protein, providing direct evidence of abundant viral transcription and viral protein expression in cells identified by our spliced vRNA assay (Fig 4A). In addition, control experiments confirmed that the majority of vRNA detected was not derived from virions bound to the cell surface (S2B and S2C Fig). Strikingly, transcriptionally active CD4+ T cell infection (env+gag+) was observed in all acute infection CSF specimens surveyed (Fig 4B). The frequency of productively infected cells ranged from 0.04%-0.45% of CSF CD4+ T cells (Fig 4C and 4D). gag vRNA+ cells were even more abundant and finer resolution cell sorting was required to determine their frequency (S4A Fig), estimated at 0.6%-4.9% of CD4+ T cells. Thus, on average >1% of CD4+ T cells in the CSF harbored vRNA during acute infection and approximately 10% of these cells supported virus replication as evidenced by spliced vRNA expression. CSF cells from PWOH were negative for vRNA. Cell-free CSF supernatant HIV-1 RNA levels did not correlate with the frequency of infected CSF CD4 T cells (gag+ or env+gag+) in this subset of six individuals. We found minimal evidence of infection in CSF monocytes, while CD8+ T cells also rarely contained vRNA (Fig 4E and 4F). As in the macaque model, monocyte sampling depth was limited due to their rarity in CSF [36].
A similar analysis of memory CD4+ T cells in PBMC specimens from the same individuals was performed to compare the infected cell frequency between CSF and PB. HIV-1 replication active (env+gag+) infected CD4+ T cells comprised 0.04-0.84% of the memory compartment in PBMC (Fig 4C), while inactive infection (env-negative, gag+) again averaged approximately ten-fold higher, ranging from 1.5%-12.6%. These values did not significantly differ from those observed in the matched CSF specimen (Fig 4D). CD4+ T cell infection in PB was not strongly correlated with that in CSF, for either gag+ only or env+gag+ cells, in this small sample set. Overall, both non-productive and productive CD4+ T cell infection was robust in both CSF and PB during acute HIV-1 infection.

Lymphocyte and monocyte activation in CSF and PB during acute HIV-1 infection
Prior study in this cohort identified dynamic alterations in CD8+ T cell responses in the CSF evolving over Fiebig stages in acute HIV-1 infection [35]. To examine the broader cellular

PLOS PATHOGENS
Cerebrospinal fluid CD4+ T cell infection during acute HIV-1 infection

PLOS PATHOGENS
Cerebrospinal fluid CD4+ T cell infection during acute HIV-1 infection composition of CSF in AHI, including CD4+ T cells and monocytes, we explored the relative proportion of different cell types in the CSF by flow cytometry. The frequency of monocytes in CSF was similar between PWOH and people living with acute HIV-1 infection (N = 5 and 9, respectively; S4B Fig). CD8+ T cell frequency increased during infection, as previously described [35], while CD4+ T cells trended lower, resulting in diminished CSF CD4+/CD8+ T cell ratios (S4B and S4C Fig). PBMC CD4+/CD8+ T cell ratios were also reduced. CD4+ T cells in both CSF and PBMC displayed a predominantly memory phenotype, based on surface protein marker expression, and memory cell frequency did not differ between acute infection and PWOH (S4D Fig). Since acute HIV-1 infection stimulates systemic immune activation and antiviral activity, we sought to determine whether monocytes and T cells in CSF also exhibit heightened activation. The frequency of activated CD4+ T cells, as measured by surface CD38 expression, increased during acute infection in both CSF and PBMC compared to PWOH (P<0.05; Fig  5A). Median activated CD4+ T cells increased by~6-fold in CSF and~2-fold in circulation, indicating considerable HIV-1-associated T cell activation within the CNS that may exceed that in PB. CD8+ T cells were also highly activated in CSF, consistent with prior analysis [35], as well as in PB. Similar results were observed for T cells co-expressing HLA-DR and CD38 (S5A and S5B Fig). CSF monocytes also exhibited marked activation during acute HIV-1 infection, with elevated surface expression of both CD169 and CD38 (Fig 5B). Median fluorescence intensities increased~16-and 9-fold for CD169 and CD38, respectively. Monocyte activation marker expression was also greater in PB, consistent with prior studies. Taken together, both CD4+ and CD8+ T cells as well as monocytes were highly activated in CSF during early acute HIV-1 infection.
As has been previously reported in this cohort [6,7,37], multiple soluble markers of inflammation were differentially expressed in CSF during acute HIV-1 infection. Among a panel of 44 pro-inflammatory cytokines and biomarkers measured by Luminex and ELISA in this subset of nine individuals, IL-10 (P<0.01), IP-10 (P<0.01), neopterin (P<0.01) and sCD163 (P<0.05) were~2-5-fold elevated in CSF compared to PWOH. Exploring the relationship between CSF cellular activation and these pro-inflammatory factors, CD4+ and CD8+ T cells co-expressing CD38 and HLA-DR were positively correlated with IP-10 (rho = 0.8, P<0.01; and rho = 0.7, P<0.05 respectively), while monocyte CD163 expression (cleaved from cell surface during activation) was negatively correlated (rho = -0.8, P<0.05) (Fig 5C). No associations were observed for the other biomarkers. Infected CD4+ T cell frequency in CSF did not correlate with the elevated factors among the six individuals for whom frequency estimates were made, though replication-active CD4+ T cells were associated with vascular endothelial growth factor C levels (rho = 0.8, P<0.05), an angiogenesis factor with known deleterious effects in the CNS [38], a finding that bears investigation in a larger study. These data indicate that activated leukocytes are associated with a pro-inflammatory response in CSF during acute HIV-1 infection.

Discussion
Early HIV-1 entry into the CNS has the potential to establish distinct viral reservoirs of infected cells as well as initiate inflammatory processes that may contribute to long-term cognitive impairment in PLWH on suppressive ART. Our studies provide direct evidence of CD4+ T cell infection in CSF during early acute infection in two independent modelsmacaque R5-tropic non-accelerated SHIV infection and human HIV-1 infection. CSF CD4+ T cell infection was consistently detected in PLWH and at frequencies ranging from~0.03-3% of T cells. Infected cell burden was similar between CSF and other anatomical sites, i.e. PB and

PLOS PATHOGENS
Cerebrospinal fluid CD4+ T cell infection during acute HIV-1 infection lymph nodes, suggesting infected cell trafficking and possible equilibration between these compartments. Both T lymphocytes and monocytes were highly activated in CSF during acute HIV-1 infection, indicating broad immune stimulation. Taken together, these findings suggest active viral replication occurs in T cells within the CNS in acute infection concurrent with a robust inflammatory response.
Some models of HIV-1 early entry into the CNS suggest that systemic inflammation disrupts the blood-brain barrier, allowing plasma virions or blood cells harboring virus to traffic into the CNS [39,40]. Circulating monocytes have been implicated in this process due to detection of infected perivascular macrophage/microglia in the CNS of macaques infected with highly pathogenic SIV. However, increased migration of activated T cells across an intact

PLOS PATHOGENS
Cerebrospinal fluid CD4+ T cell infection during acute HIV-1 infection blood-brain barrier is also possible and is supported by low-level T cell infiltrate reported in brain parenchyma of macaques infected with less aggressive SHIV and with intact barriers [16]. Here, we measured cellular infection in CSF during early acute infection directly ex vivo and observed widespread CD4+ T cell infection in all six PLWH and the majority of macaques, including evidence of active viral replication. These findings are consistent with sequence analysis of CSF viral variants during later stages of acute HIV-1 infection in humans, which identified primarily R5 T cell-tropic virus [15]. Moreover, they suggest that early HIV-1 RNA detected in the CSF is likely derived from infected CD4+ T cells residing in the CNS, rather than free virus transiting from the blood. A correlation between CSF T cell infection frequency and CSF HIV-1 RNA in future well-powered studies would support this interpretation. CD4+ T cells hosting active viral replication in the CNS during acute infection may provide an early opportunity for establishing reservoirs in non-lymphoid or neuronal cells. For example, cell-to-cell contact between infected CD4+ T cells and astrocytes, the most abundant and long-lived cell type in the brain, results in efficient HIV-1 transmission [41]. While we did not see evidence of monocyte infection in CSF (or PB) during acute infection, we were unable to probe this population in depth due to limited numbers of monocytes in CSF from both macaques and humans. Given the well-established role of myeloid lineage cells in supporting HIV-1 infection and replication in the brain during chronic infection [42][43][44][45][46][47], extending this analysis to chronic infection when myeloid cells may be more likely to harbor virus will aid in understanding their contribution to CNS pathogenesis and reservoirs.
Similar infected T cell burden between CSF and PB (and lymph nodes in macaques) during acute infection suggests equilibrated cell-associated viral replication across these anatomical sites. Genetic evidence of independent HIV-1 replication in the CNS relative to plasma as assessed by viral genomic sequencing has also been limited in primary infection [13][14][15], in contrast to later disease stages. In analyses of primary infection conducted to date, most individuals exhibit equilibrated HIV-1 sequences between blood and CSF compartments, though cases of compartmentalized CSF virus have been observed in the settings of multiple transmitted/founder viruses and beyond four months of infection. Thus, the compartmentalized HIV-1 variants detected in brains and CSF of individuals with HIV-1-associated dementia most likely reflect a process initiated after initial establishment of viral infection. Comparable CD4+ T cell infection between PB and CSF is consistent with a~2-3 log 10 difference in cellfree HIV-1 RNA levels between these compartments, as the CD4+ T cell concentration in CSF is~10 2 −10 4 -fold lower than that of PB [36]. Similar activation profiles of PB and CSF leukocytes also suggest shared inflammatory states between these sites. Taken together, our data are consistent with cell trafficking from the blood into the CSF, resulting in equilibration of infected and activated cells between these compartments during early acute infection. Related analyses of HIV-1 reservoirs in treated and untreated chronic PLWH reported a positive correlation between HIV-1 DNA levels in CSF cell pellets and HIV-1 DNA in PBMC [48], indicating that similar infected cell burdens across compartments in early infection reported here may extend to later stages as well.
Leukocyte activation and viral replication within CSF during acute HIV-1 infection likely contribute to CNS inflammation, immunopathology and HIV-1 replication [49]. In addition to increased CD8+ T cell activation as described previously [35,50,51], we report heightened activation of CSF CD4+ T cells and monocytes during acute HIV-1 infection. CSF CD4+ and CD8+ T cells were also activated during acute SHIV infection. Activated cells are more permissive to HIV-1 infection and secrete soluble inflammatory markers that increase in the CSF during early infection and are associated with neuronal injury [7,16,37]. Direct evidence of this was observed in CSF during acute HIV-1 infection in our study through elevated pro-inflammatory chemoattractants IL-10, IP-10, neopterin, and sCD163, consistent with prior reports [6,7,37], and correlation between IP-10 concentration and activated CD4+ and CD8+ T cells and monocytes. Taken together, these findings implicate these leukocytes in CNS inflammation and pleocytosis [38,[52][53][54]. We did not observe an association between cellular activation and T cell infection levels during acute SHIV infection, as has been reported previously in chronic HIV-1 infection by some studies [55][56][57]. One potential implication of this discrepancy is that cellular activation seen later in infection may be a consequence of the infected cell burden, with the important caveat that these studies do not establish causality. It is also possible CD38 expression alone may not be optimal for detecting this association, as prior associations were found primarily with HLA-DR+ CD4+ T cells and not CD38 [58].
The transient nature of PB monocyte activation and absence/low level of CSF monocyte activation in the SHIV macaque model were unexpected observations, as monocyte activation has been observed during both acute and chronic SIV infection [20,[59][60][61]. However, blood monocyte activation appears to be closely linked to uncontrolled viremia [60], and biphasic activation was reported in some SIV models, with peak activation occurring 1-2 weeks PI followed by resolution for several weeks and then another increase prior to AIDS onset [20,62]. This pattern is not inconsistent with our findings. Evaluation of additional monocyte activation markers in CSF is warranted to corroborate our CD169 findings. It is also possible that the more pathogenic nature of most SIV strains relative to SHIV-1157ipd3N4 may heighten innate immune activation.
The rhesus macaque SHIV model employed here, in two independent facilities, recapitulated several aspects of acute HIV-1 infection in humans, supporting its utility for studying early HIV-1-associated neuropathological events [16]. Widespread CSF cellular infection and immune activation, for example, were observed in both models. However, these findings were overall more prominent in people, likely due to sampling at approximately peak viral replication in plasma and CSF, while macaque CSF analyses were performed several weeks following peak plasma viremia and in the context of limited CSF viremia. Therefore, the SHIV model data reported here may underestimate events occurring earlier in the course of infection. Larger CSF cell collections from humans also enabled greater sampling depth for detection of infected cells. Few other SIV/SHIV strains have been shown to resemble early CSF findings of acute HIV-1 infection in humans. Neurovirulent SIV clones are the best characterized in the CNS compartment and while they achieve more robust viral RNA in CSF of macaques than the SHIV strain studied here [19,21], these strains exhibit accelerated neurologic disease similar to end stage disease and do not faithfully recapitulate the more subtle neurologic changes of early/acute HIV-1. Development of CNS infection models using SHIVs expressing transmitted-founder HIV-1 envelopes may more closely mimic acute HIV-1 infection in humans.
These findings add to a growing body of evidence that CD4+ T cell infection plays a central role in early viral entry of the CNS. An important implication of these findings is that even very early ART initiation may not be able to prevent establishment of CNS viral reservoirs and sources of long-term inflammation. Future studies evaluating the impact of ART on infected cell burdens in CSF will address the extent to which cell-associated virus persists during suppressed infection and investigate associations with persistent neurologic and cognitive abnormalities observed in humans despite suppressive ART. Institute of Medical Sciences (AFRIMS) for their valuable contributions to this study. We thank Mario Montero, Ming Dong, Kier Om, Ursula Tran, Desha Silsorn, Yanin Kuncharoen, Dutsadee Inthawong, and Rawiwan Im-Erbsin for expert technical assistance; Julie Ake, Dominic Paquin-Proulx, Lydie Trautmann, Hiroshi Takata, and Caroline Subra for insightful scientific discussions and input (all MHRP); Mark Lewis, Jack Greenhouse, and Jonathan Misamore (Bioqual, Inc.) for animal study execution and plasma viral load measurements; and Nancy Miller (NIAID, NIH) for the SHIV-1157ipd3N4 viral stock and scientific guidance.

Supporting information
Antiretroviral therapy for RV254/SEARCH 010 participants was supported by the Thai Government Pharmaceutical Organization, Gilead, Merck and ViiV Healthcare.
The views expressed are those of the authors and should not be construed to represent the positions of the U.S. Army or the Department of Defense (DoD). The opinions expressed in this article are the author's own and do not reflect the view of the National Institutes of Health, the Department of Health and Human Services, or the United States government.