In order to investigate the ASM/ceramide signaling in SARS-CoV-2 infection, blood samples of 23 patients and six healthy volunteers were collected. Clinical data in Table 1 summarize socio-demographic and selected clinical characteristics of the study group (day 1 of patients at intensive care unit due to COVID-19) and healthy controls. All patients received standard treatment by the time of admission, which did not include specific anti-viral or general anti-inflammatory drugs such as dexamethasone, because it was not standard of care at time of inclusion. The clinical severity assessment by use of sequential organ failure assessment (SOFA) and the simplified acute physiology score (SAPS II) were 9 (5-11, SOFA) and 41 (34-49, SAPS II) at enrollment. Predicted mortality rate of COVID-19 patients was matching that observed at ICU (26.6. vs. 30%). Parameters of healthy controls were all found in normal range. In contrast, a significant reduction in hemoglobin, hematocrit, RBC count, mean corpuscular hemoglobin concentration (MCHC), serum albumin, and lipoproteins were observed in COVID-19 patients.

First, we analyzed the pattern of sphingomyelin in lipid extracts of red blood cells and serum samples. Analysis of serum samples revealed that sphingomyelin levels in circulating lipoproteins were decreased in COVID-19 patients as compared to healthy controls ( Figure 1A ). Since host response and hypoxia have profound effects on RBCs’ morphology, rheology and functional activity (39), we thus sought to assess the composition of this cellular subpopulation. Similar results were obtained in RBCs, a reduction of sphingolipid content in these cells supports the concept of a deranged sphingomyelin balance in RBC-membranes in COVID-19 ( Figure 1B ). Next, we analyzed metabolites and degradation products of sphingomyelin. Therefore, we analyzed ceramide species with a naturally occurring sphingoid backbone (d18:1), but without any modification (n=11) with a chain length in a range between 12 and 26 carbon atoms and in part with an unsaturated double bond (i.e. 12:0, 14:0, 16:0, 18:0, 18:1, 20:0, 22:0, 24:0, 24:1, 26:0 and 26:1). Thereof, six ceramide species were found above lower limit of detection (LOD) in > 75% of all samples, which were included in subsequent analyses. Our results - presenting a specific disease pattern of concurring ceramide specimen - are in line with information from previously published cohorts (21). In serum as well as in RBC, Cer was found to be increased in COVID-19 patients and this increase was paralleled by a decrease of SM ( Figure 1 ).

Principal component analysis was then used to test whether differences in the ceramide profile reflected the clinical diagnosis ( Figure 2 ). In both compartments, serum samples as well as extracts from RBC, values clustered distinctly between patients and healthy controls. For RBC the first and second principal components parameters were 76.5 and 12.1% ( Figure 2A ), for serum 68.4 and 15.6% ( Figure 2B ), respectively.

Concentration levels of the majority of ceramide specimens were significantly changed in patients with COVID-19 with respect to controls, either in serum or RBC. The absolute amount of investigated specimen varied with respect to expected baseline levels ( Figure 3 ). An overview of resulting p-values comparing the groups is given in Table 2 . Median values of individual ceramide specimen including interquartile range (Q1/Q3) are given in Table 3 showing increased values in both compartments, either in serum or RBC.

Considering the increase of all investigated ceramide specimen in COVID-19 patients in serum and RBC, we then tested whether the corresponding stress associated enzyme ASM - converting sphingomyelin to ceramide - was more abundant and/or more active in COVID-19 patients. Indeed, in serum of COVID-19 patients the enzyme activity was markedly increased: median 4.525 (Q25% 3.827; Q75% 5.832) nmol/(mL x h) as opposed to 0.948 nmol/(mL x h) (Q25% 0.857; Q75% 0.999) in healthy controls and this effect was statistically significant ( Figure 4 ) reflecting an inflammation-driven sphingolipid reprogramming. In patients with unfavorable outcome (n=7), the highly increased activity levels remained nearly unchanged, but in patients with recovery (n=16), we observed decreasing values (data not shown).

Correlating ASM activity with clinical parameters at day of ICU admission revealed a clear association of ASM activity and severity of COVID-19 ( Table 4 ). Most prominently, we found that the concentration of high density lipoprotein (HDL) negatively correlated with ASM-activity, followed by biomarkers of metabolic dysfunction such as base excess, concentration of lactate and hydrogen carbonate. Further, the concentration of long chain sphingomyelin from serum constituents (lipoproteins) was found to be negatively correlated with ASM-activity. A positive correlation was found for plasma magnesium concentration, alanine aminotransferase (ALT), total bilirubin, ferritin and lactate dehydrogenase.

It might be speculated, that the origin of the circulating enzyme activity might be ascribed to loss of integrity of epithelial tissues in affected lungs, of which severe damage of alveolae beyond remodeling of parenchyma is manifest along histopathological findings of the disease (40, 41). Exocytosis of ASM is a common and rather protective mechanism for rapid membrane resealing and restoration of its integrity (18, 42). The appearance of ASM in circulation is a sensitive, but unspecific event in patients with severe inflammation resulting from infection such as polymicrobial sepsis (19, 20), pneumonia (21), radiation therapy (43) or chronic inflammation (44). In community-acquired pneumonia patients, Arshad et al. recently reported a nearly threefold increase of plasma ASM activity, its close correlation with severity markers (C-reactive protein, procalcitonin), a concomitant increase of SMPD1-expression in circulating white blood cells (twofold), and, most interestingly, a derangement of plasma ceramide profile (21), which is very similar to that we found in our study. There are several reports on correlations between plasma ceramide concentration and unfavorable outcome of critically ill patients (45, 46), especially in the lungs (47). On a molecular level, ceramides have been reported to activate inflammatory pathways in several abnormal physiological circumstances involving insulin resistance, mitochondrial dysfunction and endoplasmatic reticulum stress (48–51), which are attributed to worsen the clinical condition of COVID-19 patients (52–54). Despite the fact, that we are only presenting ceramide data from accessible compartments from these patients (serum and RBC), a close similarity of membrane bound ceramides in bodies’ tissues with that we analyzed is persuasive, since the hydrolyzing capacity is restricted to ASM in these conditions.

As a long-term sequelae of COVID-19, lung fibrosis is characterized by deterioration of organ function and subsequent respiratory failure (55). Since high quality data regarding long-term clinical outcomes from COVID-19 survivors are still unavailable, predictions for long-term outcome thereof are speculative at best, but it is well known that lung fibrosis as a result of other diseases closely correlates with poor prognosis (56). The underlying, irreversible process is driven by sometimes excessive release of pro-fibrotic factors (56), especially TGF-β from injured lung tissue, turning a well-controlled healing response into a pathogenic fibrotic response (56, 57). In follow-up chest imaging from severe courses, the presence of intestitial thickening, irregular interface and parenchymal bands have been suggested as predictors of COVID-19 pulmonary fibrosis (58). Generation and accumulation of ceramides were identified as pacemakers in pathogenesis of pulmonary fibrosis in cystic fibrosis in mice and men (59–61). As a consequence, inhibition of ASM resulted in normalization of pulmonary ceramide levels, inflammation and bacterial infection (62–64).

These and also previous observations from our group demonstrated that an increase of ASM-activity, ceramide formation, TGF-β circulation and - ultimately - liver fibrosis in a mouse model of polymicrobial sepsis might be abrogated by ASM-inhibition (65). These findings also support the concept that activation of the sphingomyelinase-ceramide-pathway is a universal response mechanism during host response. Furthermore, inhibition of the enzyme might have both, short- and long-term beneficial effects during the course of the disease. Especially, in high risk patients anti-fibrotic therapy is a matter of debate (66). Strikingly, data from a small series of independent retrospective studies, clearly attest a benefit for patients undergoing anti-depressive therapy precedent to hospitalization due to COVID-19 (67–70). There is an anticipation that anti-depressive drugs as functional inhibitors of sphingomyelinase (FIASMA) (71) exhibit anti-infective properties in epithelial cell with respect to SARS-CoV-2 (24, 72). Hoertel et al. showed a significant association between the prehospital use of antidepressants with subsequent inhibition of ASM and reduced likelihood of intubation or risk of death due to SARS-CoV-2 complication (67). Notably in older adults, the short-term use of FIASMA is generally well tolerated (73, 74). A trend for beneficial effects of FIASMA with respect to in-hospital mortality rate (potentially marked by older age and higher prevalence of comorbidities) was superimposed by co-medication with amlodipine (70). The later drug also exerts inhibitory capacity to ASM, but also antiviral effects in-vitro (75, 76) might be exaggerated by the Ca⁺⁺-modulating mechanism (77). In line with these results, there are similar data from small studies demonstrating a lower mortality rate after treatment with nifedipine or amlodipine (75, 78). A recent update on the potential role of either a chronic expose to FIASMA or as an interventional measure following hospitalization due to SARS-CoV-2 infection underlines the interest to evaluate these drugs as off-label therapy in SARS-CoV-2 infection (79). Interestingly, recent reports demonstrated a synergistic effect of the FIASMA fluoxetine and the direct anti-viral agent remdesivir and its metabolite in an in-vitro-model of polarized Calu-3-cells: super-additive effectiveness highlighted key advantages of a combined approach against the propagation of the viral pathogen as well as maintenance of endosomal lipid balance for entry processing into the host at low concentrations minimizing potential adverse effects of the drugs (80, 81).

RBC play an important role in oxygen transport and supply as well as they are fulfilling a plethora of metabolic activities. Within this cell population, ASM-triggered ceramide generation contributes to Ca⁺⁺-sensitivity resulting in the release of extracellular vesicles by shaping the membrane’s curvature (82) and induction of eryptosis, the suicidal death of RBC (83, 84). Also, ASM induced ceramide generation changes the biophysical properties (85) resulting in an increase of rigidity in membranes (86, 87) impairing the function of RBC. This is in line with observations from anemic COVID-19 patients, exhibiting RBC shape abnormalities and morphological changes leading to a spherocyte shape, which are all characterized by loss of elastic properties (88). After recovery, blood smear showed unremarkable morphology (88). The same is true with respect to formation and release of extracellular vesicles, since ceramide generation is pace-making as shown by studies using inhibitors of both isoforms of sphingomyelinase (i.e. GW4869, imipramine) (89). Beside the fact of increased levels of circulating extracellular vesicles that may drive thrombosis in patients undergoing COVID-19 (90), to the best of our knowledge, there are no clinical observations of RBC-borne vesicles as yet.

Data from our correlational analyses support the hypothesis that ASM activity is either a mediator or marker of a severe clinical course in our patient cohort. It is well described that patients with decreased HDL-levels are at an increased risk to develop a severe disease course compared to patients with high HDL levels (91). Here, low HDL values are found to be decreased in COVID-19 patients ( Table 1 ) and are associated with increased ASM activity ( Table 4 ). Most interestingly, ASM activity is found to be associated with biomarkers of metabolic dysfunction such as hydrogen carbonate, base excess and lactate concentration, of which the latter one was found in our patients in normal range ( Table 1 ). Unlike polymicrobial sepsis, in COVID-19 lactate levels are usually normal despite severe pneumonia or manifest ARDS without any prognostic value with respect to outcome (27, 92). In our study, ASM-activity and lactate levels are highly correlating at the day of admission to the intensive care unit. It is unclear, whether low levels of lactate are caused by increased consumption by lactate dehydrogenase (93), which is also found to be associated with ASM-activity. Increased values of lactate dehydrogenase activity (LDH) at hospitalization are positively associated with mortality (93). Considering the parallel alteration of ASM and LDH, one might speculate that both enzymes might be released probably from injured heart and lung tissue (27). Therefore, it might be interesting to determine the enzymatic activity of proteins in bronchoalveolar fluid, where a similar change might be expected. Furthermore ASM activity is positively associated with marker of impaired liver function such as ALT and bilirubin, which are both (beyond LDH) previously described as screening prognosticators of severe courses at early stages of the disease (94). The same is true with respect to ferritin as a surrogate for hyper-immune responsiveness, since baseline levels at ICU-admission are increased ( Table 1 ) (95). Established biomarkers such as procalcitonin (p= 0.065) or SOFA-values (0.138) failed to reach the level of significance due to the small cohort of patients. Interestingly, there is no association of ASM activity with values of troponin I in patients as a surrogate for damage of myocardial tissue (p=0.3365), thus the serum abundant ASM activity in COVID-19 is proposed to be released from epithelial tissue of the affected lungs as previously described in asthma (96, 97) and acute respiratory syncytial virus bronchiolitis (98).

Regardless of these promising new results, this study is also confronted with particular limitations. Our study was carried out at a single center with a quite limited number of patients, all of them admitted to the intensive care unit. Nevertheless, we are convinced, that the cohort is sufficiently powered for the presented results. However, the size of the cohort does not allow for the analysis of distinct subsets such as outcome, need for ventilation, or any prediction of the consequences of high/low levels of sphingomyelinase from circulation or in RBC. Furthermore, we cannot exclude that our results are biased by sample size or (anti-depressant) treatment strategies. We agree that data about prehospital treatment with FIASMAs would strengthen our results but due to the observational design and the confirmative character of the study of an unknown hypothesis we are not able to provide those data. Thus, also considering the small group size, a comparative analysis of our patients with respect to pretreatment with FIASMAs might over expand the interpretation of our results.

We measured sphingomyelinase activity and subsequent derangement of ceramide profile in serum/RBC and correlated the measured values with clinical and laboratory parameters. But our observations cannot explain cause-consequences at the end, especially whether the increase of enzyme activity is an epiphenomenon of or a reason for deterioration with subsequent need for intensive care treatment. Our study, without external validation, was primarily not designed to assess long-term outcomes and was therefore not feasible to screen for prognostic biomarkers for long-term sequelae. Nonetheless, our data highlight a potentially crucial signaling pathway in COVID-19 patients that warrants further investigations. Notably, the activation of circulating sphingomyelinase (ASM) and subsequent ceramide generation during host response in these patients provide a promising approach for functional inhibition by FDA-approved drugs to control resulting organ dysfunction to help the body maintaining homeostasis. Larger multicenter, interventional trials are now needed to test the potential benefit of an inhibitory strategy in critically ill patients with COVID-19.

The data from our study close the gap between retrospective observations by presenting a potential mechanism of ASM release and action in COVID-19 patients. Keeping the paucity of proven host-directed therapies in mind, the low level of evidence of the majority of all running trials (14, 99) and due to the fact that a panel of FDA-approved drugs with low risk of adverse effects is awaiting consequent investigation, the potential usefulness of anti-depressants in patients with COVID-19 taking FIASMA for other indications with now known guiding principles should be prioritized for RCT and can minimize the risk of being exposed to novel, potentially harmful of ineffective compounds or compounds with unknown mode of action. As also recommended by others, the results support the continuation of FIASMA medication in these patients (69). Moreover, the fact of pretreatment with FIASMA should be considered while interpreting the results from hundreds of running clinical trials (17). Double-blind controlled randomized clinical trials of antidepressive medication with FIASMA for transient inhibition of ASM in COVID-19 are of great interest to investigate how the ASM/ceramide-pathway affects disease severity, organ damage and improvement of clinical course.