Frontline Science: c-Myc regulates P-selectin glycoprotein ligand-1 expression in monocytes during HIV-1 infection

The identification of c-Myc as a potential PSGL-1 transcriptional regulator facilitates the determination of the physiological significance of PSGL-1 induction, and provides novel therapeutic targets.

Abstract

Leukocyte extravasation is a crucial feature of the normal immune response to disease and infection and is implicated in various pathologies during chronic inflammatory disease. P-Selectin glycoprotein ligand-1 (PSGL-1) is critical for leukocyte extravasation; however, despite extensive study, it remains unclear how its expression is regulated, which in turn, impedes a more precise understanding of how its expression level affects transmigration. To investigate the regulation of PSGL-1, 60 subjects, with or without HIV infection, were recruited and PSGL-1 expression in monocytes was measured. PSGL-1 was found to be up-regulated on leukocytes from HIV-infected individuals, and the physiologically relevant mediators soluble CD40 ligand (sCD40L) and glutamate were able to induce PSGL-1 transcription in human monocytes ex vivo. HIV-1 induced PSGL-1 induction, and its dependence on CD40L was validated further by use of the mouse-tropic HIV (EcoHIV) mouse model of HIV infection in C57BL/6 and CD40L knockout (KO) mice. To investigate crosstalk between the signaling cascades induced by CD40L and glutamate that lead to PSGL-1 induction, a network-based, discrete dynamic model was developed. The model reveals the MAPK pathway and oxidative stress as critical mediators of crosstalk between CD40L and glutamate-induced pathways. Importantly, the model predicted induction of the c-Myc transcription factor upon cotreatment, which was validated using transcriptomic data and pharmacologic inhibition of c-Myc. This study suggests a novel systems serology approach for translational research and reveals a mechanism for PSGL-1 transcriptional regulation, which might be leveraged to identify novel targets for therapeutic intervention.

Introduction

Excessive tissue infiltration of immune cells has been shown to be important to the pathology of various diseases, including chronic infections, such as HIV [1] and hepatitis C virus [2] infections, and CVDs, including atherosclerosis [3], among others. In many of these diseases, whereas excessive leukocyte extravasation may not be the underlying causal factor, it contributes significantly to the disease pathology and is often closely linked to disease morbidity and mortality. Thus, the targeting of leukocyte transmigration is a promising approach for the development of novel therapeutics to treat complications associated with these diseases [4].

Leukocyte transmigration is a multistep process [5]. Of note, leukocytes must first attach to the endothelium, mediated primarily by the binding of PSGL-1 (also known as CD162 or selectin P ligand), expressed on leukocytes and P-selectin, expressed on the surface of endothelial cells. In monocytes and granulocytes, the machinery required for the proper post-translation modification of PSGL-1 is constitutively expressed; however, in T cells, this machinery is induced upon exposure to activation signals [6]. Furthermore, the signaling through PSGL-1 has been reported to promote integrin expression and activation [7]. Additionally, various proinflammatory mediators have been reported to prime leukocytes for integrin activation [8, 9]. Finally, the various ligands for leukocyte-expressed transmigratory molecules are induced on endothelial cells upon endothelial cell activation [10]. Increased PSGL-1 expression has also been associated with various inflammatory conditions, including a mouse model of acute inflammation [11], individuals suffering from allergic asthma [12], and individuals with chronic obstructive pulmonary disease [13].

Whereas increased PSGL-1 expression would likely lead to increased adhesion and transmigration, this has yet to be formally proven. There are 2 potential difficulties in establishing this point adequately: 1) it is unclear what factor or factors are responsible for PSGL-1 induction in any of the contexts in which it has been reported, and 2) the mechanisms of transcriptional regulation for PSGL-1 induction are unknown. This work aims to establish what soluble factors and induced transcription factors are responsible for PSGL-1 induction during HIV infection.

HIV, the causative agent of AIDS, infects ∼36.7 million people worldwide [14]; however, with the advent of cART, the life expectancy of infected individuals is increased significantly compared with the pre-cART era [15]. Still, whereas the life expectancy of infected individuals is nearing that of uninfected individuals, it has become apparent that they are at increased risk for a variety of secondary disorders [16]. Importantly, these secondary complications are now significant contributors to morbidity and mortality in the HIV⁺ population, having overtaken AIDS-defining illnesses as the leading causes of death in this population [17, 18]. One complication is CVD, with HIV⁺ individuals at increased risk for myocardial infarction and displaying accelerated atherosclerotic plaque formation [19]. Another important complication is HAND, which is defined by psychomotor deficits, with the most severe form, dementia [20]. A related disorder, HIV encephalopathy, is defined, in part, by an accumulation of peripheral Mϕ in brain tissue (identified postmortem). Atherosclerosis is similarly understood to result from the accumulation of Mϕ in arterial walls, thought to be derived primarily from circulating monocytes upon entry into tissue [21], although there may be other sources in addition to circulating monocytes. Thus, a better understanding of leukocyte transmigration during HIV infection may facilitate efforts to target this process therapeutically for the mitigation of various HIV-associated disorders.

Here, we show that the PSGL-1 expression is increased in leukocytes following HIV infection, likely via the combined actions of glutamate and sCD40L, 2 well-characterized proinflammatory mediators that are implicated in the HIV pathogenesis. Elevated levels of PSGL-1 were also observed in monocytes from C57BL/6 mice but not CD40L-deficient mice infected with a novel EcoHIV, further confirming the role of CD40L in this process. Network-based, discrete-dynamic modeling of signaling cascades induced by sCD40L and glutamate suggests a role for c-Myc in mediating transcription of PSGL-1. The prediction was validated using transcriptomic data and pharmacologic inhibition of c-Myc. Thus, this work reveals novel, soluble factors that trigger PSGL-1 expression and sheds light on the molecular mechanisms that may influence tissue infiltration of immune cells during HIV infection.

MATERIALS AND METHODS

Ethics statement

The Research Subjects Review Board at the University of Rochester Medical Center approved studies involving human samples. All study participants were adults, and blood samples were obtained after written, informed consent, in accordance with the Declaration of Helsinki. Mouse experiments were carried out in accordance with the Animal Welfare Act and NIH guidelines, and the University Committee on Animal Resources of the University of Rochester Medical Center approved the animal protocol. The facilities and programs of the Vivarium and Division of Laboratory Animal Medicine of the School of Medicine and Dentistry are fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International.

Patient samples

Persons with (n = 32) and without (n = 28) HIV infection, without any occurrence of CVD for at least the preceding year, were recruited at the University of Rochester Medical Center. This sample size was determined based on initial power analysis in which it was found that recruiting n = 25 subjects in each group yields >90% of statistical power. The average age of these individuals was 55 ± 11 (means ± sd) yr. Subjects who reported current anti-platelet/nonsteroidal anti-inflammatory drug use were excluded from the study. All persons with HIV infection were on ART at the time of blood donation with undetectable viral loads and average T cell count in the range of 652 ± 405 per μl. The blood was drawn from these subjects into acid citrate dextrose-buffered vacutainers (BD Biosciences, San Jose, CA, USA), and various analyses were performed as outlined below.

LC-MS/MS

Whole blood from subjects was centrifuged at 250 g for 15 min, and platelet-rich plasma was collected. Following addition of PGI2 (1 μl/ml platelet-rich plasma; Cayman Chemical, Ann Arbor, MI, USA) to maintain platelet quiescence, platelet-rich plasma was centrifuged at 1000 g for 10 min to pellet the platelets. The resulting platelet-poor plasma was transferred to HPLC sample vials and analyzed with reverse-phase LC, coupled to a triple-quadrupole mass spectrometer running in negative mode (Thermo Quantum Ultra; Thermo Fisher Scientific, Waltham, MA, USA) with SRM-specific scans. The resulting data were analyzed by using the publicly available mzRock machine learning tool kit (http://code.google.com/p/mzRock), which automates SRM/HPLC feature detection, grouping, signal-to-noise classification, and comparison with known metabolite retention times. Standard solutions of glutamate were also prepared and run to allow for determination of the absolute concentration of glutamate in the plasma samples.

Monocyte isolation and culture

Human monocytes were isolated using a MACS Pan Monocyte Isolation Kit (Miltenyi Biotec, San Diego, CA, USA), per the manufacturer’s instructions. Monocytes were cultured in RPMI 1640, supplemented with 10% FBS and 2% penicillin/streptomycin/glutamine. For immunoblot assays and flow cytometric analyses, monocytes were treated for 24 h, whereas for PCR analysis, they were treated for 18 h. In both cases, cells were incubated overnight at 37°C and 5% CO₂ before being treated.

Flow cytometry

Human leukocytes were analyzed using a method described previously [22] within 1 h of the blood draw. In brief, 100 μl blood was fixed, and RBCs were lysed and then stained for 30 min, with 1.25 μl CD14-APC/Cy7 (BioLegend, San Diego, CA, USA), 2.5 μl CD16-Pacific Blue (BD Biosciences), and 2.5 μl CD162-PerCP-eFluor 710 (eBioscience, San Diego, CA, USA). Following the staining, cells were acquired using a flow cytometer (BD LSRFortessa 18 color; BD Biosciences). Human monocytes, granulocytes, and lymphocytes were gated based on characteristic FSC and SSC; the monocytes were identified further via CD14 staining. The monocyte subsets were defined by CD14 and CD16 expression by using the following criteria: CD14^dimCD16⁺ (nonclassic monocytes), CD14^highCD16⁺ (intermediate monocytes), and CD14^highCD16⁻ (classic monocytes). Fluorescence minus one controls were used to define gates. For mouse leukocytes, 50 μl whole blood was prepared as above, but the antibodies used were as follows: 2.5 μl CD115-APC (eBioscience), 1.5 μl Ly-6C/6G-PE/Cy7, 2.5 μl CD19-AF700, 2.5 μl CD3-BV786, and 5 μl CD162-PE (BD Biosciences).

Immunoblot assay

Immunoblot assay was performed as described previously [23]. In brief, after the indicated treatments, whole-cell lysates were prepared in erythrocytes lysis buffer [50 mM HEPES, pH 7, 250 mM NaCl, 0.1% Nonidet P-40, 5 mM EDTA, 10 mM NaF, 0.1 mM Na₃VO₄, and 50 µM ZnCl₂, supplemented with 0.1 mM PMSF, 1 mM DTT, and Phosphatase Inhibitor Cocktail 3 (P0044; Sigma-Alrich, St. Louis, MO, USA)]. After removal of cellular debris via high-speed centrifugation (13,000 rpm, 5 min), lysates were fractionalized on 7.5% SDS-PAGE gels, and protein was electrophoretically transferred to nitrocellulose membranes (GE Healthcare Bio-Sciences, Piscataway, NJ, USA). The membranes were then analyzed for immunoreactivity against the following antibodies: PSGL-1 (clone KPL1) and α-tubulin (both from Santa Cruz Biotechnology, Dallas, TX, USA). Bound antibodies were detected using species-specific infrared dye (IRDye)-conjugated secondary antibodies and subsequently developed using the Odyssey infrared imager (Li-Cor Biotechnology, Lincoln, NE, USA).

qRT-PCR

Changes in the expression of the PSGL-1 gene from treated, primary human monocytes (from healthy HIV⁻ donors) were analyzed by qRT-PCR. Total RNA was isolated using TRIzol, according to the manufacturer’s protocol (Qiagen, Valencia, CA, USA). First-strand cDNA synthesis was then performed using the iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA, USA), and qRT-PCR was completed using the SYBR Green DNA polymerase (Thermo Fisher Scientific), and the following primers were used: human PSGL-1 (forward) 5′-TCCTCCTGTTGCTGATCCTACTG-3′ and (reverse) 5′-TACTCATATTCGGTGGCCTGTCT-3′ and the housekeeping gene GAPDH (forward) 5′-TGATGACATCAAGAAGGTGGTGAA-3′ and (reverse) 5′-TCCTTGGAGGCCATFTAGGCCAT-3′.

EcoHIV virus preparation

Virus stocks were prepared by transfection of plasmid DNA into 293T human embryonic kidney cells and titered for their p24 HIV core antigen by p24 ELISA (PerkinElmer, Waltham, MA, USA). In brief, 7.5 × 10⁵ 293T cells/well were cultured in a 6-well plate and transfected with 4 μg EcoHIV DNA using polyethylenimine. 293T Cell culture supernatants were harvested 48 h after transfection and concentrated by centrifugation at 22,000 g for 2 h at 4°C. The supernatant was then removed, and the pellet was resuspended in 300 μl DMEM (Thermo Fisher Scientific), aliquoted, and stored at −80°C. The EcoHIV construct was kindly provided by Dr. David Volsky (Icahn School of Medicine at Mount Sinai, New York, NY, USA).

EcoHIV infection of mice

C57BL/6 (WT) or B6.129s2-CD40lgtm1mx/J (CD40L KO) mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). Ten- to 17-wk-old male mice (WT, n = 6; CD40L KO, n = 6) were given i.p. injections of either saline or EcoHIV (1.0 × 10⁵ pg p24), as described previously [24]. At 2 wk postinfection, blood was collected via cardiac exsanguinations and prepared for flow cytometric analyses, as outlined above.

Chemical and protein reagents

sCD40L, TNF-α, and IL-1β (R&D Systems, Minneapolis, MN, USA) were used at a concentration of 1 μg/μl and 10 and 1 ng/ml, respectively. L-Glutamic acid and carbamyl PAF (a nonhydrolyzable analog of PAF) were purchased from Sigma-Aldrich and used at a concentration of 5 ng/ml or 20 nM, respectively. ADP was purchased from Chrono-log (Havertown, PA, USA) and used at a concentration of 5 µM. MK801, CNQX, and 10058-F4 (Abcam, Cambridge, MA, USA) were used at concentrations of 10, 10, or 100 µM, respectively. S4CPG and actinomycin D (Tocris Bioscience, Avonmouth, Bristol, United Kingdom) were used at concentrations of 1 mM and 10 µg/ml, respectively.

Network assembly

The network was constructed by consulting a large number of primary research manuscripts and reviews, as well as signaling databases (KEGG [25]), to determine potential signaling pathways associated with CD40L. To synthesize signaling information upon induction by GSH, signaling mediators’ activity known to be modulated by ROS and GSH was used. The target nodes of the network were transcription factors that were chosen based on binding-site prediction within the 2 kb nucleotide sequence upstream of the PSGL-1 start site, as reported in the Encyclopedia of DNA Elements. However, only a subset of these was found to be connected to CD40L/glutamate-induced pathways, based on the available literature. Thus, network nodes represent proteins, and directional edges connecting nodes represent either activation or inhibition of the downstream nodes.

Discrete dynamic model implementation

The network was developed into discrete dynamical models by characterizing each node with a variable that can take the ON state when the concentration or activity is above the threshold level necessary to activate downstream immune processes or the OFF state when activity is below this threshold. The evolution of the state of each node was described by a Boolean transfer function. The operator AND was used to describe a synergistic or conditional interaction between 2 or more nodes that is necessary to activate the downstream node. When either of the nodes was sufficient for the activation of the downstream node we used the operator OR. An inhibitory effect was represented by an AND NOT operator. In cases where prior biologic information did not completely determine the transfer functions (e.g., there was no information whether 2 coincident regulatory effects are independent or synergistic), different alternative transfer functions were tested. The transfer functions that reproduced the qualitative features of the sCD40L-induced signaling were selected. For those nodes whose transfer function includes both positive and negative regulatory arguments and for which there was >1 of either, the total number of TRUE positive and negative regulatory arguments was balanced to determine the nodes’ updated state [26]. Thus, more TRUE positive regulatory arguments than negative regulatory arguments will result in the node’s update status being TRUE. The Boolean transfer functions applied in our model provide a mechanistic understanding of the crosstalk between CD40L and glutamate-induced signaling pathways.

The status of the system across time was simulated by repeatedly applying the Boolean rules for each node until a stationary state was found (e.g., no change in the activities of the following key nodes: p52, p65, STAT protein, CTCF, ERK, JNK, p38, GSH, GSSG, O₂, and H₂O₂). As the kinetics and time scales of the individual processes represented as edges are not known, a random order asynchronous update was selected, wherein the time scales of each regulatory process were randomly chosen in such a way that the node states were updated in a randomly selected order during each time step [27–29]. The asynchronous algorithm was X_i^t = F_i(X_a^ta, X_b^tb, X_c^tc,…), where F is the Boolean transfer function for node i, and t_a, t_b, and t_c represent the time points corresponding to the last change in the state of the input nodes a, b, c and can be in the previous or current time step. The time step (time unit) of our model corresponds approximately to 1.5 h, based on the time of PSGL-1 transcriptional induction, and predicted c-Myc activity reported here. The randomized asynchronicity of the model does not alter the steady states of the dynamic system but causes stochasticity in the trajectory between the initial conditions and the equilibria (attractors); thus, it can sample more diverse behaviors compared with traditional synchronous models. To determine the node consensus activity over time (i.e., shared trajectories with different update orders), the simulations were run 800 times. The simulated activity profiles for each node were estimated by calculating the fraction of simulations in which the nodes were in an ON state at each time step. Furthermore, network perturbations, corresponding to constitutive activation and KO simulations, were simulated by freezing the state of a node to TRUE (constitutive activation) or FALSE (KO) for the duration of the simulation. This process was conducted for each node in the network, for each treatment simulation conducted.

Our approach of using discrete dynamic modeling allowed us to sample the time scales of interactions and perform replicate simulations, as well as provide continuously varying activities of the network nodes over time, which ranged between the lower limit of 0 (below-threshold concentration in all runs) and upper limit of 1 (above-threshold concentration in all runs).

Transcriptomic data analysis

A previously published HIV infection dataset (GSE52900) was used for the transcriptomic data analysis. In brief, the study purified CD14⁺ monocytes from whole blood, collected from 5 HIV⁻ donors and 4 HIV⁺ donors with undetectable viral loads. Total RNA was isolated and analyzed for transcript expression using HumanHT-12 v3 Expression BeadChips (Illumina, San Diego, CA, USA). Log2-transformed and robust spline-normalized data were obtained from the Gene Expression Omnibus. Lists of differentially expressed genes were obtained from a prior publication [30], which was identified by a linear model fit in conjunction with empirical Bayes statistics. These data were analyzed here for transcription factor-binding site enrichment by a hypergeometric test, and transcription factor targets identified in previous publications were used [31]. A total of 722 binding sites were tested, and the P values were corrected for multiple hypothesis testing by the Benjamini-Hochberg method. These binding sites mapped to 654 transcription factors. Additionally, Gene Set Enrichment Analysis [32] was performed using all gene sets available from the Molecular Signatures Database, and P values were corrected for multiple hypothesis testing as above. All analysis was performed using the Bioconductor software package in R [33].

Statistical analysis

Descriptive statistics are reported as means ± sd throughout the manuscript. One-way ANOVA with Bonferroni post hoc test was used in analyzing data (see Figs. 2 and 3). Two-way ANOVA F-test for main effects and interaction and Bonferroni post hoc test were applied to data (see Figs. 1, 2, and 7). Student’s t test was applied to data (see Figs. 1 and 3), and the Pearson correlation test was used to evaluate correlation between PSGL-1 and CD163 levels. These statistical analyses were conducted using the Prism statistical package (GraphPad Software, La Jolla, CA, USA).

Figure 2. sCD40L and glutamate together induce monocyte PSGL-1 expression ex vivo via transcriptional up-regulation.

Monocytes were isolated from 3 HIV⁻ donors by positive selection against CD14 after density gradient isolation of PBMCs. The monocytes were treated with glutamate, sCD40L, or both for 24 h, and PSGL-1 expression was determined by (A) flow cytometry or (B) immunoblot assay. PSGL-1 expression is increased with sCD40L + glutamate but not sCD40L or glutamate alone. (A) Data were normalized to the not-treated, without CD40L + glutamate (NT) value for each replicate and analyzed by 1-way ANOVA F-test, P = 0.0335, or 1 sample t test vs. NT (1.000), glutamate + CD40L P = 0.0135. (C) Monocytes were prepared and treated as in A and B, but total RNA was collected at 18 h post-treatment, and PSGL-1 transcript levels were determined by qRT-PCR. PSGL-1 transcription is increased only with the combination treatment. One-way ANOVA F-test, P = 0.0023, or 1 sample t test vs. NT (1.000), glutamate + CD40L P < 0.01. (D) Monocytes prepared and treated as in A and B, with or without the addition of actinomycin D, and PSGL-1 expression was analyzed by flow cytometry. Transcriptional up-regulation is required for sCD40L + glutamate-mediated PSGL-1 induction. Two-way ANOVA, treatment: P = 0.0089; inhibitor: P < 0.0001; interaction: P = 0.0098; post hoc test: NT vs. treated with CD40L + glutamate (T), **P < 0.01; 1 sample t test vs. NT (1.000), P < 0.05. *P < 0.05, **P < 0.01. NI, No inhibitor, (without actinomycin D).

Figure 3. Plasma glutamate is increased with HIV infection.

Plasma from 8 HIV⁻ and 8 cART-treated HIV⁺ donors with undetectable viral loads was collected and frozen and then analyzed by LC-MS/MS for glutamate expression. Expression was normalized to known standards. Plasma glutamate is increased significantly with HIV infection. Student’s t test, **P = 0.0036.

Figure 1. Monocyte PSGL-1 and plasma glutamate are induced with HIV infection and murine EcoHIV infection.

(A) Whole blood was collected from 32 HIV⁻ donors and 28 cART-treated HIV⁺ donors with undetectable viral loads and analyzed by flow cytometry. Monocytes were identified by FSC/SSC and then CD14 staining and analyzed for PSGL-1 MFI. PSGL-1 is increased significantly with HIV infection. ***P = 0.0204, Student’s t-test. (B) Monocytes subsets were identified by CD14 and CD16 staining, and (C) PSGL-1 expression was determined in each subset and analyzed by 2-way ANOVA F-test. HIV⁺ vs. HIV⁻: P = 0.0011; subsets: P = 0.0068; post hoc tests: HIV⁺ vs. HIV⁻; Classic: **P < 0.01; Intermediate: **P < 0.01. (D) WT C57BL/6 mice (n = 6) and CD40 KO (n = 6) mice were infected with EcoHIV (1.0 × 10⁵ pg p24), and blood was collected by cardiac exsanguination at 2 wk postinfection and analyzed by flow cytometry. Monocytes were identified by FSC/SSC and then by CD115 and Gr-1 staining and analyzed for PSGL-1 expression. Monocyte PSGL-1 expression is increased with EcoHIV infection in a CD40L-dependent manner. Two-way ANOVA, interaction: P = 0.0269; post hoc test: Uninfected vs. EcoHIV Infected, *P < 0.05.

Figure 7. c-Myc-MAX dimerization inhibition by 10058-F4 prevents sCD40L + glutamate-induced PSGL-1 on monocytes ex vivo.

Monocytes were isolated from 3 HIV⁻ donors by positive selection against CD14 after density gradient isolation of PBMCs. The monocytes were treated with glutamate and sCD40L and/or the c-Myc/MAX dimerization inhibitor, 10058-F4, for 24 h, and PSGL-1 expression was determined by flow cytometry. c-Myc/MAX is required for sCD40L + glutamate-induced PSGL-1 expression on monocytes ex vivo Two-way ANOVA, Treatment: P = 0.0078, Inhibitor: P < 0.0001, Interaction: P = 0.0303, Post-test: not treated, without CD40L + glutamate (NT) vs. treated, with CD40L + glutamate (T), P < 0.01, **P < 0.01, ***P < 0.001. NI, No inhibitor, without 10058-F4.

RESULTS

Expression of PSGL-1 is increased in monocytes of HIV-infected persons

Based on previous reports of PSGL-1 induction during other inflammatory conditions, including uveitis, diabetes, and chronic obstructive pulmonary disorder [11–13], PSGL-1 expression in monocytes from HIV⁻ and HIV⁺ individuals was investigated via flow cytometry and recorded as MFI. In these experiments, monocytes were identified by FSC/SSC, followed by positive staining with CD14-specific antibodies. As shown in Fig. 1A, the monocytes derived from HIV-infected individuals exhibited greater levels of PSGL-1. Furthermore, these monocytes showed an upward trend in the expression of activation marker CD40 and a nonsignificant decrease in CD163 expression (data not shown), with no apparent correlation between PSGL-1 and CD163 levels (Supplemental Fig. 1). In addition, statistically significant increases in PSGL-1 levels were observed in lymphocytes and granulocytes of HIV⁺ subjects (Supplemental Fig. 2A and B, respectively). This observation contrasts with the findings reported by Liang et al. [34], who showed that the expression of PSGL-1 is down-regulated in monocyte subsets during HIV infection in a manner that is negatively correlated with the plasma levels of sCD163, although an inverse correlation between CD163 expression on monocytes and the plasma levels of sCD163 was not observed in this study, as reported by other groups [35].

As Liang et al. [34] also reported a decrease in PSGL-1 in all monocyte subsets, monocytes were gated, according to the expression levels of CD14 and CD16 (Fig. 1B), using the very same criteria defined by Liang et al. [34], and PSGL-1 levels were assessed. Consistent with the increased PSGL-1 expression observed in the bulk monocyte population (Fig. 1A), all 3 monocyte subtypes were found to contain elevated levels of PSGL-1 (Fig. 1C). One of the key differences between the cohort studied by Liang et al. [34] and the cohort studied here is the use of cART; therefore the following experiments were conducted to determine if infection alone could lead to an increase in PSGL-1 expression.

Expression of PSGL-1 in monocytes is increased following EcoHIV infection of mice in a CD40L-dependent manner

C57BL/6 mice were infected for 2 wk with the EcoHIV. This chimeric virus was developed previously by Volsky and colleagues [36] using HIV-1/NL4-3, where the segment encoding for the protein gp120, responsible for receptor binding, was replaced with the sequence for the corresponding murine leukemia virus type 1 protein, gp80. As previously reported, this virus is able to infect mice productively, including the induction of inflammatory markers and tissue infiltration of immune cells [24, 36–38]. Compared with control mice that were left uninfected, the PSGL-1 expression was found to be increased in CD115⁺GR1⁺ monocytes (WT; Fig. 1D), as well as in CD3⁺ T cells and CD19⁺ B cells (Supplemental Fig. 3A and B, respectively).

As HIV replication is restricted in monocytes, with productive infection requiring their differentiation into Mϕ [39–41], HIV may stimulate production of PSGL-1 in monocytes indirectly via the promotion of an abundance of proinflammatory mediators in the circulation. As CD40L signaling is reported to be perturbed following excess production of sCD40L in HIV-infected individuals, as well as being implicated in the pathogenesis of HAND [42–45] and atherosclerosis [46, 47], it was tested whether CD40L is required for HIV-induced PSGL-1 induction. CD40L-deficient mice (CD40L KO) were infected for 2 wk with EcoHIV, and PSGL-1 expression was determined. In the CD40L KO mice, EcoHIV infection failed to induce PSGL-1 expression in CD115⁺GR-1⁺ monocytes (CD40L KO; Fig. 1D), as well as in CD3⁺ T cells and CD19⁺ B cells (Supplemental Fig. 3A and B, respectively). Complementary to this observation, a significant interaction between CD40L and EcoHIV infection, with regard to PSGL-1 expression, was noted in CD3⁺ T cells of WT mice (interaction, P = 0.0269; Supplemental Fig. 3A), whereas EcoHIV-stimulated PSGL-1 expression in CD19⁺ B cells was found to be independent of CD40L (Supplemental Fig. 3B). Together, these data suggest that EcoHIV infection alone results in increased PSGL-1 expression in leukocytes, corroborating the observations noted in HIV-infected people (Fig. 1A–C), and that this induction is largely CD40L dependent.

Glutamate and sCD40L together are able to induce PSGL-1 expression, via transcriptional up-regulation, in primary human monocytes

To determine what factor(s) might mediate the observed increase in PSGL-1 levels in monocytes of HIV⁺ persons, a number of soluble factors reported to be increased with HIV infection were tested. Monocytes from healthy, HIV⁻ donors were isolated and treated with TNF-α, IL-1β, ADP, and PAF for 24 h, after which, PSGL-1 expression was determined by flow cytometry. None of these factors alone were able to induce PSGL-1 expression in monocytes (data not shown). Additionally, both sCD40L and glutamate were tested alone or in combination. Whereas 24 h treatment with either factor individually failed to induce PSGL-1 expression, as determined by flow cytometry, the combination of sCD40L and glutamate was able to induce a significant increase in PSGL-1 expression on the surface of monocytes (Fig. 2A). This increase was confirmed by immunoblotting of whole-cell lysate (24 h post-treatment; Fig. 2B) and at mRNA level by qRT-PCR (18 h post-treatment; Fig. 2C). Finally, it was determined that this induction required transcription, as actinomycin D treatment prevents PSGL-1 induction by the combination of sCD40L and glutamate (24 h post-treatment; Fig. 2D). Thus, the combination of sCD40L and glutamate is able to induce a transcription-dependent increase of PSGL-1 in human monocytes. The data may also indicate positive feedback in PSGL-1 regulation, as basal expression was also decreased with transcriptional inhibition; alternatively, this may represent rapid turnover of PSGL-1.

To investigate how glutamate may trigger synthesis of PSGL-1, pharmacologic inhibitors that block 2 major classes of glutamate receptors (NMDA subtype and AMPA/kainate receptor) were tested. The NMDA receptor channel antagonist MK801 or the AMPA receptor antagonist CNQX failed to block PSGL-1 induction (Supplemental Fig. 4A), suggesting that glutamate is acting in a receptor-independent fashion to induce PSGL-1 expression. The levels of the glutamate receptor family member GluR1 were also assessed by immunoblot assay, and no change in the total GluR1 protein levels was detected (data not shown). As the anionic amino acid transporter xCT acts as a facilitative antiporter, the excess extracellular glutamate might antagonize the normal exchange of intracellular glutamate for extracellular cystine. To test this notion, we used the xCT inhibitor S4CPG. In conjunction with CD40L, but not alone, S4CPG is able to induce PSGL-1 expression in monocytes (Supplemental Fig. 4B). In this case, the magnitude of PSGL-1 induction was similar to what was observed in glutamate plus sCD40L-treated monocytes (see Fig. 2A). As xCT serves as a major mechanism by which cells import cystine, which in turn, is important for the synthesis of the antioxidant GSH [48], if xCT activity is impaired in HIV-infected individuals, it would be expected that there is reduced GSH levels in HIV-infected individuals. The cohort studied here does have reduced plasma GSH (Supplemental Fig. 4C). Therefore, we concluded that the glutamate is likely acting in a receptor-independent fashion to induce PSGL-1 expression in monocytes, when in combination with sCD40L. Specifically, glutamate is likely acting to impair xCT-mediate cystine import and consequently, GSH synthesis. Whereas the mechanism proposed suggests that HIV might act indirectly to impair GSH synthesis, it has been reported that in HIV-infected cells, GSH is depleted and that this is required for regular viral replication [49]. Furthermore, whereas an inhibitor of xCT was able to mimic the effects of glutamate in combination with sCD40L, there are still potential alternative routes of action. There are other glutamate receptor families besides the NMDA, AMPA, and kainate families, such as metabotropic glutamate receptors, which are expressed on a variety of cell types outside of the CNS [50]. Thus, it cannot be concluded definitively that xCT inhibition mediates glutamate and CD40L-induced PSGL-1 expression.

Plasma glutamate is increased in HIV-infected individuals

Increased sCD40L in HIV-infected individuals is well characterized [51], as is increased extracellular glutamate in the CNS of HIV-infected individuals [52]; however, it is unclear if glutamate is similarly increased in the circulation of HIV-infected subjects recruited in our studies. Thus, plasma was collected from 8 HIV⁺ and 8 HIV⁻ donors, and glutamate content was determined by LC-MS/MS. Glutamate is increased significantly in HIV-infected individuals compared with healthy controls (Fig. 3). These results are consistent with previous reports in which elevated levels of glutamate were observed in the blood plasma of persons with HIV [53, 54].

Network of signaling events induced by glutamate or sCD40L reveals a highly complex signaling network

Despite the genomic organization of the PSGL-1 gene in both mice [55] and humans [56] being known for nearly 20 yr, there are no reports on the transcriptional regulation of PSGL-1. So, as to determine what factors might regulate its induction by the combination of CD40L and glutamate, a network describing signaling events downstream of CD40L and/or glutamate was constructed (see Materials and Methods). The completed network is illustrated in Fig. 4. The full network contains 155 nodes (detailed in Supplemental Table 1), or signaling mediators, and 341 edges, or regulatory interactions, and illustrates a nonassortive network (assortivity coefficient = −0.035), with an average node degree of 4.4, node clustering coefficient of 0.1982179285, and node betweeness centrality of 0.012. Thus, in this network, mediators with a large number of regulatory interactions do not preferentially interact with other mediators that also have a large number of regulatory interactions. Each mediator also has an average of 4.4 interacting partners; these partners tend not to interact with each other, and there is no 1 node that connects all other pairs of nodes. Of note, the node representing H₂O₂ is the node with both the highest degree (34) and betweeness centrality (0.23). Additionally, the nodes representing the forkhead box O transcription factor family and the IκB kinase α are among the top 22 nodes with highest degree (17 and 10, respectively), connectivity (0.58 and 0.83), and betweeness centrality (0.13 and 0.035). Finally, other nodes of special interest that are among the top 22 for both node degree and betweeness centrality include those representing factors related to redox stress and its regulation (H₂O₂, O₂, GSH), factors sensitive to redox stress or GSH levels (PP2A, GSTP), MAPK family members (p38, JNK, ERK), and others (AMPK, protein kinase A, IκB).

Figure 4. Network graph of signaling pathways induced by CD40L and glutamate and transcription factors, which may be activated by these pathways that are predicted to bind near the PSGL-1 start site.

Primary and secondary sources were consulted for signaling interactions that might mediate signaling downstream of CD40L, glutamate, or GSH depletion via manual curation of search results from PubMed. Additionally, signaling databases, such as KEGG pathways, were consulted. The ENCODE database was consulted to identify transcription factors that might bind around the PSGL-1 start site, and those for which a connection was found with the signaling pathways induced by CD40L or glutamate were encoded. All signaling interactions were coded as binary interactions in Python using NetworkX and matplotlib libraries. Negative regulatory interactions are signified by a red edge, whereas positive regulatory interactions are signified by a green edge. Directionality of the interaction is indicated by the weighting of the side of the edge closest to the regulated node. The network contains 155 nodes and 341 edges.

The network was simulated by developing a discrete dynamic model, described in Materials and Methods. The model was used to test whether it can recapitulate known, CD40L-induced dynamics. Simulations reveal that the canonical NFκB pathway, but not the noncanonical pathway, is strongly induced by CD40L (Fig. 5A and C), and this is indeed what has been reported as occurring in CD40L-treated cells (Pearson et al. [57]). Additionally, the model recapitulates robust activation of the MAPK ERK and weaker, transient activation of JNK, as described previously (Fig. 5B) [57, 58].

Figure 5. Boolean network model correctly predicts aspects of CD40L-induced signaling.

The network illustrated in Fig. 4 was implemented in Python using the Boolean net library, with logic rules describing the regulatory interaction between nodes, inferred based on the literature describing the regulation. A total of 800 simulations, with 80 time steps, of the CD40L-alone scenario were run using an asynchronous update schedule. The percentage of simulations in which a node was on or had a value of 1 is shown for (A) the NF-κB family members, p65 and p52; (B) the MAPK family members, ERK, JNK, and p38; and (C) the transcription factor complexes p50/p65, p52/RelB, cFos/cJun (AP-1), c-Myc/MAX, and MAX/MAX. (D) The steady-state values for these transcription factor complexes, taken as the 30th time step, are shown. The canonical NF-κB pathway is predicted to be more strongly activated than the noncanonical pathway, with moderate AP-1 and c-Myc/MAX activation.

Boolean network modeling predicts c-Myc as a likely mediator of glutamate and sCD40L-induced PSGL-1 induction

As PSGL-1 expression was induced in the joint glutamate and CD40L treatment, the model was used to predict transcription factors with differential activation in the glutamate and CD40L in silico stimulation (Fig. 2). Figure 6A–D shows the dynamics of select signaling mediators predicted in response to the combination treatment, whereas Fig. 6E–H shows the change in activation for the same set of intermediates in the combination treatment scenario compared with the CD40L alone scenario. These mediators include NF-κB members, MAPK members, ROS, GSH, and several transcription factors. One of the most notable features is a decrease in GSH and a corresponding increase in ROS. This, in turn, is predicted to result in altered transcription factor activity (Fig. 6I and J). c-Myc had the highest fold change in the joint treatment compared with the single treatment, suggesting its role in transcriptional regulation of the genes induced in the joint treatment. AP-1 also showed significant differential activity upon joint treatment.

Figure 6. Boolean network model reveals signaling dynamics downstream of sCD40L + glutamate treatment of monocytes and predicts c-Myc as a likely regulator of PSGL-1 induction.

The network illustrated in Fig. 4 was implemented in Python using the Boolean net library, with log rules describing the regulatory interaction between nodes inferred based on the literature describing the regulation. A total of 800 simulations, with 80 time steps, of the CD40L + glutamate-alone scenario were run using an asynchronous update schedule. The fraction of simulations in which a node was on or had a value of 1 is shown for selected signaling intermediates in A–D. The difference in the percentage of simulations in which a node was “on” between the combination scenario and CD40L-alone scenario is shown in E–H. A positive value indicates increased activation in the combination scenario. (A and E) The NF-κB family members p52 and p65 are shown. (B and F) The MAPK family members ERK, p38, and JNK are shown. (C and G) The reactive oxygen intermediates O₂ and H₂O₂ are shown. (D and H) The reduced GSH and oxidized GSSG form of GSH are shown. Shown are the dynamics of the transcription factor complexes p50/p65, p52/RelB, cFos/cJun (AP-1), c-Myc/MAX, and MAX/MAX (I) and the change in the steady-state values relative to CD40L alone (J).

Glutamate and sCD40L induce c-Myc activation in primary human monocytes ex vivo, and c-Myc inhibition prevents PSGL-1 induction but not basal expression

To validate first whether c-Myc is indeed differentially activated in monocytes during HIV infection, publicly available microarray data derived from monocytes isolated from 4 HIV⁻ donors and 5 HIV⁺ donors, responding well to cART [30], were analyzed. Particularly, enrichment of transcription factor-binding sites, 2 kb upstream of the start site of up-regulated genes, was estimated. Of note, c-Myc-binding sites are all enriched among up-regulated genes in monocytes from HIV⁺ compared with HIV⁻ individuals (P = 0.023), as are binding sites for the c-Myc-binding partner, MAX. Thus, these observations validate the model predictions, and it is likely that c-Myc is activated in monocytes during HIV.

Next, it was determined that c-Myc does regulate glutamate and sCD40L-induced monocyte PSGL-1 induction ex vivo. c-Myc acts in complex with a number of other factors to regulate transcription. One such protein is MAX, which is well described as acting in a complex with c-Myc to promote transcription [59]. As shown in Fig. 7, the treatment with glutamate plus sCD40L in the presence of the c-Myc/MAX dimerization inhibitor 10054-F8 failed to induce PSGL-1 in monocytes. Importantly, these results suggest that c-Myc/MAX dimerization is required for PSGL-1 induction. Whereas more will need to be done to confirm the role of c-Myc in PSGL-1 regulation, these data corroborate a recent report [60] and provide evidence that c-Myc is likely activated in monocytes during HIV infection and thereby, modulates PSGL-1 levels.

Network analysis reveals potential targets for therapeutic intervention

To determine which nodes or signaling intermediaries have the greatest role in c-Myc activation, simulations of systematic knockdown or constitutive activation of each node in the network was run. As seen in Fig. 8, a number of signaling intermediaries were found to have a large effect on c-Myc activation, with the top 22 being shown. Of note, the perturbation analysis suggests a crucial role for GSH, ROS, and the MAPK pathway in mediating c-Myc activation, in agreement with previous literature on oxidative stress-induced c-Myc activation [61]. Interestingly, 7 of the above nodes also have high node degree and betweeness centrality. Additionally, it would appear that c-Myc activity is primarily regulated via positive regulatory mechanisms, as only about one-third of the nodes that had a large impact on c-Myc were found to exert an effect during constitutive activation, suggesting an inhibitory role here. Interestingly, almost all of these nodes—those for which constitutive activation is predicted to lead to decreased c-Myc activity—are redox stress regulators or sensors, with the exception of the MAPK ERK and the DNA-binding protein, CTCF.

Figure 8. Knockdown and constitutive activation perturbation simulations reveal key regulatory nodes in glutamate + CD40L-induced c-Myc activation.

Each node in the network was systematically, individually forced permanently into an “off” state, knocked down (KD), or an “on” state, constitutively activated (CA), during the combination treatment scenario. Results from these simulations were then compared with the unperturbed combination treatment scenario, and changes in c-Myc/MAX activity are shown. NOX, Nitrogen oxides; Tpl2, tumor progression locus 2; PTP1b, protein tyrosine phosphatase 1b.

DISCUSSION

Here, the physiologically relevant factors CD40L and glutamate are identified as transcriptional inducers of PSGL-1 in monocytes, likely via c-Myc. That PSGL-1 is differentially regulated with HIV infection is consistent with previous reports of altered PSGL-1 expression during various inflammatory diseases [11–13]. Whereas there is another report that contradicts ours on this point [34], it is likely that this difference is associated with differences in the patient population studied.

In the work of Liang et al. [34], a cohort of men who have sex with men and acquired HIV predominantly within 12 mo of last-negative detection were shown to possess a higher proportion of nonclassic CD14^dimCD16⁺ and intermediate CD14^highCD16⁺ monocytes in the circulation. It was also demonstrated that these cellular subtypes, including their classic CD14^highCD16⁻ counterparts, expressed PSGL-1 at significantly lower levels [34]. The population studied here is older, has been on cART longer, and has been infected with HIV for a longer period of time than the population in the conflicting report, which likely accounts for the differences in the PSGL-1 expression levels observed. We have examined the effects of HIV infection on PSGL-1 expression in monocytes of older subjects (55 ± 11 yr) who were receiving ART, whereas Liang et al. [34] examined PSGL-1 levels in monocytes of younger patients (ranging from 27 to 33 yr of age) who were treatment naive. Thus, there may be age-specific differences in the effect of HIV infection on monocytes in the presence of ART. In support of this notion, previous work has demonstrated a positive correlation between the phenotypic alterations in monocytes and disease progression in HIV-infected persons on stable ART [62, 63].

The methods illustrated here may be generally applicable in the identification of the transcriptional regulators of genes differentially expressed on circulating cell types. The pipeline suggested here has several steps. First, a change in protein expression needs to be identified. Next, physiologically relevant mediators need to be screened to identify candidate factors. Following this, a Boolean network model will be generated to connect signaling pathways downstream of the candidate soluble mediators with transcription factors predicted to bind around the gene of interest’s start site. As these signaling interactions will be determined based on a review of the available literature, not all pathways may be amenable to such an analysis. Additionally, if relevant transcriptomic data are not available, then the data may be generated and analyzed for transcription factor binding-site enrichment so as to support the model’s predictions. Finally, the model’s predictions should be validated, even if only preliminarily, so as to determine if any further refinement of the model is necessary.

That glutamate is acting on leukocytes to modulate signaling is perhaps surprising. It seems likely that glutamate is acting by inhibiting the anionic amino acid transporter xCT, thereby impairing GSH synthesis and the ability of cells to buffer ROS, including those reported to be induced by CD40 signaling [64]. Whereas GSH antagonism has been reported to be a common feature of diverse viral infections, in the primary infected cell (reviewed in Aquilano et al. [65]), the data presented here suggest that GSH synthesis might be antagonized even in uninfected cells. This is important, as it suggests that even in cART-treated, virally suppressed, HIV-infected individuals, HIV may be able to induce PSGL-1 expression. Not only in active infection but also in suppressed infection, CD40L and glutamate are induced [42]. This reflects the continued expression of proinflammatory HIV genes, such as Tat and gp120, despite suppression of viral replication [66]. Importantly, as other proinflammatory cytokines have been reported to induce ROS generation, including TNF-α, IL-6, and IL-1β [67], glutamate may be able to act in conjunction with these as well to induce c-Myc-mediated PSGL-1 expression. It is worth noting that these cytokines have also been reported to be increased with HIV infection [68–70]. Whereas sCD40L and glutamate were able to induce PSGL-1 expression ex vivo, the role of oxidative stress in mediating this induction is likely more significant in vivo. It has been reported in HIV-infected individuals with prolonged cART use, particularly nucleotide RT inhibitor use, that mitochondrial replication is impaired, resulting in impaired respiration [71] and likely a corresponding increase in ROS production as a result of normal cellular respiration. This suggests that in the context of cART-treated HIV infection, monocytes might be even more sensitive to GSH depletion than the data here suggest.

A better understanding of PSGL-1 regulation could facilitate novel therapeutic strategies for the treatment of secondary disorders. In particular, many complications associated with HIV infection, including CVDs and neurocognitive disease, feature excessive leukocyte tissue infiltration as a significant component of their pathology. Importantly, sCD40L has been implicated in a variety of other disorders, including inflammatory bowel disease and atherosclerosis [72, 73], whereas glutamate dysregulation is common to a variety of neurocognitive disorders [74]. Thus, the mechanism outlined here may be common to many conditions.

To confirm the significance of PSGL-1 induction, it will be necessary to validate if increased PSGL-1 induction does, in fact, promote increased leukocyte transmigration. The identification here of c-Myc as being involved in PSGL-1 induction by sCD40L and glutamate, together with its predicted binding to a region around the PSGL-1 start site, calls for further investigation of the role of c-Myc in regulating PSGL-1 transcription. If c-Myc is found to bind directly to the promoter region of PSGL-1, then the role of increased PSGL-1 expression in leukocyte transmigration can be specifically interrogated via genetic manipulation.

Cancer is one such disease in which sCD40L and glutamate have been implicated, including glioblastomas, adenocarcinomas, and squamous cell carcinomas of the lung [75]. Furthermore, as sCD40L is thought to be primarily platelet derived [76], it is not surprising that as is seen with HIV infection, platelet activation is increased during cancer [77]. Increased levels of markers of oxidative stress also feature prominently in cancer pathogenesis. Thus, it seems likely that PSGL-1 may be similarly up-regulated on leukocytes during cancer. Importantly, it has been suggested that monocyte PSGL-1 expression is required for tumor metastasis, and this is thought to represent priming of the secondary site by infiltrating monocytes [78]. Additionally, c-Myc has been shown to regulate many of the genes associated with alternative activation-type Mϕ [79]. Thus, if the mechanism outlined here is also implicated in cancer pathogenesis, then it may contribute to both metastasis and immune evasion.

Whereas it appears likely that c-Myc regulates PSGL-1 expression in the system described here, c-Myc is not an ideal therapeutic target. c-Myc is well characterized as a regulator of the cell cycle and has been implicated in a variety of other cellular processes; thus, therapeutic targeting of c-Myc is likely to have variety of side effects. More amenable to targeting is either sCD40L- or glutamate-induced ROS. sCD40L is thought to be primarily platelet derived [76], and so, anti-platelet therapy may be beneficial. Not only is platelet activation important in the formation of platelet monocyte complexes, but also, the complexes have been described as promoting leukocyte transmigration. Furthermore, platelet activation has been implicated in the pathogenesis of disorders, such as atherosclerosis and HAND. The targeting of the increase in ROS may also be beneficial, particularly as the model developed here shows that c-Myc activation is especially sensitive to alterations in ROS and ROS sensors. That ROS can lead to c-Myc activation has been demonstrated previously [61], and ERK was similarly implicated in this study. NAC has been shown to be able to inhibit HIV replication in vitro [80], via inhibition of HIV-induced GSH antagonism. While the effects of NAC on HIV pathogenesis have not been well characterized in vivo, it is possible that they may inhibit the development of many HIV-associated complications, in part, as a result of the role of oxidative stress in PSGL-1 induction. Finally, it will be important to determine the primary source of the increased serum glutamate observed. Not only will this provide mechanistic insight, but it will also suggest additional targets for therapeutic intervention. Thus, there are a number of potential methods of targeting sCD40L- and glutamate-induced PSGL-1 induction in the context of HIV infection.

AUTHORSHIP

R.C. was involved in the design of all of the experiments and was primarily responsible for their execution and also prepared the manuscript. L.D.J. prepared the EcoHIV used in the mouse experiments and helped execute the mouse experiments. X.Q. conducted the power analysis and provided oversight on the use of statistical methods. J.T. contributed to the design and execution of the computational biology and bioinformatics experiments. S.B.M. contributed to the design of the human, mouse, and in vitro experiments, as well as provided guidance on their execution. He also edited the manuscript extensively.

Glossary

10058-F4

(Z,E)-5-(4-ethylbenzylidine)-2-thioxothiazolidin-4-one

AMPA

2-amino-5-hydroxy-5-methyl-4-isoxazolepropion acid

AMPK

AMP-activated protein kinase

APC

allophycocyanin

ART

antiretroviral therapy

cART

combined antiretroviral therapy

CD40L

cluster of differentiation 40 ligand

CNQX

6-cyano-7-nitroquinoxaline-2,3-dione

CTCF

CCCTC-binding factor

CVD

cardiovascular disease

EcoHIV

mouse-tropic HIV

FSC

forward-scatter

GSH

glutathione

GSSG

glutathione disulfide

GSTP

glutathione S-transferase

HAND

HIV-associated neurocognitive disorder

KEGG

Kyoto Encyclopedia of Genes and Genomes

KO

knockout

LC

liquid chromatography

MFI

mean fluorescence intensity

MS/MS

tandem mass spectrometry

NAC

N-acetyl cysteine

NMDA

N-methyl-d-aspartate

O₂

superoxide

p38

p38 MAPK

p52

NF-κB p52 subunit

p65

NF-κB p65 subunit

PAF

platelet-activating factor

PP2A

protein phosphatase 2A

PSGL-1

P-selectin glycoprotein ligand-1

qRT-PCR

quantitative RT-PCR

ROS

reactive oxygen species

S4CPG

(S)-4-carboxyphenylglycine

sCD40L

soluble cluster of differentiation 40 ligand

sCD163

soluble cluster of differentiation 163

SRM

selected reaction monitoring

SSC

side-scatter

WT

wild-type

xCT

functional subunit of the cystine/glutamate transporter x_c⁻ system

DISCLOSURES

The authors declare no conflicts of interest.