Decision letter: Neutrophil-mediated oxidative stress and albumin structural damage predict COVID-19-associated mortality

Abstract

Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Human serum albumin (HSA) is the frontline antioxidant protein in blood with established anti-inflammatory and anticoagulation functions. Here, we report that COVID-19-induced oxidative stress inflicts structural damages to HSA and is linked with mortality outcome in critically ill patients. We recruited 39 patients who were followed up for a median of 12.5 days (1–35 days), among them 23 had died. Analyzing blood samples from patients and healthy individuals (n=11), we provide evidence that neutrophils are major sources of oxidative stress in blood and that hydrogen peroxide is highly accumulated in plasmas of non-survivors. We then analyzed electron paramagnetic resonance spectra of spin-labeled fatty acids (SLFAs) bound with HSA in whole blood of control, survivor, and non-survivor subjects (n=10–11). Non-survivors’ HSA showed dramatically reduced protein packing order parameter, faster SLFA correlational rotational time, and smaller S/W ratio (strong-binding/weak-binding sites within HSA), all reflecting remarkably fluid protein microenvironments. Following loading/unloading of 16-DSA, we show that the transport function of HSA may be impaired in severe patients. Stratified at the means, Kaplan–Meier survival analysis indicated that lower values of S/W ratio and accumulated H2O2 in plasma significantly predicted in-hospital mortality (S/W≤0.15, 81.8% (18/22) vs. S/W>0.15, 18.2% (4/22), p=0.023; plasma [H2O2]>8.6 μM, 65.2% (15/23) vs. 34.8% (8/23), p=0.043). When we combined these two parameters as the ratio ((S/W)/[H2O2]) to derive a risk score, the resultant risk score lower than the mean (<0.019) predicted mortality with high fidelity (95.5% (21/22) vs. 4.5% (1/22), log-rank χ2=12.1, p=4.9×10−4). The derived parameters may provide a surrogate marker to assess new candidates for COVID-19 treatments targeting HSA replacements and/or oxidative stress. Editor's evaluation This submission is novel since it provides information on the structure changes of albumin in COVID-19. https://doi.org/10.7554/eLife.69417.sa0 Decision letter Reviews on Sciety eLife's review process Introduction COVID-19 pandemic continues as a global health crisis while the underlying SARS-CoV-2 virus defies all attempted treatment strategies. While writing this report, there have been more than 135 million confirmed cases including around 3 million deaths worldwide according to the World Health Organization Coronavirus Disease Dashboard (https://covid19.who.int/). Although 50% of cases are reported to be in the 25–64 age group, the percentage of deaths increases dramatically with age, and approximately 75% of deaths are in those aged 65 years and above (COVID-19 Hospitalization and Death by Age | CDC). People in the age groups 30–39 years, 40–49 years, and 50–64 years are 4, 10, and 30 times more likely to die from COVID-19 complications compared to the 18–29 years age group. Nevertheless, molecular and cellular factors contributing to mortality outcome in a homogeneous cohort of patients are not yet clear. Lack of diagnostic markers that predict mortality in COVID-19 patients impedes current efforts to siege the pandemic. It is thus critical to identify prognostic tests that can assess the risk of death in critically ill patients to guide clinical protocols and prioritize interventions. Furthermore, mechanistic clues for determining the underlying molecular factors contributing to the hypercoagulability, inflammation, and cytokine storm have been so far illusive. It is therefore imperative to intensify efforts focusing on understanding the molecular pathophysiology of COVID-19 infection and to identify prognostic markers to guide and prioritize clinical decisions. Human serum albumin (HSA) is the most abundant constituent of soluble proteins in the circulatory system. HSA has been suggested and used as a diagnostic and prognostic marker of numerous diseases and conditions including ischemia, rheumatoid arthritis, cancer, septic shock, among many others. In addition to its numerous physiological and pharmacological functions including the maintenance of blood/tissue osmotic balance (Singh-Zocchi et al., 1999), blood pH, metal cation transport and homeostasis (Bal et al., 2013; Stewart et al., 2003), nutrients and drug shuttling (Fujiwara and Amisaki, 2013; Wishart et al., 2018), and toxin neutralization (Ascenzi et al., 2006; Vorum and Honoré, 1996), HSA is suggested to be a major circulating antioxidant (Cha and Kim, 1996; Loban et al., 1997). HSA can remarkably bind with a diverse array of drugs and toxins thus controlling their bioavailability and pharmacologic effects (Fasano et al., 2005). It has been previously shown that more than 70% of the free radical-trapping capacity of serum was due to HSA (reviewed in Roche et al., 2008). Importantly, several reports indicated that inflammation enhances vascular permeability of various tissues to HSA apparently to confer antioxidant beneficial effects against reactive species released by activated neutrophils (Cross et al., 1994; Halliwell, 1988; Sitar et al., 2013). Although currently without direct experimental evidence, neutrophilia-mediated oxidative stress was implicated in the COVID-19 pathology and speculated to exacerbate the inflammatory immune response eventually causing multi-organ failure and death (Laforge et al., 2020). We hypothesized that COVID-19-mediated oxidative stress may be differentially reflected in HSA’s structure and functions and employed electron paramagnetic resonance (EPR) spin labeling spectroscopy to explore HSA’s structural changes in correlation with severity and mortality of critically ill COVID-19 patients. Spin-labeled fatty acids (SLFAs) are established probes to explore structural and functional changes in albumin by EPR spectroscopy (Ge et al., 1990; Haeri et al., 2019). This approach relies on the well-studied ability of albumin to strongly and exclusively bind with fatty acids in blood. Albumin has at least seven different specific binding sites for long-chain fatty acids located in different domains within the protein (Bhattacharya et al., 2000; Curry et al., 1999; Simard et al., 2006). Effectively, structural and functional changes in HSA may be assessed through the detection of parallel changes in mobility and binding affinity of SLFAs, in addition to the distribution of the spin labels on the albumin molecule (Haeri et al., 2019). EPR spectra of spin labels bound to different domains of the protein provide information on the local fatty acids/protein interactions, which may probe changes in the overall structure of the protein under unfolding or damaging conditions (Figure 1A; Bhattacharya et al., 2000). Here, we compare changes that occur to the mobility, binding affinity, and distribution of the HSA-bound SLFA in whole blood and plasma from COVID-19 patients in critical care unit relative to those observed in normal healthy individuals. Figure 1 Download asset Open asset Probing structual changes of serum albumin through spin labeling EPR spectroscopy. (A) HSA crystal structure containing seven copies of stearic acid. (B) Representative EPR spectra of free and HSA-bound 5-DSA (B) and 16-DSA (C) in whole blood from the same COVID-19 recovered patient. Chemical structures of the two spin-labeled fatty acids are given on the right side of the figure. EPR, electron paramagnetic resonance; HSA, human serum albumin. Results Demographic, clinical, and laboratory hematologic characteristics of COVID-19 patients Table 1 lists demographic data, comorbidities, ongoing medications, and administered anti-COVID-19 medications applied to treat current study participants that were divided into survivors (Sev-R) and deceased (Sev-D). No clinical or demographic characteristic showed statistically significant difference between Sev-R and Sev-D groups when analyzed by Pearson’s Chi-square test. In Table 2, we show and statistically compare laboratory results of survivors versus non-survivor COVID-19 groups. Although when comparing all parameters in the two COVID-19 groups, we observed changes following the same reported trends in the literature, means’ comparisons by Tukey test reported non-significant changes in all parameters except for a significant decrease in albumin level (p<0.05) and a strong trend observed for C-reactive protein (CRP) (greater levels in Sev-D group, p=0.06). Nevertheless, non-survivors’ blood carried the frequently observed hallmarks of increased CRP, D-dimer, IL-6, ferritin, and the liver enzymes ALT and AST (reviewed in: Singh et al., 2021; Velavan and Meyer, 2020). However, it is conceivable that the clinical severe category and the same ICU status of patients in the two groups in addition to relatively small sample sizes underlie the observed lack of robust statistical differences between these parameters. Table 1 Demographic and clinical characteristics of the studied subjects. Sev-RSev-DTukey 95% CIpn1623Age (mean ± SD)60.7±9.567.8±13.24.1–17.80.09Male56.25%63.16%0.677†sO2 (mean ± SD)82.1±18.776.1±18.6–20.5 to 8.40.40Hypertension12.5%42.1%0.053†Diabetes25%75%0.08†Cardiovascular disease0%15.8%0.10†Cancer0%10.5%0.18†Bronchial asthma6.25%10.5%0.65†ACE inhibitors0%7.14%0.47†ARBs9.09%7.14%0.85†calcium channel blocker14.28%7.14%0.60†Beta blockers0%7.14%0.47†Diuretics0%7.14%0.47†Sulphonylurea14.28%21.43%0.69†Other oral hypoglycemic0%21.43%0.18†Insulin42.85%21.43%0.30†Anticoagulant57.14%50.0%0.76†Steroids71.42%64.28%0.74†Hydroxychloro-quine14.28%7.14%0.60†IL-6 receptor antibody28.57%21.42%0.72†Proton-pump inhibitor28.57%42.85%0.52†Azithromycin14.28%28.57%0.47†Cephalosporin42.85%21.43%0.30†Carbapenem42.85%42.85%1.0†Oxazolidinone42.85%28.57%0.51†Fluoro-quinolone42.85%21.43%0.30†Nitrofuran14.28%0%0.15†Remdesivir14.28%28.57%0.47†Ivermectin0%28.57%0.11† sO2, blood oxygen saturation level; ACE, angiotensin-converting enzyme; ARB, angiotensin II receptor blocker; IL-6, interleukin-6. † p values obtained through Pearson’s χ2 test. Table 2 Laboratory parameters of the current study patients. WBC, white blood cell; INR, international normalized ratio; CRP, high-sensitivity C-reactive protein; ICU,intensive care unit; PLT, platelet; ALT, alanine transaminase; AST, aspartate transaminase. The Tukey’scalculated p-values as well as upper and lower 95% confidence levels for the Sev-R vs. Sev-D means’comparisons are given. Sev-R (mean ± SD)Sev-D (mean ± SD)Tukey 95% CIpWBCs (×103 /ml)10.6±4.013.9±8.0–1.47 to 8.060.17Platelets (×106 /ml)260±75.5213.7±115.8–116.7 to 24.20.19INR1.29±0.561.23±0.23–0.38 to 0.260.69CRP (mg/L)51.19±54.4103.77±86.3–2.35 to 107.50.06D-dimer (mg/ml)1.47±1.93.17±3.56–0.55 to 3.960.13IL-6 (pg/ml)314.1±527325.3±619–591 to 6140.97Ferritin922.6±5651078±578–281 to 5940.47Albumin (g/ml)31.47±7.9526.97±5.1–8.7 to –0.260.038Hemoglobin (g/dl)12.26±2.012.16±2.0–1.53 to –1.330.97ALT (U/L)33.64±24.1546.5±37.2–10.37 to 36.13 ALT, alanine transaminase; AST, aspartate transaminase; CRP, high-sensitivity C-reactive protein; ICU,intensive care unit; PLT, platelet; INR, international normalized ratio; WBC, white blood cell. The Tukey’scalculated p-values as well as upper and lower 95% confidence levels for the Sev-R vs. Sev-D means’comparisons are given. Neutrophils are a major source of reactive oxygen species It has been recently proposed that the high neutrophil-to-lymphocyte ratio (NLR) observed in critically ill COVID-19 patients may tip the redox homeostasis due to increased reactive oxygen species (ROS) production (Laforge et al., 2020). Our hypothesis implicates elevated oxidative stress as a major cause of HSA damage in severe COVID-19 patients. As a result, we started by following the dependence of clinical outcomes and mortality on ROS levels in blood cells. First, we used flow cytometry to assess percentages of neutrophils, lymphocytes, and platelets in all patients as described in Materials and methods (Figure 2A). Furthermore, we used the ROS-sensitive DCF dye to probe intracellular ROS levels in various cell populations in whole blood from all groups. Figure 2 shows that while lymphocyte counts decrease, a parallel dramatic increase in neutrophil counts (% total) was observable when going from Control (40.78±14.0, n=9) to Sev-R (64.0±20.0, n=10) to Sev-D (76.4± 6.8, n=11) groups (overall ANOVA p=3.9×10–5). Similar trend was clearly seen in the heat map depicting parameters for all patients analyzed by flow cytometry (Figure 2B). It is also clear from Figure 2B&C that changes in DCF-positive neutrophils follow similar trend observed for neutrophil counts. To confirm this relation, we compared neutrophil counts with DCF-positive neutrophil counts and found that the two parameters were strongly correlated (Pearson’s r=0.8, p=3×10–7; Figure 2C). Moreover, both parameters individually showed statistically significant increases in both of the studied COVID-19 groups when compared with the control group (Figure 2D&E). These results suggest that neutrophils are major sources of elevated oxidative stress in critically ill patients. Note that the observed trends in platelets, lymphocyte, neutrophils, and NLR are similar to reported values (Sun et al., 2020; Yang et al., 2020). Figure 2 Download asset Open asset Hematologic cellular counts and neutrophil-ROS levels reflect severity and mortality in COVID-19 patients. (A) Representative flow cytometric diagrams comparing morphologic, hematologic, and ROS levels in control (representative of n=9; upper row), Sev-R (representative of n=10; middle row), and Sev-D (representative of n=11; lower row) groups. (B) Heat diagram comparing lymphocyte, neutrophils, platelets, and DCF-positive neutrophil counts as the percentage of total cell counts in all of the studied subjects. Yellow areas are either group separators or missing data due to insufficient sample size or processing errors. (C) A diagram showing statistically positive correlation between neutrophil count and count of neutrophils stained positive for DCF dye in all groups (black dots denote controls; blue are Sev-R; and red represent Sev-D patients). (D) When neutrophil counts were compared for all groups, both Sev-R and Sev-D groups showed statistically significant neutrophilia relative to control groups. However, only a weak trend has been observed when comparing the two groups with COVID-19. (E) DCF staining revealed increased levels of ROS in Sev-R and Sev-D groups relative to control neutrophils. Sev-D showed a trend of increased ROS level relative to Sev-R group. Multiple comparisons were carried out using ANOVA followed by Tukey test and p values are given. ROS, reactive oxygen species. Figure 2—source data 1 Raw source data for Figure 2B-E. https://cdn.elifesciences.org/articles/69417/elife-69417-fig2-data1-v2.xlsx Download elife-69417-fig2-data1-v2.xlsx Hydrogen peroxide levels in plasma correlate with mortality Next, we reasoned that elevated oxidative stress in both groups with critical COVID-19 infection would be echoed in plasma levels of hydrogen peroxide. Hydrogen peroxide is the most stable ROS and is highly stable under prolonged storage at low temperatures. We used a highly specific catalase-based assay that we developed and verified in our laboratory to quantify [H2O2] in plasma samples of all groups. The assay relies on high-resolution detection and quantification of released oxygen due to hydrogen peroxide decompostion by catalase (Figure 3A&B). We constructed a calibration curve to confirm the catalase-mediated H2O2 to O2 stoichiometric conversion (Figure 3B). A linear relation was obtained with zero intercept and slope of 0.47±0.03 which closely matches the theoretically expected value of 0.5 (95% confidence interval [CI]: 0.37‒0.56, p=5.6×10–4, Pearson’s r=0.994). Indeed, we detected striking differences between groups even with relatively small sample sizes (Mean ± SD, Control, n=11: 2.95±0.77, Sev-R, n=16: 7.21±2.4, Sev-D, n=23: 9.67±2.0; overall ANOVA p=2.6×10–11; Figure 3C). The differences between groups have reached statistical significance (Sev-R vs. Control, 95% CI: 2.37‒6.13, p=4.9×10–6; Sev-D vs. Control, 95% CI: 4.95‒8.48, p=0.0; Sev-D vs. Sev-R, 95% CI: 0.90‒4.03, p=0.001). It appears from these results that a measure of oxidative stress, that is, [H2O2] in plasma, is doubled in survivors and tripled in deceased COVID-19 patients relative to controls’ plasma average levels. Figure 3 Download asset Open asset Hydrogen peroxide levels in plasma and neutrophils reflect mortality in COVID-19 patients. Catalase was used to specifically and quantitatively determine levels of hydrogen peroxide in identical plasma volumes collected from control (n=11), Sev-R (n=16), and Sev-D (n=23) groups. (A) Oxygen levels are monitored and recorded while 50 μl batches of plasma from control, Sev-R, and Sev-D subjects are sequentially infused into tightly air-controlled O2k chamber containing catalase (315 units/ml) in deoxygenated buffer. In addition to the initial rise due to residual oxygen in the added plasma samples, the decomposition of hydrogen peroxide in these samples produces oxygen quantitatively. (B) To verify the assay we measured the released oxygen upon adding an increasing volume of standard hydrogen peroxide solution in PBS buffer with 0.2, 0.8, 1.2, and 1.6 μM final concentrations; inset. Linear fitting of the plotted [O2] versus [H2O2] relation yielded a slope=0.47±0.03 (Pearson’s r=0.994, p=5.6×10–4), which is very close to the theoretically expected value of 0.5 as the catalase-mediated decomposition of one mole of H2O2 produces ½-mole O2. (C) Plasma contents of H2O2 in plasma significantly increased in the order Sev-D>Sev-R>Cont using ANOVA followed by Tukey test applied on n=11, 16, and 23 for control, survivors, and non-survivors, respectively. (D) Fluorescence imaging was used to assess levels of ROS in freshly isolated neutrophils using DCF (2,7-Dichlorodihydrofluorescein diacetate, green) staining in all groups. Hoechst binds strongly to adenine–thymine-rich regions in DNA thus mapping nuclei through emitting blue fluorescence. Merged DCF and Hoechst images are shown in the third column. Images were acquired using Cytation 5 Cell Imaging Multi-Mode Reader (Agilent) and analyzed using Gen5 Software package 3.08. Scale bar: 100 µm. Figure 3—source data 1 Raw polarographic data for released oxygen (A), calibration curve (B), and calculated plasma hydrogen peroxide levels in all groups (C). https://cdn.elifesciences.org/articles/69417/elife-69417-fig3-data1-v2.xlsx Download elife-69417-fig3-data1-v2.xlsx To confirm this finding, we performed DCF fluorescence imaging on freshly isolated neutrophils of representative group of individuals from each group (Figure 3D). We simultaneously stained neutrophils’ nuclei with Hoechst 33342 (blue stain) to follow nuclear morphologic changes and DNA diffusion in all groups. Although requiring more detailed studies, a closer look at the acquired images of Hoechst-stained neutrophils from a survivor patient showed significantly reduced average neutrophil size (control DNA area, 144.8±93.7 μm2 [133 cells analyzed] vs. Sev-R DNA area, 36.0±7.5 μm2 [241 cells analyzed], Welsh corrected two samples t-test, p=0) with tendencies toward more condensed, more segmented nuclei and frequent C-shaped chromatin. However, neutrophils from a non-survivor exhibited diffused chromatin and possessed in average, a 26% larger DNA area than normal cells and roughly five times that of Sev-R neutrophils (DNA area [97 cells analyzed], 182.7±171 μm2, p=0 vs. both control and Sev-R groups). Inspection of the DCF fluorescence images indicated that the non-survivor’s neutrophils contained larger populations of what appears to be toxic granules and cytoplasmic vacuoles that are highly ROS-positive. Analysis of mean DCF fluorescence intensities (MFI) per cell confirmed results obtained by flow cytometry and catalase assay reporting increased levels of ROS in the order control<< Sev-R<Sev-D (DCF MFI ± SD [×103]: Control [282 cells analyzed], 6.7±0.7; Sev-R [351 cells analyzed], 10.9±0.9; Sev-D [368 cells analyzed], 13.0±1.1, Welsh corrected two samples t-test, p=0 for all comparisons). EPR-determined biophysical parameters reflecting albumin conformational changes are consistent predictors of COVID-19 mortality It has been previously shown that non-survivor COVID-19 patients exhibit mild but consistent hypoalbuminemia relative to survivors (Violi et al., 2020). We started by assessing albumin levels in the studied cohort of subjects to confirm if they follow similar trends. We found that [albumin] in plasma decreased in the order Control>Sev-R>Sev-D (Mean ± SD, Control, n=8: 40.45±10.93 mg/ml, Sev-R, n=16: 31.47±7.95 mg/ml, Sev-D, n=23: 26.97±5.10 mg/ml; overall ANOVA p=2.4×10–4; Figure 4A). We detected statistically significant decrease in [albumin] in plasma of Sev-R and Sev-D groups (Sev-R vs. Control, 95% CI: –16.66 to –1.28, p=0.02; Sev-D vs. Control, 95% CI: –20.76 to –6.18, p=1.5×10–4; Sev-D vs. Sev-D, 95% CI: –10.28 to 1.28, p=0.15). The reference range of HSA concentration in serum is approximately 35–50 mg/ml, but we and other groups found that COVID-19 associated mortality correlates with lowered [albumin] <30 mg/ml (Violi et al., 2020). However, high prevalence of hypoalbuminemia in numerous disease states and the age/sex-dependent wide dynamic range of this protein concentration limits its diagnostic utility (Levitt and Levitt, 2016). As a result, we investigated biophysical parameters pertaining to HSA protein configuration in whole blood and plasma of all groups as reflectors of this critical protein functions. Figure 4 Download asset Open asset EPR spectroscopic analyses of HSA-fatty acid binding reveal strong dependence of binding on mortality in COVID-19 patients. (A) Albumin level in plasma of control (n=8), Sev-R (n=16), and Sev-D (n = 23) groups showed that both survivors and non-survivors COVID-19 patients exhibit statistically significant hypoalbuminemia. Comparisons between representative spectra showing changes in line shape that are related to mobility and microenvironmental statuses of HSA-bound 5-DSA (B) and 16-DSA (C) in whole blood of a control (black trace), a Sev-R (blue trace), and a Sev-D (red trace) patients. Calculated biophysical parameters including order parameter (D), rotational correlation time (E), and the ratio between strongly bound to weakly bound spin labels (S/W, F) as defined in Figure 1 and described in Materials and methods section. Statistical comparisons by ANOVA followed by Tukey tests were used for means’ comparisons and revealed remarkable decrease in the binding strengths and packing parameter of the local microenvironment surrounding the spin labels. All calculated parameters along with exact p values are given in the Supplementary file 1. Water accessibility into albumin/fatty acids binding pockets are followed by reacting with ascorbate, which reduces nitroxide radicals into the EPR silent hydroxylamines. Representative EPR signal decays of 16-DSA in whole blood of control (G) and Sev-D (H) are shown. Kinetic traces (n=3 per group) showing the reduction of 16-DSA (I) and 5-DSA (J) bound to HSA by sodium ascorbate in whole blood. Kinetic traces are shown as the percentage loss of the signal intensity of the middle peak (A/A0). All samples contained 0.26 mM spin label and 3 mM sodium ascorbate and measured at 37°C. Weaker and slower disappearance of the EPR signal suggests inaccessible space toward the nitroxide moiety of the spin label. EPR, electron paramagnetic resonance; HSA, human serum albumin. Figure 4—source data 1 Determined biochemical and biophysical EPR parameters used for statistical comparisons. https://cdn.elifesciences.org/articles/69417/elife-69417-fig4-data1-v2.xlsx Download elife-69417-fig4-data1-v2.xlsx Previous studies reported that allosteric changes in HSA may be utilized to reflect critical functional changes in albumin and explored diagnostic and prognostic values of these changes in cancer (Haeri et al., 2019; Kazmierczak et al., 2006). It has also been found that long-chain fatty acids binding alters the interactive binding of ligands in the two major drug binding sites of HSA (Yamasaki et al., 2017). We employed EPR spectroscopy to probe SLFAs’ binding statuses and protein configuration in all groups as detailed in the Materials and methods section above. Analysis of the 5-DSA and 16-DSA EPR spectra revealed remarkable changes in spectral features between control and COVID-19 groups (Figure 4B and Figure 5, 5-DSA; Figure 4, 16-DSA in whole blood). For example, the hyperfine coupling tensor element 2T‖ (defined in Figure 1B and used to calculate the protein packing order parameter S) is sensitive to microenvironmental effects (such as polarity, H-bonding, electrostatic interactions, etc.) on the spin probe that is localized in one of the HSA native fatty acids binding pockets. Also, the S/W ratio (see Figure 1C) corresponding to strongly bound/weakly bound populations of 16-DSA spin probe may reflect changes in protein folding that can alter protein-fatty acid interactions. Furthermore, the rotational correlation time τc which is a measure of the spin probe rotational mobility is also analyzed. Figure 5 Download asset Open asset COVID-19-associated impairment of HSA transport function. Transport function of HSA is assessed through the apparent kinetics of fatty acid uptake by following the rise in both strongly (A) and weakly (B) bound components immediately after mixing SLFA with blood from representative subjects (n=3 for each kinetic trace). These results demonstrate the hindered fatty acid uptake by HSA of critically ill patients relative to controls. (C–E) To investigate the dislodging function of HSA, increasing volumes of absolute ethanol were added to identical SLFA-plasma mixtures of all groups (n=3–5) and the EPR spectra were acquired (C) to follow weakly (D) and strongly (E) bound populations of SLFA. Redistributions of the fatty acid populations are noticeable through decreased S and increased W (includes signals of free fatty acid) peaks. This redistribution is remarkably pronounced in critically ill patients reflecting weaker association with, and easier release of fatty acids from HSA in those patients. HSA, human serum albumin; SLFA, spin-labeled fatty acid. Calculated EPR spectral parameters for all groups are listed in Supplementary file 1 including exact ANOVA and Tukey test p values along with the number of subjects analyzed. Both 5-DSA and 16-DSA were used to probe the degree of local interactions in sites where the SLFA is buried in the protein interior (high mobility, 16-DSA) and closer to the protein-aqueous interface (low mobility due to interactions with water molecules and polar amino acids, 5-DSA; Gantchev and Shopova, 1990). Indeed, 5-DSA reflected significantly greater S parameter values relative to 16-DSA both in plasma and in whole blood (Figure 4D). However, independent of the spin probe and both in plasma and whole blood samples, the order parameter has been consistently lower in Sev-R which was further decreased in Sev-D patients relative to the control group (Figure 4D). In whole blood, similar results that showed more statistically robust differences have been observed. Calculations of τc as described in Materials and methods showed rotational mobility of 16-DSA is significantly faster when bound with HSA from COVID-19 patients relative to that from control subjects in plasma or whole blood. Finally, similar trends have been observed for the S/W parameter which signifies contributions of the strongly and weakly bound components of 16-DSA in different fatty acids pockets. Taken together, these results indicate that COVID-19 pathology is associated with extensive structural changes in the HSA protein that imply the prevalence of malfunctional derivatives of this critical protein. Hampered water accessibility into HSA/fatty acid pockets in whole blood of COVID-19 patients We followed water accessibility toward deep pockets carrying the spin labels through kinetic analysis of the nitroxide radical EPR silencing by the water-soluble ascorbate anion (Pavićević et al., 2014; Figure 4(G–J)). Under matching experimental conditions, 16-DSA and 5-DSA/HSA signals in whole blood of control subjects decayed remarkably faster when compared with both Sev-R and Sev-D (n=3 per group, p<0.05). Weaker and slower disappearance of the EPR signal of COVID-19 patients by ascorbate suggests less accessible space toward the nitroxide moiety of the spin label within the protein. It is clear from these data that the HSA of COVID-19 patients is generally less water accessible relative to control ones. However, the core of the HSA of both COVID-19 groups was not significantly different in terms of water accessibility. Defective transport function of HSA in critically ill COVID-19 patients Restricted water/ascorbate accessibility in severe patients relative to controls (Figure 4I&J) along with the observed significant changes in the microenvironments surrounding SLFAs are generally viewed to implicate functional damage (Gantchev and Shopova, 1990; Pavićević et al., 2014; Junk et al., 2010; Muravsky et al., 2009; Rehfeld et al., 1978). In this paradigm, changes in the SLFA’s lineshape are taken to indirectly display functional changes in the solution shape of albumin due to spatial rearrangements of paramagnetic centers in EPR-