Introduction
Despite effective antiretroviral therapy (ART), people living with HIV (PLWH), particularly those with incomplete CD4+ T-cell recovery on ART, remain at higher risk for age-related morbidities.1–3 Although CD4 depletion is a hallmark of HIV disease progression, immune activation and inflammation also play critical roles in HIV pathogenesis, as well as in the occurrence of serious non-AIDS events in immunological non-responders (INRs).4–8 Thus, interventions aimed at reducing inflammation and immune activation may provide a clearer understanding of the underlying pathogenesis and offer potential therapeutic benefits. Since 2001, scientists have attempted to treat INRs with medications such as prednisone,9 hydroxychloroquine,10 and other non-specific candidates, including atorvastatin,11 valganciclovir,12 rifaximin,13 and maraviroc.14 However, our recent systematic review and meta-analysis showed that while some of these candidates can reduce immune activation or chronic inflammation, they rarely have the desired effect of directly recovering CD4+ T-cell counts.15 Therefore, an effective, safe, inexpensive, and convenient intervention that promotes CD4+ T-cell recovery would be highly beneficial. One potential therapeutic target for such an intervention is inflammation and immune activation.
Plants continue to be a vital source of new medicines and chemical entities. Artemisinin and its analogs, such as dihydroartemisinin (DHA), have been used for decades in malaria treatment.16 DHA has demonstrated numerous beneficial immunomodulatory and anti-inflammatory properties in animal models of lupus arthritis,17,18 inflammatory bowel disease,19 and renal fibrosis,20 conditions characterized by high levels of inflammation and immune activation. Moreover, DHA is a once-daily oral treatment with an excellent long-term safety profile and low cost.21–23 Therefore, we hypothesized that DHA might be an ideal candidate to reduce immune activation and inflammation, thereby promoting CD4+ T-cell recovery in INRs.
The Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/ ) database has greatly enhanced our understanding of immune activation and inflammation in HIV disease pathogenesis. Fortunately, efficient integrated bioinformatics and chemical informatics methods have been developed for large-scale, cross-platform high-throughput data analysis. In this study, we identified differentially expressed genes (DEGs) from microarray and RNA-seq data. Gene Ontology (GO) function and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, as well as protein-protein interaction (PPI) network analysis of DEGs, were performed. The potential binding proteins of DHA were then predicted through molecular docking. To further confirm the pharmacological effects of DHA, we cultured T cells from INRs under different concentrations of DHA in vitro.
Materials and methods
Study approval
Study materials were developed in accordance with the Declaration of Helsinki, along with ethical norms, guidelines, and HIV-related laws and regulations in China. The study was approved by the Ethics Committee of Beijing Youan Hospital, Capital Medical University (No. 2020147).
Data source
The original dataset of gene expression profiles, including immune responders (IRs), INR, and healthy controls (HCs), was downloaded from NCBI’s GEO database. The accession number was GSE106792, based on GPL10558 (Illumina HumanHT-12 V4.0 expression bead chip, Illumina Inc, San Diego, CA, USA).24 The dataset GSE106792 consists of RNA sequencing data from 36 samples: 12 HCs, 12 immune non-responders, and 12 IRs, all derived from CD4+ T cells.
Data normalization and differential gene expression analysis
The robust multi-array average approach, including quantile normalization and log2 transformation, was performed for background correction and normalization to reduce variability. The gene probes in the expression profiles were annotated through the GPL10558 platform, and duplicate gene probes were combined, retaining the higher values. The Limma R package (version 3.44.3) was subsequently employed to identify DEGs,25 with patient status set as an independent variable while cell subset was not considered as an independent or dependent variable. Absolute log2FC greater than 1.0 and adjusted P-value, based on the Benjamini & Hochberg procedure, less than 0.05 were selected as threshold values. Heatmaps were generated in the pheatmap R package (version 1.0.12) with z-score normalization within each row (gene).
Go and KEGG enrichment analysis
GO and KEGG enrichment analyses were performed using the clusterProfiler R Package (version 3.16.1) and the Database for Annotation,26 Visualization, and Integrated Discovery (https://david.ncifcrf.gov/ ).27 The enrichment analyses, including biological process (BP), cellular component (CC), molecular function (MF), and KEGG pathway, were conducted to investigate the DEGs at the functional level. P-value < 0.05 and gene counts ≥ 2 were set as the cutoff criteria.
PPI network construction
The Search Tool for the Retrieval of Interacting Genes (Version 10.5, https://string-db.org ) database and Cytoscape software (Version 3.7.2) were employed to construct and visualize the PPI network of the identified DEGs.28,29 The MCODE module of Cytoscape was used for clustering sub-networks. MCODE scores greater than three and node counts greater than four were applied to screen for the most critical sub-networks in the PPI network.
Chemical association networks
The STITCH (Version 5.0) online server was used to predict the potential target proteins of DHA, and proteins with scores above 0.7 were selected as the threshold value.30 Meanwhile, an enrichment analysis of the protein-protein interaction network was performed.
Molecular docking
The chemical structure of DHA was processed using the LigPrep module in Schrödinger 10.2 software (Schrödinger, LLC, NY, USA).31 The Epik module in Schrödinger was used to generate ionized states and tautomers/stereoisomers at pH = 7.0 ± 2.0.32 The original chiralities were preserved, and the maximum number of stereoisomers was set to 32, with one low-energy ring conformation generated. The OPLS3 force field was used to perform energy minimization.
The crystallographic structures of the protein complexes were retrieved from the RCSB Protein Data Bank (PDB) and prepared using the Protein Preparation Wizard module in Schrödinger 10.2 software. Proteins were assigned bond orders, hydrogens were added, protonation was performed, crystallographic water molecules were removed, and restrained minimization was carried out until the root-mean-square deviation was lower than 0.3 Å based on the OPLS3 force field.
For each selected crystal structure, DHA was docked into the original ligand-binding site, which had dimensions of 10 Å × 10 Å × 10 Å, using the Glide module in Schrödinger 10.2 software.33,34 The extra precision option was selected, and all other parameters for Glide were kept at their default values.
ADMET properties prediction of DHA
Animal studies investigating the anti-inflammatory effects of DHA have been published; however, pharmacokinetic data on DHA from clinical trials is still lacking. Therefore, it is necessary to predict the ADMET properties of DHA using the ADMET predictive module of Pipeline Pilot (Version 8.5), which includes aqueous solubility, blood-brain barrier (BBB), cytochrome P450 2D6 binding (CYP2D6), human hepatotoxicity, human intestinal absorption, plasma protein binding (PPB), lipid-water distribution coefficient, and polar surface area. The methods of gradual screening ADMET are as follows: Aqueous solubility: 0 = extremely low; 1 = very low; 2 = low; 3 = good; 4 = optimal; 5 = very soluble. BBB: 0 = very high; 1 = high; 2 = medium; 3 = low; 4 = undefined, the molecule is outside the confidence range of the LogBB regression model). CYP2D6: TRUE = an inhibitor; FALSE = a non-inhibitor. Human hepatotoxicity: TRUE = the compound predicted hepatotoxicity; FALSE = the compound predicted non-hepatotoxicity. Human intestinal absorption: 0 = good; 1 = moderate; 2 = poor; 3 = very poor. PPB: TRUE = compound predicted strong binding (>90%); FALSE = compound predicted weak or no binding (<90%).
DHA validation experiments in vitro
Study subjects
In this study, INRs were defined as HIV-infected patients who had been on ART for more than two years, with plasma viral load completely controlled (plasma HIV RNA levels < 40 copies/mL), and CD4+ T cell counts < 350 cells/µL.4 We enrolled INRs (n = 18), who were treated with Tenofovir + Lamivudine + Efavirenz for 96 weeks. All the patients were male. The average age was 38.4 ± 7.6 years, and the average CD4+ T cell count was 280.7 ± 53.7 cells/µL. All participants provided written informed consent. Peripheral blood mononuclear cells (PBMCs) were separated from venous blood and used for research. This study was approved by the Beijing YouAn Hospital Research Ethics Committee.
Cell culture
PBMCs were freshly separated and seeded in a 48-well plate at a concentration of 1 × 106 cells/mL. Cells were stimulated with PHA (5 µg/mL) and simultaneously treated with different doses of DHA (1,000 µM, 500 µM, 100 µM, and 0 µM). After 48 h of culture, cells were harvested and used for flow cytometry analysis.
Flow cytometry
Cells were washed with 1% bovine serum albumin in PBS and then incubated with the cell viability marker fixable viability stain 510 (BD Biosciences, San Jose, CA) and labeled with specific surface antibodies for CD3 (HIT3a), CD4 (OKT4), CD8 (SK1), CD38 (HIT2), and HLA-DR (L243). Flow cytometry was performed using a FACScan flow cytometer. Data were analyzed using FlowJo software (Version V10.6.2; Tree Star Inc., Ashland, OR, USA). Cells were sequentially gated on lymphocytes, live cells, single cells, and CD3+ T cells. Then, CD38 and HLA-DR expression on CD4+ T cells and CD8+ T cells were analyzed separately. For further details, see the Supplementary Method for cell proliferation detection.
Statistical analysis
CD38 and HLA-DR expression on CD4+ and CD8+ T cells were presented as percentages and compared across groups using one-way ANOVA. Post-hoc comparisons between groups were conducted using Tukey’s test to identify significant differences. All statistical analyses were performed using SPSS 21.0. A P-value of < 0.05 was considered statistically significant.
Results
Identification of DEGs among the IR, INR, and HC groups
After data normalization (Fig. S1), as shown in Figure 1a, 119 DEGs were identified in the INR vs. HC groups, 56 DEGs in the INR vs. IR groups, and 189 DEGs in the IR vs. HC groups. Among the identified DEGs, 30, 22, and 22 genes were downregulated, while 89, 34, and 167 genes were upregulated in the INR vs. HC groups, INR vs. IR groups, and IR vs. HC groups, respectively (Figs. S2–S4).
Functional enrichment analysis
To investigate the enrichment and pathway distribution of the aforementioned DEGs, GO and KEGG enrichment analyses were performed using the clusterProfiler R package and the database for Annotation, Visualization, and Integrated Discovery. In the INR vs. HC groups, GO analysis enriched 119 DEGs into 375 BPs, 8 CCs, and 21 MFs (Fig. 1b). In the BP term, the DEGs were primarily involved in leukocyte migration, cell chemotaxis, positive regulation of cell activation, regulation of response to biotic stimulus, negative regulation of phosphorylation, leukocyte chemotaxis, regulation of the innate immune response, response to bacterial molecules, positive regulation of leukocyte activation, and positive regulation of defense response (Fig. 1b).In the CC term, the DEGs were mainly enriched in the secretory granule lumen, cytoplasmic vesicle lumen, vesicle lumen, leading edge membrane, mast cell granule, tertiary granule membrane, specific granule membrane, and ruffle membrane (Fig. 1b).In the MF term, the DEGs were primarily related to cytokine receptor activity, immune receptor activity, carbohydrate binding, cytokine binding, serine-type peptidase activity, hydrolase activity acting on acid-phosphorus-nitrogen bonds, serine hydrolase activity, non-membrane spanning protein tyrosine kinase activity, antioxidant activity, protein tyrosine kinase binding, phospholipase activity, RAGE receptor binding, C-C chemokine receptor activity, C-C chemokine binding, and G protein-coupled chemoattractant receptor activity (Fig. 1b).KEGG pathway analysis revealed that the DEGs were mainly associated with cytokine-cytokine receptor interaction, chemokine signaling pathway, and HIF-1 signaling pathway (Fig. 1b).
In the INR vs. IR groups, GO analysis enriched 56 DEGs into 119 BPs, 14 CCs, and three MFs terms (Fig. 1c). In the BP term, the DEGs were mainly involved in positive regulation of GTPase activity, regulation of translation, regulation of cellular amide metabolic processes, regulation of lymphocyte activation, modulation of processes of other organisms involved in symbiotic interactions, smoothened signaling pathway, viral transcription, viral gene expression, translational initiation, nuclear-transcribed mRNA catabolic processes, cellular response to transforming growth factor beta stimulus, and response to transforming growth factor beta (Fig. 1c). In the CC term, the DEGs were primarily enriched in the transport vesicle membrane, intrinsic component of organelle membrane, transport vesicle, nuclear envelope, endosome membrane, recycling endosome membrane, melanosome, pigment granule, inner mitochondrial membrane protein complex, early endosome membrane, RNA polymerase II transcription regulator complex, anchored component of the membrane, recycling endosome, and coated vesicle membrane (Fig. 1c). In the MF term, the DEGs were primarily related to phosphotransferase activity (alcohol group as acceptor), enzyme activator activity, and endopeptidase activity (Fig. 1c). KEGG pathway analysis demonstrated that the DEGs were mainly associated with tight junctions, endocytosis, rheumatoid arthritis, Fc gamma R-mediated phagocytosis, choline metabolism in cancer, Toll-like receptor signaling pathway, T cell receptor signaling pathway, and TNF signaling pathway (Fig. 1c).
In the IR vs. HC groups, GO analysis enriched 189 DEGs into 727 BP, 53 CC, and 78 MF terms (Fig. 1d). In the BP term, the DEGs were mainly involved in positive regulation of cell activation, positive regulation of cytokine production, positive regulation of leukocyte activation, leukocyte migration, regulation of immune effector processes, T cell activation, leukocyte cell-cell adhesion, response to bacterial molecules, phagocytosis, positive regulation of defense response, and response to lipopolysaccharide (Fig. 1d).In the CC term, the DEGs were mainly enriched in the secretory granule membrane, tertiary granule, secretory granule lumen, cytoplasmic vesicle lumen, vesicle lumen, external side of plasma membrane, tertiary granule membrane, endocytic vesicle, membrane raft, membrane microdomain, membrane region, ficolin-1-rich granule, ficolin-1-rich granule lumen, collagen-containing extracellular matrix, and ficolin-1-rich granule membrane (Fig. 1d). In the MF term, the DEGs were mainly related to carbohydrate binding, immune receptor activity, peptide binding, amide binding, phosphoric ester hydrolase activity, carboxylic acid binding, organic acid binding, hydrolase activity acting on acid-phosphorus-nitrogen bonds, serine-type peptidase activity, serine hydrolase activity, DNA-binding transcription repressor activity (RNA polymerase II-specific), cargo receptor activity, cytokine receptor activity, cytokine binding, and antigen binding (Fig. 1d). KEGG pathway analysis revealed that the DEGs were mainly associated with hematopoietic cell lineage, transcriptional misregulation in cancer, Fc epsilon RI signaling pathway, phagosome, cytokine-cytokine receptor interaction, NF-kappa B signaling pathway, HIF-1 signaling pathway, and cell adhesion molecules (Fig. 1d).
Key candidate genes identification within DEGs’ PPI network
To identify the key DEGs in the aforementioned groups, the PPI networks of the DEGs were constructed using the Search Tool for the Retrieval of Interacting Genes database and Cytoscape software. In the INR vs. HC groups, the PPI network of 119 identified DEGs contained 97 nodes and 92 edges, with the top ten highly connected genes being FGR, CCR1, HK3, NCF2, S100A12, BTK, CLEC12A, S100A8, FCN1, LY86, MCEMP1, MS4A14, MS4A6A, and VNN2 (Fig. 2). In the INR vs. IR groups, the PPI network of 56 identified DEGs contained 35 nodes and 28 edges, with the top ten highly connected genes being JUN, GLE1, HLA-C, SFRP1, TAF15, ACO1, CCL5, RPL9, ANXA8L1, ARPC1B, EXOSC6, HIPK2, HNRNPUL2, LRRN3, LSAMP, RAB35, and WASF3 (Fig. 3). In the IR vs. HC groups, the PPI network of 189 DEGs contained 154 nodes and 711 edges (Figs. 4 and S5).
Potential target proteins of DHA
To investigate the mechanism of DHA in the treatment of PLWH, the potential binding proteins of DHA were predicted using the STITCH online server and molecular docking. As indicated in Table S1 and Figure S6, the results from the STITCH database suggested that DHA might interact with 10 proteins, including TNFSF10B, UGT1A9, CYP3A4, CYP2C19, CASP3, CYP1A2, TP53, UGT1A7, UGT1A10, and UGT1A1.
As a natural product, DHA could potentially bind to more proteins. To explore this further, DHA was docked into all available protein crystal structures in the PDB database using the Glide module in the Schrodinger platform. The docking scores of proteins with DHA below -7 kcal/mol were retained, resulting in the selection of 127 proteins (Table S2). Among the top-ranked 20 results (Table 1), six docking scores of the predicted target proteins with DHA were higher than those of natural ligands, including retinol-binding protein 2, odorant-binding protein (PDB ID: 1DZM), odorant-binding protein (PDB ID: 1HN2), retinaldehyde-binding protein 1, bile acid receptor, and pheromone-binding protein asp1. The bioactivities of these natural ligands with target proteins are generally in the nM and µM range, suggesting that DHA may have strong binding abilities to these target proteins.
Table 1Docking results of DHA with target proteins (top-ranked 20 results)
| Predicted proteins | PDB ID | Bioactivities of natural ligands | Docking score (kcal/mol)
|
|---|
| Natural ligands | DHA |
|---|
| HIV-1 reverse transcriptase | 3M8P | IC50 = 2.1 nM | −13.63 | −9.86 |
| Muscarinic acetylcholine receptor m2, redesigned apo | 5YC8 | Kd = 6.4 nM | −13.75 | −9.75 |
| Retinol-binding protein 2 | 4GKC | Kd = 150 nM | −8.47 | −9.71 |
| Odorant-binding protein | 1DZM | IC50 = 3.9 uM | −8.42 | −9.61 |
| HTH-type transcriptional repressor kstr | 5CW8 | Kd = 60 nM | −12.67 | −9.60 |
| Estrogen receptor | 2FAI | Ki = 570 nM | −9.66 | −9.51 |
| Odorant-binding protein | 1HN2 | Kd = 1.0 uM | −8.88 | −9.50 |
| Orphan nuclear receptor PXR | 2O9I | IC50 = 40 nM | −11.12 | −9.50 |
| Antibody fab fragment mor03268 heavy chain | 2JB6 | Kd = 11 pM | −15.52 | −9.45 |
| Histamine n-methyltransferase | 1JQD | Ki = 6.9 uM | −9.77 | −9.42 |
| Oxysterols receptor LXR-alpha | 3IPU | Ki = 48 nM | −13.46 | −9.36 |
| Retinaldehyde-binding protein 1 | 4CIZ | Kd ∼ 51 nM | −7.22 | −9.32 |
| Beta-glucosidase | 1E55 | Ki = 76 uM | −9.81 | −9.28 |
| Vitamin D3 receptor A | 6FOB | IC50 = 23.5 nM | −12.13 | −9.26 |
| Estrogen receptor beta | 2Z4B | Ki = 0.44 nM | −10.89 | −9.24 |
| Bile acid receptor | 5Q11 | IC50= 7.19 uM | −8.68 | −9.24 |
| Envelope glycoprotein gp160 | 5U7O | Kd = 73 nM | −13.12 | −9.18 |
| Mineralocorticoid receptor | 5L7G | Ki = 16 nM | −11.90 | −9.17 |
| Vitamin D3 receptor | 3VTB | IC50 = 0.5 nM | −14.45 | −9.16 |
| Pheromone-binding protein asp1 | 3D78 | Kd = 23 nM | −7.84 | −9.14 |
HNMT encodes histamine N-methyltransferase, which was one of the DEGs between INRs and HCs. The docking score of DHA with HNMT was −9.42 kcal/mol, which is comparable to the natural ligand AdoHcy (−9.77 kcal/mol, experimental Ki = 6.9 µM).35 This suggests that DHA may have a similar binding affinity to that of the original ligand AdoHcy. Further binding pattern analysis showed that DHA binds to the inhibitor pocket of HNMT (Fig. 5a and b), where DHA is embedded in hydrophobic pockets formed by Phe19, Phe22, Tyr146, Tyr147, Trp179, Trp183, and Phe243. The dioxo bridge structure of DHA also forms a hydrogen bond interaction with the phenolic hydroxyl group on the benzene ring of Tyr147’s side chain (Fig. 5c and d). These results indicate that DHA might serve as an inhibitor of HNMT. Another DHA molecule docking target protein of interest, HIV-1 reverse transcriptase, is shown in Figure S7.
The ADMET properties of DHA
The ADMET properties of DHA were predicted using the ADMET predictive module of Pipeline Pilot (Version 8.5). As shown in Table 2, the water solubility of DHA was −4.39, indicating that DHA has low water solubility. The blood-brain barrier was −0.19, suggesting that DHA can pass through the blood-brain barrier to some extent. Cytochrome P450 2D6 binding (ADMET_EXT_CYP2D6) was −33.72, indicating that DHA might not bind to cytochrome P450 2D6. The human hepatotoxicity (ADMET_EXT_Hepatotoxic) was −3.88, suggesting that DHA may have hepatotoxic effects. The ADMET absorption level was 0, indicating good intestinal absorption. Plasma protein binding (ADMET_EXT_PPB) was −14.257, suggesting weak or no binding to plasma proteins. ADMET_AlogP98 and ADMET_PSA_2D were both within Lipinski’s rule of five.
Table 2The predicted ADMET properties of DHA
| ADMET properties | DHA |
|---|
| ADMET_Solubility/LogSw (mol/L) | −4.39 |
| ADMET_Solubility_Level | 2 (Low) |
| ADMET_BBB/LogBBB | −0.19 |
| ADMET_BBB_Level | 2 (Medium) |
| ADMET_EXT_CYP2D6 | −33.72 |
| ADMET_EXT_CYP2D6#Prediction | FALSE (a non-inhibitor) |
| ADMET_EXT_Hepatotoxic | −3.88 |
| ADMET_EXT_Hepatotoxic#Prediction | TRUE (toxic) |
| ADMET_Absorption_Level | 0 (Good) |
| ADMET_EXT_PPB | −14.257 |
| ADMET_EXT_PPB#Prediction | FALSE (weak or non-binder) |
| ADMET_AlogP98 | 2.762 |
| ADMET_PSA_2D | 56.535 |
Flow cytometry detection
In this experiment, we treated PBMCs with different doses of DHA for 48 h and measured CD38 and HLA-DR expression on CD4+ and CD8+ T cells by flow cytometry (Fig. 6a). We found that the frequency of CD38−HLA-DR+CD4+ T cells was significantly decreased after treatment with 1,000 µM DHA compared to the control (0 µM). However, no significant difference in the frequency of CD38+HLA-DR+CD4+ T cells was observed at any DHA dose (Fig. 6b). Additionally, we observed that the frequency of CD38+HLA-DR+CD8+ T cells was significantly decreased after treatment with 1,000 µM and 500 µM DHA compared to the control. However, although the frequency of CD38−HLA-DR+CD8+ T cells gradually decreased after DHA treatment, no significant difference was observed (Fig. 6c). Furthermore, we did not observe any significant changes in CD38+HLA-DR−CD4+ T cells or CD38+HLA-DR-CD8+ T cells before and after DHA treatment (data not shown). Additionally, we found that DHA inhibited T cell proliferation in vitro in INRs (Fig. S8)
Discussion
Incomplete immune reconstitution is characterized by chronic immune activation and systemic inflammation. However, the underlying physiological molecular mechanisms remain unclear. This study identified common significant potential mechanisms and pharmacological intervention strategies from seven independent studies (DEG analysis, enrichment analysis including GO and KEGG pathway analysis, PPI networks analysis, chemical association networks, molecular docking, ADMET, and flow cytometry detection).
Based on GO and KEGG enrichment analyses of the DEGs between INRs and HCs or IRs, “positive regulation of leukocyte activation”, “leukocyte chemotaxis”, and “regulation of lymphocyte activation” showed significant enrichment in the “biological process” category. In the “molecular function” category, the significantly affected molecular functions included “cytokine receptor activity”, “immune receptor activity”, “cytokine binding”, “C-C chemokine receptor activity”, and “C-C chemokine binding”. The role of cytokines in HIV infection and T lymphocyte activation is complex and depends on various factors, including the stage of HIV infection, ART effects, and immune reconstitution.36–38 For the KEGG pathway analysis, the canonical pathways associated with common DEGs are related to “cytokine-cytokine receptor interaction”, “chemokine signaling pathway”, and “T cell receptor signaling pathway”. Interestingly, the DEGs are involved in the upregulation of known T lymphocyte activation pathways and incomplete immune reconstitution,5,39–43 such as the “Toll-like receptor signaling pathway”, “HIF-1 signaling pathway” and “TNF signaling pathway”.
The RNA sequencing analysis aimed to uncover the molecular mechanisms of immune activation and inflammation in INRs compared to HC and IR groups. By focusing on DEGs from the NR vs. HC and INR vs. IR comparisons, we identified key pathways such as “cytokine-cytokine receptor interaction” and “T cell receptor signaling” that are relevant to immune dysregulation. These insights not only shed light on the potential therapeutic effects of DHA but also suggest that further exploration of these pathways could help develop new treatments for patients with poor immune recovery on ART.
PPI network analysis provides detailed interactions among the DEGs. Compared with the DEGs of HCs or IRs, FGR, CCR1, S100A12, CLEC12A, S100A8, LY86, MCEMP1, and CCL5 were the top highly interacted/connected genes in INRs. S100A8 and S100A12, collectively known as myeloid-related proteins, are calcium- and zinc-binding proteins that play a prominent role in regulating inflammatory processes and immune responses. High levels of these myeloid-related proteins in the serum of HIV-1-infected patients correlate with disease progression and low CD4+ T-cell counts.44 CCL5, a chemoattractant for memory T-helper cells and eosinophils, causes the release of histamine from basophils, activates eosinophils, and binds to several chemokine receptors, including CCR1, CCR3, CCR4, and CCR5.45 CCL5 has been associated with resistance to HIV infection,46 delayed AIDS progression,45 and immune recovery status.40 Interestingly, other DEGs in the PPI network also contribute to immune response regulation. Their proinflammatory activity involves the recruitment of leukocytes, promotion of cytokine and chemokine production, and regulation of leukocyte adhesion and migration. Further studies are warranted to elucidate the role of muscle-related genes in T lymphocyte activation in INRs. The PPI analysis also revealed key proteins, such as HNMT, which may play an important role in the immune dysregulation observed in INRs. While the PPI network provides insights into potential interactions between these proteins, further investigation into the extent of protein alterations at both RNA and protein levels is needed. This will allow us to better understand the functional impact of these interactions in the context of incomplete immune reconstitution.
So, how can we suppress inflammation and abnormal immune activation to recover CD4+ T-cell counts? Much has been learned about potential treatments for incomplete immune reconstitution over the 30 years since the use of prednisone in HIV.9 However, our recent review indicated that the majority of current candidates may not be ideal treatments for increasing CD4+ T-cell counts in PLWH.15 The immunological benefits and adverse events depend on factors such as safety, dosage, duration of use, and whether the treatment is combined with ART. Therefore, we have proposed the “SCAL” principles—safe, combined, adequate, and long—for CD4+ T-cell recovery in PLWH and INRs.15 Clearly, other safe, efficient, convenient, and affordable treatments are needed to stimulate this process. DHA possesses a broad range of bioactivities, including anti-inflammatory, immunosuppressive, and anti-tumor properties.16,47 Protein docking results from our study indicated that DHA exerts anti-inflammatory effects via multiple potential target proteins, such as HNMT and HIV-1 reverse transcriptase. Crucially, we found that the expression of immune activation markers on CD4+ and CD8+ T cells was significantly decreased after treatment with DHA compared to the control.
We are particularly interested in HNMT, which was one of the DEGs between INRs and HCs. DHA molecular docking in our study indicates that HNMT might be a potential therapeutic target in INRs. HNMT is responsible for histamine degradation, a biogenic amine involved in inflammation.48,49 Previous studies have shown that gp120 of HIV is a powerful stimulus for the release of histamine and cytokines from basophils.50 The histamine release caused by HIV might be involved in the development of HIV infection.51 Interestingly, single nucleotide polymorphisms of HNMT are associated with the risk of common inflammatory and immune activation diseases, such as allergic asthma,52–55 chronic urticaria,56,57 and atopic dermatitis.58 Moreover, artesunate, which is metabolized to DHA in vivo, possesses anti-allergic activity by blocking IgE-induced mast cell degranulation and histamine release.59,60 Given that DHA might reduce histamine levels by regulating HNMT, thereby exerting anti-inflammatory effects, we speculate that HNMT may be another potential target of DHA for treating INRs. The docking results suggest that DHA may interact with HNMT and other proteins involved in immune regulation. Future studies will focus on how DHA treatment influences signaling pathways downstream of HNMT, with particular emphasis on immune-related pathways. Understanding these molecular effects will provide deeper insights into the therapeutic potential of DHA for immune non-responders.
On the other hand, interestingly, the antiviral activity of DHA against HIV-1 has not been reported. In this paper, we predict a new class of potent anti-HIV agents, indicating that DHA might inhibit HIV-1 replication by exerting anti-HIV-1 reverse transcriptase activity for the first time. The virally encoded gag protein is essential for efficient reverse transcription and plays a central role in sustaining viral replication. Moreover, considering that long-term low-level HIV-1 replication is one reason for poor immune reconstitution,4–6 DHA is also expected to promote CD4+ T-cell count recovery by inhibiting virus production from chronically and latently infected cells. Therefore, the clinical potential of DHA, targeting both anti-HIV-1 reverse transcription and anti-inflammatory mechanisms, is an attractive approach due to the safety and effectiveness of DHA. Furthermore, the ADMET properties of DHA indicate that it has good pharmacokinetic properties, druggability, and potential clinical application value.
Though DHA shows great potential, there are some limitations. GSE106792 data was generated with sort-purified CD4+ memory T cells rather than PBMCs. In our study, we used PBMC samples, which include several types of cells, and this may impact the effects of DHA. Moreover, the gene expression profiles were from retrospective cohorts, and therefore, these findings must be prospectively validated in future studies. We confirmed that DHA could reduce T cell activation in vitro, but the complexity of incomplete immune reconstitution demands a more comprehensive analysis of clinical trials. Further studies should be conducted to determine whether DHA also provides therapeutic benefits in anti-HIV-1 treatment and promotes CD4+ T-cell count recovery.
While our study examined DHA’s effects on T cell activation and proliferation in INRs, it’s important to assess whether similar effects occur in healthy controls. This would clarify whether DHA’s actions are specific to the immune dysregulation in INRs. Future studies will address this to further validate our conclusions. Although we focused on HNMT based on the INR vs. HC RNA data and its position in the PPI network, we acknowledge the importance of investigating protein expression levels in actual patient samples. Differences in RNA expression do not always directly correlate with protein levels, which are often more functionally relevant. Future studies will aim to explore the expression of HNMT at the protein level to validate its role in immune non-responders, providing a more comprehensive understanding of its potential as a therapeutic target. While our study focuses on the immune-modulatory effects of DHA, we recognize the importance of thoroughly evaluating its safety profile. In future experiments, we plan to perform Annexin-V assays to assess cytotoxicity at each concentration of DHA. This will provide more robust evidence that DHA is non-toxic to cells at the concentrations used, further supporting its potential as a safe therapeutic option for immune non-responders. We recognize that PMA activates a broad range of cells, which may not fully reflect natural T cell activation. In future studies, we will use anti-CD3/anti-CD28 stimulation to more accurately mimic TCR engagement and assess activation markers like CD69 on CD4+ and CD8+ T cells.