T4 apoptosis in the acute phase of SARS-CoV-2 infection predicts long COVID

Background As about 10% of patients with COVID-19 present sequelae, it is important to better understand the physiopathology of so-called long COVID. Method To this aim, we recruited 29 patients hospitalized for SARS-CoV-2 infection and, by Luminex®, quantified 19 soluble factors in their plasma and in the supernatant of their peripheral blood mononuclear cells, including inflammatory and anti-inflammatory cytokines and chemokines, Th1/Th2/Th17 cytokines, and endothelium activation markers. We also measured their T4, T8 and NK differentiation, activation, exhaustion and senescence, T cell apoptosis, and monocyte subpopulations by flow cytometry. We compared these markers between participants who developed long COVID or not one year later. Results None of these markers was predictive for sequelae, except programmed T4 cell death. T4 lymphocytes from participants who later presented long COVID were more apoptotic in culture than those of sequelae-free participants at Month 12 (36.9 ± 14.7 vs. 24.2 ± 9.0%, p = 0.016). Conclusions Our observation raises the hypothesis that T4 cell death during the acute phase of SARS-CoV-2 infection might pave the way for long COVID. Mechanistically, T4 lymphopenia might favor phenomena that could cause sequelae, including SARS-CoV-2 persistence, reactivation of other viruses, autoimmunity and immune dysregulation. In this scenario, inhibiting T cell apoptosis, for instance, by caspase inhibitors, could prevent long COVID.


Introduction
In addition to potentially resulting in COVID-19 at its acute phase, SARS-CoV-2 infection may thereafter provoke sequelae, socalled long COVID, in 1 in 10 individuals (1, 2).Long COVID may be defined by the persistence of symptoms, 3 months after the acute infection, for at least 2 months (3).These symptoms encompass asthenia, shortness of breath, neurocognitive disorders, insomnia, anxiety, cardiovascular dysfunction, muscular weakness, and depression (3).As more than three-quarters of a billion cases of COVID-19 have been reported to the World Health Organization (4), post-acute COVID-19 sequelae are a major concern.
One way to address this concern is to better understand the pathophysiology of long COVID by identifying the markers able to predict it.The severity of acute SARS-CoV-2 infection is clearly one of them.Indeed, over 70% of patients hospitalized for COVID-19 presented sequelae (5), particularly those admitted for long stays (6) or to the intensive care unit (ICU) (7).Accordingly, lactate dehydrogenase (LDH), a marker of tissue lesions, has been associated with the development of long COVID (8).C-reactive protein (CRP), Tumor Necrosis Factor-a (TNF-a) (9), interleukin-6 (IL-6), IL-8, CXCL10, peripheral inflammation biomarkers linked to COVID-19 severity, also predict sequelae (8,(10)(11)(12)(13), as well as Type I and Type III interferons (IFN) (14).Concerning lymphocytes, the persistence of a low level of naïve T cells ( 14) and a high level of exhausted T cells (15) was correlated with the presence of sequelae.An increase in inflammatory monocytes was reported in patients who later developed long COVID (16), and CD57-expressing mature/senescent NK cell frequency was identified as predictive for pulmonary sequelae (17).
To further identify the biomarkers predictive of long COVID, we determined the immune profiles of 29 patients recently hospitalized for COVID-19 and monitored them for one year.We then compared the initial immune profile between patients who later developed sequelae or not.Our immune profiling included 19 soluble markers, quantified in plasma and in the supernatant of non-stimulated peripheral blood mononuclear cells (PBMCs), and a panel of cell surface markers analyzing T and NK cell differentiation, activation, exhaustion and senescence, monocyte subpopulations as well as T cell apoptosis.This latter phenomenon is of interest, since we have previously shown that programmed T cell death is linked to disease severity (18).The main form of programmed cell death in the COVID-19 patients we analyzed is apoptosis rather than pyroptosis, necroptosis or PANoptosis, since transcripts of the proapoptotic Bcl-2 family members, Bax and Bak, were up-regulated in T cells, CD95 was over-expressed at the T cell surface, plasma level of soluble Fas was increased, and the pancaspase inhibitor Q-VD was more efficient in preventing T cell death than the inflammasome/pyroptosis inhibitors VX765, IDN6556, and MCC950 and the necroptosis inhibitors GSK-872 and Dafrafenib (18).

Study design
Volunteers were recruited between April and October 2020 on the very first day of their hospitalization in the intensive care unit (ICU) or non-ICU at Nımes University Hospital, France, for SARS-CoV-2 infection confirmed by RT-PCR performed on a nasopharyngeal swab.All participants provided written informed consent and the study was approved by the I ̂le-de-France 1 Ethics Committee.The trial was registered on ClinicalTrials.govunder the reference NCT04351711.
Whole blood collected on EDTA was labeled and stored in the dark for 10 minutes at room temperature and fixed (IMMUNOPREP reagent system kit and TQ Prep automate, Beckman C oulter ) according t o t he manufacturer' s recommendations.Samples were run on a Navios flow cytometer (Beckman Coulter) and results were analyzed using Kaluza software (Beckman Coulter).A minimum of 20,000 lymphocytes were analyzed per subpopulation.The flow cytometer's inter-run stability was verified using the same batch of FlowSet Pro Beads (Beckman Coulter).

Soluble immune activation markers
We used TruCulture tubes (Myriad RBM) which allow blood collection and culture in a single tube.To analyze the soluble factors spontaneously released by PBMCs in culture, we chose TruCulture tubes devoid of immune stimulant.After drawing 1mL of blood, the tube was inserted into a dry block incubator maintained at 37°C for 48 h.The tube was then opened, and culture supernatant was separated from the sedimented cells and maintained at −20°C.

Statistical analysis
The d'Agostino and Pearson test was used to assess normality.Group comparisons were performed with a two-sided unpaired Student t-test or Mann-Whitney test as appropriate.Adjusted pvalues were calculated using the Benjamini-Krieger-Yekutieli method when necessary.A p-value < 0.05 was considered as statistically significant.

Patient characteristics
We recruited 29 PCR-positive SARS-CoV-2-infected patients on the very first day of admission to the ICU or the Infectious Diseases Department (non-ICU) at Nımes University Hospital, France.ICU patients presented an oxygen saturation of less than 90% and/or PaO2 below 60 mmHg in room air, or an oxygen saturation of less than 90% while receiving 5L/min of oxygen.Non-ICU patients had an oxygen saturation of less than 96% in room air and/or deterioration in their general condition.The bioclinical characteristics of these patients during the acute phase of the infection are shown in Table 1.These 29 volunteers were monitored over a year.At Month 12, 19 (66%) of them presented long COVID.Sequelae included asthenia (58%), neurologic disorders (42%), breathing difficulties (37%), insomnia (37%), anxiety (21%), cardiovascular disorders (16%), muscular weakness (10%) and depression (10%).In line with the literature (5-7), the severity of COVID-19 was a risk factor for developing post-acute syndrome.Thus, the incidence of long COVID was 11-fold higher in ICU than in non-ICU patients (Table 1).Likewise, individuals with long COVID tended to have been hospitalized for a longer period than those without post-acute syndrome (mean ± SD, 9.6 ± 6.9 vs. 5.9 ± 6.6 days).As previously reported (8), biomarkers of severity, like high lactate dehydrogenase concentration in plasma (mean ± SD, 381 ± 147 vs. 195 ± 53 U/L) and low levels of circulating monocytes (mean ± SD, 0.42 ± 0.22 vs. 0.78 ± 0.33 x 10 9 /L), were also associated with the development of long COVID.

Immune activation in the acute phase of infection: soluble markers
As it has been reported that patients who develop long COVID present more marks of immune activation during the acute phase of their infection than patients who do not (8, 10-17), we measured the plasma levels of a panel of soluble markers on the first day of hospitalization in the 29 volunteers recruited.We chose to measure inflammatory cytokines IFN-a, IFN-g, TNF-a, IL-1a, IL-1b, IL-6, IL-8, anti-inflammatory cytokine IL-10, TH1 cytokine IL-12, TH2 cytokines IL-4 and IL-13, TH17 cytokine IL-17A, growth factor GM-CSF, inflammatory chemokines IP-10, MIP-1a and MIP-1b, as well as the endothelium activation markers ICAM-1, E-and Pselectins.Comparing the plasma levels of these markers during the acute SARS-CoV-2 infection in individuals who later presented long COVID or not, we observed no difference (Table 2).We quantified the same markers in the supernatant of PBMC cultured for 48 hours without stimulation.Table 2 shows that the PBMC of patients with sequelae initially spontaneously released higher amounts of the soluble endothelium activation marker P-selectin.Yet, after false discovery rate correction, this difference did not appear significant.

T4 apoptosis in the acute phase of the infection predicts sequelae
We had previously observed that programmed T cell death was a characteristic of severe COVID-19 (18,19).The destruction of T cells during acute infection might pave the way for persistence of SARS-CoV-2, reactivation of other viruses such as cytomegalovirus (CMV) and Epstein-Barr virus (EBV), autoimmunity, and immune dysregulation, phenomena which are all thought to potentially provoke long COVID.We therefore explored the hypothesis that T cell apoptosis might favor sequelae.For this purpose, we tested whether the intensity of T4 and T8 apoptosis that we measured in 21 patients at the acute phase of infection predicted sequelae.As we previously reported, compared with the healthy state, the acute phase of infection was characterized by a high level of T4 (31.6 ± 13.9 vs. 19.5 ± 6.3%, p < 10 -4 ) and T8 (46.5 ± 15.5 vs. 29.5 ± 10.6%, p < 10 -4 ) apoptosis (Figure 2).Interestingly, we noted that long COVID was preceded by a high level of programmed T4, but not T8, cell death at the acute phase (Figure 2).Indeed, patients who later suffered sequelae presented a higher level of T4 apoptosis than those who did not (36.9 ± 14.7 vs. 24.2± 9.0%, p = 0.016).This was particularly obvious for non-ICU patients, with a threshold of 25% T4 apoptosis distinguishing individuals doomed to suffer sequelae from individuals without (Figure 2A).Nonetheless, this was not the case for T8 apoptosis (46.2 ± 16.4 vs. 43.9± 17.34%, p = 0.611) which was similar between participants who developed post-acute syndrome or not (Figure 2B).

Discussion
In the present study of 29 individuals hospitalized for acute SARS-CoV-2 infection, we measured a large panel of biomarkers in order to unveil those capable of predicting the subsequent onset of long COVID.As a non-significant predictor, we identified the peripheral blood level of soluble P-selectin, a marker of endothelium and platelet activation, already positively associated with severe COVID-19 (20).Nonetheless, we did not identify any inflammation markers previously linked to the onset of sequelae, such as TNF-a, IL-6 or IFNI, as predictors of long COVID (8)(9)(10)(11)(12)(13)(14).This may be explained by the limited number of patients in our cohort.
However, we did identify an increase in T cell differentiation, activation, and exhaustion as well as perturbations in and NK monocyte subpopulations in patients with severe COVID-19 as observed by many groups (21)(22)(23)(24)(25).But these markers of COVID-19 severity did not differentiate patients who progressed towards long COVID from those who did not.Here again, the low number of patients we analyzed may have masked differences.By contrast, the intensity of programmed T4 cell death in the acute phase of the infection was able to perform this differentiation.Note that the relatively small number of participants in this study emphasizes the strength of this link between initial T4 cell death and subsequent sequelae.
In a previous work, we showed that T cell apoptosis was a major feature of severe COVID-19, more frequent in ICU than in non-ICU patients, and accounting for lymphopenia (18).One consequence of T4 apoptosis is the poor quality of the anti-spike humoral response from patients with critical disease (26).Given the ability of cytotoxic T cells to kill activated immune cells in the course of viral infections, T8 apoptosis might also favor a cytokine storm (27).Our data now raises the additional hypothesis that T4 apoptosis might be a cause of sequelae in SARS-CoV-2 infection.
Various pathophysiological hypotheses have been proposed to explain long COVID (28, 29).First, sequelae could be the In addition, T4 deficiency is known to provoke immune activation and autoimmunity (30).Immune activation is also known to provoke endothelial cell activation and coagulation (31).Finally, given the interactions between, on the one hand, the immune system and microbiota (32), metabolism (33), as well as hormones (34) on the other, T4 loss may provoke dysbiosis, hormonal and metabolic perturbations.Thus, in addition to impairing anti-SARS-CoV-2 immune response and participating in a cytokine storm during the acute phase, programmed T4 cell death could thereafter contribute to sequelae.
We recently reported that a low frequency of perforin-positive T8 cells may be observed at the acute phase of SARS-CoV-2 infection in patients who later presented sequelae.This observation aligns with our present data, arguing for a role of T cell deficiency in long COVID.
Our results are a supplementary argument for trying to prevent T cell apoptosis in SARS-CoV-2 infection.In this perspective, we have also shown that the pan-caspase inhibitor Q-VD inhibited ex vivo programmed death of COVID-19 T cells (18).Moreover, we have also reported that angiotensin II-induced monocytic reactive oxygen species production caused DNA damage and T cell apoptosis in hospitalized patients, a cascade which may be blocked by an angiotensin-II-receptor antagonist (sartan) and the antioxidant N-acetylcysteine (19).Therefore, sartans and Nacetylcysteine might be good candidates for impeding T cell apoptosis in the course of SARS-CoV-2 infection and the development of long COVID.
FIGURE 1 (A-M) Cell surface activation markers which differed between patients in the acute phase of SARS-CoV-2 infection and healthy donors.Patients were split into those who developed long COVID one year later and those who did not.Means ± SD are indicated.

TABLE 1
Characteristics of the study volunteers at the acute phase of the infection.
Bold p-values are below or equal to 0.05.

TABLE 2
Soluble activation marker concentrations in plasma and PBMC supernatant in the acute phase of infection in patients who suffered from long COVID one year later or not.