Mitochondrial oxidative phosphorylation in peripheral blood mononuclear cells is decreased in chronic HIV and correlates with immune dysregulation

Background Cellular immunometabolism among people living with HIV (PLWH) on antiretroviral therapy (ART) remains under investigated. We assessed the relationships between mitochondrial oxidative phosphorylation (OXPHOS) in peripheral blood mononuclear cells (PBMCs) and blood parameters associated with HIV immune dysregulation. Methods PLWH ≥40 years old and on stable ART ≥3 months were enrolled (N = 149). OXPHOS complex I (CI, NADH dehydrogenase) and complex IV (CIV, cytochrome c oxidase) protein levels in PBMCs were quantified using immunoassays. Monocyte subsets and markers of T-cell activation, senescence, and exhaustion were measured on PBMC by flow cytometry. Plasma inflammatory mediators were quantified using a multiplex assay. HIV-uninfected group (N = 44) of similar age, gender, and ethnicity had available OXPHOS levels. Results PLWH had a median age of 51 years. Majority were male (88.6%), Caucasian (57.7%), and with undetectable plasma HIV RNA <50 copies/mL (84.6%). Median CI level was lower in PLWH compared with the HIV-seronegative group (65.5 vs 155.0 optical density/μg protein x 103, p <0.0001). There was no significant difference in median CIV levels. Lower OXPHOS levels correlated with lower CD4% and CD4/CD8 ratio. On multivariable linear regression adjusted for age, current use of zidovudine/didanosine, and HIV RNA (detectable versus undetectable), lower OXPHOS levels were significantly associated with higher MPO, SAA, SAP, and sVCAM, and higher frequencies of intermediate (CD14++CD16+) monocytes and TIGIT+TIM3+ CD4 T-cell (p<0.01). Conclusion CI PBMC protein levels were decreased in PLWH on ART. Decreased OXPHOS correlated with disease severity and inflammation. Further studies on the relationship between immunometabolism and immune dysregulation in HIV are warranted.

ART for �3 months. Participants were recruited between the years 2009 and 2012. PLWH with active malignancy, acute infection, or AIDS-defining illness at the time of enrollment were excluded. A cohort of HIV-seronegative individuals were recruited as a comparator group. IRB approval was obtained from the University of Hawaii Human Studies Program. All participants provided written informed consent. All banked specimens and data collected from participants were anonymized and de-identified prior to analysis.

Mitochondrial assessments
Quantitation immunoassays (Abcam, PLC, Cambridge, MA) were performed to quantify OXPHOS CI and CIV protein levels in viable PBMCs, as previously reported [9]. Cell viability was between 90-95%, determined using AOPI (acridine orange/propidium iodide). Each vial of viable PBMCs was thawed and washed in 0.5 ml of phosphate-buffered saline (PBS) twice before addition of 0.5 ml of ice-cold extraction buffer [1.5% lauryl maltoside, 25 mM HEPES (pH 7.4), 100 mM NaCl, plus protease inhibitors (Sigma, P-8340)]. Samples were mixed gently and kept on ice for 20 min, and then they were spun in a microcentrifuge at 16,400 rpm at 4˚C for 20 min to remove insoluble cell debris. The supernatant, an extract of detergent-solubilized cellular proteins, was then assayed with the OXPHOS immunoassays. All samples were loaded on the immunoassays with equal amounts of total cell protein using an amount previously established with control samples to generate signals within the linear range of the assay. Therefore, the resulting signal was directly proportional to the amount of OXPHOS protein or enzyme activity in the sample. Quantitation of the signal was done by densitometric scanning with a Hamamatsu ICA-1000 reader.
Mitochondrial DNA-specific 8-oxo-2'-deoxyguanosine (mt-specific 8-oxo-dG) was measured by the Gene Specific Repair Assay [10,11], which identifies the frequency of oxidative changes in guanine nucleotides within the mitochondrial DNA-16 kb molecule. Quantitation of the break frequency of mt-specific 8-oxo-dG was based on comparing the undigested and digested mitochondrial DNA break frequencies calculated using the Poisson distribution, with units reported in break frequency (BF). Mt-specific 8-oxo-dG BF was available only among PLWH.

Statistical analyses
Median CI and CIV protein levels were compared between HIV-positive patients and seronegative controls using independent samples Mann-Whitney U test. The correlation of various immune parameters with CI and CIV were assessed using Spearman's correlation (rho, r).
Immunologic parameters that were found to correlate with CI and CIV at the level of significance of p<0.01 were further analysed using linear regression analyses. Immune parameters were log-transformed to improve normal distribution of values. Multiple linear regression models were adjusted for age, current use of zidovudine or didanosine, and undetectable plasma HIV RNA. Differences in the median levels of various immune parameters were compared among patients with detectable versus undetectable mt-specific 8-oxo-dG using independent samples Mann-Whitney U test. Statistical analyses were performed using the IBM SPSS statistics version 25.0 (Armonk, NY).

Results
The demographic and immunologic parameters of the 149 PLWH are summarized in Table 1 Median CI protein level was significantly lower in PLWH compared with the HIV-seronegative group (65.5 vs 155.0 optical density/μg of protein x 10 3 , p < 0.0001; Fig 1). In contrast, there was no significant difference in median CIV protein levels (49.2 vs 46.0 optical density/ μg of protein x 10 3 , p = 0.20).
On multivariable linear regression analyses, lower CD4% and lower CD4/CD8 ratio remained significantly associated with lower CI and CIV protein levels ( Table 3). OXPHOS parameters did not correlate with absolute CD8 count, CD8%, or percentage of activated CD38+HLA-DR+ CD8 T-cell.

Discussion
In this cohort of older PLWH on stable ART, we found that mitochondrial CI protein levels were significantly decreased compared to HIV-seronegative persons. Lower CI and CIV protein levels in PBMCs correlated with disease severity among PLWH, as assessed by CD4 percent and CD4/CD8 ratio. Decreased PBMC CI and CIV protein levels were associated with higher plasma inflammatory markers and increased frequencies of intermediate monocytes and TIGIT+TIM3+ CD4 T-cells.
HIV proteins induce direct viral toxicity to the mitochondria. HIV-1 trans-activator (Tat) protein has been shown to cause rapid dissipation of the mitochondrial transmembrane potential and inactivates cytochrome c oxidase in mouse liver, heart, and brain [14]. Similarly, gp120 and Tat proteins induce mitochondrial fragmentation in neurons [15]. In our PLWH cohort, mitochondrial OXPHOS protein levels in PBMC, particularly CI, were significantly decreased compared with the seronegative group. This observation is likely due to several factors, such as direct mitochondrial toxicity of HIV, antiretrovirals, comorbidities, and persistent inflammation. Mitochondrial DNA, which is necessary for the synthesis of the components of the OXPHOS system, has also been reported to be significantly decreased in chronic HIV [16].
Persistent inflammation and immune dysregulation have been well documented among PLWH despite the use of suppressive ART [17]. We found that lower PBMC OXPHOS levels correlated with lower CD4/CD8 ratio, higher pro-inflammatory cytokine levels (MCP-1, MPO, SAA, SAP, and sVCAM), and higher percentages of the pro-inflammatory intermediate monocyte subset. These inflammatory mediators are known to induce mitochondrial dysfunction, which increases reactive oxygen species (ROS), leading to a vicious cycle of mitochondrial damage and inflammation [18]. ions, resulting in a pro-inflammatory state [20]. Modulation of the inflammatory response via redox-sensitive pathways or direct activation of the inflammasome by mitochondria-derived free radicals may result in the production of cytokines and orchestrate a vigorous inflammatory response [21]. Mitochondrial dysfunction also increases cellular responsiveness to pro-

PLOS ONE
inflammatory cytokines through an increase in ROS production [22] and NF-κB activation [23], and has been shown to cause accumulation of cells bearing pro-inflammatory phenotypes Table 2

PLOS ONE
in lung tissue [24]. These inflammatory mediators could lead to further mitochondrial dysfunction [25]. The HIV proteins gp120, Tat, Nef, Vpr, and reverse transcriptase have been shown to enhance ROS production and dysregulate oxidative stress pathways [26]. CI levels in PLWH was decreased but CIV levels were similar to HIV-seronegative controls. The decreasing trend in dysfunction as the electrons progress through the respiratory chain has been previously reported. Among ART-naïve subjects, there was a 41% decrease in mitochondrial respiratory chain complex II activity, 38% decrease in complex III, and 19% decrease in CIV compared to seronegative controls [19].
We found that the presence of mitochondrial oxidative damage, as assessed by mt-specific 8-oxo-dG, was associated with higher CI and CIV protein levels. Interestingly, 8-oxo-dG, did not correlate with other immune parameters that correlated with CI and CIV, such as CD4 percent, monocyte phenotypes, or NCR-bearing CD4 T-cells. Dysfunctional OXPHOS leads to increased ROS generation, which damages mitochondrial DNA, membrane lipids, and proteins [27]. In particular, CI has been identified as a common site of superoxide generation [28,29]. Our analyses were limited by the lack of mt-specific 8-oxo-dG levels in the HIV-seronegative group. In addition, we assessed OXPHOS protein levels and not function. One possible mechanism for our observation may be that increased ROS production is a consequence of stoichiometric mismatches in the electron transport chain complexes. This may result in increased residence time of electrons on sites of the complexes that mediate electron reduction of O 2 resulting in the increased production of H 2 O 2 and superoxide [30]. We further hypothesize that the lack of changes in CIV protein levels in the setting of low CI may be due to the electron transport compensation via Complexes II-IV, which has been previously reported in computer modelling studies and other disease states [31,32]. We found that higher percentage of TIM3+TIGIT+ CD4 T-cells was strongly associated with lower PBMC OXPHOS levels. TIM-3 and TIGIT are co-inhibitory receptors expressed by functionally exhausted T-cells, exhibiting decreased proliferation and suppressed T-cell responses [33,34]. Aged T-cells upregulate the expression of co-inhibitory receptors and Table 3. Linear regression analyses of immunologic parameters associated with mitochondrial complex I and complex IV levels.

Complex I Complex IV Complex I Complex IV
Absolute CD4 count 0.19 (p = 0.020) 0. 13  demonstrate reduction in respiratory metabolism and electron transport chain activity [35].
We have previously reported that TIGIT expression on CD4 T-cells correlates with total HIV DNA and residual immune activation despite suppressive ART [13]. CD4 T-cells expressing TIGIT, as well as those expressing PD-1 or lymphocyte activation gene 3 (LAG-3), have been identified as a major contributor to the pool of inducible HIV genomes [36]. In the SPARTAC study, the expression on CD4 or CD8 T-cells of TIM-3 as well as expression of PD-1 or LAG-3 measured prior to ART predicted the time to viral rebound after treatment interruption [37]. PD-1 signaling has been shown to switch T cell metabolism from glycolysis to fatty acid metabolism. These metabolic reprogramming mediated by PD-1 may lead to mitochondrial depolarization, reduction of mitochondrial biogenesis, and higher rate of ROS production [38]. The effects of TIM-3 and TIGIT on the OXPHOS system warrant further investigation.
Several limitations of our analyses should be mentioned, including the cross-sectional nature of our study. Mt-specific 8-oxo-dG and other immune parameters (cytokines, monocyte subsets, and NCRs) were not available for the HIV-seronegative controls. In our cohort, CI and IV analyses were performed on PBMCs and not on individual cell populations. Although it has been described in the literature that T-cells constitute up to 90% of immune cells in PBMCs [39], the correlations between PBMC CI and CIV protein levels may not accurately reflect OXPHOS levels in T-cells. Furthermore, a recent review has described that various T-cell and monocyte subsets have unique metabolic profiles, with metabolically active effector CD4 T-cells predominantly utilizing glycolysis instead of OXPHOS [40]. Nonetheless, this study recruited a modest number of PLWH and identified key immune correlates of dysfunction important in inflammation and HIV pathogenesis.
In summary, we found that PBMC CI levels were decreased in PLWH on stable ART and correlated with HIV disease severity and inflammation. Decreased PBMC OXPHOS protein levels were associated with higher plasma inflammatory markers and increased frequencies of intermediate monocytes and TIGIT+TIM3+ CD4 T-cells. Further studies are needed to investigate the relationship between CD4 T-cell exhaustion, immunometabolism, and HIV persistence.