The present study aims to analyze the effect of COVID-19 severity and mortality on platelet functions in the most severe cases of COVID-19 in comparison with mild/moderate cases and control subjects. This is a prospective observational cohort study of patients with confirmed RT-PCR positive COVID-19. Nasopharyngeal swab RT-PCR results and Lung CT scans were combined to classify severe symptomatic COVID-19 cases. All patients were recruited from Kasr Alainy Cairo University Hospital-ICU-facility at the Internal Medicine Quarantine Hospital. Supportive therapy including supplemental oxygen and symptomatic treatment were administered as required. Hospitalized patients with moderate to severe hypoxia (defined as requiring fraction of inspired oxygen [FiO2] ≥40%) were transferred to the intensive care (ICU) for further management including invasive mechanical ventilation when necessary. ICU-patients recruited in the current study were divided into two arms based on future mortality outcome: Those who survived (ICU-S), and those who died within 20 days of samples collection (ICU-NS).

Patient cohort in this study is an expanded cohort for which we recently reported on the biophysically determined serum albumin damage as an accurate mortality predictor (24). With the initial aim being the assessment of metabolic alterations in platelets, no statistical method was employed to predetermine the number of subjects due to lack of previously reported similar studies on COVID-19 patients. Instead, sample size was based on sample availability and collection continued until a total of N=110 has been reached for COVID-19 patients during the period from October 4, 2020 to December 7 2021. Within the follow-up time 60 patients had died. All available samples were investigated by blinded operators and were all included in the final analysis.

Written informed consents were obtained from participants in accordance with the principles of the Declaration of Helsinki. Children’s Cancer Hospital’s Institutional Review Board (IRB) has evaluated the study design and protocol, IRB number 31-2020 issued on July 6, 2020 and renewed on July 28, 2021.

10 mL of fresh peripheral venous blood samples were collected from each subject on ACD tubes. The samples were divided into 2 portions; 2 mL of citrated whole blood that was incubated with RBCs lysis buffer containing NH₄Cl (ammonium chloride), NaHCO₃ (sodium bicarbonate), and EDTA (disodium) for 15 min (25) and used for flow cytometry measurements. To be used in metabolic and molecular studies, the remaining 8 mL blood were processed to isolate platelets by differential centrifugation as follows. Platelet-rich plasma (PRP) was obtained by centrifugation at 300× g for 15 min with decreased acceleration and no brakes. PRP was then centrifuged at 4000× g for 10 min. The platelet pellets were resuspended in 1 mL platelet poor plasma and the count was determined using hemocytometer (24, 26). In a few cases, blood samples were disposed because they were clotted either during transfer or during processing.

Lysed cell suspension was centrifuged at 500xg for 5 minutes, washed twice with phosphate-buffered saline (PBS) and finally, obtained lysed cells were suspended in PBS. Whole blood was assessed for the distribution status of platelets, neutrophils, monocytes and T-T-lymphocytes using 13-color flow cytometer CytoFLEX system (Beckman Coulter Life Sciences CytoFLEX benchtop flow cytometer). Obtained cell suspension was incubated for 30 min in the dark at room temperature with anti-human monoclonal antibodies for identification of cellular subsets: CD-42b-PE (Beckman Coulter Life Sciences, IM1417U) for platelets, CD14-PC7 (Beckman Coulter Life Sciences, A22331) for monocytes, CD66b-APC-Alexa Fluor 750 for neutrophils (Beckman Coulter Life Sciences, B08756), and CD3-ECD (Beckman Coulter Life Sciences, IM2705U) for T-T-lymphocytes. Following incubation, cells were washed with PBS to be suspended in 300 μL and analyzed by flow cytometry for gating platelet-specific CD42b-PE positive population, neutrophil-specific CD66b-APC-Alexa Fluor 750 positive population, monocytes-specific CD14-PE positive population, and T-T-lymphocytes specific CD3-ECD positive population. Data was acquired for 20,000 events and analyzed using CytExpert software to assess the percentage and mean fluorescence intensities (MFIs) of cellular subsets.

Distribution status of activated platelets was measured in whole blood samples by 13-color flow cytometry as described using CytoFLEX system (Beckman Coulter Life Sciences CytoFLEX benchtop flow cytometer). Suspended cells were incubated for 30 min in the dark at room temperature with combinations of anti-human monoclonal antibodies for subset identification as follows: CD-42b-PE (Beckman Coulter Life Sciences, IM1417U) for platelets, PAC1- AF647 (Biolegend, 362806) and CD62P-BV785 (Biolegend, 304942) for activated platelets. Following incubation, cells were washed with PBS to be suspended in 300 μL and analyzed by flow cytometry for gating platelet-specific CD42b-PE positive population and activated platelets PAC1 and CD62P positive population. A number of 20,000 events were acquired and analyzed using CytExpert software to determine the percentage and mean fluorescence intensities (MFIs) of analyzed cell subsets.

Whole blood samples were assessed for the generation of ROS by platelets using 2′,7′-dichlorofluorescein-diacetate (DCF, Sigma-Aldrich, D6883). Suspended cells were incubated with DCF (20 μM) and monoclonal antibody; CD-42b-PE (Platelets) for 30 min at room temperature in the dark. Cells were then washed with PBS and resuspended in 300 μL of PBS. A total of 20,000 events were recorded and analyzed using CytExpert program.

The transmembrane potential of different cell populations was measured in whole blood samples using Tetramethylrhodamine methyl ester perchlorate (TMRM, Sigma, T5428). Suspended cells were incubated with TMRM (1 μM) and combinations of monoclonal antibody; CD-42b-BV650 (Platelets) for 30 min at room temperature in the dark. Cells were then washed with PBS and resuspended in 300 μL of PBS. A total of 20,000 events were recorded and analyzed using CytExpert program.

Intracellular calcium of different cell populations was measured in whole blood samples using Fluo-4-AM, cell permeant (Fluo-4-AM, ThermoFisher scientific, F14201). Suspended cells were incubated with Fluo-4-AM (10 μM) in the presence of pluronic F-127 (0.02%) for 15 min followed by addition of combinations of monoclonal antibody; CD-42b-PE (Platelets) in HEPES buffer for another 15 min at room temperature in the dark. Cells were then washed with HEPES buffer and resuspended in 300 μL of HEPES buffer and incubated for 20 min. A total of 20,000 events were recorded and analyzed using CytExpert program.

Assessment of mitochondrial respiratory functions was carried out at 37 °C using high-resolution respirometry system Oxygraph-2K (Oroboros Instruments, Innsbruck, Austria) as previously described (26)). Prior to starting the experiment, oxygen calibration was done by permitting the respiration medium MIR05 to equilibrate with air in the oxygraph chambers till the detection of a stable signal. Platelets were then transferred to chambers and permeabilization was performed by addition of 50 µg/mL saponin. Substrate–uncoupler–inhibitor titration (SUIT) protocol was applied as follows: pyruvate (5 mM), malate (5 mM), (PM) adenosine diphosphate (ADP, 1 mM), glutamate (G, 5 mM), succinate (S, 10 mM), oligomycin (1 μg/mL) carbonyl cyanide-4-(trifluoromethoxy) phenylhydrazone (FCCP) (multiple 0.5 µM infusions), rotenone (R, 2 µM), Antimycin A (AmA, 1.25 µM), N,N,N’,N’-tetramethyl-p-phenylenediamine (0.5 mM)/Ascorbate (2 mM), (TMPD/Asc). The rates of oxygen consumptions were normalized to platelets count and estimated as the negative time derivative of oxygen concentration. Mitochondrial respiratory parameters were assessed as follow (1): Oxidative phosphorylation (OXPHOS) rate after adding saturating ADP in the presence of pyruvate, malate and/or glutamate (OXPHOS-I) or in the presence of succinate as a complex II substrate (OXPHOS I + II) (2). Electron transfer system (ETS) capacity maximum respiration rate: in the presence of FCCP (3). Complex IV activity after addition of TMPD/ASC. DatLabVR software version 7.4.0.4 (Oroboros Instruments, Innsbruck, Austria) was utilized for data acquisition and analysis.

Samples from the Oroboros O2k chambers were collected at the end of each experiment and stored in -80°C until measurement. Citrate synthase activity was assessed as previously described (27) using 300 μg of platelets protein. Since the irreversible chemical reaction CoA-SH + DTNB → TNB + CoA-S-S-TNB is catalyzed by citrate synthase, the amount of thionitrobenzoic acid (TNB) was quantified by measuring absorbance at 412 nm using Cytation5 Cell Imaging Reader (Agilent BioTek, CA, USA).

Platelets bioenergetics were measured through oxygen consumption rate (OCR) and proton production using XF analysis (XF96, Seahorse analyzer, Agilent) as previously described (28–30). In brief, For the mitochondrial stress test: 2 × 10⁷ platelets/well were plated on 96 well format XF plates in unbuffered Dulbecco’s Modified Eagle’s Media (DMEM; with 1 mM pyruvate, 5.5 mM D-glucose, 4 mM L-glutamine, pH 7.4 at 37°C). For the glycolysis stress test, the same seeding density was applied but in unbuffered Dulbecco’s Modified Eagle’s Media (DMEM; with 4 mM L-glutamine, pH 7.4 at 37°C). Plates was subsequently centrifuged (800g for 5 minutes) to sediment platelets’ monolayers in wells. Plates were allowed to settle and were incubated for 30-40 min at 37 °C in non- CO₂ incubator. OCR and the extracellular acidification rate (ECAR) were measured simultaneously for 30 min to establish a baseline measurement. The mitochondrial stress test was performed by measuring OCR during sequential injection of oligomycin (2.5 µM), carbonyl cyanide p-(trifluoro-methoxy) phenyl-hydrazone (FCCP) (2 µM) and rotenone: antimycin A mixture (2.5 µM). For the glycolysis stress test, ECAR measurements were analyzed during the following injections sequence: glucose (5.5 mM), oligomycin (2.5 µM), 2-deoxy-D-glucose (2-DG) (50mm). All measurements were normalized to seeding cell density. Proton leak was quantified as the rate of respiration in the presence of rotenone and antimycin A subtracted from the rate of respiration in the presence of oligomycin. ATP-linked respiration was the difference of rate in basal respiration and oligomycin. Non-mitochondrial oxygen consumption is the rate of respiration in the presence of rotenone. Glycolysis, glycolytic capacity, and glycolytic reserve were calculated by subtracting the average rates before and after the addition of glucose, ATP synthase inhibitor oligomycin and 2-DG.

10 mL fresh peripheral citrated blood samples were centrifuged at 800g for 5 minutes without brakes. The plasma top layer was removed leaving less than 1 mL on top of the buffy coat. Cold fixative containing 82 mM sodium monophosphate, 20 mM sodium hydroxide, 4% formaldehyde and 0.01% glutaraldehyde, pH 7.2, was added drop wise on top of the sample and left for 30 minutes. The coat was then transferred into the same fixative for short storage at 4°C. Fixed samples were processed for the transmission electron microscopy (TEM) as described previously (31). The ultrathin sections were post-stained in saturated uranyl acetate and lead citrate, and examined by JSM1400 plus-JEOL transmission electron microscope (Alexandria university microscopy unit). Fixed samples that were stored for scanning electron microscopy (SEM) were processed and coated using Sputter Coating Evaporator as described in (32, 33). JSM-5300 – JEOL scanning electron microscope (Alexandria university microscopy unit) was used to obtain SEM images.

Platelets were incubated with protein lysis buffer containing a proteinase and phosphatase inhibitor cocktail for 15 minutes to prepare protein lysates. 60 µg of lysed proteins were loaded, resolved by 15% SDSPAGE and transferred to PVDF membranes. Membranes were blocked with 5% nonfat dry milk for 1 h and then incubated with mouse Total OXPHOS Human WB Antibody Cocktail (1: 500) (Abcam, ab110411) at 4°C overnight. Membranes were then probed with anti-mouse HRP conjugated secondary antibody for 1 h (1:5000) and developed using the enhanced chemiluminescence (ECL) reagent (Pierce). Membranes were stripped for the detection of β-actin protein using anti- b-actin (1:1000). Bands were quantified by densitometric analyses using Image Lab Software version 6.1.0 (Bio-rad Laboratories, California, USA). The density of each band was normalized to β-actin band density.

qPCR was performed to assess complex I and complex IV gene expression in platelet samples isolated from healthy controls and COVID-19 infected patients. RNA was isolated using Trizol reagent (ThermoFisher, 15596026) followed by RNA purification using miRNeasy Mini Kit (Qiagen, 217004). RNA concentration and purity was measured using nanodrop and cDNA reverse transcription was carried out for total RNA using Quantitect Reverse Transcription kit (Qiagen, 205311). qPCR was carried out using primer-specific annealing temperature on cDNA templates. Primers spanning exon-exon junctions specific for complex I (NDUFA4), complex IV (MTCO2) and β-actin were used as listed in Table 1 . 2x Maxima™ SYBR™ Green Master Mix (Applied Biosystems, k0251) was used to perform qPCR on QuantStudio Real-Time PCR (QuantStudio 12K Flex Real-Time PCR System). Levels of RNA were normalized to β-actin levels and estimated as delta-delta threshold cycle (ΔΔCT). Each sample was measured 3 times and average was calculated.

Statistical analysis and data graphing were performed using OriginPro 2021 (OriginLab Corporation, Northampton, USA). Graphical representations of data utilized distribution and rug analysis which visualize prevalence of a given parameter among studied subjects (subject counts) while showing the actual scatter of data points over vertical-dash rugs on the x axis. Violin plots were also utilized to outline statistical analyses outcomes and to demonstrate data scatter, distribution, and mean ± SD or SEM as stated in figure legends. Exact p-values are given for each comparison on most graphs and in text. Continuous variables were expressed as mean and standard deviation or as median (IQR). Following tests for normality (Shapiro–Wilk test), variables that fulfilled the normality test were analyzed using the ANOVA test while non-normally distributed data was analyzed using the Kruskal–Wallis test. Both ANOVA and Kruskal–Wallis test were followed by Tukey post-hoc tests to compare the differences between the three groups. Categorical variables are reported as counts and percentages while continuous variables are expressed as mean ± SD. Differences between percentages were assessed by Pearson’s χ² tests or Fisher exact tests when the number of observations per group were less than 5. The χ² tests provided results that tested the hypothesis that the mortality and a given variable (e.g. sex or a comorbidity) are independent. When p is less than the significant level of 0.05, there is significant evidence of association between mortality and the variable.

Original datasets and detailed protocols are available upon request from the corresponding author sameh.ali@57357.org.

The demographic and clinical characteristics of the studied patients are shown in Table 2 . In the present study, participants were divided into three groups; Non- ICU patients (n=15), ICU- Survivors (ICU-S, n=35) and ICU-Non survivors (ICU-NS, n=60). No clinical or demographic characteristic showed statistically significant difference between the three groups except for the age (p =0.004), and blood saturation level (p=0.003). The number of patients who are suffering from diabetes and/or cardiovascular diseases are significantly higher in the ICU-NS group (p <0.05) than the other groups. Moreover, the number of patients who were treated with steroids, remdesivir, hydroxychloroquine and/or carbapenem antibiotics was significantly higher in the ICU-NS group than in the other 2 groups (p <0.05).

Table 3 shows laboratory results obtained for participants at the site of sample collection during hospitalization. Although when comparing all parameters in the three groups, we observed changes following similar reported results in previous studies, mean comparisons by Tukey test reported non-significant changes in all parameters except for a significant decrease in platelet counts (p <0.001), and albumin levels (p <0.001). Moreover, blood analyses showed significant increases in the white blood cell counts (p <0.001), neutrophil counts (p=0.015), C-reactive protein levels (p=0.02), ferritin levels (p=0.004),D-Dimer levels (p=0.001), and Creatinine levels (p=0.01) in critically-ill patients relative to mild/moderate patients.

COVID-19 pathology involves the impairment of the circulatory system including vascular occlusion, hypercoagulability, and reduced oxygen capacity. Platelet count, volume, and morphology are established risk factors for cardiovascular events. We explored if these parameters exhibit systematic changes in COVID-19 patients with variable severity and mortality outcomes. Platelet count was significantly lower in critically-ill patients when compared with mild/moderate (Non-ICU) group (Non-ICU vs. ICU-S, p=0.008; Non-ICU vs. ICU-NS, p=0.006) but those patients exhibited a weak trend of decreasing platelet populations relative to control group as determined by the proportion of CD42b-positive cells (% Total); Figures 1A, B (Mean ± SEM, control, n=31: 11.37 ± 1.68, non-ICU, n= 15: 15.22 ± 2.76, ICU-S, n=34: 7.63 ± 0.94, ICU-NS, n=56: 7.89 ± 0.87). This result is similar to those obtained through absolute platelet counts carried out at the site of sample collection ( Table 3 ). We then compared the mean platelet volume as estimated by the forward scatter analysis using flow cytometry (34). Figure 1C shows a distribution and rug plot comparing subject counts and FSC mean value distributions. In the lower part of the graph, the rugs reveal individual subjects’ distribution in terms of their platelet FSC-H parameter. It is clearly noticeable that the mean platelet size increases systematically with COVID-19 severity (Mean ± SEM, control, n=31: 3.60x10⁵ ± 0.20x10⁵, non-ICU, n=15: 4.02x10⁵ ± 0.19x10⁵, ICU-S, n=35: 4.30x10⁵ ± 0.12x10⁵, ICU-NS, n=57: 4.51x10⁵ ± 0.12x10⁵). The inset within Figure 1C is a violin plot showing the probability density of the FSC data for different groups as smoothed by a kernel density estimator. The violin plot includes all analyzed data with a marker for the mean FSC along with whiskers indicating the standard deviations. Our data analysis revealed a statistically significant increase in the mean FSC (platelet size) in ICU-hospitalized patients relative to control group (ICU-S vs. control, p=0.02; ICU-NS vs. control, p=7.5x10^-5).

We then used side scatter (SSC-H) measurement to provide information about, and compare the internal complexity (i.e. granularity) of platelets in all groups, Figure 1D . Platelets of non-survivors exhibit particularly increased granularity relative to control platelets (ICU-NS vs. control, p=1.4 x10^-5; Mean ± SEM, control, n=31: 4.40x10⁵ ± 0.33x10⁵, non-ICU, n=15: 4.77x10⁵ ± 0.33x10⁵, ICU-S, n=35: 5.32x10⁵ ± 0.20x10⁵, ICU-NS, n=57: 6.09x10⁵ ± 0.22x10⁵). These results enticed us to further analyze our flow cytometry data looking for populations of giant platelets. The presence of large platelets is usually associated with enhanced platelet turnover which is a hallmark of inflammatory response and correlates with an increased ischemic risk (35, 36). We compared the percentage of platelets exhibiting 2-5 times larger mean sizes as reflected in their FSC values ( Figure 1E ) and found that non-survivors’ platelets include significantly greater subset of giant platelets in Figure 1F (ICU-NS vs. control, p=0.03; Mean ± SEM, control, n=30: 8.13 ± 1.29, non-ICU, n= 15: 6.59 ± 1.03, ICU-S, n=35: 12.51 ± 1.54, ICU-NS, n=56: 14.02 ± 1.50).

To confirm the presence of these morphological changes we used transmission electron microscopy (TEM) to assess ultrastructural features of peripheral blood cells (buffy coats) from control and ICU-hospitalized survivors and non-survivors COVID-19 patients. Fields populated with platelets were imaged and representative images are given and analyzed in Figure 1G (Mean ± SEM, control, n of platelets analyzed=54: 4.39 ± 0.41, ICU-S, n=53: 4.59 ± 0.49, ICU-NS, n=34: 7.98 ± 1.35). As can be seen in this figure, giant platelets are more frequently encountered in the ICU-NS group (upper-right image and quantification panel, ICU-NS vs. control, p=0.0048, where n=3 subjects from each group). Moreover, platelets generally show a shift from a resting discoid shape in control subjects to an activated state with numerous pseudopodia in severe COVID-19 patients. In tune with the flow cytometry results in Figure 1D , enlarged TEM images in the lower row of Figure 1G indicate a remarkably increased granularity; i.e. platelet internal complexity especially in non-survivors relative to both control and survivor groups.

Metabolic characteristics dictate platelet function and activation (15, 16), but this has not been unambiguously demonstrated in the context of COVID-19 pathology. We asked whether severity and mortality of COVID-19 patients are associated with altered glucose metabolism including oxidative phosphorylation and glycolysis. Seahorse metabolic analysis of freshly isolated platelets from representative subsets from all groups revealed an overall COVID-19-associated depression in aerobic and anaerobic pathways, Figures 2A–C . That is, despite large variability of basal metabolic activities; i.e., without adding external substrates ICU-hospitalized COVID-19 patients exhibit lower glycolytic as well as mitochondrial respiratory activities ( Figure 2C ). This metabolic depression reached statistical significance during ATP-linked oxygen consumption even in non-ICU patients indicating defective mitochondrial bioenergetics at early onset of SARS-CoV-2 infection ( Figure 2D , Non-ICU vs. control, p=0.046; ICU-S vs. control, p=0.009; ICU-NS vs. control, p=0.004, where Mean ± SEM, control, n=8: 23.31 ± 4.08, non-ICU, n=11: 13.16 ± 3.05, ICU-S, n=8: 9.85 ± 1.23, ICU-NS, n=15: 10.56 ± 1.39). Relative to control platelets, glycolytic metabolism and glycolytic capacity were also lower in patients’ platelets ( Figure 2E , ICU-S vs. control, p=0.004; ICU-NS vs. control, p=0.001, where Mean ± SEM, control, n=12: 11.37 ± 1.37, non-ICU, n=11: 5.17 ± 1.56, ICU-S, n=15: 3.09 ± 0.8, ICU-NS, n=23: 4.92 ± 1.03) ( Figure 2F , Non-ICU vs. control, p=0.02; ICU-S vs. control, p=0.007; ICU-NS vs. control, p=0.008, where Mean ± SEM, control, n=12: 18.22 ± 2.41, non-ICU, n=11: 8.42 ± 1.58, ICU-S, n=15: 7.85 ± 1.74, ICU-NS, n=23: 8.88 ± 1.72).

A relevant aspect of platelets’ metabolic phenotype is their metabolic flexibility; i.e. the capacity to switch among energy substrates to generate ATP in response to changes in functional bioenergetic demands. For instance, transition of platelets to an activated state was described to involve a shift to a glycolytic phenotype (15). We analyzed the association of COVID-19 infection severity and mortality outcomes in terms of the ability of glucose-supplemented platelets to switch to glycolytic pathway under the inhibition of mitochondrial ATP-synthesis by oligomycin, Figures 2G, H . While platelets of normal control subjects were able to remarkably increase their ECAR with > 90% in response to around 60% oligomycin-induced mitochondrial inhibition, those from COVID-19 patients showed significantly diminished ability to compensate for the induced bioenergetic shortage following ATP synthase inhibition, Figure 2H (Non-ICU vs. control, p=5x10^-4; ICU-S vs. control, p=0.044; ICU-NS vs. control, p=0.013, where changes in Mean ± SEM (% of ECAR without oligomycin), control, n=7: 93.03 ± 14.4, non-ICU, n=6: 26.36 ± 6.9, ICU-S, n=7: 53.14 ± 11.36, ICU-NS, n=19: 53.98 ± 5.47).

Since mitochondrial ATP synthesis in activated platelets is consistently reported to regulate granule secretion [e.g (17).], we dissected mitochondrial respiratory function using high-resolution respirometry ( Figure 3 ). In tune with the Seahorse data ( Figure 2 ), mitochondrial respiration of ICU-hospitalized patients’ platelets was remarkably reduced whether at basal ( Figure 3B , ICU-S vs. control, p=7x10^-4; ICU-NS vs. control, p=1.5x10^-4, where Mean ± SEM, control, n=15: 15.31x10^-2 ± 2.54 x10^-2, non-ICU, n=4: 8.33 x10^-2 ± 0.68 x10^-2, ICU-S, n=17: 7.28 x10^-2±,0.74 x10^-2 ICU-NS, n=29: 7.32 x10^-2 ± 0.66 x10^-2) or upon Complex-I ( Figure 3C , ICU-S vs. control, p=8.7x10^-5; ICU-NS vs. control, p=1.7x10^-5, where Mean ± SEM, control, n=15: 18.98x10^-2 ± 3.37x10^-2, non-ICU, n=4: 9.28x10^-2 ± 1.25x10^-2, ICU-S, n=17: 7.19x10^-2 ± 0.8x10^-2, ICU-NS, n=29: 7.37x10^-2 ± 0.76x10^-2) or complex-II stimulations ( Figure 3D , ICU-S vs. control, p=8x10^-4; ICU-NS vs. control, p=2x10^-4, where Mean ± SEM, control, n=15: 18.53x10^-2 ± 3.38x10^-2, non-ICU, n=4: 10.44x10^-2 ± 1.9x10^-2, ICU-S, n=17: 7.39x10^-2 ± 0.8x10^-2, ICU-NS, n=29: 7.57x10^-2 ± 0.77x10^-2). However, complex-IV activity was independent of the disease state (not shown) hinting that the observed respiratory depression is not due to levels of expression of mitochondrial ETC complexes. This was further demonstrated as shown in supplementary Figure S1 through the comparisons of gene and protein expression profiles of mitochondrial ETC complexes in representative subsets (n=3-9) from all groups. We also observed a significant decrease in citrate synthase activity, a marker often used to reflect mitochondrial density, in ICU-hospitalized patients ( Figure 3E , ICU-S vs. control, p=4x10^-4; ICU-NS vs. control, p=0.0034, where Mean ± SEM, control, n=15: 2.37x10^-2 ± 0.5x10^-2, non-ICU, n=2: 1.26x10^-2 ± 0.13x10^-2, ICU-S, n=15: 0.29x10^-2 ± 0.12x10^-2, ICU-NS, n=21: 0.75x10^-2 ± 0.27x10^-2).

To explore if COVID-19-induced decline in mitochondria performance is related to qualitative mitochondrial morphological changes we analyzed TEM images of representative control (n=2, 96 mitochondria), ICU-S (n=1, 48 mitochondria), ICU-NS (n=3, 144 mitochondria); Figures 3F–H . Mitochondria from non-survivors exhibited significantly increased average cross-sectional area ( Figure 3G , ICU-NS vs. control, p=0, where Mean ± SEM, Control: 157.81 ± 7.28, ICU-S: 157.22 ± 5.91, ICU-NS: 221.67 ± 6.73) and circularity ( Figure 3H , ICU-NS vs. control, p=0.022, where Mean ± SEM, Control: 6.48x10^-2 ± 0.43x10^-2, ICU-S: 5.7x10^-2 ± 0.23x10^-2, ICU-NS: 7.77x10^-2 ± 0.31x10^-2) relative to controls’ and survivors’ mitochondria. These results suggest that non-survivors’ mitochondria are functionally and morphologically stressed with swollen and more-circular phenotypes.

Previous studies suggested that mitochondrial hyperpolarization represents a key event in platelet activation and remodeling (21) in addition to increased formation of cytosolic reactive oxygen species (37). Furthermore, numerous studies highlighted a pivotal role for intracellular calcium in exocytosis through the direct activation of synaptic proteins with defined calcium-binding domains and/or the activation of protein kinases controlling granule secretion in activated platelets [e.g (38, 39).]. However, these molecular parameters were not adequately addressed in the context of SARS-CoV-2 viral infection. We used flow cytometry to explore if the observed mitochondrial dysfunction in critically-ill COVID-19 patients affects platelet function through modulating intracellular levels of ROS and calcium. First, CD42b-positive platelet populations that were simultaneously positive for TMRM staining were quantified to qualitatively reflect ΔΨ_m, while intracellular ROS were assessed through DCF-positive cells, and intracellular [Ca²⁺] through Fluo4-positive staining as detailed in the methods’ section. Distribution of TMRM mean fluorescence intensity indicates hyperpolarized mitochondria in both ICU-S and ICU-NS in Figures 4A, B , ICU-S vs. control, p= 0.03; ICU-NS vs. control, p=0.0145. (Mean ± SEM, control, n=23: 0.42x10⁴ ± 0.05x10⁴, non-ICU, n= 12: 0.37x10⁴ ± 0.05x10⁴, ICU-S, n=16: 0.78x10⁴ ± 0.15x10⁴, ICU-NS, n=29: 0.76x10⁴ ± 0.08x10⁴). This was associated with increased ROS-positive platelets (% parent, Figure 4C , ICU-NS vs. control, p=0.001; Mean ± SEM, control, n=17: 53.48 ± 5.89, non-ICU, n= 13: 70.25 ± 4.82, ICU-S, n=16: 68.58 ± 4.80, ICU-NS, n=28: 76.22 ± 2.90). However, both counts of calcium-positive cells and levels of intracellular Ca²⁺ were significantly reduced in critically-ill patients’ platelets ( Figure 4D , ICU-NS vs. control, p=0.003; Mean ± SEM, control, n=23: 8.0 ± 1.44, non-ICU, n=13: 5.98 ± 2.16, ICU-S, n=16: 3.81 ± 1.07, ICU-NS, n=38: 2.99 ± 0.45) ( Figure 4E , ICU-S vs. control, p=0.036; ICU-NS vs. control, p=0.017; Mean ± SEM, control, n=23: 2.33x10⁴ ± 0.18x10⁴, non-ICU, n= 13: 1.66x10⁴ ± 0.26x10⁴, ICU-S, n=15: 1.66x10⁴ ± 0.12x10⁴, ICU-NS, n=37: 1.74x10⁴ ± 0.10x10⁴). We also analyzed giant platelet populations for the same parameters including transmembrane potential, ROS, and intracellular calcium. This analysis confirmed that giant platelets preserve profiles of these parameters similar to the general platelet populations. In Figures 4F, G we show that COVID-19 significantly reduces the number of Ca²⁺-positive giant platelets as well as diminished intracellular calcium levels especially in critically ill patients ( Figure 4F , non-ICU vs. control, p=0.02; ICU-S vs. control, p=0.003; ICU-NS vs. control, p=1x10^-4; Mean ± SEM, control, n=23: 58.90 ± 4.39, non-ICU, n=13: 39.59 ± 6.98, ICU-S, n=16: 36.88 ± 4.04, ICU-NS, n=38: 36.33 ± 2.54) ( Figure 4G , ICU-S vs. control, p=0.001; ICU-NS vs. control, p=1.8x10^-6; ICU-NS vs. non-ICU, p=0.042; Mean ± SEM, control, n=23: 2.10x10⁵ ± 0.16x10⁵, non-ICU, n=13: 1.80x10⁵ ± 0.14x10⁵, ICU-S, n=16: 1.46x10⁵ ± 0.11x10⁵, ICU-NS, n=38: 1.36x10⁵ ± 0.04x10⁵).

Our results so far highlight cellular characteristics suggestive of increased platelets’ activation in COVID-19 patients. To confirm this, we employed flow cytometry technique to analyze platelet activation in their natural environment; i.e. in whole blood immediately after collection. We explored the dependence of the level of expression of platelet activation marker CD62P (P-selectin) and the activation of integrin receptor GP IIb/IIIa (PAC1) on disease severity and mortality outcome ( Figures 5A–D ). Although both markers are frequently used in the literature to assess platelet activation, CD62P was reported to be a more reliable marker compared with PAC1 for measuring platelet activation (40). Indeed, while platelets of both ICU-S and ICU-NS groups showed a trend of increase in PAC1 expression ( Figure 5B ; Mean ± SEM, control, n=11: 0.50x10⁴ ± 0.03x10⁴, ICU-S, n=12: 0.57x10⁴ ± 0.04x10⁴, ICU-NS, n=13: 0.81x10⁴ ± 0.14x10⁴), they significantly exhibited an increased CD62P mean fluorescence intensity when compared with control subjects ( Figure 5C , ICU-S vs. control, p= 0.020; ICU-NS vs. control, p=0.025; Mean ± SEM, control, n=11: 0.34x10⁴ ± 0.04x10⁴, ICU-S, n=12: 0.62x10⁴ ± 0.07x10⁴, ICU-NS, n=13: 0.61x10⁴ ± 0.08x10⁴). Only platelets of non-survivors showed significant increase in MFI of double positive populations ( Figure 5D , ICU-NS vs. control, p=0.011; Mean ± SEM, control, n=11: 0.52x10⁴ ± 0.05x10⁴, ICU-S, n=12: 0.84x10⁴ ± 0.1x10⁴, ICU-NS, n=13: 0.98x10⁴ ± 0.13x10⁴). In tune with this observation, TEM images in Figures 1G and 5E demonstrate that platelets exhibit a shift from a resting discoid shape prevailing in control to an activated state with numerous pseudopodia in critically ill patients. Inspection of the same images hinted at increased levels of granules that are packed within platelets of COVID-19 patients, which, along with reduced intracellular calcium, are hallmarks of affected exocytosis (41, 42).

We therefore identified and counted α-granules, dense-granules, open canalicular structures (OCS), and lysosomes in magnified TEM images of randomly selected platelets from all groups (n= 3-4 from each group; n=31, 21, and 16 platelets for control, ICU-S, and ICU-NS; respectively); Figures 5F–H . To avoid ambiguity regarding granule assignments we analyzed total granule counts (α- + dense-granules) in TEM-acquired platelets cross-sectional images which showed significant increase in ICU-NS group relative to control group ( Figure 5F , ICU-NS vs. control, p=0.011; Mean ± SEM, control, n=31: 18.84 ± 1.56, ICU-S, n=21: 24.67 ± 3.48, ICU-NS, n=16: 30.69 ± 3.81). For the same analyzed platelets, we observed an increased trend in counts of OCS/platelet in critically-ill patients that reached significance only in the ICU-S group ( Figure 5G , ICU-S vs. control, p=0.012; Mean ± SEM, control, n=29: 4.79 ± 0.59, ICU-S, n=19: 10.95 ± 2.31, ICU-NS, n=16: 9.31 ± 1.95). Finally, platelets of the ICU-NS group showed significantly increased count of lysosomes per platelet relative to control and ICU-S groups ( Figure 5H , ICU-NS vs. control, p=3x10^-4; ICU-S vs. ICU-NS, p=0.03; Mean ± SEM, control, n=27: 1.56 ± 0.145, ICU-S, n=17: 2.24 ± 0.46, ICU-NS, n=14: 3.71 ± 0.59). These results are taken to suggest an impaired exocytosis in critically-ill COVID-19 patients especially those that didn’t survive.

Activated platelets may modulate inflammation via their receptor-dependent interactions with leukocytes and the subsequent exocytosis of bioactive mediators such as inflammatory cytokines. We asked if the above-described COVID-19-associated changes in molecular, mitochondrial, morphological, and activation profile exacerbate heterotypic platelet-leukocyte interactions especially in critically-ill patients. As previously reported, neutrophils-lymphocyte-ratio (NLR) and platelets-lymphocytes-ratio (PLR) showed systemic increase with severity and mortality (43); supplementary Figure S2 . Additionally, in spite of administering anticoagulants in nearly half of the studied hospitalized patients ( Table 2 ), flow cytometry analyses of whole blood samples freshly collected from ICU patients showed remarkable increase in the percentage of CD42b/CD66b double positive populations (platelet-neutrophil aggregates; Figures 6A, B , ICU-S vs. control, p=0.008; ICU-NS vs. control, p=3.5x10^-7; Mean ± SEM, control, n=31: 41.77 ± 3.92, non-ICU, n=15: 57.75 ± 4.97, ICU-S, n=35: 57.89 ± 3.54, ICU-NS, n=57: 67.48 ± 2.48). Ironically, heterotypic platelet-monocyte aggregation (CD42b/CD14 double positive populations) was actually significantly lower in non-survivors’ case not only relative to controls, but also to survivors’ group ( Figures 6C, D ); ICU-NS vs. control, p=0.012; ICU-NS vs. ICU-S, p=0.001; Mean ± SEM, control, n=31: 13.86 ± 1.90, non-ICU, n=15: 8.14 ± 1.11, ICU-S, n=35: 15.23 ± 2.72, ICU-NS, n=57: 6.43 ± 0.95). No tendency for altered platelet-lymphocyte aggregation (CD42b/CD3 double positive populations) was observed between any of the studied groups, Figure 6E . Indeed, transmission ( Figure 6F ) and scanning ( Figure 6G ) electron microscopy images of fixed buffy coats show the increased tendency of homo- and heterotypic aggregate formation especially in the ICU-NS group.