Ceramide Levels and Covid-19 Respiratory Distress, a Causal Relationship

: A causal relationship between plasma ceramide concentration and Covid-19 patients with respiratory distress symptoms is presented. Hence, monitoring of plasma ceramide concentration, targeting ceramide synthesis, its salvage and its regulatory mechanisms, are validated approach towards enhancing survival from Covid-19 respiratory distress. In this study, plasma samples of 52 individuals infected with Covid-19 were utilized in a lipidomic analysis. Lipids belonging to ceramide class exhibited a 400-fold increase in total plasma concentration in infected patients. Further analysis lead to demonstration of concentration dependency, for severe Covid-19 respiratory symptoms, in a subclass of ceramides. The subclasses Cer(d18:0/24:1), Cer(d18:1/24:1), and Cer(d18:1/22:0) are shown to be increased by 48, 40, and 33–folds respectively in infected plasma samples, and to 116, 91 and 50-folds in plasma samples with respiratory distress.


INTRODUCTION
Covid-19 virus penetrates vital organs such as the lung with a range from asymptomatic or mild infections restricted to the upper respiratory tract to severe respiratory syndromes hallmarked by disseminated spread to the lower airways. Observation of local inflammation and pneumonia, especially in patients with preconditions such as diabetes, hypertension, and cardiovascular disease are also reported (1). Supportive supplemental therapy for alleviation of these respiratory distress symptoms is the most viable strategy and the goal of most current efforts for improving Covid-19 patient survival.
Recent studies have highlighted that after infection some viruses can highjack and utilizes the host lipid metabolism for their own propagation to counteract cellular response (11,12). To explore evidences for similar virus host exchanges, and with the aim of better understanding the mechanisms in which Covid-19 virus infect host cells a number of metabolomics studies have been reported (2)(3)(4)(5)(6)(7)(8)(9). Schwarz et al. described the progression from moderate to severe disease to be marked by loss of specific immune regulatory lipid mediators (LMs) and increased pro-inflammatory constituents (3). Similarly, Wu et al. reported that in addition to dyslipidaemia major changes to carbamoyl phosphate and guanosine monophosphate (GMP) levels are also associated with the progress and severity of Covid-19 (4). Shen et al. in a metabolomics and proteomic study revealed characteristic protein and metabolite changes in the sera of severe Covid-19 patients which might be used in the selection of potential biomarker for the evaluation of the severity of disease (5). All these studies have reported unbiased analysis of patient plasma samples, and all have shown that changes in the Lipidome to be a factor for severe Covid-19 infection.
In an attempt to explore changes in patient's plasma metabolites as well as in the lipidome upon infection with Covid-19 we utilizing a lipidomics approach. We found associations of distinct lipids and hence, tested their modulation in relation to the severity of the infection. Our findings have demonstrated that Covid-19 infection causes modulation of the sphingolipid balance in the host plasma. A concentration dependent association between increases in the production of ceramides and the onset of respiratory distress symptoms of Covid-19 infection is observed.
Ceramides (Cer) are the products of metabolism of fats and lipids. They have been reported to accumulate in humans with obesity and hyperlipidemia. They are the central metabolite of the sphingolipid family and have also been reported to be the signaling molecule which can regulate ER stress, apoptosis, insulin sensitivity and inflammation (13,15). Increased Cer levels are heavily implicated in the pathogenies of insulin resistance, neurodegeneration conditions, lung diseases including asthma, chronic obstructive pulmonary disease (POD), pulmonary fibrosis (16).
Our observations of association and concentration dependency in combination with current understanding of Cer and their role in the pathobiology of lung inflammation, provide reasonable support for defining Cer and Covid-19 infection mediated respiratory distress, as a causal type relationship. By gaining greater understanding of this relationship, we shed light on the possible treatment options for which further research is needed.

Subhead 1: Identification of differential metabolites
The ions corresponding to d7-Androstenedione, d4-Cortisol, and d5-DHEAS, used as internal standards, were selected for extraction of the ion chromatographic peaks in a LC/MS/MS based comparative metabolomics analysis. Figure 1a shows the obtained score plots from metabolomics profiles of plasma from Covid-19 infected and control uninfected group. The clear separation shows that significantly different metabolites are present within each group. The loading plots shown in Figure 1b demonstrate these differences at the component level. The ions that are furthest from the origin of the plots, are the ones that make the most contribution to the separation seen in the score plots and were selected for further analysis. Figure 1c shows a profile of a metabolite chosen from the loading plot.
Each component of the loading plot was investigated and a list composed of the m/z of selected ions with a define profile of abundance and characteristics was formed. The accurate molecular ion masses were used for composition, calculation, and collection of data dependent MS/MS data. MS/MS data was collected for structural deduction and data base searches. Subhead 2: Identification and structural deduction of metabolites via data dependent MS/MS Figure 2 shows an overview of the results from analysis of plasma from uninfected and Covid-19 infected individuals using a data dependent MS/MS method. The combined monitored m/z ion profile from this analysis confirms the principal component analysis (PCA) directed selection process. Structural deduction and data base searches demonstrate that >283 of the metabolites identified belong to the lipid family.    Figure  4c shows that, when only comparing the metabolic differences in the plasma of the Covid-19 infected with mild to the severe symptoms, the observed changes in lipid profiles become limited to Cer (average more than 1.5 -2 fold increase this study) and its derivatives HexCer, Hex2Cer.

Subhead 4: Covid-19 respiratory distress and Cer
The compositional makeup of the subclasses of lipids present in the plasma of Covid-19 infected patients with mild symptoms vs those with respiratory distress shows no change at the subclass levels in each class of the identified lipids (Supplementary Figure 1). The overall change in plasma concentration of each lipid class is estimated by multiplying the average fold increases in peak area of each subclass by the number of identified lipid subclasses. Figure 5a demonstrates the concentration of each lipid class in relative terms of fold induction to the uninfected control group. Figure 5b shows the calculated plasma concentration of Cer and Sphingomyelin (SM) class of lipids in µmol/l in the uninfected, infected with mild, and infected with severe respiratory distress symptoms (SRM 1950 -metabolites by LIPID MAPS consortium, was the basis for comparison and establishment of concentration (14)). SM, not being identified to be modulated in plasma by Covid-19 infection, serves as control in this study. These figures show that the total levels of Cer are increased in plasma of Covid-19 infected individuals with mild symptoms by more than 250-folds or to 500 µmol/l. In cases of Covid-19 infected individuals with severe respiratory distress symptoms Cer levels are increased by over 450-folds or to 720 μmol/l (Figure 5a, 5b). In summary an increase of 220 µmol/l plasma Cer concentration is the change seen in between a Covid-19 infection with mild symptoms to an infection with severe respiratory distress response.  (14)). The overall levels of Cer(d18:1/16:0) to be 2.79 μmol/l, and Cer(d18:1/24:1) to be 39.57 μmol/l in the plasma of Covid-19 infected individuals with no symptoms of distress and levels in Cer(d18:1/16:0) to be increased to 3.63 μmol/l, (30 % increase) and Cer(d18:1/24:1) to be at 95.63 μmol/l (142% increase) in the plasma of respiratory distress individuals.
Ratios of lipids can be very informative tool for the process of identification of more selective targets for regulation of each Cer subclass. Figure 6c, and 6d show comparative ratios of the identified subclasses Cer with their associated dihydroceramide and monohexosylceramide species in the plasma of Covid-19 infected individuals with mild vs severe respiratory distress symptoms. These figures demonstrate both the presence and relative ratio of the Cer substrates to Cer in a subclass dependent manner. In figure 6c the almost 2 fold greater Cer(18:1/22:0) to its Hexosyl derivative ratio in respiratory distress patients shows a tool to predict severe respiratory distress and a possible mechanism for modulation. Figure 6d, demonstrated presence with relative abundance of substrates for de novo synthesis of Cer(d18:1/20:0), Cer(d18:1/16:0), and Cer(d18:1/24:1) in plasma of Covid-19 infected individuals.

DISCUSSION
A causal relationship between plasma levels of Cer and the onset of Covid-19 respiratory distress symptoms is claimed. Cer involvement have been highlighted in reports of lung diseases including: acute lung injury, cystic fibrosis, emphysema, lung infections, and asthma (13,16). These reports show pulmonary manifestation to be the main symptoms of Cer mediated toxicity. That Cer's can regulate major aspects of lung endothelial cell function and are involved in the pathogenesis of several conditions associated with pulmonary vascular dysfunction (16). These pulmonary manifestation symptoms are common with the symptoms also seen in patients with Covid-19 associated respiratory distress (17). The reported high levels of plasma Cer concentration in the obese, and asthmatic individuals is consistent with the observed sensitivity of this group to Covid-19 infection supporting its biological relevance (25).
In this study, we show that in addition to the described biological relevance, association, and concentration dependency, of specific subclasses of Cer in plasma of Covid-19 infected individuals, with respiratory distress symptoms are observed. These findings provide the support for existence of the claimed causal type relationship.
Covid-19 infection causes profound changes in the metabolome of infected individuals, including plasma lipids levels. In this study, changes in plasma levels of 283 lipids covering 8 lipid classes, including PS, PE, PG, Cer, HexCer, Hex2Cer, Hex3Cer has been identified to be associated with Covid-19 infection. Figure 3a, 3b, shows the levels of subclasses of the above classes, as represented by peak area, to be markedly increased in the plasma of Covid-19 infected individuals, describing an association between the identified subclasses of lipids to the Covid-19 infected plasma. Figure 5a, shows changes in concentration of Cer as a class, in terms of fold induction, the total levels of Cer are increased in plasma of Covid-19 infected individuals with mild symptoms by more than 250-folds or to 500 µmol/l (figure 5b). In cases of Covid-19 infected individuals with severe respiratory distress symptoms this total Cer level is increased by over 450-folds or to 720 µmol/l. Demonstrating concentration dependency, the difference between a Covid-19 infection with mild symptoms to an infection with severe respiratory distress response at the lipid class level to be an increase of 220 µmol/l plasma in total Cer class concentration. The predominant Cer subclasses (figure 6a and 6b) detected to be associated with these plasma levels in respiratory distress Covid-19 infection are Cer(d18:1/16:0) at 3.63 μmol/l, and Cer(d18:1/24:1) at 95.63 μmol/l. The concentrations of which are 2.79 umol/l (30% less) and 39.57 umol/l (142% less) respectively in plasma of Covid-19 infected individuals with no symptoms of distress.
The Cer(d18:1/16:0), Cer(d18:1/24:1) subclasses were reported to be involved in a cardiac mortality study and to be associated with increased risk of cardiac death outcomes, in patients with stable coronary artery disease (18). In this study we report that Covid-19 infection causes a significantly greater mortality risk with a 10 and a 30-fold increase in plasma levels of these subclasses over the reported cardiac mortality predictions level.
Causality assessment, leads us to believe that monitoring and therapeutic aim of reducing Cer(d18:1/16:0), and Cer(d18:1/24:1) levels in the plasma of patients is required for enhancing survival from Covid-19 respiratory distress. A reduction in plasma levels of the above Cer subclasses (bringing down the observed plasma levels to the level seen associated with mild symptoms) can be an aim to aid in the recovery process of the severe symptoms associated with Covid-19 infection.
Cer are generated by de novo synthesis, salvage of sphingosine, and breakdown of complex sphingolipids, including sphingomyelin. In which the necessary substrates are: monohexocylceramide, sphingomyelin, and dihydroceramide (20). An increase in all three of these substrates is observed in plasma of Covid-19 infected individuals (figure 3a, figures 6a, and 6b).
To effectively modulate toxic effects of Cer, in the context of Covid-19 infection, is to reduce its circulating concentrations. This can be achieved by inhibiting its biosynthesis or by chemical modification of the circulating Cer in the plasma. The Cer de novo synthesis pathway includes a series of enzymes that produce Cer from the starting components serine and palmitoyl CoA. Researchers seeking to pharmacologically inhibit Cer synthesis in vivo have generally used myriocin, which inhibits serine palmitoyltransferase, the rate limiting initial step in the biosynthesis of all sphingolipids (20).
Reduction of circulating concentration of Cer in the plasma can be achieved by causing chemical modifications with introduction of enzyme or agonist to at least one ceramidemodifying enzymes such as Glucosylceramide synthase, Ceramidase (SMase), Ceramide kinase, and Sphingomyelin synthase. Both acid and neutral SMase inhibitors, as well as inhibitors of the de novo pathway of Cer synthesis, effectively inhibited Cer-induced apoptosis in the lung in various acute or chronic injury models in vivo, as recently reviewed by Uhlig and Gulbins (22).

Carpinteiro et al. reported that by pharmacological and genetic inhibition of acid SMase enzyme they can prevent infection of cells with Covid-19, VSV and PP.VSV viruses (21).
Indirect examples of regulators or circulating Cer levels have also been described to include TNF-alpha inhibitors, TLR-4 inhibitors, adiponectin, FGF21, apoptosis inhibitors, and mitophagy inhibitors (23,24).
Future studies should address more effectively, the limitations of this study, the third criterion for causality which requires that alternative explanations for the observed relationship between two variables to be ruled out (non-spuriousness, or "not false."). To better address these criteria one would need to directly monitor the effect of lowering plasma concentrations of Cer, treatment with Cer inhibitors, on the Covid-19 mediated respiratory distress. Furthermore, our investigations have focused on the detected subclasses of Cer isolated through a none comprehensive metabolomics analysis paradigm, a more comprehensive approach may lead to identification of more sensitive subclasses of these lipids.
In summary, uncovering causal relationships is an important first step towards understanding disease and predicting the course of future treatments. In this study a causal relationship between plasma Cer plasma concentration and respiratory distress symptoms in Covid-19 patients is presented. Specific subclasses of Cer has been identified that can be used for monitoring Covid-19 infection severity and progression. The causality relationship also defines that modulating ceramide synthesis pathways, and its salvage and its regulatory mechanisms, to be a validated approach towards enhancing survival from Covid-19 respiratory distress. For the metabolomics evaluation directly before analysis, the samples were thawed on ice. Then, 75 μL of pre-cooled ISP precipitant solution (ClinMass® Steriods in Plasma LC-MS/MS Complete Kit (RECIPE Chemicals + Instruments GmbH, Munich, Germany)) was added to 50 μL of plasma, the mixture was vortexed for 30 s, and left for 10 min before centrifuged at 14,000 rpm for 10 min at 4°C. The supernatant was separated and 40 μl of it was subjected to metabolite analysis by SCIEX X500R UPLC−QTOF/MS. The ISP contains for internal standards six steroids in its panel included d7-Androstenedione, d4-Cortisol, d5-DHEAS, d5-11-Deoxycortisol, d8-17-hydroxyprogesterone, and d3-Testosterone.

Instrumentation
HPLC separation was performed using the SCIEX ExionLC AC system on a Phenomenex: Luna® Omega 3 m Polar C18 100A column. Data was collected on a SCIEX X500R QTOF mass spectrometer with SCIEX OS software: TOF -MS survey scan with Information Dependent Acquisition (IDA) -triggering of up to 16 product ion scans. Data processing was also performed in SCIEX OS software with simultaneous identification and quantification being accomplished in single software (all devices from AB SCIEX, Framingham, U.S.A.). 40 ul of the sample was injected for chromatography separation at 30 °C, with a flow rate of 0.8 mL/min, 0.1% formic acid and 2 mM Ammonium Formate in HPLC water as mobile phase A and HPLC acetonitrile and Methanol (50:50) plus 0.1% formic acid and 2 mM Ammonium Formate as mobile phase B. A 26-min linear gradient was set as follows: 0 min: 2% B, 0-1 min: 2% B, 1-16 min: 98% B, 16-20 min: 98% B. The SCIEX X500R QTOF system with a Turbo V™ source and capable of electrospray ionization (ESI) was used in positive polarity. The ion source temperature was set to 500ºC and the ion source voltage was set to 5500 V. An information dependent acquisition (IDA) method, consisting of a TOF-MS survey (250-1100 Da for 350 msec). The declustering potential (DP) was set to 80V. The collision energy (CE) was set to 10 eV with a collision energy spread (CES) of ± 0 eV. To achieve the most complete MS/MS coverage, the dynamic background subtraction (DBS) function was activated.

Data processing
MarkerView™ Software 1.3 (SCIEX) was used to process, align, deconvolute, and normalize (sum of total area) the obtained raw data in which the retention time (RT) was from 0.5 min to 26 min. Mass and RT tolerance values were set to 10 ppm and 0.15 minutes. Mass and RT of internal standards used for analysis include: d7-Androstenedione, d4-Cortisol, d5-DHEAS, d5-11-Deoxycortisol, d8-17-hydroxyprogesterone, and d3-Testosterone introduced via the ISP mixture during sample preparation. Principal component analysis (PCA) was used to visualize system stability of the system and sample distribution. The orthogonal partial least squares discriminant analysis (OPLS-DA) was used to identify the variables responsible for the discrimination. The system stability (RSD %) of m/z, peak areas and retention times were as follows 0.02 -0.14%, 0.0003 -0.0007%, 5.9 -8.5%).
A list of the intensities for each detected peak was generated, using retention time and the massto-charge (m/z) ratio data pairs as the parameters for each ion. Manual scanning of the 10,000 spectral features represented by a unique m/z, retention time, and peak area allowed for generation of a list of 670 peaks of interest used for further evaluation. The Formula Finder algorithm was used to identify potential differential metabolites and generate a group of probable formulas on an unknown ion based on the secondary fragment information, mass error, and isotope distribution patterns.
After the data preprocessing, the LipidCreator workbench software in Skyline (available at https://lifs.isas.de/lipidcreator) was used for both quantification and qualification of the lipid species present within the plasma samples (10).