Reduced Energy Production in Immune Cells Linked to Death Risk and Long COVID Severity in COVID-19 Patients

Table 1 summarizes the demographics and clinical characteristics of the population, composed of twenty COVID-19 patients and eight healthy volunteers. The 20 patients with confirmed COVID-19 diagnosis were admitted in the intensive care unit at the Hospital of Strasbourg (France). The mean age was not significantly different between these two groups (59.49 ± 1.10, vs. 65.69 ± 1.69 years). The body mass index (BMI) of the COVID-19 patients was significantly higher compared to that of the healthy volunteers (25.00 ± 1.60, vs. 29.50 ± 1.33 kg/m², p = 0.03). As expected, the healthy subjects showed no risk factors, the most frequent being systemic hypertension, diabetes and dyslipidaemia in the COVID patients.

To go further, we classified the COVID patients depending on the severity of the disease, ranging from long-COVID with few (n = 6) or many symptoms (n = 6), to mortality (deceased patients, n = 4).

The main biological characteristics of the patients are presented in Table 2, as compared to the normal laboratory values. As compared to controls, COVID patients presented mainly with anemia and enhanced inflammatory markers, including increased C-reactive protein (CRP) and neutrophil-to-lymphocyte ratio (NLR) and decreased leucocyte-to-lymphocyte ratio (LLR).

Three years after hospital discharge, we conducted phone calls to determine eventual symptoms corresponding to long COVID. We asked the patients to rank their symptoms on a scale ranging from 0 to 10 (0 means no symptoms and 10 means the highest intensity of symptoms). Out of the 20 initial patients, four had died and four did not respond. In the 12 responding patients, the total score was calculated by summing each individual symptom score. We classified the patients based on COVID severity. Patients with a total score of post-COVID symptoms <10 were classified as no or few post-COVID symptoms. Two patients presented no remaining symptoms three years after hospital release. Patients with a total score of post-COVID symptoms >10 were classified as many post-COVID symptoms.

Table 3 shows that anosmia/agueusia, dyspnea, tiredness and muscle pain were the most common and the strongest symptoms in the patients presenting with long COVID.

Interestingly, we observed a positive correlation between the length of stay in the intensive care unit and the post-COVID severity (p = 0.0025, Figure 1).

Considering the entire population, the OXPHOS state by CI was decreased in the COVID-19 group compared to the control group (8.24 ± 1.68 vs. 4.03 ± 0.62 pmol/s/10⁶ cell, p = 0.007, Figure 2). A similar impairment was observed in the COVID-19 group compared to the control group during OXPHOS CI + II respiration (18.34 ± 3.33 vs. 8.24 ± 1.06 pmol/s/10⁶ cell, p = 0.0008). The OXPHOS CII state also decreased in the COVID-19 group compared to the control group (14.13 ± 2.35 vs. 6.21 ± 0.88 pmol/s/10⁶ cell, p = 0.0006).

To investigate in more detail a potential relationship between COVID severity and PBMC mitochondrial respiration, we classified the patients in three subgroups, depending on COVID severity (patients with few or many post-COVID symptoms and deceased patients).

OXPHOS CI was significantly reduced in the COVID group with high post-COVID symptoms, compared to the control group (8.24 ± 1.69 vs. 3.04 ± 0.51 pmol/s/10⁶ cell, p = 0.037, Figure 3a).

We observed the same profile for the OXPHOS CI + II state. A significant mitochondrial respiration decrease was found in the patients with many post-COVID symptoms group and in patients who died, compared with the control group (6.93 ± 1.08 and 6.19 ± 1.5 vs. 18.34 ± 3.33 pmol/s/10⁶ cell in the control group, p < 0.05 for each comparison with the control group, Figure 3b).

The OXPHOS state by CII decrease was also significant in the many post-COVID symptoms group compared to the control group (14.13 ± 2.35 vs. 4.91 ± 0.75, p = 0.008). Such a significant impairment was also observed in the deceased patients (4.94 ± 1.11, p = 0.04, Figure 3c).

To investigate the potential mechanisms involved in the greater PBMC mitochondrial dysfunction, we determined whether inflammatory and oxidative stress markers are enhanced in the more severe COVID patients. Since enhanced inflammatory pathways are known to participate in COVID severity, we examined in PBMC the expression of important genes involved in inflammation such as interferon-stimulated gene 15 (ISG15), chemokine ligand-9 (CXCL9), and the gene expression of the succinate receptor 1, SUCNR1 (Figure 4).

All markers tended to be increased in relation to COVID severity, the increase in ISG15 transcript being significant in patients who died (1.0 ± 0.08 vs. 14.45 ± 2.07, p = 0.02).

Finally, we observed significant negative correlations between the ISG15 transcript and PBMC mitochondrial respiration (Figure 5).

In our COVID-19 patients, we observed impaired PBMC mitochondrial respiration, involving several complexes of the mitochondrial respiratory chain. This is consistent with previous data in circulating cells [19]. Indeed, mitochondrial dysfunction has been observed in acute SARS-CoV-2 infection [20]. Ajaz et al. reported that COVID patients were characterized by mitochondrial dysfunction with energy deficit, leading to a metabolic switch toward glycolysis [21]. Similarly, De Vitis et al. also observed decreased PBMC mitochondrial respiration in COVID patients, as compared to healthy controls [25]. Particularly, the authors suggested that the mitochondrial respiration impairment, associated with reduced FT3 serum levels, may have hampered their reaction to viral infection [25]. Interestingly, although the degree of mitochondrial dysfunction likely depends on the COVID-19 stage and associated organ damages [22,26], this might have clinical significance since the virus reduces mitochondria-related antiviral-signaling pathways [19,20,21,22,27,28,29,30]. Although all mechanisms involved in severe and/or long-COVID symptoms still need investigation, mitochondrial complications are likely related to spike protein. Thus, SARS-CoV-2 spike protein decreases energy production, disrupts mitochondrial dynamics and inhibits recycling of the mitochondria by mitophagy through a possible downregulation of TOM20 [31,32,33,34]. Nevertheless, whether such mitochondrial dysfunction is associated with increased mortality and/or occurrence and severity of long COVID still deserves further investigation.

The finding of prognostic factors is important to adapt medical therapy aiming to reduce disease-induced mortality. Besides clinical risk factors, such as clinical severity and the related hospital length of stay, several parameters were linked with increased mortality in COVID-19 patients. Thus, hospital-based mortality was associated with solid organ tumors, diabetes, renal, liver and cardiorespiratory diseases [35].

Increased inflammatory markers and/or factors related to endothelial and other organ dysfunctions were also associated with a worse prognosis [36,37,38]. Accordingly, increased NLR was shown to be associated with a greater risk of intra-hospital mortality [36,39].

In our study, inflammatory markers were augmented in the most severe COVID patients, and particularly ISG15 was significantly increased in COVID patients who eventually died. This is in line with previous data showing an enhanced expression of ISG15 in the lungs and blood of severe COVID patients [40]. Similarly, CXCL9, a chimiokine associated with gamma interferon and SUCNR expression, stimulated during hypoxia [41], tended to be increased as compared to controls and to patients with no or few post-COVID symptoms.

Concerning PBMC, peripheral blood monocytes from patients with COVID-19 pneumonia demonstrated altered bioenergetics and reduced basal, maximal and spare mitochondrial respiratory capacity [20]. Fibroblast growth factor 21, proposed as a biomarker of mitochondrial dysfunction, was also shown to be associated with COVID-19 severity and mortality [21]. We therefore challenged the hypothesis that impaired PBMC mitochondrial respiration might be associated with COVID-19 patients' mortality. Indeed, the patients who died had a lower level of PBMC mitochondrial respiration. This result further supports the involvement of the PBMC metabolism in COVID pathophysiology.

Interestingly, long COVID was associated with lactic acidosis and chronic muscle weakness, suggesting alterations in mitochondrial oxidative phosphorylation. De Boer et al. observed decreased fat oxidation and increased lactate production in a cohort of long-COVID patients [37,38]. Such abnormalities are likely to be involved in patients' tiredness and muscle damages associated or not with sarcopenia [42].

Concerning PBMC, several pieces of data supported modified mitochondrial respiration in long-COVID patients [38,43], but whether impaired PBMC mitochondrial respiration might be a biomarker of later occurrence and severity of long COVID was unknown.

Although a causal relationship remains to be proven, our data supports a link between PBMC mitochondrial respiration and long-COVID occurrence and severity. Indeed, the patients demonstrating the greater number and severity of symptoms were the patients with the lower PBMC mitochondrial respiration. This was true for both OXPHOS I and II-related mitochondrial respiration. To the best of our knowledge, such findings between impaired PBMC mitochondrial respiration and subsequent occurrence and severity of long COVID have never been reported.

To investigate further potential molecular mechanisms, we determined whether PBMC inflammatory and interferon pathways might be related to mitochondrial respiration. Although not reaching statistical significance, such parameters tended clearly to be increased in patients with many post-COVID symptoms (i.e., long-COVID patients). Further, ISG15 negatively correlated with PBMC mitochondrial respiration.

Supporting our results, Marzetti et al. recently reported an improvement in mitochondrial quality in long-COVID patients supplemented with beetroot juice, associated with increased physical performance and endothelial function [44].

First, the BMI and age of the controls were slightly higher than that of COVID patients, which may be considered a limitation of the study. Second, several patients suffered from chronic respiratory failure, associated with more symptoms. Although chronic situations are different from acute situations, this characteristic could have modulated our results. Indeed, we previously reported that PBMC mitochondrial respiration was enhanced in severe asthma patients in exacerbation [44]. In contrast, PBMC respiration was decreased in patients with chronic obstructive pulmonary disease [45,46]. Nevertheless, the small number of patients in this subgroup prevents further analysis and justifies a larger-scale study to particularly examine the participation of respiratory failure in PBMC mitochondrial dysfunction.

Twenty patients, hospitalized in the intensive care unit for COVID-19, participated in this prospective study. COVID-19 was diagnosed by positive real-time reverse-transcriptase polymerase chain reaction from a nasal and/or throat swab together with clinical signs, symptoms, and/or radiological findings suggestive of COVID-19 pneumonia. Data regarding COVID-19-related hospitalization were collected by systematic medical chart review, and long-COVID symptoms were obtained by performing a phone call three years after hospital discharge. Eight healthy volunteers, matched for age, served as a control group when considering clinical characteristics and investigations on PBMC analysis.

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Faculties of Medicine, Dentistry, Pharmacy, Nursing Schools, Physiotherapy, Midwifery and Hospitals of Strasbourg (CE-2020-34, 27 March 2020).

Total RNA was isolated from 5 × 10⁶ PBMC using RNeasy mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. RNA was quantified by spectrophotometry (Nanodrop, Thermo Fisher). cDNA was synthesized with Maxima™H Minus cDNA synthesis Master Mix (Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer's protocol. Quantitative PCR was performed with 5 ng cDNA, using the QuantStudio 3 Real-Time PCR System (Applied Biosystem, Thermo Fisher Scientific, Waltham, MA, USA) and PowerTrack SYBR Green Master Mix (Applied Biosystem). YWHAZ was used as a housekeeping gene. The expression of genes implicated mainly in inflammatory pathways were analyzed: CXCL9, ISG15, and SUCNR1.

Primer sequences of the genes analyzed in RT-PCR are shown in Table 4 below.

Values are expressed as mean ± standard error of the mean (SEM). Statistical analysis was performed using Prism 6.0 (Graph Pad Software Inc., San Diego, CA, USA). Values were not distributed normally (using the Shapiro–Wilk test), so we performed a non-parametric test (Mann–Whitney test and Kruskal–Wallis test depending on the number of groups to compare). Correlations were analyzed using a non-parametric Spearman's test. A p value < 0.05 was considered significant.