Human peripheral blood mononuclear cells (PBMCs) were obtained from leukopaks of healthy donors (New York Blood Center and Texas Gulf Coast Regional Blood Center) using Ficoll-Paque PLUS (Fisher Scientific) density gradient centrifugation. CD14⁺ monocytes were magnetically labeled using CD14 MicroBeads (Miltenyi) and positively selected with magnetic nanoparticle cell separation (STEMCELL). CD14⁺ monocytes were resuspended at 2×10⁶ cells/ml in monocyte media consisting of RPMI 1640 supplemented with 10% human AB serum (Sigma), 5% fetal bovine serum (FBS) (Gibco, Thermo Fisher), 1% HEPES (Teknova), 1% Pen-Strep (Gibco, Thermo Fisher), and 10 ng/ml macrophage colony-stimulating factor (M-CSF) (Peprotech). Cells were cultured non-adherently for 3 days at 37°C with 5% CO₂ in Teflon-coated flasks (Thermo Fisher) to increase the population of CD14⁺CD16⁺ monocytes, as previously described (16, 17, 19, 30). In our previously published study (31), uninfected monocytes from two donors were matured in vitro and sorted for CD14⁺CD16⁺ intermediate monocytes, followed by scRNA-seq analysis. While this analysis was performed on cell populations highly enriched for intermediate monocytes, there may have been a very small percentage of non-classical monocytes after sorting. In the current study, this previously published scRNA-seq data set was re-analyzed and used for DEG analysis to identify Group 1 and 2 monocytes. Monocytes from new sets of donors were matured in vitro to obtain CD14⁺CD16⁺ monocytes for subsequent experiments to characterize Group 1 and 2 monocytes.

We re-analyzed scRNA-seq data from CD14⁺CD16⁺ monocytes of two donors matured in culture from our previously published study (31) using Seurat (v3.2.2) package on R programming language (v4.1.0). Cells were confirmed for CD14 and CD16 positivity by flow cytometry before and after scRNA-seq analysis (31). Quality control was performed on each individual sample to remove cells with high mitochondrial gene expression and cells with less than 200 expressed genes (31). Samples from both donors were merged to yield a total of 26,977 cells for downstream analyses. Merged data was normalized using the SCtransform function (31, 32). The merged data were then normalized and integrated, as previously described (31, 32). To identify cell clusters, we first performed dimensionality reduction by principal component analysis (PCA) using the top 19 principal components (PCs). Clusters were generated from PCs using the FindClusters function with resolution set at 0.3. The clusters were then visualized using the uniform manifold approximation and projection (UMAP) dimensionality reduction algorithm.

DEGs for each cluster of CD14⁺CD16⁺ monocytes were identified by the FindMarkers function in the Seurat package, as previously described (31, 32). For this study, expression of genes in each individual cluster were compared to the expression of the same genes in all the remaining clusters, to quantify an average difference (avg diff) in expression for each DEG. We performed DEG analysis on Pearson’s residuals after regularized negative binomial regression, employing t-tests per gene for all genes identified in at least 3 cells in the groups being compared. Genes were considered differentially expressed if they had an adjusted p-value<0.005 and FDR<0.01 (31, 32) ( Supplementary Table S1 ). Significant DEGs were then analyzed using IPA software (Qiagen). We used a cut off Z-score of log₂fold-change ± 0.3 and FDR<0.05 to correlate significant avg diff for DEGs in each cluster with pathways and functions involved in motility, chemotaxis, migration, invasiveness, and inflammation as determined by Z-score.

For analysis of monocyte surface markers, CD14⁺CD16⁺ monocytes from ten new independent donors were cultured as described above, washed once with 1X PBS, and stained with Zombie NIR fixable viability dye (Biolegend, 1:1000) for 30 minutes at room temperature (RT) in the dark. Cells (0.2-1×10⁶ cells/tube) were washed in FACS staining buffer consisting of 1X PBS with 1% bovine serum albumin (BSA), and stained with surface antibodies Alexa-700 anti-human CD14 (BD Biosciences, clone M5E2, 1.2μg), PE-Cy7 anti-human CD16 (BD Biosciences, clone 3GB, 3.2μg), BB515 anti-human ALCAM/CD166 (BD Biosciences, clone 3A6, 1.8μg), PE anti-human CD52 (Biolegend, clone HI186, 0.45μg), PerCP-Cy5.5 anti-CD63 (BD Biosciences, clone H5C6, 1.2μg), and APC anti-human SDC2/CD362 (R&D systems, clone 305515, 0.07μg) for 30 minutes at 4°C in the dark. These antibodies were used in all experiments involving staining of CD14⁺CD16⁺ monocytes, unless stated otherwise. Cells were then washed with FACS staining buffer and fixed with 2% paraformaldehyde (PFA) in 1X PBS supplemented with 5mM EDTA. Fluorescence minus one (FMO) controls were prepared concomitantly. The AbC Total Antibody Compensation Bead Kit (Invitrogen) was used to prepare bead compensation controls. Samples were acquired on the Attune NxT flow cytometer (Thermo Fisher). Data was analyzed using FlowJo v10.10.0 software (BD Biosciences) with only CD14⁺CD16⁺ monocytes gated in Quadrant 2 used for subsequent analyses. Additionally, cells excluded from singlets and live/dead gates were backgated and determined to be dying cells and/or debris as determined by their FSC/SSC positions and were not included in subsequent analyses.

Single cell suspensions of matured CD14⁺CD16⁺ monocytes from 13 new independent donors were stained in polypropylene FACS tubes that were previously coated with PBS with 4% FBS (Gibco, Thermo Fisher) overnight at 4°C. Flow cytometry staining was performed for CD14, CD16, ALCAM, and viability, as described above. Monocytes to be sorted were seeded at 5-10×10⁶ cells/tube, stained, and then pooled and re-suspended at 15×10⁶ cells/ml in FACS sorting buffer consisting of 1X PBS supplemented with 2mM EDTA, 25mM HEPES (Teknova), 4% FBS (Gibco, Thermo Fisher), and 1% Pen-Strep (Gibco, Thermo Fisher). Samples were sorted on the BD FACS Aria in the Albert Einstein College of Medicine (AECOM) Flow Cytometry Core. Gates were set using FMO controls to sort CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo populations. Cells were sorted by ALCAM expression without the use of other surface proteins that define Group 1 to obtain sufficient number of cells for subsequent RNA extraction and qRT-PCR analyses. Sorted cells were collected in BD Falcon tubes containing monocyte media supplemented with M-CSF and kept at 4°C until RNA isolation, which was performed the same day.

To compare CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo populations, RNA was extracted from sorted populations using TRIzol (Invitrogen), per the manufacturer’s protocol. The number of cells isolated after sorting CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo monocyte populations from different donors varied greatly, with monocytes from some donors yielding large and others yielding small numbers of sorted cells. As a result, not enough RNA was isolated to quantify expression of all the markers by qRT-PCR for cells from every donor. RNA concentration and purity was evaluated using a NanoDrop 2000 spectrophotometer (Thermo Fisher). Only samples with 260/280 values at ~2 were used. RNA quality was assessed with the Agilent 2100 Bioanalyzer (AECOM Genomics Core), with RNA quality numbers all above 8. One to 2μg of RNA was converted to cDNA using the SuperScript VILO Master Mix (Invitrogen), per the manufacturer’s protocol. QRT-PCR was performed using the QuantStudio 3 real-time PCR system and software (Applied Biosystems). Relative expression of target genes between CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo cells was quantified using TaqMan^® probes (Applied Biosystems) and normalized to internal control gene 18S using the 2^-ΔΔCt method. Each comparison was performed in triplicate using 1μl of cDNA. A fold change of expression between CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo populations was generated by setting CD14⁺CD16⁺ALCAM^-/lo populations to one. The following TaqMan^® primer probes from Thermo Fisher were used (catalog: 4331182): 18S (Hs99999901_s1), ALCAM (Hs00233455_m1), APOC1 (Hs00155790_m1), APOE (Hs00171168_m1), CHI3L1 (Hs01072228_m1), CLEC7A (Hs01902549_s1), CTSB (Hs00947433_m1), DUSP1 (Hs00610256_g1), FCGR2B (Hs01634996_s1), LGALS3 (Hs00173587_m1), SPP1 (Hs00959010_m1), and TPT1 (Hs02621289_g1). PCR primer efficiencies were evaluated using 5-fold serial dilutions of cDNA isolated from mature monocytes to generate standard curves, with all target efficiencies between 90-110%.

For intracellular staining, CD14⁺CD16⁺ monocytes from ten new independent donors were matured as described previously, with 100 ng/ml LPS (Sigma-Aldrich) added for the last 24 hours, and 10 μg/ml Brefeldin A (eBioscience) added for the last 4 hours, of the three-day incubation. Cells (1×10⁶ cells/sample) were then added to a 96-well U bottom plate, washed, stained for viability with Zombie NIR (1:4000), and then stained for surface markers CD14, CD16, and Alexa-488 anti-human ALCAM/CD166 (Bio-Rad, clone 3A6, 0.4μg), as described above. Cells were fixed and permeabilized using the Foxp3 Fixation/Permeabilization working solution from the Foxp3/Transcription Factor Staining Buffer Set (eBioscience) for 1 hour in the dark on ice. Cells were then washed with 1X Permeabilization buffer (eBioscience), and stained intracellularly for one hour in the dark on ice with the following antibodies diluted in 1X Permeabilization buffer and Brilliant Stain Buffer (BD Biosciences, 1:20): PE mouse anti-human IL-1R antagonist (BD Biosciences, clone AS17, 0.05μg), PerCP-Cy5.5 mouse anti-human IL-6 (Biolegend, clone MQ2-13A5, 0.125μg), PE-CF594 mouse anti-human IL-8 (R&D, clone 79018, 0.013μg), BV711 mouse anti-human IL-10 (BD Biosciences, clone JES3-9D7, 0.325μg), RB780 mouse anti-human TNFα (BD Biosciences, clone Mab11, 0.325μg). Cells were acquired using spectral cytometry on a Cytek Aurora in the AECOM Flow Cytometry Core. For these experiments, the percent expression of intracellular mediators was compared between CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo cells, without the use of other surface proteins that define Group 1, because of the extensive antibody panel used to assess all the mediators in each cell per sample, and the number of cells needed for necessary controls.

To measure ROS, the CellROX Deep Red Flow Cytometry Assay kit (Invitrogen) was used according to the manufacturer’s instructions. Briefly, matured CD14⁺CD16⁺ monocytes from 11 new independent donors were added to FACS tubes (0.4-1×10⁶ cells/sample). Positive controls were treated with 200μM tert-butyl hydrogen peroxide (TBHP) and negative controls with 1mM N-Acetylcysteine (NAC) for 1 hour in the dark at 37°C, prior to measurement of ROS. After incubation, cells were washed with 1X PBS and subsequently incubated with 5μM CellROX Deep Red reagent for 1 hour at 37°C in the dark. Cells were then washed and stained with Zombie NIR fixable viability stain (Biolegend) for dead cell exclusion, followed by staining with antibodies to cell surface ALCAM, CD14, CD16, CD52, and CD63, as described above. ROS was identified by the corresponding median fluorescence intensity (MFI) detected at emission/excitation wavelength 644/665 nm by the red laser. Samples were acquired on the Attune NxT flow cytometer (Thermo Fisher). Baseline ROS MFI was compared between CD14⁺CD16⁺ALCAM⁺CD52^hiCD63^hi and CD14⁺CD16⁺ALCAM^-/loCD52^loCD63^lo populations. Designations of high and low expression of CD52 and CD63 were defined by the top and bottom 20^th percentiles in that fluorophore.

We used a human in vitro BBB model to study transmigration of CD14⁺CD16⁺ monocytes. This model is a co-culture system of human BMVEC (Cell Systems) and human cortical astrocytes separated by a gelatin-coated tissue culture insert with 3μm pores (BD Falcon) that are placed in the wells of a 24-well plate containing media (baseline transmigration) or media with CCL2 (200ng/ml; R&D systems), as described previously (16, 30, 33). BMVEC seeded on the upper surface of the tissue culture represent the peripheral side of the BBB. Primary human astrocytes seeded on the underside represent the CNS side. A portion of matured CD14⁺CD16⁺ monocytes from six new independent donors (pre-transmigrated cells) was stained with antibodies to CD14, CD16, ALCAM, CD52, CD63, SDC2, with FMO controls, as described above. Three × 10⁵ cells were added to each BBB co-culture and allowed to transmigrate to CCL2 for 24 hours, with each condition performed in quadruplicate. Permeability of co-cultures was assessed using Evans blue dye coupled to BSA (16, 19, 33) and all co-cultures were impermeable. Transmigrated cells were collected and stained with the same panel of antibodies as the pre-transmigration cells. At least 20,000 events were collected for pre-transmigrated cells, and the entire sample of post-transmigrated cells was collected for analysis by flow cytometry. Positivity for each marker was determined by FMO controls on the pre-transmigration population, with the same gating applied to the post-transmigrated cells. The number of cells that transmigrated from each Group was expressed as the percent of transmigrated cells relative to the input population of each Group of monocytes, as described previously (19, 34). Group 1 monocyte surface proteins were evaluated individually and sequentially on CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo populations. CD14⁺CD16⁺ monocytes were first gated for ALCAM⁺ and ALCAM^-/lo cells, and then high and low expression of each additional surface protein marker was assessed separately on these two populations. High and low expression was defined as the top and bottom 20^th percentiles in that fluorophore. Additionally, the proportion of Group 1 monocytes that were CD14⁺CD16⁺ALCAM⁺CD52⁺CD63⁺SDC2⁺ was compared pre- and post-transmigration using sequential gating. Overall positivity, and not high expression, of each marker was used to identify Group 1 monocytes because of the limited number of transmigrated cells available for analysis of high expression of all six markers in each cell by sequential gating.

Statistical analyses and graphs were generated with GraphPad Prism software v.10.1.0 (GraphPad software). Normality was assessed using D’Agostino & Pearson’s tests. For normally distributed data, paired student t-tests and one-sample t-tests were used to detect significant differences between groups. For non-normally distributed data, Wilcoxon matched-pairs signed rank and Wilcoxon signed rank test were used to detect significant differences between groups. Data was presented as means ± standard error of the mean (SEM).

CD14⁺CD16⁺ monocytes are implicated in various pathological processes and diseases (10, 11, 24). To characterize different subpopulations within this heterogenous monocyte subset, we performed new analysis of scRNA-seq data from a previous study that focused on characterizing gene expression in CD14⁺ monocytes matured in vitro to CD14⁺CD16⁺ monocytes, and subsequently infected with HIV (31). In this previous study, uninfected CD14⁺CD16⁺ monocytes from two independent donors were analyzed by scRNA-seq as a control. In the current study, we reanalyzed this scRNA-seq data to examine subpopulations within uninfected CD14⁺CD16⁺ monocytes. We assessed an average of 13,489 cells per donor, with a total of 26,977 cells ( Supplementary Table S2 ). Clustering analysis and UMAP embedding showed that CD14⁺CD16⁺ monocytes segregated into nine clusters ( Figure 1A ), indicating heterogeneity within the intermediate monocyte subset. The data was then analyzed to quantify expression of genes in each individual cluster compared to that of all the remaining clusters to identify DEGs. IPA was then performed to indicate molecular pathways that were enriched in each cluster compared to all other clusters. IPA identified increased migratory and inflammatory pathways, including integrin signaling, migration of cells and of tumor cell lines, leukocyte extravasation signaling, actin cytoskeleton signaling, and inflammatory response, in clusters 2, 4, and 8 ( Figure 1B ). We termed cells in clusters 2, 4, and 8 as Group 1 monocytes and cells in clusters 0, 1, 3, 5, 6, and 7 as Group 2 monocytes. Of the CD14⁺CD16⁺ monocytes analyzed by scRNA-seq, approximately 19% were Group 1 cells, as determined by assessing the number of cells in clusters 2, 4, and 8, compared to the total ( Supplementary Table S2 ). IPA suggested that Group 1 monocytes exhibited heightened motility, neurotoxicity, and inflammatory functions that could contribute to the pathogenesis of neuroinflammatory diseases such as HIV-NCI.

We then examined DEGs specific for the molecular pathways identified by IPA. A heat map of some of these genes showed that Group 1 monocytes exhibited a similar gene expression signature with consistent upregulation of genes specific for the different molecular pathways, indicating that these clusters may share functional properties ( Figure 1C ). Conversely, Group 2 monocytes exhibited decreased expression of genes in those same pathways, indicating they may be more similar to each other and contribute less to neurotoxic and inflammatory processes. To characterize gene expression in monocytes within Group 1, we focused on DEGs that encoded cell surface proteins involved in migratory and inflammatory processes. ALCAM is a junctional adhesion protein that is critical to BBB transmigration of CD14⁺CD16⁺ monocytes (30, 34), and CD52, a GPI-anchored glycoprotein, is known to facilitate transendothelial migration in lymphocytes (35). Although the role of CD52 in monocytes remains incompletely characterized, it may contribute to adhesion (36). CD63 is a tetraspanin that is highly expressed on CD14⁺CD16⁺ monocytes and is associated with recruitment of monocytes to tissues during an inflammatory response (37). SDC2, a member of the syndecan family, regulates cell adhesion and migration (38), and has been linked to tumor cell invasiveness (39). The DEG data indicated that these genes were highly expressed in Group 1, with lower expression in Group 2 ( Figure 1C ). This was confirmed by violin plots showing relative expression of these DEGs ( Figure 1D ).

In addition to genes involved in migratory and inflammatory pathways, we also examined the expression of disease-associated microglia (DAM) genes. While some DAM are considered neuroprotective (40), many are implicated in the pathogenesis of several neurodegenerative diseases, including AD and amyotrophic lateral sclerosis (ALS). These are referred to as neurodegenerative microglia (MGnD) (41). A myeloid cell population similar to DAM has been identified in the cerebrospinal fluid (CSF) of PWH on antiretroviral therapy (ART), with these cells proposed to play a role in HIV-NCI (42). DAM markers such as SPP1 and APOE (40, 41, 43) were among the most upregulated DEGs in monocytes from Group 1 ( Figure 1C ). Proteins encoded by these genes are reported to contribute to various CNS diseases (40, 41).

To validate our scRNA-seq analysis and further characterize Group 1 monocytes, we examined, by flow cytometry, expression of cell surface proteins specific for genes enriched in Group 1 that play important roles in migration and inflammation. Freshly isolated CD14⁺ monocytes from ten independent donors were cultured non-adherently for three days to obtain mature CD14⁺CD16⁺ monocytes, as described previously (16, 17, 19, 30). Gene expression of several surface markers was higher in Group 1 monocyte clusters. Therefore, we stained for these in our flow cytometry experiments. CD14⁺CD16⁺ monocytes were labeled with fluorescent antibodies to ALCAM, CD52, CD63, and SDC2. ALCAM expression was used as an initial marker to identify a population enriched for Group 1 monocytes, with minimal or absent ALCAM expression identifying mainly Group 2 monocytes. These populations were distinguished by gating on ALCAM⁺ cells defined by FMO controls. We then used an additional gating strategy by examining ALCAM⁺ and ALCAM^-/lo cells for levels of CD52, CD63, and SDC2, which were expected to be elevated in Group 1 monocytes based on our scRNA-seq data ( Figure 2A ). MFIs of CD52, CD63, and SDC2 were compared between CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo populations. Results showed that CD14⁺CD16⁺ALCAM⁺ cells have significantly higher MFIs for CD52 (p=0.0021), CD63 (p=0.0002), and SDC2 (p=0.003), than CD14⁺CD16⁺ALCAM^-/lo cells ( Figures 2B–D ), indicating that ALCAM⁺ cells have higher expression levels of these cell surface proteins and represent a monocyte population enriched for Group 1 monocytes.

To validate and characterize further expression of various intracellular markers in Group 1 monocytes identified from scRNA-seq data, we performed FACS and subsequent qRT-PCR analyses of sorted cells enriched for Group 1 or Group 2 monocytes. We selected six DEGs that encode intracellular proteins associated with neurodegenerative and neuroinflammatory diseases, all of which were significantly increased in Group 1 compared to Group 2 monocytes. The genes examined were CHI3L1, and the DAM markers APOC1, APOE, CTSB, LGALS3, and SPP1. APOC1 and APOE encode the apolipoproteins C1 and E, with APOE contributing significantly to AD development (44). CHI3L1 encodes chitinase-3-like protein 1 and is upregulated in macrophages, microglia, and astrocytes associated with MS, AD, and HIV-associated dementia (45). CTSB encodes for cathepsin-B, a lysosomal protease involved in protein degradation and neurodegeneration (46, 47). LGALS3 and SPP1, which encode galectin-3 and osteopontin proteins, respectively, have various signaling properties and can serve as chemoattractants for monocytes (48, 49). Collectively, APOE, CTSB, LGALS3, and SPP1 are hallmark genes of DAM and MGnD (40, 41). We also analyzed CLEC7A, a C-type lectin and DAM marker, FCGR2B, which encodes the Fc fragment of IgG receptor IIb, and DUSP1, a dual-specificity protein phosphatase, as controls for detecting genes that were increased in Group 2 compared to Group 1 monocytes. We also quantified TPT1, a regulator of cell growth with homeostatic function often used to normalize qRT-PCR gene expression in monocytes (50, 51) as a control expected to be the same in both Group 1 and 2 monocytes.

Monocytes from 13 independent donors were matured in culture as previously described, and stained for CD14, CD16, and ALCAM. Cells were sorted based on ALCAM positivity, and CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo populations collected for RNA isolation. Analyses by qRT-PCR demonstrated that Group 1 monocytes (ALCAM⁺) exhibited higher expression of APOC1 (p=0.001), APOE (p=0.0002), CHI3L1 (p=0.014), CTSB (p=0.008), LGALS3 (p=0.0018), and SPP1 (p=0.016) compared to Group 2 monocytes (ALCAM^-/lo) ( Figures 3B–G ). CHI3L1 and SPP1 showed large differences in expression between Group 1 and Group 2 monocytes, with fold changes at 25.64 and 11.34, respectively ( Figures 3D, G ). In addition, higher expression of ALCAM in Group 1 monocytes confirmed the effectiveness of FACS of CD14⁺CD16⁺ monocytes ( Figure 3A ). CLEC7A (p=0.043) and FCGR2B (p=0.01) were higher in Group 2 monocytes ( Figures 3H, I ). DUSP1 (p=0.1078) was not significant but showed a trend for increased expression in Group 2 monocytes ( Figure 3J ). TPT1 showed no significant difference (p=0.227) in expression between Group 1 and Group 2 monocytes ( Figure 3K ). These results confirm the use of ALCAM positivity alone to identify populations enriched in Group 1 or Group 2 monocytes that can be used to validate further the DEGs identified by our scRNA-seq data and characterize the gene expression profiles of Group 1 and Group 2 monocytes.

We evaluated functional differences between Group 1 and Group 2 monocytes. CD14⁺CD16⁺ monocytes produce mediators that can either promote or suppress inflammation (52–55). To characterize inflammatory mediator production in Group 1 and Group 2 monocytes, we performed intracellular staining and flow cytometry. CD14⁺CD16⁺ monocytes were exposed to LPS for 24 hours, and then stained with antibodies to surface markers CD14, CD16, ALCAM, and intracellular CXCL12, IL-6, IL-8, and TNFα, chemokines and cytokines that contribute to inflammatory diseases, including HIV neuropathogenesis (56–59). We also evaluated IL-1Ra and IL-10, as these mediators possess anti-inflammatory properties, or are produced to downregulate and limit an ongoing inflammatory response (60, 61). Group 1 and Group 2 monocytes were defined as only CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo for these experiments without including expression of the different surface proteins examined previously, due to the antibody panel complexity needed to analyze all the intracellular mediators in each cell, and the number of cells needed for proper controls. Percent expression of each mediator was then compared between CD14⁺CD16⁺ALCAM⁺ and CD14⁺CD16⁺ALCAM^-/lo populations ( Figure 4A ). Our results for ten independent donors demonstrated that Group 1 monocytes produced significantly more CXCL12 (p=0.002), IL-6 (p=0.002), TNFα (p=0.0103), IL-1Ra (p=0.0008), and IL-10 (p=0.0008), compared to Group 2 monocytes ( Figures 4B–F ). Interleukin-8 was similar between Groups (p=0.814, Figure 4G ). Group 1 monocytes produced more IL-1Ra and IL-10 than Group 2 monocytes, contrary to our initial hypothesis that Group 2 monocytes may produce more of these anti-inflammatory mediators ( Figures 4E, F ).

CD14⁺CD16⁺ monocytes generate ROS (62) that can contribute to neuropathogenic diseases, such as HIV-NCI (63–65). Our scRNA-seq analysis demonstrated that Group 1 monocytes have decreased levels of superoxide dismutase 2 (SOD2) ( Figure 1C ), an important antioxidant that neutralizes ROS (66). Thus, we examined whether Group 1 monocytes from eleven independent donors produce more ROS than Group 2 monocytes using flow cytometry. CD14⁺CD16⁺ monocytes were incubated with CellROX, a dye that detects cytoplasmic ROS. As controls, cells were incubated with TBHP, a compound that induces ROS, and NAC, an antioxidant that neutralizes ROS, prior to addition of CellROX. To delineate Group 1 from Group 2 monocytes further, we performed sequential gating where ALCAM⁺ and ALCAM^-/lo cells were gated for high and low expression of CD52 and CD63, respectively ( Figure 5A ). Using this gating strategy, Group 1 monocytes were classified as CD14⁺CD16⁺ALCAM⁺CD52^hiCD63^hi, and Group 2 monocytes as CD14⁺CD16⁺ALCAM^-/loCD52^loCD63^lo. We then compared basal ROS between the two groups of cells by MFIs ( Figure 5B ). Our results showed that Group 1 monocytes produce significantly more ROS at baseline compared to Group 2 monocytes (p=0.007, Figure 5C ). NAC effectively inhibited ROS production in both Group 1 (p=0.031) and Group 2 monocytes (p=0.027, Figure 5C ).

CD14⁺CD16⁺ monocytes have been shown to preferentially transmigrate across the BBB compared to classical and non-classical monocytes (16–19, 30). Our findings in this study indicate that Group 1 monocytes are a subpopulation of CD14⁺CD16⁺ monocytes that produce more ROS and inflammatory mediators, suggesting that entry of these cells into the CNS would contribute significantly to HIV-NCI and other neuroinflammatory diseases. Thus, we examined whether Group 1 monocytes preferentially transmigrated across the BBB compared to Group 2 monocytes using a human in vitro co-culture BBB model that has been used extensively to characterize mechanisms of monocyte transmigration (16, 19, 30, 33, 34, 67). Group 1 monocytes were defined as CD14⁺CD16⁺ALCAM⁺CD52^hiCD63^hiSDC2^hi, and Group 2 monocytes as CD14⁺CD16⁺ALCAM^-/loCD52^loCD63^loSDC2^lo. Pre- and post-transmigrated CD14⁺CD16⁺ monocytes were first gated for ALCAM⁺ or ALCAM^-/lo cells, and then sequential gating was performed to assess high or low expression of each individual additional marker on ALCAM⁺ and ALCAM^-/lo cells, respectively. Since the number of cells post-transmigration is limited, we did not have a sufficient number of cells to perform sequential gating to analyze high expression of all Group 1 markers on transmigrated ALCAM⁺ cells. Thus, in a separate analysis, after gating for ALCAM⁺ monocytes, we performed sequential gating for overall positive expression of CD52, CD63, and SDC2. Our results using gating for each marker individually showed that Group 1 monocytes defined by ALCAM positivity and high expression of CD52 (p=0.0312) or CD63 (p=0.0312) preferentially transmigrated across the BBB compared to Group 2 monocytes, as expressed by the percent of transmigrated Group 1 cells relative to their input ( Figure 6A ). While there was no significance in transmigration of cells with high compared to low expression of SDC2 (p=0.219), there was a trend toward increased transmigration of Group 1 cells with high expression of this marker. This preferential transmigration of Group 1 monocytes was confirmed by sequential gating for expression of all markers on ALCAM⁺ cells pre-transmigration compared to post-transmigration (p=0.031, Figure 6B ). These findings suggest that Group 1 monocytes have a higher propensity to transmigrate across the BBB into the CNS than Group 2 monocytes, indicating that they are primed to enter the CNS and be major contributors to neuropathogenesis.