Persistent Monocytic Bioenergetic Impairment and Mitochondrial DNA Damage in PASC Patients with Cardiovascular Complications.

The baseline characteristics of the entire cohort of 24 patients are shown in Table 1. For the purposes of this study, 14 patients were defined as long COVID patients with cardiovascular symptoms, while 10 were age-matched controls with similar cardiovascular risk factors. A demographic breakdown of the study groups is outlined in Table 1. The long COVID group and the control group exhibited a comparable proportion of male participants, with 57.1% and 50.0% of the respective groups falling into this category (p = 0.666). The mean age was 54 ± 16 years in the long COVID group and 53 ± 16 years in the control group (p = 0.944). The distribution of cardiovascular risk factors between the groups was found to be similar, with a history of smoking (57.1% vs. 30.0%, p = 0.228), dyslipidemia (42.9% vs. 30.0%, p = 0.228), hypertension (57.1% vs. 50.0%, p = 0.666), and family history of cardiovascular disease (50% vs. 40.0%, p = 0.472) being equally prevalent. No patients in either group had been diagnosed with diabetes. The most striking differences were in symptomatology: Dyspnea was more prevalent in long COVID patients compared to controls (92.9% vs. 0%), and angina pectoris was reported by 50% of long COVID patients but none of the control subjects. These findings underscore the pronounced cardiovascular symptom burden in long COVID patients, despite their comparable baseline cardiovascular risk profiles to the control group.

Since there are reports of mitochondrial dysfunction in various inflammatory and post-viral conditions, we sought to evaluate whether PASC is associated with persistent alterations in monocytic mitochondrial function.

To assess mitochondrial function, we isolated CD14⁺⁺ monocytes from both PASC and control patients and exposed them to buthionine sulfoximine (BSO) for two hours to induce oxidative stress, followed by comprehensive bioenergetic profiling using the Seahorse Agilent Analyzer (Figure 1). BSO (buthionine sulfoximine) is a specific inhibitor of γ-glutamylcysteine synthetase (γGCS), the rate-limiting enzyme in glutathione (GSH) synthesis [25]. By inhibiting this enzyme, BSO depletes cellular GSH, a major antioxidant that neutralizes reactive oxygen species and maintains cellular redox balance. This depletion leads to increased oxidative stress as cells lose their primary defense mechanism against free radicals, allowing us to examine how monocytes from different groups respond to this oxidative challenge. Our results revealed significant differences in mitochondrial adaptation to oxidative stress between groups.

As shown in Figure 1A, BSO treatment significantly increased basal respiration in control monocytes (from baseline to 30.4 pmol/min), whereas long COVID monocytes failed to upregulate their basal respiration in response to oxidative stress (long COVID + BSO: 9.5 pmol/min vs. control + BSO: 30.4 pmol/min; p = 0.0043). Similarly, proton leak, an indicator of mitochondrial membrane integrity, showed a striking 26-fold increase in control monocytes following BSO treatment (from 0.9 to 23.8 pmol/min; p = 0.0003), while long COVID monocytes exhibited a non-significant baseline elevation and minimal additional response to BSO challenge (Figure 1B). Maximal respiratory capacity was substantially reduced in long COVID monocytes compared to controls (15.8 vs. 44.9 pmol/min; 65% reduction), though this difference did not reach statistical significance (p = 0.4035). Notably, control monocytes responded to BSO treatment with a robust increase in maximal respiration, whereas long COVID monocytes showed no enhancement (Figure 1C). Spare respiratory capacity, representing the bioenergetic reserve critical for adaptation to stress, was 70% lower in long COVID monocytes at baseline (8.7 vs. 29.1 pmol/min; p = 0.4143) and showed significant impairment in response to BSO (long COVID + BSO: 9.9 pmol/min vs. control + BSO: 54 pmol/min; p = 0.0091) (Figure 1D). While non-mitochondrial respiration showed modest differences between groups (Figure 1E), ATP production appeared reduced in long COVID monocytes compared to controls (6.6 vs. 15.9 pmol/min), although this difference did not reach statistical significance (p = 0.2043) (Figure 1F). Collectively, these data indicate that monocytes from long COVID patients exhibit substantial mitochondrial dysfunction, particularly in their ability to adapt their bioenergetic profile in response to oxidative stress, which may contribute to the persistent cardiovascular symptoms observed in these patients.

To further characterize the differences in mitochondrial adaptation to oxidative stress between control and long COVID monocytes, we analyzed the relative BSO-associated changes in mitochondrial function parameters. This approach allowed us to quantify the magnitude of response to oxidative stress in each individual sample, controlling for baseline variations. As shown in Figure 2A, control monocytes demonstrated a robust increase in basal respiration following BSO treatment (mean increase of 62%), whereas long COVID monocytes showed virtually no change in basal respiration (0.3% decrease). Although this differential response appeared substantial, it did not reach statistical significance (p = 0.1995).

The most striking difference was observed in proton leak dynamics (Figure 2B). Control monocytes exhibited a dramatic induction of proton leak in response to BSO (13-fold elevation), while long COVID monocytes showed a significantly attenuated response (3-fold elevation), resulting in a statistically significant difference between groups (p = 0.0294). Impaired proton leak response in long COVID monocytes likely compromises their ability to mitigate oxidative stress, as proton leak serves as a critical protective mechanism that reduces mitochondrial ROS production when activated. Analysis of maximal respiration (Figure 2C) revealed that control monocytes responded to BSO with a substantial 5.3-fold increase in respiratory capacity, whereas long COVID monocytes showed a blunted response (1.3-fold increase). Despite this apparent difference, the comparison did not reach statistical significance (p = 0.1543), likely due to the variability observed within groups. Similarly, spare respiratory capacity (Figure 2D) was markedly enhanced in control monocytes following BSO treatment (4.4-fold increase), while long COVID monocytes not only failed to increase their spare capacity but actually showed a 20% decrease. This differential response, though pronounced, was not statistically significant (p = 0.205). The relative BSO-associated changes in non-mitochondrial respiration (Figure 2E) and ATP production (Figure 2F) did not differ significantly between the groups, suggesting that these parameters may be less susceptible to BSO-induced stress or that both groups maintain similar adaptability in these specific aspects of mitochondrial function.

These data demonstrate that long COVID monocytes exhibit impaired adaptive responses to oxidative stress, particularly evident in their inability to appropriately modulate proton leak, a critical mechanism for maintaining mitochondrial membrane integrity during cellular stress.

Similar to our findings in long COVID patient monocytes, we hypothesized that direct exposure to SARS-CoV-2 components might impair mitochondrial function in healthy monocytes. Since mitochondrial dysfunction appears to be a hallmark of long COVID syndrome, we investigated whether the viral spike protein alone could recapitulate these effects in vitro. We tested this by exposing CD14⁺⁺ monocytes from healthy donors to recombinant SARS-CoV-2 spike protein S1 for 20 h, with or without subsequent BSO treatment to induce oxidative stress.

As shown in Figure 3A, basal respiration was comparable between control monocytes and spike-protein-treated monocytes, with both groups showing similar reductions following BSO treatment. This suggests that spike protein exposure does not significantly alter baseline oxygen consumption. Similarly, non-mitochondrial respiration remained consistent across all treatment conditions (Figure 3B), indicating that spike protein does not affect oxygen consumption through non-mitochondrial processes. Proton leak, a critical indicator of mitochondrial membrane integrity, showed significant induction in control monocytes following BSO treatment, but this response was notably attenuated in spike-protein-treated cells (Figure 3C, p < 0.05). This finding suggests that while spike protein alone does not alter proton leak, it may interfere with the normal proton leak response to oxidative stress. ATP production was comparable between control and spike-protein-treated monocytes (Figure 3D), further supporting that spike protein alone does not directly impair energy production. The maximal respiratory capacity appeared somewhat reduced in spike-protein-treated monocytes compared to controls, though this difference did not reach statistical significance (Figure 3E). However, coupling efficiency, which reflects the proportion of respiratory activity devoted to ATP synthesis, was significantly reduced in spike-protein-treated cells compared to controls (Figure 3F, p < 0.05), suggesting subtle impairments in respiratory chain efficiency despite normal ATP production. Spare respiratory capacity following BSO treatment was similar between control and spike-protein-treated monocytes (Figure 3G), although percentage changes in spare capacity showed high variability (Figure 3H). This suggests that while spike protein may not impair the absolute capacity for increased respiration under stress, it might affect the consistency of this response between cells.

Taken together, these data indicate that direct exposure to SARS-CoV-2 spike protein alone does not dramatically alter baseline mitochondrial function in healthy monocytes, but it may subtly modulate specific aspects of respiratory chain activity and stress responses. These modest in vitro effects contrast with the profound mitochondrial dysfunction observed in monocytes from long COVID patients, suggesting that the persistent bioenergetic impairments in long COVID syndrome likely result from complex in vivo processes rather than direct viral protein effects.

Since we observed impaired mitochondrial function and bioenergetic adaptation in long COVID monocytes, we next investigated whether these cells also display altered reactive oxygen species (ROS) dynamics in response to oxidative stress. ROS production is tightly linked to mitochondrial function and represents a critical indicator of cellular stress responses. To assess this, we isolated CD14⁺⁺ monocytes from long COVID patients and cardiovascular risk-matched controls using negative selection, then exposed these cells to BSO for two hours to induce oxidative stress, and ROS accumulation was measured using FACS.

As shown in Figure 4, basal ROS levels did not differ significantly between control and long COVID monocytes under resting conditions, suggesting that the baseline oxidative state is comparable between groups. However, when challenged with BSO, a striking difference in ROS dynamics emerged. Control monocytes responded to BSO with the expected increase in ROS production (1.2-fold elevation), indicating a normal oxidative stress response. In contrast, long COVID monocytes not only failed to increase ROS production but actually showed a slight reduction in ROS levels following BSO treatment (6% decrease). This differential response was highly significant (p = 0.0015).

Mitochondrial membrane potential is a critical indicator of mitochondrial health and function in immune cells. As shown in Figure 5, CD14⁺⁺ monocytes from long COVID patients display altered mitochondrial characteristics compared to control subjects. Monocytes isolated from long COVID patients exhibit significantly elevated mitochondrial membrane potential compared to those from control subjects (157 vs. 113.7 mean TMRE fluorescence, p = 0.0179), suggesting a fundamental alteration in mitochondrial bioenergetics that persists after the acute infection. This elevation in membrane potential may reflect a compensatory response to underlying mitochondrial dysfunction or inefficient electron transport chain activity in long COVID monocytes.

To determine whether this altered membrane potential phenotype could be modified by oxidative stress, we treated monocytes with BSO, a glutathione synthesis inhibitor that induces cellular oxidative stress. Interestingly, BSO treatment had a minimal effect on mitochondrial membrane potential in both groups. Control monocytes maintained their baseline membrane potential (113.5 after BSO vs. 113.7 at baseline), while long COVID monocytes similarly maintained their elevated membrane potential (158 after BSO vs. 157 at baseline). The significant difference between groups persisted after BSO treatment (p = 0.0179), indicating that the altered mitochondrial membrane potential in long COVID monocytes represents a stable phenotype that is not readily normalized by acute oxidative stress.

This resistance to BSO-induced changes in membrane potential contrasts sharply with our previous findings regarding respiratory parameters and ROS production, where long COVID monocytes showed impaired responses to oxidative stress. The elevated but stable membrane potential in long COVID monocytes, regardless of oxidative challenge, suggests persistent remodeling of mitochondrial function that may contribute to the long-term immune dysregulation and cardiovascular symptoms observed in long COVID syndrome.

Similar to dysfunction observed in various cellular systems under pathological conditions, long COVID syndrome may alter mitochondrial DNA integrity in immune cells, potentially contributing to their impaired function. Since mitochondrial dysfunction is frequently associated with alterations in mitochondrial DNA (mtDNA), we investigated whether long COVID patients exhibit changes in monocytic mtDNA content and integrity. We isolated CD14⁺⁺ monocytes from long COVID patients and age-matched controls, extracted total DNA, and performed detailed analysis of the mtDNA copy number and regional damage. As shown in Figure 6A, long COVID monocytes displayed a dramatic reduction in mtDNA copy number compared to controls. Quantification of the mitochondrial-specific gene tRNA^LEU relative to nuclear beta-globin revealed that long COVID monocytes contain approximately 80% less mtDNA compared to control cells (p < 0.001). This substantial depletion of mitochondrial genomes likely contributes to the bioenergetic deficits we previously observed in these cells.

We next investigated whether the remaining mtDNA in long COVID monocytes exhibited signs of damage by analyzing the amplification efficiency of three different long mtDNA fragments relative to the short reference fragment tRNA^LEU. Interestingly, the analysis of region B (Figure 6B) showed no significant difference in amplification efficiency between long COVID and control samples, suggesting that this specific mitochondrial genome region remains relatively intact. However, analysis of region C (Figure 6C) revealed significant damage in long COVID monocytes, with an approximately 75% reduction in amplification efficiency compared to controls (p < 0.05). Similarly, region D (Figure 6D) showed significant damage in long COVID samples, with an approximately 70% reduction in amplification compared to controls (p < 0.05).

The observed pattern of region-specific mtDNA damage, combined with overall mtDNA depletion, provides molecular evidence for persistent mitochondrial compromise in long COVID monocytes. The regional selectivity of mtDNA damage suggests that certain mitochondrial genome segments may be particularly vulnerable to SARS-CoV-2-associated oxidative stress, potentially affecting the expression of specific mitochondrial genes involved in electron transport chain function and cellular bioenergetics.

Patients recruited in this study had a proven COVID-19 infection three months to one year before admission. COVID-19 disease was previously diagnosed using real-time qPCR. Patients were admitted to our department due to persisting cardiac symptoms, such as angina or dyspnea, in their daily activities. Patients with a medical history of cancer or autoimmune diseases, persistent acute kidney injury, or infections during admission were excluded. Pulmonary diseases were ruled out by clinical examination, pulmonary function test, and, where required, pulmonary imaging. The diagnostic workup further included obtaining patient history and clinical examinations. Through a cardiac work-up covering electrocardiography, echocardiography, functional stress testing, and, when necessary, coronary angiography, structural heart disease was excluded. All recruited patients and matched unaffected controls provided informed consent during their admission to our hospital. The patients were matched with unaffected controls for age, sex, and cardiovascular risk factors (history of smoking, diabetes mellitus, dyslipidemia, hypertension, and family history of cardiovascular diseases).

Our study focused on circulating CD14⁺⁺CD16⁻ monocytes as these cells play a critical role in cardiovascular homeostasis and can serve as an accessible biomarker of systemic immune and metabolic dysfunction in long COVID patients with cardiovascular manifestations. CD14⁺⁺CD16⁻ human monocytes were isolated from healthy donors and from control and long COVID patients according to a published protocol [50] using magnet-assisted cell sorting (MACS) with negative selection using Monocyte Isolation Kit II human from Miltenyi Biotec. The study was approved by the scientific and ethics committee of the University of Münster and conforms to the principles of the Declaration of Helsinki. Written informed consent was obtained from all donors by the blood bank, and thrombocyte reduction filters were provided anonymously without sharing personal and detailed information. The purity of isolated cells was confirmed by FACS, using CD14/CD16 staining, and they were around 98% pure.

Primary human monocytes were maintained in RPMI-1640 medium (+L-Glutamine, −D-Glucose, Thermo Scientific, Waltham, MA, USA) supplemented with 5 mM glucose, 10% fetal bovine serum (FBS), and 1% Penicillin/Streptomycin. For migration experiments and signaling studies, cells were starved for 2–4 h in FBS-free medium. Monocytes were kept in an incubator at 37 °C and 5% CO₂. Normoglycemic medium: RPMI-1640 medium (Gibco, Thermo Scientific, Waltham, MA, USA), Penicillin/Streptomycin, 5 mM glucose, and 20% fetal bovine serum.

The Cell Mito Stress Test kit was performed using a Seahorse analyzer according to the manufacturer's protocol. A total of 300,00 monocytes were seeded per well into a poly-L-lysed-coated XF cell culture mini plate and incubated for 3 h at 37 °C and 5% CO₂. The cells were cultured in XF-medium (XF Base Medium Minimal RPMI; Agilent Technologies, Santa Clara, CA, USA) containing 10 mM glucose, 2 mM l-glutamine, and 1 mM sodium pyruvate (all from Sigma-Aldrich, St. Louis, MO, USA). After three measurements of baseline recording, oligomycin A, FCCP, and rotenone/antimycin A were injected, followed by three measurement intervals after each injection. For BSO pretreatment, the test compound was added for 2–2.5 h before the assay. The data were analyzed using Agilent Seahorse Analytics (Agilent). The results were plotted using GraphPad Prism software, version 8.

To assess the cellular redox status, a previously used protocol was used [51]. Roughly, 100,000 cells per donor were used. For redox modulation, cells were treated with BSO for 2 h. CellROX Reagent (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) was then added to a final concentration of 5 μM. The cells were incubated for 30 min at 37 °C in the dark. After incubation, the medium was removed, and the cells were washed three times with PBS. Fluorescence intensity was measured using a fluorescence plate reader with an excitation/emission of 485/520 nm (Victor, PerkinElmer, Shelton, CT, USA) and flow cytometry (FITC; excitation/emission: 485/520 nm, Guava easyCyte, Millipore, Darmstadt, Germany).

A total of 125,000 monocytes per donor were used for the assay. The FCCP (carbonyl cyanide-p-trifluoromethoxy phenylhydrazone), an uncoupler of mitochondrial oxidative phosphorylation, was applied at a concentration of 20 μM for 10 min. The cells were incubated with 1 μM TMRE for 20 min at 37 °C in the dark, followed by washing once with 100 μL of PBS containing 0.2.% bovine serum albumin. A volume of 200 µL of PBS containing 0.2% bovine serum albumin was added to each well, and the fluorescence was measured with excitation/emission: 549/575 nm (Victor, PerkinElmer, Shelton, CT, USA).

The total DNA was isolated from 3–5 million monocytes using a Quick DNA mini prep kit (Zymo Research, Irvine, CA, USA), and, after enrichment of the DNA, the measurement of mitochondrial and nuclear DNA damage was performed by RT-qPCR as described previously [52]. In each of the genomic or mitochondrial gene loci examined, a large DNA fragment (3000 to 4000 base pairs) was employed as a sensor for DNA damage, and an internally nested small fragment (50 to 70 base pairs), which is assumed to be undamaged, as a reference. The nested intact reference fragment is effectively amplified by PCR, but the amplification of the large sensor is obstructed by other DNA lesions such as abasic sites, thymine dimers, strand breaks, and oxidative damage. Following the induction of DNA damage, the exponential amplification phase for the damage-sensitive large fragment is achieved at a later time point relative to the nested intact reference. The reactions were carried out using a CFX Connect Real-Time PCR Detection System (Bio-Rad, Feldkirchen, Germany). Samples were measured in duplicates. Cq values were calculated using CFX manager software 3.1, and replicates were averaged. For mitochondrial copy number analysis, the levels of mitochondria-specific gene tRNA^LEU(UUR) were measured compared to nuclear DNA specific gene beta globin using RT-qPCR. All primers used are given in Supplementary Table S1.

To analyze the significance of differences in experiments with monocytes isolated from long COVID or healthy individuals, the Mann–Whitney Rank Sum Test (for intergroup comparisons) or Kruskal–Wallis One-Way Analysis of Variance on Ranks with Tukey or Dunn's post hoc correction was used. For all the other experiments, two-sample independent t-tests or, when multiple comparisons were made, Kruskal–Wallis One-Way Analysis of Variance on Ranks with Tukey or Dunn's post hoc correction was performed. The level of significance was defined as p < 0.05. All statistics and graphs were generated using GraphPad Prism 8 software.