Samples from twenty-five pigtailed macaques (Macaca nemestrina; PTMs) were included. Fifteen of the twenty-five were intravenously infected with 300 TCID50 of SIVsab (47). Starting from ~1.5 months postinfection (mpi), six received coformulated ART (48) throughout the remainder of the follow-up. Coformulated ART consisted of a combination of first line antiretrovirals used in PWH (Tenofovir+Emtricitabine+Dolutegravir). The following blood samples were available for lipidome analyses: preinfection samples for twenty-five uninfected PTMs; infection time point (D0) and early acute (EA) infection [~10 days postinfection (dpi)] for fifteen PTMs; late acute (LA) infection (~1.5 mpi) for fourteen PTMs; early chronic (EC) infection [~2.5 mpi] for seven PTMs; and late chronic (LC) infection (~6 mpi) for six PTMs. Assessment of the impact of SIV infection on the lipidome profile was performed by comparing each timepoint to the overall preinfection baseline levels ( Figure 1 ; Supplementary Table 1 ).

The assessment of the overall effect of SIV and ART on lipidome was performed by comparing the lipidomics profiles in the blood samples collected at <6 mpt (~2.5 mpt or >6 mpt (~10 mpt) to the pretreatment baselines from six SIV-infected PTMs on ART. In the text, we refer to <6 months as early treatment (ET) and >6 months as late treatment (LT) timepoints ( Figure 1 ; Supplementary Table 1 ).

Similar to humans, the lipidomic analysis identified 14 lipid classes. We first screened for significant variations in concentration of the various lipid classes and then of species in each class throughout infection, and with ART. The lipid species that yielded altered concentrations during SIV infection or ART were next correlated with key biomarkers of SIV/HIV disease progression and comorbidities.

Changes in the overall lipidome were observed at every stage of SIV infection, and with ART ( Figure 2A ), with the key lipid classes being modified as follows ( Figure 2B ): (i) The levels of sphingomyelins (SM) and lactosylceramides (LCER) increased only during acute SIV infection, returning to baseline preinfection during chronic infection; (ii) Hexosylceramides (HCER) increased throughout SIV infection and returned to baseline with prolonged ART; (iii) phosphatidylinositols (PI) and lysophosphatidylcholines (LPC) decreased with SIV infection and ART did not normalize their overall levels during the follow-up; (iv) Phosphatidylethanolamines (PE), lysophosphatidylethanolamines (LPE) and phosphatidylcholines (PC) did not change with SIV infection, yet they were reduced throughout ART.

Conversely, dihydroceramides (DCER), free fatty acids (FFA), cholesterylesters (CE), triacylglycerols (TAG), diacylglycerols (DAG) and monoacylglycerols (MAG) remained unaltered with SIV and ART ( Figure 2B ).

We therefore further focused the analyses just on the lipid species belonging to lipid classes that showed significant variation with SIV and/or ART.

Sphingolipids are important signaling molecules, which are involved in cell growth and apoptosis (73). They are components of the cell membranes and their interaction with cholesterol is important for signal transduction (73). Sphingomyelin is a major membrane sphingolipid, that is a precursor of the bioactive ceramide and lysophospholipid (73) which modulates growth factor receptors and extracellular matrix proteins. Sphingolipids can be used as binding sites by microorganisms, viruses, and toxins (74).

Total SMs were significantly increased during acute SIV infection ( Figure 2B ) being reversed during chronic infection and ART.

On principal component analysis of SM species, acute infection shifted most significantly from the preinfection, with chronic infection cluster moving towards preinfection and the ART points forming distinct clusters ( Supplementary Figure 7A ).

Individual SM species were differently modified by both SIV and ART ( Supplementary Figure 7B ): 3/12 SM species increased throughout SIV infection and were minimally restored by ART; meanwhile, 2/12 species increased only during the acute SIV infection; one single SM species decreased with SIV infection, and was not significantly restored with ART; two SM species were not impacted by the SIV infection, but decreased on ART; finally, one species was not altered by the SIV infection, but significantly increased on ART.

SM (24:1), which increased with SIV infection and was normalized by prolonged ART, positively correlated with apoA1, coagulation markers like VWF and sICAM-1, and lymphocyte counts ( Supplementary Figure 8A ). SM(22:0), which decreased on ART when compared to pretreatment levels, negatively correlated with several markers of inflammation and immune activation ( Supplementary Figure 8B ). Thus, some SMs that are modified by SIV may promote hypercoagulation, while ART itself may impact other SMs that contribute to inflammation and immune activation.

Lipid changes could trigger CVD, which are specifically associated with progressive HIV/SIV infection (17). Here, SIV infection and/or ART induced changes in lipid species previously associated with atherosclerosis and CVD (83, 84), indicating that HIV-associated CVD could be linked to: (a) Increases in SMs, known to be associated with hypercholesterolemia, atherosclerotic lesions (85–88), and correlated with coronary disease severity (89, 90); (b) Decreases in total LPCs and specific LPCs species that are known biomarkers of coronary heart disease (91, 92), of incident myocardial infarction (92) and associated with increased CVD risk (81); (c) Increases of total LCERs and certain individual LCER species, known to induce hypertrophy in cardiomyocytes (93); (d) ART-induced reductions in PCs and CEs, known to relate to incident CVD (94, 95); (d) Changes by both SIV or ART of PE-plasmalogens, cholesteryl esters, dihydroceramides, hexosylceramides, lactosylceramides, and sphingomyelins which positively correlated with coagulation markers, pointing to possible involvement in the PWH’s and SIV-infected PTMs hypercoagulation. ART did not reverse many of these changes, and further induced deleterious lipid modifications, pointing to a persistent risk of CVD in HIV/SIV-infected individuals, irrespective of successful virus suppression with ART. As such, specific lipid changes induced by both the virus and ART could be responsible for CVD in PWH and for the high incidence of hypercoagulation and epicardial and myocardial inflammatory lesions observed in our model.

Other lipid changes that occurred in SIV-infected PTMs were reported in other models to be associated with insulin resistance and diabetes, two highly prevalent conditions in PWH (96, 97). These changes: (a) occurred with SIV infection, and persisted with ART: decreases in LPC species, which are highly predictive for impaired glucose tolerance and/or type 2 diabetes mellitus (98, 99); (b) occurred during acute SIV infection and further increased with ART: increases of DCER species that are emerging biomarkers of insulin resistance, altered glucose homeostasis (100), diabetes risk (101) and metabolic dysfunction (102); (c) increased throughout SIV infection and ART of the ceramide species associated with prediabetes (103) or higher insulinemia (103), and which are also independently and strongly predictive for CVD and mortality (104). All these are highly prevalent conditions in PWH (96, 97). Increases of plasma ceramides and their infiltration in tissues were reported to promote hypertension, atherosclerosis, fibrosis, apoptosis, diabetes, mitochondrial dysfunction and heart failure (104–107). Interventions to normalize ceramides in PWH might be useful to prevent diabetes, reduce the incidence of CVD, and possibly ameliorate other comorbidities. Finally, measuring plasma ceramide levels may represent a promising diagnostic tool for the PWH at risk for CVD and metabolic disturbances.

Changes of lipid species seem to play a role in weight gain or loss: (a) Increases in PE species that positively correlates with obesity and decreases in the LPC and PC species that negatively correlate with the BMI were detected throughout infection and ART, suggesting that these lipid species may be involved in the weight gains seen in our PTMs on ART. Similar changes were reported in humans (108); (b) A reduction in PC(18:2/18:3), a subtype of PC(36:5), which was normalized by ART; and a decrease in long chain, polyunsaturated PCs during chronic SIV infection, many of which were not normalized by ART. This low PC signature may be due to reduced phosphatidylethanolamine methyltransferase (PEMT) activity, a metabolic alteration that may be responsible for the major weight loss observed in untreated PWH and SIV-infected PTMs. (c) Decreases in total and specific LPCs species which negatively correlate with the BMI in the PTMs gaining significant weight. These changes that persisted throughout SIV infection and ART were associated with obesity (109). (d) A profound and significant decrease of total and specific species of PI (that were reported to have anti-obesity effects (110)) particularly on ART, when most weight gain occurred in PTMs.

Our model is characterized by high levels of gut dysfunction and inflammation that persist under ART (47, 111). It is thus not surprising that lipid species with direct impact on inflammation were modified in SIV-infected PTMs: (a) A significant decrease in total PC levels occurred with prolonged ART. Several PC species, were also consistently decreased during chronic infection and after prolonged ART. Loss of PCs (known to have anti-inflammatory effects in intestinal cells (50)), with infection and ART suggests that this lipid class could play a role in the intestinal dysfunction and inflammation associated with HIV/SIV, and ART (112, 113). Therapeutic administration of PC could thus ameliorate gut health in PWH, as reported for patients with ulcerative colitis; (b) A profound and sustained decrease of both total LPCs and LPCs species occurred throughout SIV infection and ART. LPC levels decrease with inflammation and aging (114), their decrease being associated with poor prognostic and high mortality. LPCs reduce IL-6 production from macrophages post-LPS administration (115) and the platelet-activating factor (PAF) (116). LPCs can also inhibit neutrophil and eosinophil migration and activation and prevent tissue damage (117–119), enhance the suppressive platelet activity of Tregs (120, 121), increase extracellular antioxidant superoxide dismutase (122) and natriuretic peptide, which have anti-inflammatory and vasoprotective roles by binding C-reactive protein, thus decreasing its proatherogenic effects (123). LPCs also play a significant role in maintaining homeostatic T cell proliferation and support cytotoxic cell function postinfection (124). Their loss during SIV/HIV infection and ART could contribute to residual inflammation and immune dysfunction in PWH and SIV-infected PTMs. Our results suggest that LPC are excellent predictors of HIV/SIV disease progression, residual inflammation, immune dysfunction and comorbidities, particularly CVD. Efforts to revert the LPC losses via therapeutic interventions are thus warranted. (c) A significant decrease in the total levels of PC, PE and LPE was specifically induced by ART. These lipids have anti-inflammatory effects and inhibit macrophage-related inflammation (125–127). PC reduction in the intestinal mucosa was reported to occur in ulcerative colitis (49). Our data show that, in addition to the negative direct impact of SIV/HIV on the anti-inflammatory lipids, ART can also have a detrimental impact on lipids that are associated with gut integrity and systemic inflammation; (d) A significant decrease of phosphatidylinositol (PI), a pluripotent inhibitor of T-cell proliferation and activation and of inflammatory cytokines (Il-2, Th17) (128), throughout SIV infection and therapy. Dysregulation of PI signaling occurs in gastrointestinal malignancies and endoplasmic reticulum stress-mediated mucosal inflammation (129) and in both acute and chronic colitis (130). In PWH and SIV-infected PTMs on ART, PI dysregulation may be thus associated with the gut dysfunction and inflammation. PIs may therefore be potent biomarkers of residual gut dysfunction and inflammation in PWH. PI targeting is already a promising new, nontoxic nontherapeutic approach for inflammatory bowel disease (IBD) and their replenishment should be also targeted to improve clinical status and prevent comorbidities in PWH (131); (e) Increases in the ceramides (involved in multiple inflammatory processes, including obesity, diabetes, COPD, IBD and neuroinflammatory diseases (80, 132, 133)) by both SIV infection and ART. Altogether, our results illustrate the complex and vital role that lipids play in inflammatory pathways in HIV/SIV.

Phospholipids, such as PC, PE and PI, which are the main components of the inner mitochondrial membrane (134), were significantly reduced in SIV-infected PTMs on ART. ART also reduced total and specific PE species, which maintain mitochondrial functions (49). As mitochondrial dysfunction plays a role in CVD/metabolic syndrome (135), diabetes (136) and neurodegenerative diseases (137), ART may be contributing to the CV, metabolic and neurodegenerative comorbidities seen in PWH and SIV-infected PTMs. The increases of LCER and other ceramides species which induce lipotoxicity in the pancreas and heart via mitochondrial damage (138–140) strongly support a role of this lipid class of lipids in promoting HIV/SIV-related comorbidities.

Disturbances in the lipid levels induced by both virus infection and ART may contribute to the HIV/SIV-associated neurocognitive dysfunction. We report increases in the plasma levels of SM(18:0) with ART. SM(18:0) is a lipid species that was found to be increased in the cerebrospinal fluid of patients with Alzheimer’s disease-like pathology (141). Increases of TAG and MAG species occurred during early SIV infection. These lipids also increase in Alzheimer disease patients with mild cognitive impairment (142). Other neurologic abnormalities described in PWH can be triggered by changes in the lipid species described here: headache (143), which is associated with decreases in LPC species that also decreased in SIV-infected PTMs (144); depression, which associates PE, PC and PI decreases, and Cer, TAG and DAG increases, similar to those reported here (145, 146). It is thus conceivable that currently available or experimental therapies such as administration of high doses of Omega-3 fatty acids, of docosahehaenoic acid or of fermented soy bean lipid that can slow progression of Alzheimer disease, improves autisms, depression and schizophrenia or alleviate cognitive dysfunctions due to chemotherapy in human patients (147, 148) can be used as adjuvants to alleviate neurocognitive defects induced by infection and treatment in PWH.

In conclusion, SIV infection and ART have significant and distinctive impact on lipidome, which could contribute to non-AIDS comorbidities. By using a metabolically clean model, we identified specific lipidomic signatures of SIV infection or ART. We also described multiple lipid species alterations previously not reported by any of known diseases or clinical conditions, which may provide insight into the potential impact of lipid alterations in human diseases. Additional studies employing only antiretrovirals alone or in combinations could further help confirm the sole effects of ART on the lipidome, as previously shown with protease inhibitors (149). Our study strongly suggest that interventions aimed to supplement lipid deficits or block production of inflammatory lipids may help preventing comorbidities and may be employed as adjuvant treatments for HIV.

Archived samples were used from animals included in several previous studies conducted in the lab. Samples from twenty-five pigtailed macaques (Macaca nemestrina; PTMs) were used in this study. From ten PTMs we used only preinfection timepoints. Fifteen were intravenously infected with 300 tissue culture infectious doses (TCID50) of SIVsab92018. In eight of these PTMs SIV infection followed its natural course, while six PTMs received, starting at 1.5 mpi, the coformulated ART regimen with emtricitabine, tenofovir, and dolutegravir (48). ART was then administered throughout the follow-up ( Supplementary Table 1 ).

All PTMs were housed and maintained at the Plum Borough Research animal facility of the University of Pittsburgh according to the standards of the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). The experiments were approved by the University of Pittsburgh Institutional Animal Care and Use Committee (IACUC) (protocols 12040408, 15045829, 18021354, 20087378). The animals were fed and housed according to regulations set forth by the Guide for the Care and Use of Laboratory Animals and the Animal Welfare Act (150). The animals were paired and housed indoors in stainless steel cages. They had 12/12 light cycle, were fed twice daily, and water was provided ad libitum. Some of the environmental enrichment strategies that were employed include housing of animals in pairs, providing toys to manipulate, and playing entertainment videos in the animal rooms. The animals were observed twice daily, and any signs of disease or discomfort were reported to the veterinary staff for further evaluation. For sample collection, animals were anesthetized with 10 mg/kg ketamine HCl (Park-Davis, Morris Plains, NJ, USA) or 0.7mg/kg tiletamine HCl and zolazepan (Telazol, Fort Dodge Animal Health, Fort Dodge, IA) injected intramuscularly. At the completion of the study, the animals were sacrificed by intravenous administration of barbiturates.

Blood was collected in EDTA anticoagulant collection tubes for use in this study. Preinfection and other critical time points after infection and treatment were selected, as shown in Figure 1 and Supplementary Table 1 : early acute infection (10 dpi), late acute infection ~1.5 months post infection (mpi), early chronic infection (~2.5mpi), and late chronic infection (~6mpi). For analysis of lipidome in SIV-infected PTMs on ART, they were sampled during early treatment (<6 months post-treatment; mpt) and late treatment (>6mpt). Plasma separated from all these samples by blood centrifugation for 20 minutes at 2,200 rpm was used for viral load measurements, lipidomic assays and assessment of inflammatory biomarkers. Whole blood samples were also used for the assessment of the complete blood counts (CBCs) by the Marshfield Laboratories.

To assess the lipidomic changes associated with the SIV infection and ART each timepoint was compared to the overall preinfection baseline. The samples collected from the PTMs on ART, were compared to both their preinfection baseline, as well to their pretreatment baseline (~1.5 months postinfection) to isolate effects primarily due to ART. In addition, comparisons were made between late and early ART to identify the effects of long-term ART. Furthermore, cross-sectional comparisons were done between treated and untreated groups at around ~6mpi.

The data acquisition and analysis were performed by Metabolon. In brief, samples were prepared using the automated MicroLab STAR^® system from Hamilton Company. The method utilized a Waters ACQUITY ultra-performance liquid chromatography (UPLC) and a Thermo Scientific Q-Exactive high resolution/accurate mass spectrometer interfaced with a heated electrospray ionization (HESI-II) source and Orbitrap mass analyzer operated at 35,000 mass resolution. The MS analysis alternated between MS and data-dependent MSⁿ scans using dynamic exclusion. The hardware and software foundations for the informatics components were the LAN backbone, and a database server running Oracle 10.2.0.1 Enterprise Edition. Raw data was extracted, peak-identified and QC processed using Metabolon’s hardware and software. Peaks were quantified using area-under-the-curve. A data normalization step was performed to correct variation resulting from instrument inter-day tuning differences. Each compound was corrected in run-day blocks by registering the medians to equal one (1.00) and normalizing each data point proportionately. Lipids were extracted from the bio-fluid in the presence of deuterated internal standards using an automated BUME extraction according to the method of Lofgren et al. (151). The extracts were transferred to vials for infusion-MS analysis, performed on a Shimadzu LC with nano PEEK tubing and the Sciex SelexIon-5500 QTRAP. The samples were analyzed via both positive and negative mode electrospray. Individual lipid species were quantified by taking the ratio of the signal intensity of each target compound to that of its assigned internal standard, then multiplying by the concentration of internal standard added to the sample. Lipid class concentrations were calculated from the sum of all molecular species within a class, and fatty acid compositions were determined by calculating the proportion of each class comprised by individual fatty acids.

The names of the individual lipid species are based on the length of the acyl chain(s) found in each lipid along with the number of unsaturated bonds found in each. Each acyl chain is denoted as (A:B) where: A= The total number of carbon atoms that are present in the chain. B= number of unsaturated bonds.

Whole blood was stained for flow cytometry, as described (152). Intracellular staining for Ki-67 was performed as described (153). Stained cells were acquired on an LSR-II flow cytometer (Becton Dickinson, BD, Franklin Lakes, NJ) and analyzed with FlowJo version 10 (Treestar, OR). Antibodies used were as follows (with clone in parenthesis). All antibodies are from BD unless otherwise noted: CD25-PE (2A3), CD8-PE-CF594 (RPA-T8), CD4 APC (L200), CD3 V450 (SP34-2), CD38 FITC (AT-1 Stemcell, Canada), CD69 APC-Cy7 (FN50), Ki-67 FITC (B56), HLA-DR PE-Cy7 (L243), CD45 PerCP (D058-1283). Trucount (BD) was used to quantify absolute CD4⁺ T cell counts, as per the manufacturer protocols.

The following biomarkers were quantified with commercially-available kits as listed below: Leptin (MBS705354), Adiponectin (MBS069505), Omentin (MBS736583) all from MyBioSource (San Diego, CA). For sICAM-1 we used the kit BMS648INST from ThermoFisher (Waltham, MA); CRP was tested with the kit 2210-4 from Life Diagnostics (West Chester, PA); sCD163 was tested with the kit IQP-383 from Trillium Diagnostics (Brewer, ME); sTF was tested with the kit 845 from Sekisui Diagnostics (Burlington, MA); sCD14 was quantified with the kit DC140 from R&D (Minneapolis, MA), P-selectin was tested with the kit BMS650-4 from eBioscience (San Diego, CA), Apolipoprotein A1 and B were tested with the kits 3710-1HP-2 and 3715-1HP-2, both from Mabtech (Nacka Strand, Sweden); Oxidized HDL was tested with the kit STA-8880 from Cell Biolabs (San Diego, CA), Platelet Factor 4 was tested with the kit ELK-PF4 from RayBiotech (Peachtree Corners, GA); von Willebrand Factor and HDL & LDL/VLDL were tested with the kits ab223864 and ab65390, respectively, both from Abcam (Cambridge, United Kingdom). All these assays were run as per manufacturer’s protocols and our previous experience (42, 47, 154–158).

Was done on frozen plasma using a 29-plex luminex (Thermofisher, MA) as per manufacturer protocol and as previously described (155, 159, 160); results were read on a Bio-Plex reader (Bio-Rad Laboratories, CA).

SIV pVLs were measured using quantitative real-time PCR, as described (152, 161, 162).

For the lipidomics data, standard statistical analyses were performed in ArrayStudio on log transformed data, by Metabolon. p<0.05 was considered statistically significant and p<0.1 was considered to indicate a trend toward significance. For lipidomic changes in SIV infection, statistical analysis between each timepoint and preinfection was done using ANOVA contrasts, and fold changes were calculated. For lipidomic changes on ART, statistical analysis between each timepoint and preinfection baseline, between each timepoint and pretreatment levels, and between late and early ART were done using ANOVA contrasts, and fold changes were calculated. For cross-sectional analysis between treated and untreated groups, statistical analyses were done using Welch’s Two-Sample t-Test. All heat maps were generated using Prism 8 (GraphPad, CA). All correlations were performed using Spearman’s ranked correlation in Prism 8 (GraphPad). All chord diagrams were generated using R v3.6 using the circlize package (163). All PCA calculations were performed using the prcomp function and plots were performed using the factoextra package of R.