Introduction

Proteomic analysis of COVID-19 patient plasma has previously indicated that AZGP1, B2M, CRP, HP, HPR, ORM, RBP4, and some SAA may be repurposed as biomarkers of COVID-19 [1,2,3,4,5,6,7,8,9,10,11]. Infection with SARS-CoV-2 resulting in COVID-19 may have some symptoms similar to patients experiencing Acute Respiratory Distress Syndrome (ARDS) [12]. The SARS-CoV-2 spike protein may be cleaved by a protease presumed to be TMPRSS2 to gain entry to the cell through the ACE2 receptors [13]. Subsequently, the spike protein may act as a trigger of apoptosis via a mitochondrial pathway [14]. The release of mitochondrial cytochrome is known to trigger apoptosis of cells [15]. Viral infection resulted in major re-arrangements of cellular compartments including mitochondrial perinuclear clustering, association with the ER and Golgi apparatus, and the fission of mitochondria with the release of reactive oxygen species [16]. The mitochondria may play a central role in the physical interactions of SARS-CoV-2 during the viral replication cycle [17]. Expression of mitochondrial porin forms a channel for the release of macromolecules from the mitochondria organelle into the cytoplasm of COVID-19 patients and thereby acts as a cell death regulator [18]. Viral infection may permit the escape of large molecules like mitochondrial DNA and proteins into the cytosol [8, 19,20,21]. The mitochondrial NADH dehydrogenase complex (MT-ND) is the main source of reactive oxygen species like H2O2 in the cell [22]. The release of reactive oxygen species from the mitochondria apparently impacts the capacity for viral replication and may help regulate a pathway towards apoptosis in response to viral infection [23]. Genetic knockout, overexpression and radical scavengers indicate that the MT-ND complex and the production of reactive oxygen species in the mitochondria play a role in governing viral replication [16, 18, 24, 25]. The Tumor Necrosis Factor (TNF) response to viral infection triggers the mitochondrial activity that oxidizes luminol in the presence of reactive oxygen species [26, 27]. Infection by SARS-CoV-2 may lead to the loss of mitochondrial components into the extracellular space [28]. The cytochrome complex members including cytochromes like CYB, CYC, CYTB, cytochrome oxidase (MT-COX) and MT-ND from the mitochondria may play a role in H2O2 dependent luminol oxidase activity in vivo or in vitro [27, 29,30,31,32].

Plasma from COVID-19 patients was compared to ICU patients in Acute Respiratory Distress (ICU-ARDS) and Normal Human plasma EDTA using manual biophysical and biochemical sample preparation. The detection of human plasma proteins by precipitation, preparative quaternary amine chromatography, tryptic digestion and collection by micro C18 disposable resin (ZipTip) for immediate acid dilution and manual injection for nano spray liquid chromatography and tandem mass spectrometry of peptides (LC–ESI–MS/MS) is a laborious but effective means to attain sensitivity in blood plasma analysis [33, 34]. Here, the analysis of protein observation frequency of tryptic peptides by tandem mass spectrometry [35] alongside random and noise MS/MS spectra led to the discovery of new prognostic molecules specific to severe lung damage from COVID-19 infection and ICU-ARDS, such as mitochondrial cytochrome components. The experiments discovered components of the cytochrome electron transport system including CYCB, CYTB, CYC, CYP, MT-ND5 and MT-COX that were increased in the plasma of COVID-19 patients compared to normal controls. PCR analysis of mitochondrial MT-ND1 DNA, LC–ESI–MS/MS of plasma proteins and luminol oxidase assays were consistent to indicate the presence of macromolecule components associated with the cytochrome system in the plasma of COVID-19 patients.

Materials and methods

Materials

The Dionex UltiMate 3000 series UHPLC, C18 Acclaim PepMap NanoLC column (75 μm ID, 25 cm length C18), Fusion Lumos Q-Orbitrap-LTQ Tribrid MS (OIT) tandem mass spectrometer [36], LTQ XL linear ion trap (LIT) mass spectrometer [37] and Trizol reagent was from Thermo Fisher Scientific (Waltham, MA, USA). The 1100 HPLC was from Agilent (Santa Clara, CA, USA). Trypsin, salts, buffers, and luminol, H2O2, 4-iodophenylboronic acid (4IPBA) were obtained from Sigma Aldrich (St. Louis, MO, USA). The quaternary amine (QA) resin on a ceramic support was from BIORAD (Hercules, CA, USA). The HPLC grade water, ethanol, acetone and acetonitrile were obtained from Caledon Laboratories (Georgetown, Ontario, Canada). The 0.45 micron PVDF was from Millipore (Burlington, MA, USA). The imaging workstation was from BIORAD (Hercules, CA, USA).

Plasma sample collection

The presence of SARS-CoV-2 infection was confirmed by nasal PCR and serology assays at St. Michael’s Hospital, Toronto. EDTA plasma tubes were rapidly inverted 5 times before packing in ice [38]. The ice-cold plasma was then separated from blood cells at ≤ 3,000 RCF for 15 min at 4 °C prior to aliquoting to ≥ 1 ml prior to freezing at − 80 °C prior to analysis. Plasma from COVID-19 patients was compared to acute respiratory distress (ARDS) patients in the ICU and normal healthy volunteers under ethics protocol REB# 20-078. There were 16 COVID positive patients by nasal PCR and serology that were sampled three times, 2 individuals sampled twice and 1 individual sampled once, for a total of 19 individuals. A total of 16 ICU patients in acute respiratory distress were each sampled three times. Normal human plasma (NHP) contained 31 samples including 15 healthy normal volunteers from St. Michael’s hospital and a reference set of 16 normal human plasma from the International Biobank of Luxembourg of the Luxembourg Institute of Health (LIH) collected under a Comité National d’Ethique de Recherche (CNER) Protocol 201107/02 “Biospecimen Research” at the Centre Hospitalier de Luxembourg [39]. Five individuals with COVID and 4 normal plasma were analyzed by high-resolution OIT as an analytical reference. Subsequently, aliquots were thawed on ice, centrifuged at 12,000 RCF for 5 min at 4 °C to separate any cellular debris, and aliquoted to 25 µl samples on ice and refrozen for following analysis. Plasma samples of 25 μl were stored on ice and used for protein analysis.

Mitochondrial PCR

The MT DNA was extracted from human plasma using the phenol/guanidine isothiocyanate method with the commercial product Trizol. The ND1 forward primer was ND1F 5′-ACTACAACCCTTCGCTGACG-3′ and the reverse primer ND1R was 5′-GAAGAATAGGGCGAAGGGGC-3′ that together yield an expected product of 538 bp [19]. The PCR [40] conditions were: 40 cycles, lid temp 105 °C, 25 µL reaction volume, 94 °C melting (30 s), 58 °C annealing (1 min), 72 °C extension (1 min). The PCR products were separated by a 1.5% Agarose gel run at 100 V for 2 h and stained with GelRed by a minor modification of the method of Huang et al. [41].

Plasma precipitation

Plasma samples (25 µL) were precipitated with 9 volumes of acetonitrile 90% final (v/v) and centrifuged at 12,000 RCF for 15 min at room temperature. The acetonitrile was removed with a pipette and pelleted material dried under vacuum overnight. The proteins were then resuspended in 250 µL of 20 mM Tris pH 8.85 on ice with occasional vortex, brief centrifugation at 14,000 RCF, and the dissolved proteins were collected from insoluble components with a pipette. The resuspended proteins were assayed for protein content using the Dumbroff assays against BSA standards [42].

Quaternary amine (ammonium) chromatography

The precipitated, dried and re-dissolved plasma proteins were diluted in 250 µL of 20 mM Tris pH 8.85 and loaded on quaternary amine resin, washed with 5 column volumes of loading buffer and eluted in 200 µL of 300 mM NaCl with 20 mM Tris pH 8.85 [43].

Tryptic digestion

Tryptic digestion was performed in 600 mM urea and 5% ACN at 1/100 trypsin to protein overnight in 20 mM Tris pH 8.5. The samples were then reduced in 2 mM DTT for 20 min at 50 °C. The samples were digested again at 1/100 trypsin to protein for 2 h and quenched with 5% acetic acid.

LC–ESI–MS/MS

The plasma from COVID-19, ICU patients and the normal samples were analyzed in technical triplicate. Proteins from human blood fluid that were precipitated in acetonitrile (ACN) [44], re-dissolved 20 mM tris pH 8.85 and collected over preparative quaternary amine ion exchange resin [43] and digested to fully tryptic peptides, collected over preparative C18 ZipTip resin in 5% formic acid and were diluted in 18 µL of 5% formic acid immediately prior to injection via a 20 µL loop [45]. A total of ~ 5 µg of extracted and purified peptides was injected for each analytical HPLC separation over a 150 micron ID column (15 cm) with inline filter frits. The peptides were ionized by nano spray of the solvent gradient generated at 2 µL per minute split to a flow of ~ 200 nL per minute with a transfer capillary temperature of ≤ 250 °C into a Thermo Electron Corporation LTQ XL linear ion trap (LIT) mass spectrometer [37]. The peptides were randomly and independently sampled from 150 to 2000 m/z as the peptides eluted from the HPLC column into the nano electrospray source. A reference database was created using high-resolution trihybrid mass spectrometry where identical COVID-19 and NHP samples were analyzed by UPLC using Dionex UlitMate 3000 series over an Acclaim PepMap 100 C18 HPLC column (Thermo) C18, 2uM, ID: 0.075 mm × 250 mm for the orbital ion trap (OIT) [36, 46].

Peptide MS/MS spectra correlation analysis

The LC–ESI–MS/MS spectra and the results of the correlation algorithms were parsed into an SQL Server Database for analysis with the R system [33]. A physical filter of at least one thousand (E3) intensity counts for peptide parent ions was used to limit noise MS/MS spectra. The MS/MS spectra were fit to the peptides of the non-redundant human UNIPARC human proteins. The MS/MS spectra were fit by fully tryptic enzyme specification with a charge state of 2+or 3+ with up to three missed cleavages by SEQUEST [47], and X!TANDEM [48] using the default ion trap setting of fragments within 0.5 Da and within ± 3 m/z for the calculated peptide [M + H]+ [33, 34, 49,50,51,52]. The MS/MS spectra were fit to fully tryptic or phosphotryptic peptides by SEQUEST and to optional phospho/tryptic peptides by X!TANDEM. All the Gene Symbols presented showed p-values and FDR corrected q-values of < 0.01 from the Chi Square comparison of authentic observation frequency to that of the Monte Carlo simulation with computer generated random MS/MS spectra and random MS/MS from blank injection noise [34, 39, 50,51,52,53,54].

Computational analysis in SQL and statistical analysis with R

The LC–ESI–MS/MS results from samples and blank controls together with the results of the X!TANDEM and SEQUEST algorithms were collected and redundancy filtered out in SQL Server followed by statistical analysis performed with the R statistical system and biological connections represented using STRING algorithm [55]. The total number of MS/MS spectra from precursors greater than E3 intensity was used to normalize the observation frequency from the SEQUEST algorithm between the COVID-19, ICU-ARDS and NHP treatments for summation and Chi Square χ2 comparisons. The number of manually prepared samples and manual injection LC–ESI–MS/MS runs for each treatment were: Covid positive patients (COVID-19), 165 LC–ESI–MS/MS runs with a sum of 3,116,582 MS/MS spectra ≥ E3 intensity; Normal human plasma (NHP), 93 LC–ESI–MS/MS runs with a sum of 1,846,168 MS/MS spectra ≥ E3 intensity; and ICU acute respiratory distress syndrome (ARDS) control, 144 LC–ESI–MS/MS runs with a sum of 2,746,085 MS/MS spectra ≥ E3 intensity. The observation frequency for the ICU-ARDS and NHP treatments from the random and independent sampling of all plasma in triplicate by the LIT was corrected by the equations:

$$ {\text{ICU Observation frequency}} = {\text{ICU Count}}*\left( {\Sigma {\text{COVID MS/MS/}}\Sigma {\text{ICU MS/MS}}} \right) $$
(1)
$$ {\text{NHP Observation frequency}} = {\text{NHP count}}*\left( {\Sigma {\text{COVID MS/MS / }}\Sigma {\text{NHP MS/MS}}} \right) $$
(2)

Luminol oxidase activity

Cytochrome such as CYB, CYC, CYTB, CYP and the electron transport components such as the MT-COX and MT-ND proteins observed in the plasma have been previously shown to play a role in the reaction of luminol in the presence of H2O2 to yield a chemiluminescent signal [27, 29,30,31,32]. One µL sample of the plasma proteins, alongside known amounts of Cytochrome C preparation in the same buffer, was spotted directly onto PVDF using a pipette [56]. The total luminol oxidase enzyme activity was measured using ECL solution containing 4-iodophenylboronic acid (4IPBA) (100 mM Tris/HCl pH 8.85, 2.5 mM luminol, 0.4 mM 4IPBA, 2.6 mM Hydrogen peroxide) that generated strong specific signals and low background [57] on a BIORAD image analysis workstation.

Results

Mitochondrial PCR

The polymerase chain reaction (PCR) [40] showed mitochondrial DNA encoding the cytochrome electron transport complex component NADH dehydrogenase 1 (MT-ND1) was detectable in the plasma of most COVID-19 as well as ICU-ARDS patients. The presence of MT-ND1 DNA in the plasma was also detected in some normal human plasma (Fig. 1A). MT-ND1 DNA showed variation across individual COVID-19 and ICU-ARDS patients after PCR amplification, agarose electrophoresis and fluorescent staining (Fig. 1B) in agreement with previous results [19]. Thus macromolecule components from the mitochondria were clearly observed in the plasma of COVID-19 and ICU-ARDS patients as well as NHP controls.

Fig. 1
figure 1

Agarose gel electrophoresis and GelRed Staining of PCR products from mitochondrial DNA in human plasma with mitochondrial MT-ND1 specific primers. A PCR analysis of representative COVID-19, ICU-ARDS and NHP plasma samples; B the variation in COVID-19 and ICU-ARDS PCR amplification. The PCR conditions were 40 cycles, lid temp 105 °C, 25 µL reaction volume, 94 °C melting (30 s), 58 °C annealing (1 min), 72 °C extension (1 min). Template DNA was extracted from plasma with Trizol. Serum from the Luxembourg cohort was used as Normal Human Plasma (NHP) healthy controls

Full size image

Comparison of COVID-19 vs NHP and ICU-ARDS

The observation frequency of proteins from COVID-19 plasma was compared to those of ICU-ARDS and/or Normal Human Plasma (NHP) by One Way ANOVA, and the Chi Square test χ2, which revealed some proteins showed significant variation in the plasma across the disease and control treatments. All proteins reported herein were shown to have a low rate of type I false positive identification of p ≤ 0.01 versus the Monte Carlo simulation of random MS/MS from computer random numbers or random physical noise with FDR values q ≤ 0.01 [50, 52,53,54]. Analysis of COVID and normal samples by nano electrospray with a tri-hybrid orbital trap, a highly resolving mass spectrometer, showed good agreement on many of the proteins identified with the linear ion trap (LIT).

Plasma proteins

The majority of human protein gene symbols from serum proteins showed nearly identical observation frequency between COVID-19 versus normal controls (NHP) or severe respiratory distress ICU-ARDS. Plasma proteins including AAT, ABO, APP, FGA, HPX, ITIH4, PON and others showed similar or higher observation frequency in NHP compared to COVID-19 (Table 1).

Table 1 A comparison of corrected mean observation frequency per gene symbol for common serum proteins and acute phase response proteins in EDTA plasma across COVID-19, ICU-ARDS and Normal Human Plasma (NHP)
Full size table

Acute phase plasma proteins

A small minority of proteins that included acute phase markers increased in COVID-19 versus the controls based on counts of tryptic (TRYP) or phospho/tryptic (STYP) peptides with Chi-Square score of greater than 800 (χ2 ≥ 9, p ≤ 0.01) (Table 1). The observation frequency of the acute response serum proteins such as AZGP1, B2M, CRP, HP, HPR, ORM, RBP4, and some SAA was dramatically increased in COVID-19 compared to NHP plasma and showed agreement with previous studies [1,2,3,4,5,6,7,8,9,10,11]. In contrast AAT and transferrin (TF) showed a sharp decline in COVID-19. However, the observation frequency of CRP, AAT, FGA, S100, SAA1 and others was often greater in ICU-ARDs or COVID-19 compared to normals, therefore these proteins were not specific markers of COVID-19 infection but rather reflected lung damage (Table 1).

Apolipoproteins

Many apolipoproteins were dramatically reduced in COVID-19 compared to NHP or ICU-ARDS plasma. The apolipoproteins APOA1, APOA2 and APOA4 showed a 30% to 90% decline in COVID-19 and/or ICU-ARDS observation frequency compared to Normals (NHP). The deficiency in apolipoproteins levels in COVID-19 compared to normal human plasma (NHP) and ICU-ARDS patients was most pronounced for APOA1, APOA2, APOA4, APOC3 and APOE (Table 2). In contrast, APOBR was apparently increased in ICU-ARDS. Some proteins such as APOA1BP binding protein were constant over all treatments.

Table 2 A comparison of corrected mean observation frequency per gene symbol of apolipoproteins across COVID-19, ICU-ARDS and Normal Human Plasma (NHP)
Full size table

Mitochondrial and cytochrome proteins

Mitochondrial and cytochrome proteins were observed to be elevated in the plasma of COVID-19 patients versus the ICU-ARDS and NHP controls. The observation frequency of ICU-ARDS and NHP treatments was computed after correcting for the number of MS/MS spectra in each treatment with precursor intensity ≥ E3 detector counts to ensure a balanced comparison [58]. Mitochondrial proteins such as ATP5A1, CYB561D1, several CYP accessions, L2HGDH, two MRP, MRPL37, NDUFS1 and others showed increased observation frequency across COVID-19 and ICU-ARDS versus NHP individually by the Chi Square test (χ2 ≥ 10, p ≤ 0.01) and as a group by one way ANOVA (p ≤ 0.003) (Table 3). Chi Square analysis of corrected observation frequency indicated that cytochrome complex members including CYB, CYTB, CYP, CYBR, MT-COX, NDUF, MT-ND5 and other structurally or functionally related proteins were significantly elevated (χ2 ≥ 21, DF = 1, p ≤ 0.0001) in COVID-19 plasma (Table 4). The observation frequency of mitochondrial or cytochrome proteins from BFPS peptides typically showed a two fold to three fold increase that was significant by the Chi Square test (p ≤ 0.01). For example, MRPL37 showed an increased observation frequency from 216 tryptic peptides from SEQUEST in the NHP samples to 760 observations in COVID-19 that is an increase of more than threefold with a χ2 value of χ2 = 1364 where χ2 = 9 is the cut off for significance. The cytochrome CYP3A43 was observed in NHP 75 times from BFPS tryptic peptides by the SEQUEST algorithm but was observed 192 times in COVID-19 plasma that is a greater than two-fold increase with a highly significant Chi Square value (χ2 = 180) where a χ2 of ≥ 9 is significant.

Table 3 A comparison of corrected mean observation frequency of tryptic (TRYP) versus phosphotryptic (STYP) mitochondrial and cytochrome proteins per gene symbol from SEQUEST across COVID-19, ICU-ARDS and Normal Human Plasma (NHP)
Full size table
Table 4 Mitochondrial proteins identified from SEQUEST was compared by Chi-Square between COVID vs ICU and COVID vs NHP from the non-redundant sum of tryptic (TRYP) and optional phosphotryptic (STYP) peptides
Full size table

Mitochondrial protein interactions

The mitochondria may contain more than 1100 proteins [59]. However only a small subset of 131 mitochondrial proteins was observed to be sharply increased in the plasma of COVID-19 patients. STRING analysis indicated that there were many known functional or structural interactions between the mitochondrial molecules observed (Fig. 2). There were 652 previously established interactions between these 131 mitochondrial proteins indicating that these mitochondrial proteins may form structural or functional protein complexes [60].

Fig. 2
figure 2

The mitochondrial and cytochrome proteins specific to COVID-19 with a Chi Square value greater than 21 compared to ICU Acute Respiratory Distress Syndrome (ARDS) and/or Normal Human Plasma. The results reported in Table 4 with COVID-19 versus NHP and/or ICU-ARDS of χ2 ≥ 21 (p ≤ 0.001) were automatically analyzed by the STRING algorithm. Network Statistics: number of nodes, 131; number of edges, 652; average node degree, 9.95; avg. local clustering coefficient, 0.51; expected number of edges, 86; PPI enrichment p-value, < 1.0e-16

Full size image

Venn diagram of Chi Square analysis

A comparison of the proteins in COVID versus ICU-ARDS and normal controls from tryptic or optional phospho/tryptic peptides showed that CYTB, ND5, MRPL37 and ALDH2 were the most specifically elevated mitochondrial proteins in COVID-19 plasma. Organizing the results from the plasma discovery and Chi Square analysis into a Venn Diagram showed that some mitochondrial components specific to viral infection were up-regulated in COVID-19 versus both ICU-ARDS and/or NHP samples (Fig. 3). Central members of the cytochrome system including CYTB and ND5 as well as mitochondrial ribosome components (MRPL) and central metabolic enzymes such as Aldehyde dehydrogenase (ALDH2) and L-2-Hydroxyglutarate dehydrogenase (L2HGDH) were increased in COVID-19 compared to normal human plasma and/or ICU-ARDS patients.

Fig. 3
figure 3

The Venn diagram of the top Chi-Square (χ2 > 21, p ≤ 0.0001) results from COVID-19 versus ICU-ARDS and COVID-19 versus NHP at the level of full tryptic or optionally phospho/tryptic peptides shown in Table 4

Full size image

Luminol oxidase activity

Proteins of the cytochrome electron transport chain of the mitochondria including MT-ND, NDUF, complex IV (COX), CYBR, CYC, CYB and CYTB, but also CYP, have all been previously observed to participate or contribute to luminol or lucigenin oxidation in vivo and/or in vitro [15, 22, 27, 29,30,31,32, 61, 62]. Similarly other heme containing oxidases or peroxidases have also been implicated in the response to infection [63]. Cytochrome and cytochrome oxidase enzymes have been previously shown to react with luminol or lucigenin in the presence of H2O2 to yield chemiluminescence in vitro [27, 29,30,31,32]. No signal was observed in the absence of H2O2 from the luminol test solution. Testing plasma for H2O2 (Hydrogen peroxide) dependent cytochrome-like activity using luminol (Fig. 4A) showed a significant increase in the ECL signal intensity in COVID-19 and ICU compared to Normal plasma (Fig. 4B). There was apparently the equivalent of ng/µL (µg/mL) levels of cytochrome c-like activity in COVID-19 and ICU-ARDS plasma while NHP was near the lowest detection levels.

Fig. 4
figure 4

The presence of a luminol-oxidase activity in the plasma of COVID-19 and ICU Acute Respiratory Distress Syndrome patients (ICU-ARDS) compared to Normal Human Plasma (NHP). A One microlitre (1 µL) of a representative set of COVID-19 plasma was spotted onto PVDF alongside a specifically selected set of ICU-ARDS plasma (positive control) and two sets of representative NHP plasma (negative control) ECL dot blots. Cytochrome c from 0 to 100 ng/µL served as a detection standard. Arrows show the sample lanes; B Quantile boxplot of the intensity of luminol oxidase activity from the complete set of COVID-19, ICU-ARDS and NHP plasma samples in the study. The results from the two independent batches of normal that showed similar results were combined for graphic and statistical analysis. Significant difference by the Tukey Kramer Honestly Significant Difference (HSD) test at the p ≤ 0.05 level are shown by lower case letters

Full size image

Discussion

The aim of this study was to orchestrate standard biochemistry techniques followed by LC-ESI-MS/MS with classical computation and statistical methods for primary discovery of plasma proteins from COVID-19 and ICU-ARDS versus normal controls. Traditional protein precipitation and standard partition chromatography were followed by tryptic digestion and LC–ESI–MS/MS. The resulting peptide identifications were analyzed by the classical statistical practice of random and independent sampling for Chi Square comparison of corrected peptide observation frequency. The type I error rate of protein identification with respect to random expectation was estimated using the Monte Carlo controls of computer random and noise random MS/MS spectra. Herein, the orchestration of well-established laboratory approaches and classical statistical methods has revealed the mitochondrial components and heme-containing oxidases such as cytochromes released from the cells of severe COVID-19 or ICU-ARDS patients that can be directly measured by a rapid and simple enzyme assay.

Mitochondrial DNA PCR assay

The PCR amplification of mitochondrial ND1 was a sensitive method to detect mitochondrial components in COVID-19 or ICU-ARDS plasma and demonstrated that macromolecules from the mitochondria may be released from cells in agreement with previous results [19]. However, the PCR assay was so sensitive it detected mitochondrial DNA in many of the NHP control samples and so apparently did not show much discrimination in the detection of MT DNA.

Analysis of human COVID-19 plasma proteins

The COVID-19 plasma proteome was recorded using manual plasma precipitation, isolation of proteins by disposable chromatography for tryptic digestion, manual collection and injection of peptides for random sampling by tandem mass spectrometry alongside random MS/MS spectra controls. The laborious manual procedure enabled the detection and quantification of cellular proteins that were released into the plasma in COVID patients. The significant difference between authentic sample observation frequency versus computer random and noise MS/MS spectra, and the significant difference between treatments by ANOVA and Chi Square, were consistent with an apparently low type I error rate of protein discovery from the fit of MS/MS spectra. The results of this study showed good agreement with previously reported analysis of COVID-19 plasma regarding common acute phase markers such as AZGP1 ORM, CRP, SAA1, HP, ORM, HPR and B2M that were increased in COVID-19 patients [1,2,3,4,5,6,7,8,9,10,11], that confirmed the sampling scheme, instrumental analysis and computations were successful. The reduction in apolipoproteins in COVID-19 was consistent with ELISA assays [64]. The sensitivity to the spectrum of acute phase markers and apolipoproteins is a clear demonstration of the efficacy of random and independent sampling by LC–ESI–MS/MS with frequency-based analysis using classical statistical methods in R alongside the Monte Carlo statistical controls of computer random and noise MS/MS spectra.

Mitochondrial proteins in plasma

There were 652 previously established structural or functional interactions between the small subset of mitochondrial proteins observed in COVID-19 plasma that may indicate the mitochondrial proteins take the form of protein complexes in circulation [60]. The results of LC–ESI–MS/MS showed excellent agreement with the recent literature on the mitochondrial apoptosis pathway that is activated in response to viral infection [8, 14,15,16,17,18, 20, 21]. An increasing repertoire of cytochrome p450s (CYP) is now known to be targeted to the mitochondria, especially CYPs associated with sterol and steroid synthesis [65, 66]. The release of mitochondrial protein enzymes such as oxidase components MT-CO2 and CYB, CYTB or CYP into circulation were consistent with cellular damage from viral infection that might be quantified by enzyme activity to serve as a prognostic marker for the severity of COVID-19 infection [67]. Cytochrome enzymes such as MT-COX, CYB, CYTB, or CYP are known to react with luminol to provide an ECL signal in vitro [29,30,31, 62]. The increased presence of the cytochrome oxidase MT-COX2, cytochromes such as CYB, CYTB, or CYP or associated factors in the plasma of ARDS and COVID-19 patients as discovered by plasma proteomics was consistent with the measurements of increased cytochrome-like activity in the plasma of patients with respiratory distress [29,30,31,32].

Biology of SARS-CoV-2

Infection by the SARS-CoV-2 virus is known to result in the clustering of the mitochondria and the endoplasmic reticulum (ER) with the loss of normal compartments and integrity of the cells [8, 14,15,16,17,18,19,20,21, 68]. The SARS-CoV-2 spike protein may be cleaved by the protease TMPRSS2 to gain entry to the cell through the ACE2 receptors [13] (Fig. 5). Once inside the cell, the SARS-CoV-2 spike protein may act as a trigger of apoptosis via a mitochondrial pathway [14]. The increased expression of porins in the mitochondrial outer membrane in response to SARS-CoV-2 infection [8] provides a clear and established mechanism for the release of mitochondrial proteins from the organelle into the cytoplasm of cells infected with SARS-CoV-2 [18]. The porin VDAC may permit the loss of cytochrome enzymes from the mitochondria to the cytosol with eventual loss from the compromised cells [27, 29, 32, 61]. The release of cytochrome is a crucial part of the process that triggers cell death and apoptosis [15]. SARS-CoV-2 infection may lead to the fission of membrane-bound organelles and cellular lysis consistent with the escape of large molecules like mitochondrial components [8, 16, 20, 21, 28] including cytochromes and MT-COX proteins that might be detectable by luminol oxidase assays [29,30,31,32]. All of the data presented herein supports a mitochondrial pathway of apoptosis in SARS-CoV-2 infected cells that resulted in the release of mitochondrial proteins and DNA into the plasma that was similar to that observed in ICU-ARDS patients.

Fig. 5
figure 5

A cartoon summary of the mitochondrial proteins and enzyme activities discovered by LC–ESI–MS/MS from COVID-19 versus ICU and NHP sample populations. The SARS-CoV-2 spike protein is apparently cleaved by extracellular TMPRSS5 to gain entry via ACE2 and is unpacked in the cytosol releasing the spike and triggering expression of VDAC that releases the contents of the mitochondria including cytochrome, electron transport complex, ribosomal and others components such as MRPL, MT-ND, MT-COX, NDUF, CYB5RL and CYTB (see Table 4). Created with BioRender.com

Full size image

Conclusion

The SARS-CoV-2 virus shows clear structural and functional associations with the mitochondria during viral infection and replication [17]. Four independent lines of evidence including the PCR against the oxidase component ND1, the increased observation frequency of mitochondrial proteins in plasma by LC–ESI–MS/MS, the previously established structural and functional interactions between the increased proteins and the presence of a cytochrome-like ECL activity in COVID-19 plasma were all consistent with the loss of mitochondria and cytochrome components from cells in severe SARS-CoV-2 infection. Here it was demonstrated for the first time that the plasma from clinical populations may be analyzed in the protein discovery laboratory by LC–ESI–MS/MS to reveal new cellular proteins and the results translated into a rapid and simple biochemical test for an enzyme activity in COVID-19 and ICU-ARDS patients.