Long-term treated HIV infection is associated with platelet mitochondrial dysfunction

HIV infection and antiretroviral therapy have been linked to mitochondrial dysfunction. The role of platelet mitochondrial dysfunction in thrombosis, immunoregulation and age-related diseases is increasingly appreciated. Here, we studied platelet mitochondrial DNA content (mtDNApl) and mitochondrial function in people living with HIV (PLHIV) and related this to platelet function. In a cohort of 208 treated PLHIV and 56 uninfected controls, mtDNApl was quantified, as well as platelet activation, platelet agonist-induced reactivity and inflammation by circulating factors and flow cytometry. In a subgroup of participants, the metabolic activity of platelets was further studied by mitochondrial function tests and the Seahorse Flux Analyzer. PLHIV had significantly lower mtDNApl compared to controls (8.5 copies/platelet (IQR: 7.0–10.7) vs. 12.2 copies/platelet (IQR: 9.5–16.6); p < 0.001), also after correction for age, sex and BMI. Prior zidovudine-use (n = 46) was associated with a trend for lower mtDNApl. PLHIV also had reduced ex vivo platelet reactivity and mean platelet volume compared to controls. MtDNApl correlated positively with both platelet parameters and correlated negatively with inflammatory marker sCD163. Mitochondrial function tests in a subgroup of participants confirmed the presence of platelet mitochondrial respiration defects. Platelet mitochondrial function is disturbed in PLHIV, which may contribute to platelet dysfunction and subsequent complications. Interventions targeting the preservation of normal platelet mitochondrial function may ultimately prove beneficial for PLHIV.


Platelet mitochondrial dysfunction in PLHIV.
Energy demand for platelet ATP production and other metabolic processes that are essential for platelet activation is met by the combined actions of glycolysis and mitochondrial OXPHOS 41 . To validate our findings that the lower platelet mtDNA content is associated with platelet mitochondrial dysfunction, and to investigate platelet glycolysis activity, we assessed the metabolic activity of washed platelets of five PLHIV and five age-sex matched controls to assess membrane potential (Δψ m ), mitochondrial superoxide production (ROS m ) and real-time glycolysis and mitochondrial respiration using the Seahorse Extracellular Flux Analyzer 41 Fig. 3A) and a similar trend was found for basal mitochondrial respiration of platelets (R = 0.61, p = 0.14, n = 7, Fig. 3B). Taken together, our data suggest that the lower platelet mtDNA in PLHIV in associated with a concurrent reduction in mitochondrial respiration capacity (OXPHOS) without a compensatory increase in glycolysis.
Next, platelet activation and function and function were determined using multiple methods. First, plasma markers of in vivo platelet activation (chemokines released from alpha-granules; CCL5, CXCL4, CXCL7) were comparable between PLHIV and controls (Table 2). Second, using flow cytometry, the activation status of circulating platelets, as well as their reactivity to ex vivo stimulation by platelet agonists was assessed. In unstimulated platelets, the expression of the alpha-granule marker P-selectin (measure of platelet degranulation) and the binding of fibrinogen to the activated integrin αIIbβ3 (measure of aggregation; Table 2) were also similar across groups. When analysis was restricted to individuals above 40 years of age or male only, unstimulated platelet activation was lower in PLHIV compared to controls (Supplemental Tables S1, S2). In line with this observation, fibrinogen binding to αIIbβ3 in response to stimulation by adenosine diphosphate (ADP)-and collagen related peptide (CRP-XL) was reduced in PLHIV-induced) (Fig. 4A). Differences in P-selectin reactivity across the groups were smaller with only a significant difference with a high dose (125 µM) of ADP stimulation showed a significant difference between PLHIV and controls (Fig. 4B). We observed no correlations with cART regimens containing either NNRTI, PI or INSTI-use as well as ABC and platelet reactivity indices (all P > 0. 15). In addition, persistent immune activation did not correlate with platelet indices in this cohort (Supplemental Table S4). In summary, these data show that platelet reactivity, and especially αIIbβ3 activation is reduced in PLHIV.
Association of mtDNA pl with platelet function. Next, given the role of platelet mitochondria in platelet function, we assessed associations of mtDNA pl with platelet reactivity in PLHIV and controls. MtDNA pl copies did neither correlate with ADP-induced P-selectin expression, nor with binding of fibrinogen to platelets or Table 2. MPV: mean platelet volume. Unstimulated platelet aggregation measured as Fibrinogen binding by flowcytometry in median fluorescence intensity (MFI). Unstimulated platelet degranulation measured as P-selectin expression by flow cytometry by MFI. Data were analyzed using Mann-Whitney U test. IPF immature platelet fraction as a percentage of platelet count (Sysmex, Kobe, Japan); sCD14 serum levels of CD14, a marker of monocyte activation; sCD163 serum levels of CD163, a monocyte-and macrophage-specific scavenger receptor; hsCRP high sensitive C-reactive protein. www.nature.com/scientificreports/ www.nature.com/scientificreports/ mean platelet volume (MPV) in PLHIV (Fig. 5A,B). In controls, MtDNA pl copies were significantly correlated with MPV (Fig. 5C) and a positive trend was observed with fibrinogen binding (Fig. 5D).
To further explore the link between platelet reactivity and platelet activation, we performed a principal component analysis (PCA) to summarize both platelet activation (plasma markers of platelet activation and unstimulated P-selectin expression and fibrinogen binding) and platelet reactivity (P-selectin expression and fibrinogen binding after ex vivo ADP and CRP-XL stimulation). This PCA showed that Principal component (PC) 1 mainly represented platelet reactivity (P-selectin expression and fibrinogen binding after stimulation) whereas PC2 mainly represented in vivo platelet activation (plasma markers and unstimulated P-selectin expression and fibrinogen binding; Supplemental Fig. S7a,b). We used these derivatives to correlate mtDNA pl with platelet reactivity (coordinates on PC1) and platelet activation (coordinates on PC2). mtDNA pl correlated with platelet reactivity (Supplemental Fig. S7c; PC1 of platelet parameters vs mtDNA pl , R = 0.14, p = P = 0.024), but not with in vivo platelet activation (Supplemental Fig. S7d; PC2 of platelet parameters vs mtDNA pl , R = 0.05, P = 0.41). These data suggest that mtDNA depletion is associated with platelet dysfunction with a reduced platelet reactivity capacity, but not with increased platelet activation status.
Whereas mitochondrial dysfunction can result in platelet dysfunction, platelet activation itself may also contribute to loss of mitochondria from platelets through the formation of platelet microparticles (PMP) formation 42 . We therefore studied PMP in a subgroup of 20 PLHIV and age-sex matched controls. No differences in total PMP number (Supplemental Fig. S8a), nor PMPs containing mitochondria (Supplemental Fig. S8b) were observed across both groups.

Discussion
The present data show that PLHIV on long-term cART have reduced platelet mitochondrial content (mtDNA pl ) which was associated with platelet mitochondrial dysfunction and reduced energy supply. Platelet mitochondria play a key role in platelet metabolism, ATP production and platelet activation and lifespan, and we propose that the observed abnormalities in mtDNA pl and platelet mitochondrial function contribute to platelet dysfunction in PLHIV.
The literature on platelet function in cART treated individuals is contradictory, with some studies reporting increased platelet reactivity [33][34][35][36][37] , while others reporting reduced reactivity 28,38,39 . This heterogeneity in study results may be partly explained by differences in the characteristics of study participants, including the enrolment of PLHIV with detectable plasma HIV-RNA, the degree of persistent immune activation, cART regimens and timing of treatment initiation, factors that all have changed considerably over the years. In accordance with our present findings, Mesquita et al. 28 recently reported decreased platelet reactivity and platelet mitochondrial dysfunction in 36 PLHIV on stable cART. In contrast to our findings, platelets in the PLHIV in their cohort exhibited increased P-selectin expression. In the present study, not only platelet P-selectin expression, but also soluble markers of platelet activation were similar in PLHIV and controls. In addition, PLHIV exhibited a lower immature platelet fraction and a similar number of mitochondria-containing platelet microparticles. Together, these findings argue against excessive platelet activation being primarily responsible for reduced platelet reactivity www.nature.com/scientificreports/ in PLHIV. Whether mitochondrial depletion contributes to the observed platelet dysfunction in PLHIV remains uncertain. A recent study reported that chemotherapy-associated platelet hyporeactivity was caused by mitochondrial dysfunction and subsequent reduction in mitochondrial respiration 27 . Consistent with these observations, the lower platelet mtDNA content in PLHIV was associated with a reduction in mitochondrial respiration capacity, which may negatively impact platelet reactivity. In our study, however, mtDNA pl levels in PLHIV did not correlate with ex vivo platelet reactivity, whereas a positive trend was observed in HIV uninfected controls. The fact that all PLHIV exhibited reduced mtDNA pl with little variation in the absolute values may explain the absence of a correlation with platelet reactivity measures. Future studies focusing on platelet dysfunction should incorporate mitochondrial dysfunction to corroborate our findings. Age, history of zidovudine (AZT)-use and innate immune activation were all associated with decreased mtDNA pl . NRTI-use is a well-known cause of mitochondrial dysfunction and mtDNA depletion [6][7][8] . We found that prior AZT-use was a possible risk factor for reduced mtDNA pl , a trend that remained after correcting for age and CD4 nadir, whereas total duration of cART-use or duration of HIV infection were not. As platelets are short-lived, it is conceivable that mitochondrial mass is reduced during thrombopoiesis and that the known bone marrow toxicity of AZT 1 is still present even after switching to newer NRTIs such as TDF or ABC. Even though these NRTIs are known to have lower mitochondrial toxicity 1 , it is unclear whether long-term treatment does not exert any cumulative reduction in mtDNA pl too. While ABC-use has been linked to platelet perturbations in multiple studies [43][44][45] , others could not confirm ABC associated platelet dysfunction 35,46 . In our study, neither mtDNA pl content nor platelet function were associated with current or prior ABC-use. Unfortunately, we could not dissect the link between the mtDNA pl , overall NRTI exposure and duration of HIV infection itself, as exposure to NRTIs was high in the total study group. Still, as mtDNA pl content was lower in PLHIV than in controls, it is plausible that both NRTIs and HIV itself exert long-term changes in mitochondrial function 2 . This possible long-term NRTI effect on mitochondrial function in platelets supports recent efforts to implement NRTI sparing regimens as viable treatment options for long-term HIV treatment 47 . It would be interesting to assess mtDNA pl content, as well as platelet function, in PLHIV who are switched to a NRTI sparing regimen. www.nature.com/scientificreports/ Mitochondrial dysfunction and depletion have been associated with many diseases such as dementia, neuropsychiatric diseases, and cardiovascular diseases 22,23,30,48 . In PLHIV, these (non-AIDS related) co-morbidities have also been linked to persistent inflammation 20,21 . In our cohort, we indeed found increased levels of hsCRP, sCD14 and sCD163, but only the latter parameter was associated with mtDNA levels in platelets. Importantly, sCD163 was shown to be independently correlated with overall mortality in PLHIV and the incidence of non-AIDS related comorbidities 49,50 .
Targeting mitochondrial dysfunction and inflammation may help reduce excess mortality and morbidity that is associated with HIV infection, even when treated successfully with cART 22,27 . Reducing inflammation, besides reducing NRTI exposure, could indirectly reduce oxidative stress and mitochondrial dysfunction while treatment with ROS scavengers may also have beneficial effects in reducing non-AIDS related co-morbidities.
Multiple methods of mtDNA quantification have been used in whole blood or peripheral blood mononuclear cell (PBMC) fractions in HIV and other diseases 14,32 . However, different methods of quantification, and heterogeneity of the cell composition of whole blood or PBMCs may hamper its interpretation 32,51 . It is conceivable that platelet mtDNA content could mirror mitochondrial toxicity in other cell types as well. As a single cell-type source of mtDNA, platelet mtDNA may indeed serve as a possible biomarker for mitochondrial toxicity and non-AIDS related comorbidities associated with mitochondrial dysfunction, such as neurocognitive impairment and cardiovascular disease 22 .
Our study has limitations associated with its cross-sectional design. Even though sample size was large enough to explore the link between inflammation, cART-use, mtDNA pl and platelet function, it lacked power to confirm a possible link between mtDNA pl levels and clinical outcomes such as non-AIDS co-morbidities and NRTI-related adverse events. Although we observed reduced oxygen consumption in individuals with reduced mtDNA pl , the overall reduction of mtDNA pl in PLHIV prevented to investigate functional consequences of mtDNA pl depletion in PLHIV. Furthermore, we did not perform immunofluorescence confocal or transmission electron microscopy experiments in the current study. In addition, age and sex differed substantially between cohorts, with an effect of age on mtDNA content in the uninfected cohort. We explored multiple methods to account for these differences using adjusted models and subgroup analyses. The subgroup of above > 40 years revealed a significant difference between PLHIV and controls supporting the independent correlation of HIV infection in the age-sex adjusted model. Finally, our study included mostly Caucasian men limiting generalization of the findings to women and non-Caucasians.
In conclusion, PLHIV under long-term cART have reduced platelet mtDNA content and abnormalities in platelet mitochondrial respiration, which may possibly contribute to platelet dysfunction. Given the key role of www.nature.com/scientificreports/ platelets and mitochondria in the pathophysiology of long-term complications of HIV, interventions targeting platelet mitochondria, such as introducing NRTI sparing regimens, should be considered.

Methods
Patient selection. This cross-sectional, single center, prospective study was performed at the Radboud university medical center, a tertiary teaching hospital in The Netherlands. This study is part of the Human Functional Genomics Project (HFGP; www. human funct ional genom icspr oject. org) and was conducted in accordance with the Declaration of Helsinki after approval of the ethics committee (CMO Arnhem-Nijmegen, The Netherlands; NL42561.091.12, 2012/550). No animal experiments were performed in the current study. Adult HIV-1-infected individuals receiving cART for at least six months were included after providing written informed consent. Other inclusion criteria were a suppressed viral load (< 200 copies/mL). Exclusion criteria included, use of P 2 Y 12 receptor antagonists (platelet inhibitor), an active hepatitis B or C infection and/or signs of other active intercurrent infection other than HIV-1 (e.g. fever in last week or antibiotic-use in last 4 weeks). Healthy controls were concurrently included throughout the duration of inclusion of PLHIV. Exclusion criteria were use of medication (excluding oral contraceptives or paracetamol) and/or signs of an active infection in the last month. Clinical data was collected by extracting data from electronic medical record (Epic Systems, Verona, WI, USA). History of cART use was extracted from the Dutch HIV registry (Stichting HIV-monitoring). In a separate validation experiment for mitochondrial function, five virally suppressed male PLHIV (45-60 years) who were not using statins or acetylsalicylic acid (ASA), were enrolled together with five age and sex matched controls.
Platelet count and function. Platelet count and parameters were determined using an automated hematology analyzer (Sysmex, Kobe, Japan). Platelet reactivity was determined in citrated whole blood (3.2% sodium citrate, Becton Dickinson, Franklin Lakes, NJ, USA) using a flow cytometry based assay as previously described between 1 and 3 h after blood collection 52  Platelet isolation. Platelet rich plasma (PRP) was obtained from citrated plasma (Vacutainer, Beckton-Dickinson) after centrifugation 156g for 15 min without brake at room temperature (RT). Samples were processed within 2 h of blood collection. Platelet count in PRP was measured using an automated hematology analyser (Sysmex, Kobe, Japan). Washed platelets were obtained as previously described 53 . In short, PRP was supplemented with acid citrate dextrose (10%) and prostaglandin I 2 and washed twice with Hepes tyrode's buffer using centrifugation (330g, 20 min).
Platelet microparticles. Peripheral blood was centrifuged at 1000g for 5 min. Plasma was then centrifuged at 1500g for 20 min to obtain platelet poor plasma (PPP). www.nature.com/scientificreports/ (complex V), reducing mitochondrial respiration), Carbonyl cyanide-4-(trifluoromethoxy) phenylhydrazone (FCCP; 1 μM, uncoupling agent causing maximum mitochondrial respiration), antimycin A (2.5 μM, Complex III inhibitor inhibiting mitochondrial respiration) and 2-deoxy-d-glucose (2DG; 40 mM, inhibits glycolysis) after ex vivo stimulation (all Sigma). Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were determined before and after injection of platelet agonists. Samples were measured in five replicates with three measurements after every injection. For mitochondrial respiration measurements using Seahorse extracellular flux analyzer, significant outliers were excluded from analysis if measurement was 2× SD below the mean value in every timepoint with a maximum of one of the five replicates. Mitochondrial membrane potential was determined using Tetramethylrhodamine ethyl ester (TMRE, Sigma; 100 nM at 37 °C for 20 min). A negative control was generated for every sample using FCCP (1 μM at 37 °C for 10 min). Mitochondrial reactive oxygen species (ROS m ) were detected using the cationic probe MitoSOX Red (Invitrogen, Carlsbad, CA, USA (2.5 μM at 37 °C for 30 min). Both probes were quantified in CD61+ cells using a Cytoflex flow cytometer (Beckman Coulter).
Plasma markers. Concentrations of three plasma markers of platelet activation: chemokine (C-X-C motif) ligand 4 (CXCL4 also known as platelet factor 4), CXCL7 (also known as beta-thromboglobulin) and chemokine (C-C motif) ligand (CCL5; also known as RANTES) were determined by ELISA (R and D systems, Minneapolis, USA) according to the manufacturer's instructions in citrated PPP 52 . In addition, markers of persistent inflammation, high-sensitive C-reactive protein (hsCRP), sCD163 and sCD14, were measured using ELISA (Quantikine, R and D systems) according to the manufacturer's instructions in EDTA plasma.
Statistical analyses. Data were analyzed by independent T-test or Mann-Whitney U test. Pearson's correlations coefficient was used for univariate correlation analyses, unless otherwise stated. An inverse rank-based normalization was performed for non-normal data (e.g. mtDNA in platelets). A multivariate linear regression model was used to correct for age, body mass index (BMI) and sex. Several sensitivity analyses using subgroups (males only and above 40 years) were performed to explore for possible confounding. Principal component analysis (PCA) was performed using singular value decomposition to summarize platelet function and correlate with mtDNA pl . cART-use was calculated as days on a certain drug, cumulative use of multiple drugs of the same class were combined. R studio (CRAN project) and Graphpad Prism version 5.03 were used for analyses. www.nature.com/scientificreports/