Opioid-induced transcriptional reprogramming of cerebrospinal fluid immune cells drives neuroinflammation in SIV-infected rhesus macaques
Arpan Acharya, Anoop T Ambikan, Ujjwal Neogi, Benjamin G Lamberty, Shannon Callen, Shilpa Buch, Howard S Fox, Siddappa N Byrareddy
TL;DR
Opioid use in SIV-infected macaques alters immune cells in cerebrospinal fluid, leading to neuroinflammation and worsened brain function, even when virus is suppressed.
Contribution
This study reveals how opioids reprogram immune cells in cerebrospinal fluid, contributing to neuroinflammation in SIV-infected macaques, even under antiretroviral therapy.
Findings
Opioid use increases CD4+ TCM and Treg cells while reducing CD8+ TEM cells in cerebrospinal fluid before infection.
Chronic opioid exposure reprograms monocytes toward a DAM state, altering cell communication and signaling pathways.
Dysregulation of genes related to T-cell signaling, apoptosis, and neurodegeneration is observed in opioid-dependent animals.
Abstract
Opioid use is disproportionately high among People with HIV (PWH). Although combined anti-retroviral therapy (ART) can dampen HIV-associated dementia, a large fraction of PWH continue to experience neurocognitive deficits which are further exacerbated by opioid use. In the present study, we performed single cell transcriptomic profiling of cerebrospinal fluid (CSF) immune cells to explore their functional characteristics in opioid mediated neurological disorders among PWH using the SIV/rhesus macque model. In this study, we utilized CSF cells from morphine- and saline-administered, SIV-infected, ART-treated rhesus macaques (RMs). The CSF scRNA-Seq was performed longitudinally at baseline, post ramp-up with morphine (pre-infection), during acute infection, and after suppression of viremia to profile cell-specific transcriptomic signatures that mirror the CNS pathogenesis observed in…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsNeuroinflammation and Neurodegeneration Mechanisms · HIV Research and Treatment · Single-cell and spatial transcriptomics
Background
Although combined antiretroviral therapy (ART) has transformed Human Immunodeficiency Virus infection from a deadly disease to a manageable chronic disorder, prevalence of HIV-associated neurocognitive deficits among people with HIV (PWH) is on a rise [1]. In addition to HIV and antiretrovirals, substance use also significantly worsens neurological disease outcomes [2]. Despite substantial efforts, the molecular mechanism(s) of Neuro-HIV, including the role of viral reservoirs and HIV-associated neuroinflammation, remains incompletely understood [3]. This knowledge gap hinders the development of a therapeutic strategy for Neuro-HIV [4]. Chronic immune activation of the central nervous system (CNS) is the hallmark of HIV infection that persists even after suppressive ART [5, 6]. The cellular subset and genes responsible for this inflammatory disorder remain largely unknown, due to the inherent problem of accessing brain tissue samples. Cells of the myeloid lineage, including the CNS resident microglia, perivascular macrophages and circulating monocytes that egress from the periphery to the brain, play a crucial role in HIV-associated neuroinflammation [7, 8]. In the recent past, advances in single cell/nucleus multiomic sequencing (snRNA+snATAC) technologies have enabled the characterization of transcriptomic and epigenomic changes at the single-cell level in both health and disease [9, 10]. Despite these technological advances, recent efforts using archived brain tissues from PWH to characterize region and specific cell type-specific effects and to elucidate the molecular mechanism(s) of Neuro-HIV have had limited success [11, 12]. This can be attributed to multiple confounding factors in these studies, including heterogeneous host genetic background, use of different combinations of antiretrovirals, differences in the time interval between HIV diagnosis and initiation of ART, use of highly toxic nature of first generation antiretrovirals in a subset of PWH, withdrawal of ART before death in terminally ill PWH, and random use of different substances etc among others.
SIV/SHIV-infected non-human primates (NHP) are routinely used as a model of HIV infection for HIV cure studies, vaccine development, and for other associated comorbid conditions like HIV-associated premature aging and neurocognitive disorders [13–15]. In a recent study using morphine-dependent, SIV-infected ART-suppressed rhesus macaques (RMs), we demonstrated that the size of the SIV reservoirs in the CNS of CD11b+ microglia/perivascular macrophages was higher in morphine-dependent macaques compared to control animals [16], and that, morphine-mediated alterations of CNS myeloid cell transcriptomics potentiated SIV-associated neuropathogenesis [17]. In a follow-up study, as part of the Single Cell Opioid Responses in the Context of HIV (SCORCH) consortium [18], we demonstrated that CNS myeloid cells have a dampened anti-viral gene expression response in morphine-dependent, SIV-infected, ART-treated RMs [19]. But these studies utilizing the brain tissues are only able to capture the molecular signature of the disease pathogenesis at the study endpoint due to the inaccessibility of the brain tissue during life.
Cerebrospinal fluid (CSF) produced within the brain choroid plexus envelopes and protects the compartmentalized immune privilege site of the CNS [20, 21]. The cellular fraction of CSF derived from hematopoietic linages is distinct from that of blood, although the underlying mechanisms remain unknown [22, 23]. Conventionally, CSF is used as a surrogate marker in the diagnosis of neuroinflammatory disorders [24, 25]. During various neuroinflammatory and neurodegenerative disease states, as well as therapeutic interventions, the composition and function of CSF immune cells are considerably altered, necessitating their comprehensive characterization [26, 27]. However, in both healthy and diseased conditions, the cellular density of CSF is very low (1 to 5 cells per microliter) [28], and the volume of spinal fluid that can be collected from an individual or animal disease model is limited [29]. Therefore, the utility of immunophenotyping and functional characterization of CSF immune cells using conventional flow cytometry is limited.
Along with conventional RNA sequencing for analysis of large-scale differential gene expression analysis, single-cell RNA sequencing (scRNA-seq) technology enables scientists to identify and characterize rare cell types present in heterogeneous tissue types using an unbiased, high-throughput transcriptomic analysis of individual cells [30]. In the last decade, several studies have employed scRNA-Seq to explore the composition and functionality of CSF immune cells [31–35]. To overcome the limited accessibility of CNS tissues, recent studies have employed scRNA-seq of CSF cells from PWH, demonstrating that HIV transcripts persist in central memory CD4 + cells despite ART-mediated suppression of plasma viremia, and identifying CD204 + microglia like cells, a rare population implicated in HIV neuropathogenesis. Although most HIV transcript-positive central memory CD4 + T-cell clones were compartment specific, a subset persisted in both blood and CSF indicating trafficking of HIV reservoirs between tissue compartments [36–38].
To investigate the disease spectrum of opioid use, SIV-infection, and ART-therapy, we performed longitudinal scRNA-seq of CSF cells from morphine dependent and saline administered macaques at multiple time points: baseline, post-morphine administration, during acute SIV infection, following ART-mediated plasma and CSF viral suppression, and at necropsy after long term ART and chronic morphine exposure. Our study demonstrated that morphine-dependent animals has an increased proportion of CD4 + T_CM_ cells and Treg cells at the pre-infection time point, which was substantially higher than in saline-administered macaques. In contrast, morphine-dependent animals exhibited a marked reduction of CD8 + T_EM_ cell population at the pre-infection time point relative to baseline, with decline persisting during acute infection. Additionally, similar to our earlier reports on morphine-mediated Th1/Th2 imbalance in peripheral blood of RMs [16], we also observed Th1/Th2 imbalance in CSF CD4 + T-cell population, characterized by Th2 dominance. Gene set enrichment analysis (GSEA) showed dysregulation of genes involved in oxidative phosphorylation (OXPHOS) and neurodegenerative pathways, with longitudinal shifts in contributions from distinct immune cell populations. Furthermore, cell-cell communication analysis, indicated differences in both the number and strength of interactions in morphine-dependent RMs compared to saline-administered animals. Notably, signaling pathways mediated by chemokines of the CCL family (CCL signaling) consistently differed between the groups, with varying contributions from different cell population-specific contributions over time. In future, an integrated comparative analysis of CSF, peripheral blood and the brain region-specific scRNA-seq datasets could identify key therapeutic targets and biomarkers for the identification/monitoring of disease pathogenesis in the context of opioid use among PWH.
Material and Methods
Animal samples and ethical statement
Indian origin, outbred, pathogen-free male rhesus macaques were used in this study as reported earlier [16, 19]. The macaques were housed in an AAALAC-approved facility at the Department of Comparative Medicine, University of Nebraska Medical Center (UNMC), Omaha, Nebraska, USA. Animals were maintained in a temperature-controlled (72°F) indoor climate with a 12-hour light/dark cycle. The monkeys were observed twice daily for the development of distress or disease by the animal care staff and veterinary personnel. The animals were daily fed monkey diet (Purina) supplemented with fresh fruit or vegetables and water ad libitum. At the end of the study, all RMs were humanely euthanized using a high dose of ketamine/zylazine and then opened the thoracic cavity and perfused/exsanguinated according to the guidelines of the American Veterinary Medical Association. This study was reviewed and approved by UNMC Institutional Animal Care and Use Committee (IACUC) under protocol number ‘16-073-07-FC’.
Study design.
The details of the original study were published earlier [16, 19], in which 11 rhesus macaques (RMs) were included. Six of them were administered morphine, and five of them received saline and served as controls. CSF samples were collected at the indicated time points (Supplementary Fig. 1) and processed for scRNA-seq as described below.
CSF single cell suspension preparation.
CSF was collected in a 2 ml conical tube on ice by cisterna magna puncture with a 22-gauge needle after anesthetizing the macaques with ketamine-HCl (10 mg/kg). The sample was centrifuged at 2000 rpm for 15 min at 4° C, CSF was aspirated, and the cell pellet was resuspended in 50 μl plain RPMI, and immediately transferred to UNMC genomic core for cell capture using 10x genomics assay technology. At longitudinal time points (baseline, pre-infection, acute infection, and after viral suppression) cells from the same group were pulled together before processing, while at necropsy, cells from individual animals were processed for single-cell sequencing.
Single cells capture and RNA sequencing.
Cell suspensions were evaluated by light microscopy for debris and viability, and given the relatively low counts of cells, the remaining cells (46 μl) was loaded onto the 10x Genomics Chromium GEM Chip and inserted into the Chromium controller. Single cells were captured, lysed, and RNA was reverse transcribed and barcoded using a 10x Genomics Chromium instrument and the Chromium Single Cell 3’ Reagent Kit v3 reagents. (10x Genomics, Pleasonton, CA). Illumina compatible sequencing libraries were generated using the resultant cDNA following the recommendations of the manufacturer. Next the cDNA underwent fragmentation followed by A-tail repairing and a double-sided bead cleanup. After that adapters are ligated to the cDNA fragments and the fragments are PCR amplified using unique sample index primers per manufacturer’s recommendations. Libraries were quantified by qPCR using the KAPA Library Quant Kit (Illumina) from KAPA Biosystems (Roche, Pleasonton, CA). Libraries were loaded on an Illumina flow cell at a concentration of 1.3 pM and sequenced using NovaSeq 6000 sequencer.
Flow cytometry evaluation of CSF cells
CSF was processed as described above and the cell pellet was resuspended in 50 μl PBS. After that, a 1:1,000 dilution of live/dead dye was added to the cell suspensions and incubated for 30 min in the dark. Then the cells were washed, and Fc blockade was carried out using polyclonal anti-human Fc receptor binding inhibitor (eBioscience, USA). Next, surface staining was performed using anti-CD45, anti-CD3, anti-CD4, anti-CD8, anti-CD20, anti-CD14, anti-CD16, anti-HLADR antibodies. Cells were incubated for 1 h at room temperature in the dark. Thereafter, cells were fixed using a 2% paraformaldehyde (PFA) solution for 30 min. All events were acquired using an LSR-II FACS analyzer (BD Biosciences), and data were analyzed using FlowJo version 8. The details of the antibodies used in this study are described in Supplementary Table 1.
Bioinformatic analyses
The sequencing samples were demultiplexed using the 10x Genomics cellranger v.3.1.0 analysis pipeline. The Macaca mullatta reference genome (Mmul 10) downloaded from Ensembl was used as a reference. Genes annotated as protein-coding or lncRNA are selected as the reference set. The genome of Simian Immunodeficiency virus (SIV) (Genbank M33262.1) was also incorporated into the reference genome to detect viral transcripts. The SIV genome was divided into five segments to avoid multi-mappings as described previously [39]. The count matrices of all samples were then separately merged using aggr utility in cellranger v6.1.2 [40].
All the downstream analysis were performed using R package Seurat v5.0.1 [41]. The doublets were removed using R package DoubletFinder v2.0.4 [42]. Then low-quality cells and empty droplets were removed from the data set (min.features < 100 and min.cells < 3). The number of cells per group was determined based on the number of unique cellular barcodes detected. This was followed by the number of UMIs/transcripts per cell and the number of genes/cells determined. Further, cells expressing abnormally high and low numbers of genes were removed (nGene > 3000 and nGene < 500). The data were then normalized using SCtransform [43, 44], and perform cell clustering using Uniform Manifold Approximation and Projection (UMAP).
The number of principal components (PC) for UMAP was computed by generating an elbow plot of standard deviations of each PCs and the threshold was chosen by taking the data point where the percent of change in variation between the consecutive PCs is less than 0.1%. Cell clusters exhibiting similar gene expression patterns were merge and the final cell type annotation was done based on cell type-specific marker gene expression. Cell types were annotated using the R package Azimuth v0.5.0 [45] and manually confirmed based on the canonical marker gene expression of the assigned cell types. Next, differential gene expression analysis was performed to find genes significantly regulated in each cell cluster using the Model-based Analysis of Single-cell Transcriptomics (MAST) method [46]. Statistical significance was determined using a False Discovery Rate (FDR) of 5%.
Cell-cell communication analysis
Intercellular communication networks are quantitatively inferred and analyzed using the R package CellChat v2.1.2 [47, 48]. CellChat utilizes network analysis and pattern recognition methodologies to anticipate primary signaling inputs and outputs within cells, elucidating how these cellular components and signals synchronize to perform various functions.
Results
Characterization of CSF cells in morphine-and saline-administered SIV-infected, ART-treated rhesusmacaques.
This study included samples from two cohorts of macaques with an identical study design. The longitudinal samples were part of the study described in SupFig. 1A [19], and the necropsy time point samples were from another study previously reported by our group [16]. The details of the macaque cohorts are described in Supplementary Table 2. As illustrated in SupFig. 1B-D, and reported by us previously [16], we did not observe any significant differences in plasma and CSF viral loads between the morphine and saline-administered macaques.
The scRNA-seq workflow is described in material and Methods section (Fig. 1A). After initial QC, a total of 8762 cells were pooled together for unsupervised cluster analysis. Based on the UMAP projection, initially 22 cell clusters were identified in the UMAP. After merging similar cell clusters based on gene expression pattern, 13 major cell clusters were identified, and cell types were assigned based on cell-specific canonical marker gene expression as outlined in Fig. 1B. Additionally, a heatmap displaying the top ten differentially expressed genes in each cluster is displayed in SupFig. 2, the expression of selected canonical marker genes used for cell type annotations are displayed on the UMAP feature plots (SupFig. 3). The cell types identified includes, B cells (0.98%), CD14 + monocytes (7.24%), CD16 + monocytes (2.72%), CD4 + central memory T cells (CD4 + T_CM_; 46.59%), CD4 + effector memory T cells (CD4 + T_EM_; 5.14%), CD8 naïve (2.39%), CD8 + central memory T cells (CD8 + T_CM_; 1.85%), CD8 + effector memory T cells (CD8 + T_EM_; 19.21%), dendritic cells (DC; 1.28%), double negative T cells (dnT; 8.24%), mucosal-associated invariant T cells (MATI; 1.15%), NK cells (2.24%), and regulatory T cells (Treg; 0.99%) respectively (Fig. 1C). The percentage of each cell type out of the total cells analyzed was included in parentheses. These CSF cell compositions are in line with the proportion of cells present in the CSF of HIV-infected, ART-treated patients with suppressed viremia [37, 38]. Distinct from the human cohort studies, we did not detect the presence of SIV transcripts in our studies, most probably due to the overall low number of cells recovered from the macaque CSF. The proportion of each cell type in saline and morphine-dependent animals at baseline, pre-infection, acute infection, after viral suppression, and at necropsy time point is illustrated in the stacked bar plots (Fig. 1D). There was a spike in the frequency of CD4 + T_CM_ cell population in the morphine-dependent animals at the pre infection time point (58.99%) from the baseline (41.48%), which was substantially higher from the saline-administered macaques (46.04%) (Fig. 2A). Distinct from this, there was a noticeable reduction in the frequency of CD8 + T_EM_ cell population in morphine-dependent animals at the pre infection (16.61%) time point from baseline (26.37%), which continues to fall at acute infection (9.19%) before recovery post initiation of the ART (Fig. 2B). It is known that chronic use of morphine increases the frequency of circulating Treg cells in the peripheral blood [49]. Surprisingly, in the CSF of the morphine-dependent macaques, we observed similar trends of an increase in the frequency of Treg population (3.28%), which came back to the baseline level (0.28%) post SIV infection and initiation of ART (Fig. 2C). During necropsy, in a subset of animals from cohort 2 (saline: n = 3; morphine: n = 5), we performed flow cytometry analysis of the CSF cells to compare them with the single cell transcriptomic data. The gating strategy for flow cytometry data analysis is illustrated in SupFig. 4. Similar to the single cell transcriptomic data, we detect the presence of major immune cells in CSF including T cells, B cells, NK cells, and three different subsets of monocytes (classical: CD14 + CD16-/HLA-DR+; intermediate: CD14 + CD16+/HLA-DR+; and nonclassical: CD14-CD16-/HLA-DR+) in all animals without a significant difference in their frequency in morphine versus saline administered macaques (SupFig. 5).
Chronic use of morphine resulted in a Th1 to Th2 phenotypic shift in the circulating T-cell population [50]. Similarly, in our earlier study, we reported a significant reduction in CD4 + Th1 cells in circulation as well as in the lymph nodes in morphine-dependent, SIV-infected, ART-treated macaques compared to the saline-administered controls [16]. Herein, we looked at the Th1 and Th2 populations out of total CD4 + and CD8 + T cells in the CSF. Canonical markers CXCR3, CCL5, IFNG, TBX21, and GZMK used for Th1, and GATA3, IL5, PLAC8, IGFBP7, and IL4 used for Th2 classifications. The UMAP plot showing the distribution of Th1 and Th2 populations is illustrated in (Fig. 3A), and the proportion of CD4 + and CD8 + Th1/Th2 cells in saline- and morphine-dependent animals at baseline, pre-infection, acute infection, after viral suppression, and at necropsy time point is illustrated in the stacked bar plots (Fig. 3B). At necropsy, we observed a substantially lower frequency of CD4 + Th1 cells in the CSF of morphine-dependent macaques compared to saline-administered animals (4.1% versus 10.5%); whereas the frequency of CD4 + Th2 cells were 12.2% and 7.2% in morphine versus saline-administered animals, resulting in a Th1:Th2 ratio of 0.33 vs 1.46. Overall, like PBMCs and LNs, in CSF, we observed a Th1/Th2 imbalance with a dominance of the Th2 population in morphine-dependent, SIV-infected, ART-treated macaques (Fig. 3B).
Gene expression analysis revealed cell-type-specific regulation of signaling pathways
Differential gene expression analysis showed cell type-specific up- and down regulation of genes correlating with different stages of disease pathogenesis. The total number of differentially expressed genes (DEGs) in each cell type and the number of overlaps and uniquely regulated genes at pre-infection, acute infection, after viral suppression, and at necropsy time point are illustrated in an upset plot (SupFig. 6 A-D). At morphine post-administration, pre-infection timepoint, gene set enrichment analysis (GSEA) showed dysregulation of multiple genes involved in C-type lectin receptor signaling pathways, apoptosis, platelet activation, Th1/Th2 differentiation, T-cell receptor signaling pathways, Th17-cell differentiation, HIV infection, Ras signaling pathways, and cellular senescence in CD4 + T_CM_ cells. In CD4 + T_EM_ cells on the other hand, we observed significant dysregulation of genes involved in the PI3K-Akt signaling pathway, ECM-receptor interaction, and hematopoietic cell lineage. Next, during the acute infection phase, GSEA revealed dysregulation of multiple genes involved in OXPHOS, retrograde endocannabinoid signaling, Prion disease, Alzheimer’s disease, Huntington disease, Parkinson’s disease, and Amyotrophic lateral sclerosis in majority of the cells, including CD16 + monocytes, NK cells, CD4 + T_CM_ cells, CD8 + T_CM_ cells, and CD8 + T_EM_ cells. Surprisingly, even after complete suppression of plasma and CSF viremia, majority of these pathways remained dysregulated; although the contribution of the cell types shifted to CD4 + T_CM_ cells, CD8 + T_CM_ cells, and CD8 + T_EM_ cells only. Finally, at necropsy during the chronic stage of SIV infection and long-term dependence on morphine, GSEA identified dysregulation of genes involved in similar pathways as observed at earlier timepoints, but in this case driven by CD14 + monocytes. Additionally, we also observed dysregulation in the Toll-like receptor signaling pathway, apoptosis, and Platelet activation mediated by CD14 + monocytes (Fig. 4). Additionally, at the necropsy time point, we observed several diseases associated microglia (DAM) genes differentially regulated within the myeloid cell clusters between morphine versus saline-administered RMs. There are 8 DAM-associated genes significantly up-regulated in the morphine group, which includes FCER1G (5.31-fold increase), PLAUR (4.79-fold increase), CSTB (4.11-fold increase), ARRDC4 (3.55-fold increase), NPC2 (3.42-fold increase), FUCA1 (3.08-fold increase), TREM2 (2.26-fold increase), and SOAT1 (2.00-fold increase). On the other hand, we observed 6 genes significantly down-regulated in morphine-dependent animals, which included SYNGR1 (11.46-fold decrease), LGALS3BP (3.12-fold decrease), MAFF (2.96-fold decrease), ATF3 (2.88-fold decrease), CXCR4 (2.84-fold decrease), and RAMP1 (2.06-fold decrease). In summary, these data indicated that the majority of the macrophage/myeloid cells present in CSF express the canonical brain-specific DAM genes observed during chronic immune activation responsible for neurodegenerative process. Overall, we observed morphine mediated CD4 + T_CM_ and CD4 + T_EM_ cell dysregulation in CSF of RMs impacting the TCR signaling and Th1/Th2 differentiation, which was transient to CD16 + monocytes, NK cells, and CD4 + T cells mediated energy metabolism dysregulation and genes involved in multiple neurodegenerative disease pathways at acute infection, which was not completely resolved post ART.
Morphine-mediated alterations in CSF cell-cell communication in SIV-infected ART-treated macaques
Next, we performed cell-cell communication analysis using CellChat based on the gene expression data, to examine the differential communication patterns across cell types between morphine administered and control study timepoints. The differential number of interactions between different cell types in morphine-dependent RMs compared to control animals is illustrated in Fig. 5A–D, and the differential number of interactions is depicted in Fig. 5E–F. At the pre-infection time point, we not only observed a relative increase in the number of interactions between CD16 + monocytes and other cell types in morphine-dependent RMs (Fig. 5A), but also that the strength of the interactions was relatively stronger between all cell types (Fig. 5E). At acute infection, the relatively higher number of cell-cell interactions was dominated by CD8 + T_CM_, CD8 + T_EM_ cells, and CD14 + monocytes (Fig. 5B), whereas the relatively stronger cell-cell interaction strength was mainly mediated by CD8 + T_EM_ cells and CD14 + monocytes in morphine-dependent macaques (Fig. 5F). After ART-mediated viral suppression, during the chronic stage of disease, differentially higher number of cell-cell interactions (Fig. 5C–D) along with the relatively higher strength of interaction (Fig. 5G–H) was primarily dominated by CD14 + and CD16 + monocytes in morphine-dependent macaques.
Next, using CellChat by comparing the informational flow for the signaling pathways between morphine and saline-administered RMs, which is the sum of communication probability among all pairs of cell types for the specific pathway, the conserved pathways between the groups are identified at the indicated study timepoints. As illustrated in Fig. 6A, CD39, TRAIL, NECTIN, LIGHT, Prostaglandin, ANNEXIN, LAMININ, and CD48 signaling communication pathways are turned off while the SPP1, IL16, and Cholesterol pathways are turned on in morphine dependent macaques. Some pathways, like TGFß are increased in morphine dependent RMs compared with control, whereas some pathways, like CCL remain comparable between groups. After that, based on the computed four network (sender, receiver, mediator, influencer) centrality measurements, we measured the relative importance of each cell types present in the CSF for the specific signaling pathways mentioned above (SupFig. 7). From the heatmaps it is evident that SPP1 is only turned on in morphine dependent RMs, whereas for TGFß and CCL signaling pathways, the relative contribution of individual cell types differs substantially between the groups. At acute infection, PECAM1, DHT, NCAM, cholesterol, IL-16, TGFß, and APP signaling communication pathways are turned off, while the prostaglandin, SELPLG, CD48, CSF, and TRAIL pathways are turned on in morphine dependent macaques. Some pathways, like CCL are increased in morphine-dependent RMs compared with control (Fig. 6B). The heatmap illustration of the relative importance of each cell types present in the CSF for CD48 and CCL signaling pathways is shown in SupFig. 8, from which it is evident that CD48 is only turned on in morphine-dependent RMs, whereas for CCL signaling pathways, the relative contribution of individual cell types differs substantially between the groups. After the viral suppression time point, NCAM, cholesterol, TGFß, prostaglandin, among other signaling communication pathways, are turned off while the CSF, CD80, LT, and other pathways are turned on in morphine-dependent macaques. Some pathways, like CCL are increased in morphinedependent RMs compared with controls (Fig. 6C). The heatmap illustration of the relative importance of each cell type present in the CCL signaling pathways is shown in SupFig. 9, from which it is evident that the relative contribution of individual cell types differs substantially between the groups. At necropsy, post-long term ART mediated suppression of plasma and CSF viremia and chronic exposure to morphine, PCDH, FN1, CypA signaling communication pathways among others are turned off, while the MIF, CD48, ICAM, CD45 and other pathways are turned on in morphine-dependent macaques. Some pathways like CCL, ApoE are increased in morphine-dependent RMs compared with controls, whereas pathways like APP remain comparable between the groups (Fig. 6D). The heatmap illustration of the relative importance of each cell types present in the CD48, CCL, ApoE, and APP signaling pathways is shown in SupFig. 10, from which it is evident that the CD48 is only on for morphine-dependent macaques, for APP the relative contributions were similar for both groups, while for ApoE and CCL relative contributions of individual cell types differ substantially between the groups.
Next, by computing the Euclidean distance, we measured the dissimilarity between the shared signaling pathways between the morphine- and saline-administered animals at different time points of the study. The signaling pathways, having a larger Euclidean distance than the others, are implicated to be substantially altered in morphine-dependent RMs compared with controls. At pre infection, the complement signaling pathways have the highest Euclidean distance from the others (Fig. 7A). At acute infection, ADHRE and CCL signaling pathways have considerably higher Euclidean distance from the others (Fig. 7B). After viral suppression, RESISTIN and CCL signaling pathways has considerably higher Euclidean distance from the others (Fig. 7C), and at necropsy ApoE signaling pathways has the highest Euclidean distance from the others (Fig. 7D).
We also computed the outgoing signaling patterns by leveraging pattern recognition approaches in CSF cells of the morphine- and saline-administered RMs. The outgoing signaling patterns indicated how sender cells coordinated among themselves and certain signaling pathways to regulate cell-cell communications. At pre-infection time point, CD39, Prostaglandin, and CD48 outgoing signaling pathway communication pathways were downregulated in morphine-dependent macaques (Fig. 8A). During acute infection, outgoing prostaglandin signaling pathway communication was upregulated in morphine-dependent macaques (Fig. 8B). After viral suppression, cholesterol, TGFß, prostaglandin outgoing signaling pathways communication were found to be downregulated in morphine-dependent macaques (Fig. 8C). While at necropsy, FN1, and CypA signaling pathway communications were downregulated, while the MPZ, CD48, ICAM, and CD45 signaling pathway communications were upregulated in morphine-dependent macaques (Fig. 8D). Among the other outgoing significant communication signaling pathways, although the overall signaling strengths were similar between the groups, the relative contribution of individual cell population varied considerably, indicating morphine-mediated alterations of outgoing signaling patterns from the sender cells. Moreover, these findings are in line with the overall informational flow reported in Fig. 6.
Next, we performed cell-cell ligand-receptor analysis across the cell types longitudinally at different stages of the disease. At pre infection time, we observed a strong interaction involving APOE–TREM2, SPP-ITGA4, SPP-ITGA1, TGFB1–(TGFBR1 + TGFBR2), CCL3 – CCR1 in morphine dependent macaques (Fig. 9A). At acute infection, we observed a strong interaction involving CCL8 – CCR1, CLEC2B-KLRG1, MIF–(CD74 + CXCR2), SELPLG–SELL, and SEMA4A-PLXNB2 in morphine-dependent macaques (Fig. 9B). After viral suppression time point, we observed a strong interaction involving RETN–TLR4, RETN–CAP1, APP–SORL1, CCL2 – CCR2, and CCL8 – CCR2 in morphine-dependent macaques (Fig. 9C). Finally at necropsy time point, we observed a strong interaction involving TGFB1–(TGFBR1 + TGFBR2), RETN–TLR4, RETN–CAP1, MIF–(CD74 + CXCR4), MIF–(CD74 + CD44), CCL3 – CCR1, APP–SORL1, APP–(TREM2 + TYROBP), APOE – TREM2 in morphine-dependent macaques (Fig. 9D).
Discussion
In this study using scRNA-seq analysis, we delineated the CSF immune cell landscape in morphine-dependent, SIV-infected, ART-treated RMs across the disease spectrum. We observed a morphine-mediated relative increase in CD4 + T_CM_, T_reg,_ and a reduction in CD8 + T_EM_ cell population before SIV infection. Chronic use of morphine imparted a Th1/Th2 T-cell imbalance with a dominance of the Th2 population. In morphine-dependent RMs, there as dysregulation of genes involved in T-cell receptor signaling pathways, apoptosis, PI3K-Akt signaling pathway, cellular senescence, OXPHOS, and multiple neurodegenerative disorders, including Alzheimer’s, Huntington’s and Parkinson’s diseases, and Amyotrophic lateral sclerosis, among others. Surprisingly, the contribution of different cell populations in this process keeps evolving along with different stages of the disease pathogenesis. Additionally, cell-cell communication analysis revealed morphine mediated differential numbers of interactions and strength of interaction in ligand-receptor pairs in different cell populations. The cell-cell communication involving CCL signaling pathways remained differentially regulated in morphine-dependent RMs compared to controls.
The use of the current generation of ART has significantly reduced the incidence of HIV-associated dementia (HAD). However, PWH continued to suffer from neurocognitive impairments, which are exacerbated by comorbid conditions like substance use disorders [51, 52]. In the quest to identify a biomarker of Neuro-HIV, we performed scRNA-Seq analysis of CSF immune cell populations from SIV-infected ART-treated rhesus macaques, a widely used model for HIV cure research and vaccine development studies. We detected all major cell types in the CSF, which included T cells, B cells, Monocytes, Dendritic cells, and NK cells. We observed a substantial increase in the central memory CD4 + T cells (T_CM_) in the CSF of morphine-dependent animals, and these trends continued until the start of the ART initiation (acute infection time point). CD4 + T_CM_ cells serve as the primary reservoirs of HIV/SIV. Their expansion at the time of infection and persistence until the initiation of ART could result in a higher proportion of virally infected cells in morphine-dependent RMs compared to controls. As reported previously, during routine CNS surveillance, HIV/SIV infected T_CM_ cells could potentially seed the CNS reservoirs, including microglia, perivascular macrophages, and possibly CNS-resident CD4 + T cells [53]. This could be a possible mechanism for observing a higher level of replication-competent SIV reservoirs in the CNS of morphine-dependent RMs compared to the controls, as reported in our previous study [16]. These findings are in line with the recent report from our group demonstrating CD4 + cytotoxic T-cell-mediated CNS entry of SIV [54].
In our previous study, we reported opioid-mediated Th1 to Th2 polarization of T cells in PBMC and lymph nodes of SIV-infected, ART-treated RMs. Herein, using the scRNA-seq data, we observed a substantial reduction in Th1 + CD4 T-cell population in morphine-dependent RMs vs. saline-administered controls, with a subsequent increase in Th2 + CD4 T-cell population. As it is well established that in the periphery, clonal expansion of the memory Th1 + T cells plays a crucial role in maintaining replication-competent HIV reservoirs [55], the role of this opioid-mediated Th1 to Th2 polarization in CSF immune cells, which continuously surveil the CNS in modulating the brain HIV/SIV reservoirs, remains to be identified. Remarkably, it has been reported that in aged mice, Th2 cells keep accumulating within the choroid plexus, resulting in an overproduction of IL-4 that prompts the choroid plexus epithelial cells to generate the chemokine CCL11 [56]. This increase in CCL11 accumulation has been linked with the agingassociated cognitive impairments [57–59]. It can thus be envisioned that chronic morphine-mediated accumulation of Th2 cells in the choroid plexus could likely be a potential cause of opioid mediated potentiation of cognitive impairments among PWH; needing further investigations.
There were several DAM genes significantly up-regulated (FCER1G, PLAUR, CSTB, ARRDC4, NPC2, FUCA1, TREM2 and SOAT1) or down-regulated (SYNGR1, LGALS3BP, MAFF, ATF3, CXCR4 and RAMP1) in morphine-dependent macaques versus controls. The dysregulated DAM genes observed in morphine-administered animals could likely be linked with various CNS-associated pathologies. For example, elevation of CSF soluble TREM2 has a linkage with CNS microglia/macrophage activation and neuronal injury in untreated HIV patients [60]. FCER1G codes for Ig-E Fc receptors γ-subunit, responsible for phagocytosis and microglia activation [61]. It plays a significant role in aging, neurodegeneration, and has a link to Alzheimer’s Disease (AD) [62–64]. Down-regulation of FGER1G is associated with lower activation/inflammation and HIV reservoirs in elite controllers [65]. The PLAUR codes for urokinase plasminogen activator receptor (uPAR), and there is a correlation between PLAUR expression and immunosuppressive checkpoint markers [66]. uPAR is present in the plasma membrane of CNS infiltrating macrophages, whereas the presence of soluble uPAR in the CSF is associated with multiple CNS pathologies [67]. These findings are supported by in vitro observations wherein the urokinase plasminogen activator, a natural ligand for uPAR, is able to block HIV replication [68]. NCP2 codes for a protein involved in intracellular cholesterol trafficking. Abnormalities in this pathway are responsible for the accumulation of cholesterol and other lipids in endosomal/lysosomal compartments [69]. FUCA1 codes for a lysosomal enzyme α-l-Fucosidase 1 (FUCA1) that takes part in the enzymatic degradation of amyloid β [70]. SOAT1 codes for Sterol O-acyltransferase, which is implicated in the formation of beta-amyloid and atherosclerotic plaques. Further, HIV TAT protein induced up-regulation of SOAT1 leads to beta-amyloid deposition in neurons [71]. On the other hand, down-regulation of SYNGR1 is implicated in schizophrenia (SCZ) and bipolar disorder (BPAD) [72] and CXCR4 is associated with age-associated neurodegeneration [73], LGALS3BP codes for Galectin 3 Binding Protein (Gal-3BP). It stimulates host defense against viral infection. Gal-3BP suppresses TAK1-dependent NF-κB activation, and its level decreases with progression of AD [74, 75]. Toll-like receptor (TLR) ligands induce ATF3 and negatively regulate TLR signaling associated with innate immune response. Down-regulation of ATF3 results in overexpression of the chemokine CCL2, which, in turn, results in the activation of the TLR4/NF-κB signaling [76]. Distinct from circulating monocytes, within the majority of the cells, the NK cell coreceptor ligand CD48 is overexpressed in morphine-dependent animals compared to controls, but was found to be downregulated during the progression of HIV infection [77].
Cell-cell communication analysis demonstrated that at pre-infection, strong interactions involving APOE–TREM2 signaling pathway were present in CD14 + monocytes, which serves as a master regulator of microglial functional phenotype in neurodegenerative diseases [78]. Moreover, there was a strong interaction involving SPP-(ITGA1 + ITGA4) signaling pathways between CD14 + monocytes and CD4 + T_CM_, CD4 + T_EM_, and CD8 + T_EM_ linked to inflammation, cell migration, and implicated in chronic neuroinflammation [79, 80]. Furthermore, TGFB1 – TGFBR1/2 is known to cause tissue injuries by promoting inflammation and tissue remodeling [81]. Additionally, there was a strong ligand-receptor interaction involving CCL3-CCR1 signaling pathway in morphine dependent RMs, which drives immune cell infiltration in the CNS and results in chronic neuroinflammations and has a role in neurodegenerative disorders [82]. During acute infection, we observed strong intracellular CCL8 – CCR1 interaction in CD14 + monocytes among morphine dependent macaques, which promotes recruitment of inflammatory cells to the site of infection and results in tissue fibrosis [83]. Moreover, binding of CCL8 with CCR1, reduces the sensitivity of μ-opioid receptors towards opioids known as desensitization [84].
At necropsy, after long-term viral suppression and chronic use of morphine, we observed strong ligand receptor interactions involving APOE–TREM2 and TGFB1 – TGFBR,1/2 similar to pre-infection, as well as MIF–(CD74 + CXCR4), MIF–(CD74 + CD44), CCL3 – CCR1. The binding of pro-inflammatory cytokine CCL3 to its cognate receptor CCR1 to recruit immune cells to the CNS during chronic neuroinflammations [85]. Macrophage migration inhibitory factor (MIF) promotes inflammatory responses in astrocytes through activation of CD74 [86]. Whereas MIF signals through CXCR4 result in the activation of NF-κB, causing chronic inflammation [87].
The limitation of this study includes a limited number of live cells recovered from each RMs due to restrictions in the maximum volume of CSF that can be collected from RMs during life, which necessiated pooling CSF cells from the same group for single-cell analysis. Moreover, in the future, a comparative analysis of peripheral blood and CSF would be interesting to understand the impact of opioid use on circulating immune cells versus immune cells in the CSF that continuously traffic the CNS. Despite these limitations, our unique RM single-cell dataset from CSF provides the basis for new research approaches investigating the utility of circulating immune cells in CSF as a biomarker of neuroinflammation and Neuro-HIV in the context of opioid use. Although these data need further validation using a large number of animals, and comparative analysis with various regions of the brain, single-cell data sets. Our data suggested that future use of scRNA-seq technology and multiomic data analysis has the potential to predict microglia gene expression abnormalities using CSF immune cells as a biomarker without brain biopsies.
In summary, to the best of our knowledge, this is the first study to perform longitudinal scRNA-seq analysis of CSF immune cells from SIV-infected, ART-treated RM in the context of opioid use. Our findings demonstrate the presence of all major immune cell populations in the CSF, a relative increase in CD4 + T_CM_ and T_reg_ populations following morphine administration, and a Th1/Th2 imbalance characterized by Th2 dominance after chronic morphine exposure, which may exacerbate HIV/SIV-associated neurocognitive impairments. It further demonstrate that chronic morphine exposure reshapes the cellular composition, transcriptomic programs, and intercellular communication networks of CSF immune cells in SIV-infected, ART-treated macaques. The emergence of disease-associated microglia-like signatures within monocyte populations highlights a previously unrecognized mechanism by which opioids could exacerbate neuroinflammation and neurodegenerative processes despite viral suppression, offering new insights into how the persistent burden of neurocognitive impairment can recapitulate what happens in PWH. Furthermore, cell type-specific dysregulation was associated with OXPHOS pathways and multiple neurodegenerative disorders. Finally, cell-cell communication analysis revealed strong ligand-receptor interactions that promote immune cell infiltration into the CNS and neuroinflammation in morphine administered animals.
Supplementary Material
Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.
SupplementaryFigurelegends.docx
OnlineMorphineCSFSINGLECELLMANUSCRIPTSupFig1.png
OnlineMorphineCSFSINGLECELLMANUSCRIPTSupFig10.png
OnlineMorphineCSFSINGLECELLMANUSCRIPTSupFig2.png
OnlineMorphineCSFSINGLECELLMANUSCRIPTSupFig3.png
OnlineMorphineCSFSINGLECELLMANUSCRIPTSupFig4.png
OnlineMorphineCSFSINGLECELLMANUSCRIPTSupFig5.png
OnlineMorphineCSFSINGLECELLMANUSCRIPTSupFig6.png
OnlineMorphineCSFSINGLECELLMANUSCRIPTSupFig7.png
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Weiss JJ, Calvi R, Naganawa M, Toyonaga T, Farhadian SF, Chintanaphol M, Chiarella J, Zheng MQ, Ropchan J, Huang Y, Preliminary In Vivo Evidence of Reduced Synaptic Density in Human Immunodeficiency Virus (HIV) Despite Antiretroviral Therapy. Clin Infect Dis. 2021;73:1404–11.34050746 10.1093/cid/ciab 484PMC 8528400 · doi ↗ · pubmed ↗
- 2Murphy A, Barbaro J, Martinez-Aguado P, Chilunda V, Jaureguiberry-Bravo M, Berman JW. The Effects of Opioids on HIV Neuropathogenesis. Front Immunol. 2019;10:2445.31681322 10.3389/fimmu.2019.02445 PMC 6813247 · doi ↗ · pubmed ↗
- 3Chan P, Hellmuth J, Spudich S, Valcour V. Cognitive Impairment and Persistent CNS Injury in Treated HIV. Curr HIV/AIDS Rep. 2016;13:209–17.27188299 10.1007/s 11904-016-0319-7PMC 4977199 · doi ↗ · pubmed ↗
- 4Wang Y, Liu M, Lu Q, Farrell M, Lappin JM, Shi J, Lu L, Bao Y. Global prevalence and burden of HIV-associated neurocognitive disorder: A meta-analysis. Neurology. 2020;95:e 2610–21.32887786 10.1212/WNL.0000000000010752 · doi ↗ · pubmed ↗
- 5Spudich SS. Immune activation in the central nervous system throughout the course of HIV infection. Curr Opin HIV AIDS. 2016;11:226–33.26760827 10.1097/COH.0000000000000243 PMC 4773903 · doi ↗ · pubmed ↗
- 6Tozzi V, Balestra P, Bellagamba R, Corpolongo A, Salvatori MF, Visco-Comandini U, Vlassi C, Giulianelli M, Galgani S, Antinori A, Narciso P. Persistence of neuropsychologic deficits despite long-term highly active antiretroviral therapy in patients with HIV-related neurocognitive impairment: prevalence and risk factors. J Acquir Immune Defic Syndr. 2007;45:174–82.17356465 10.1097/QAI.0b 013e 318042 e 1ee · doi ↗ · pubmed ↗
- 7Herz J, Filiano AJ, Smith A, Yogev N, Kipnis J. Myeloid Cells in the Central Nervous System. Immunity. 2017;46:943–56.28636961 10.1016/j.immuni.2017.06.007PMC 5657250 · doi ↗ · pubmed ↗
- 8Greter M, Lelios I, Croxford AL. Microglia Versus Myeloid Cell Nomenclature during Brain Inflammation. Front Immunol. 2015;6:249.26074918 10.3389/fimmu.2015.00249 PMC 4443742 · doi ↗ · pubmed ↗