Increase in venous thromboembolism in SARS‐CoV‐2 infected lung tissue: proteome analysis of lung parenchyma, isolated endothelium, and thrombi

COVID‐19 pneumonia is characterized by an increased rate of deep venous thrombosis and pulmonary embolism. To better understand the pathophysiology behind thrombosis in COVID‐19, we performed proteomics analysis on SARS‐CoV‐2 infected lung tissue.


Introduction
The pandemic of COVID-19 was characterized by an excessive mortality rate.In 2020, the World Health Organization reported approximately 1.8 million confirmed deaths attributable to the COVID-19 pandemic, but estimated this number to be at least 3 million due to underreporting. 1 Influenza, a viral disease with similar transmission and pulmonary complaints, has lower annual deaths, presumed to be between 0.2 and 0.5 million. 2In fact, mortality of hospitalized patients was 10 times higher for COVID-19 compared to influenza. 3The high frequency of thrombotic complications seen in COVID-19 contributes significantly to the high mortality, especially when compared with influenza. 4,5In situ pulmonary arterial thrombosis as well as classic pulmonary thromboemboli-mostly originating in the leg veins-have been associated with SARS-CoV-2. 5,6An increase in complement activation, cytokine storms, and vascular permeability play a vital role in COVID-19 thrombosis, 7 although activation of these pathways is also present in influenza. 8,9The reason why this increase of the abovementioned pathways in thrombosis seems more activated in COVID-19 than in influenza is unclear.We studied protein composition through use of liquid chromatography-mass spectrometry (LC-MS) in thrombosis of COVID-19 postmortem SARS-CoV-2 infected lung tissue.We compared profiles with postmortem lung tissue samples from influenza cases.Sequentially, LC-MS was performed on isolated thrombi and endothelium of the same groups to improve the identification of proteins related to thrombosis.Lastly, analyses were performed on isolated thrombi and surrounding endothelium based on thrombus histomorphology (in situ thrombus or thromboembolus) independent of underlying disease to gain insight into -mechanisms behind both phenotypes of thrombosis.

T I S S U E A C Q U I S I T I O N A N D H I S T O L O G I C A L E X A M I N A T I O N
Postmortem lung tissue from 19 autopsies in the Erasmus University Medical Center was selected for LC-MS.These included eight COVID-19 autopsies and 11 consecutive influenza pneumonia autopsies (Table 1).The COVID-19 cases had been tested positive for SARS-CoV-2 by polymerase chain reaction (PCR), both by nasal swab up to 1 month before death as well as postmortem swabs of both main bronchi.Influenza cases were included if the autopsy report mentioned influenza as (contributory) cause of death.All influenza cases were tested positive for influenza type A or B through partial sequencing of reverse-transcriptase-PCR of lung tissue or nasopharyngeal swab; one influenza case had an increased titre of influenza antibodies.COVID-19 autopsies were from 2020, during the first wave of the alpha variant in the Netherlands.Influenza autopsies were carried out between 1993 and 2019.All autopsies were carried out in the Erasmus University Medical Center in Rotterdam.Consent was by the next of kin to utilize postmortem organ tissue from autopsies for research purposes.Permission for utilization of COVID-19 tissue samples was granted by the Medical Ethics Review Committee (MEC-2020-0322).

W H O L E -T I S S U E A N A L Y S I S
For whole-tissue analysis, formalin-fixed and paraffinembedded (FFPE) samples of lung tissue were selected from five available COVID-19 cases and all 11 influenza cases.FFPE sample selection was performed on corresponding haematoxylin and eosin (HE)-stained slides, of which one slide with the most extensive diffuse alveolar damage was selected for proteomics analysis.Each slide was then scored for the presence of histomorphological features indicative for diffuse alveolar damage (DAD) in a semiquantitative 3-point scoring system indicating absence (0), low, 1 medium, 2 or high 3 presence (Table 2).These features include hyalin membranes, haemorrhage, alveolar histiocytes, fibrosis, acute fibrinous, and organizing pneumonia (AFOP) and pneumocyte hyperplasia (Figure 1).Slides available per case for this selection varied between 2 to 20 slides.Then, for each selected FFPE lung tissue sample a consecutive slide following the HE-stained slide was cut and processed for LC-MS (see: Processing, below).Cases for whole-tissue analysis were grouped based on disease as either COVID-19 or influenza.

L A S E R C A P T U R E M I C R O D I S S E C T I O N
For the laser capture microdissection (LCM) analyses, all lung tissue slides from all cases were examined for the presence of thrombi.Samples from four COVID-19 cases and five influenza cases used for wholetissue analysis had (micro)thrombi and were eligible for LCM analysis.Samples from three other COVID-19 cases with (micro)thrombi were added for LCM analyses to increase the group size.Each case was labelled for disease and thrombus type.Protein     profiling analyses performed on isolated thrombi and endothelium were then stratified by disease (either "COVID-19" or "influenza") and thrombus type (either "in situ thrombus" or "thromboembolus").The distinction between in situ thrombi and thromboemboli was based on histomorphological appearance independent of underlying disease.Thrombi were considered to be in situ if the thrombus showed eccentric inward growth from the vessel wall towards the lumen.If the majority of the thrombus surface was not connected to the vessel wall, but consisted of a considerable amount of erythrocytes and/or showed layered formation of fibrin and erythrocytes (layers of Zahn), the thrombus was categorized as a thromboembolus (Figure 2).To meet the minimum requirement of material for LCM analysis, a minimum surface of 800,000 lm 2 of material was dissected from 10-lm-thick slides from both thrombi and endothelium, prepared on polyethylene naphthalate (PEN) membrane-covered microscopic slides (Carl Zeiss, Zwijndrecht, the Netherlands).For LCM analysis of endothelium, strips of vessel wall containing endothelium were dissected, extending up to 50 lm into the intima and media, as single endothelial cells could not be dissected because of the thickness of the laser cut.Thrombi and surrounding endothelium were captured from the same slide.For both thrombus and endothelium isolation, additional consecutive slides were cut for LCM if the material from one slide was deemed to be insufficient.

P R O C E S S I N G
Material from both whole-tissue and LCM analyses were analysed with LC-MS after deparaffinization, xylene removal, de-crosslinking, solubilizing, and Increase in venous thromboembolism in SARS-CoV-2 971 digestion.Samples were loaded onto a trap column (PepMap C18, 300 lm ID, 5 mm length, 5 lm particle size, 100 A pore size; ThermoFisher Scientific, Waltham, MA, USA), washed, and desalted for 8 min using 0.1% trifluoroacetic acid as the loading solvent at a flow rate of 20 ll/min.The trap column was switched in-line with the analytical column (PepMap C18, 75 lm ID 9 500 mm, 2 lm particle size, and 100 A pore size, ThermoFisher Scientific) and peptides were eluted with a 90-min acetonitrile gradient ranging from 3% to 30% (and formic acid concentration from 0.1% to 0.08%, respectively).All LC solvents were purchased from Biosolve (Valkenswaard, the Netherlands).The column flow rate was set to 250 nl/min, and eluting peptides were measured at 214 nm in a 3 nl nano flow cell (ThermoFisher Scientific), coupled online to the mass spectrometer.
For electrospray ionization nano ESI emitters (New Objective, Woburn, MA, USA) were used and a spray voltage of 1.7 kV was applied.For MS detection, a data-dependent acquisition method was used with a survey scan from 350 to 1650 Th at 120,000 resolution (AGC target 400,000) and consecutively isolated and fragmented by collisional induced dissociation (CID) at 35% normalized collision energy (AGC target 10,000) of the most abundant precursors in the linear ion trap until a duty cycle time of 3 s was reached ('Top Speed' method).Precursor masses that were selected once for MS/MS were excluded from further fragmentation for the next 60 s.Proteins from the LCM-derived samples were assigned by exporting features, for which MS/MS

W H O L E -T I S S U E A N A L Y S I S
Whole-tissue analysis identified a total of 4452 different proteins among all samples after removal of proteins or peptides with zero counts after normalization.On average, 2231 different proteins and peptides were identified per sample, ranging between 1794 and 2451 without statistical difference between COVID-19 and influenza cases.In total, 356 proteins were differentially expressed between COVID-19 and influenza samples (P < 0.050), of which 27 were upregulated in COVID-19 (Table 3).
A full list of identified proteins per analysis is displayed in Table S1.
Whole-tissue analysis shows significant upregulation of 37 identified pathways in COVID-19 compared to influenza, including upregulation of ephrin receptor signalling and histamine degradation.A significant portion of these pathways were related to metabolic processes taking place in the liver, among which the urea cycle pathway (Figures 3A and 4).Nearly half of the identified upregulated pathways (14/37) had involvement of aldehyde dehydrogenases.A full list of up-and downregulated pathways per analysis is displayed in Table S2.No significant upregulated pathways were identified in influenza compared to COVID-19.

L A S E R C A P T U R E M I C R O D I S S E C T I O N A N A L Y S I S : T H R O M B I
Protein profiling of isolated thrombi identified 2463 different proteins and peptides with on average 1026 per sample, ranging between 356 and 1421 different proteins and peptides.LCM analysis of thrombi based on underlying disease showed 127 differentially expressed proteins between COVID-19 and influenza thrombi.LCM analysis of thrombi based on histomorphology showed 54 differentially expressed proteins between in situ thrombi and thromboemboli (Table 3).Pathway analysis showed five upregulated pathways and 87 downregulated pathways in COVID-19 thrombi compared to influenza thrombi.Integrinand interleukin-8 signalling were among the most downregulated pathways, with respective z-scores of À3.05 and À3.00 (Figures 3B and 5).Other pathways related to inflammatory response (signalling pathways of: interleukin 3/8/17, Fc Epsilin RI.HMGB1, phospholipase C, STAT3, NF-kB) and platelet activity (signalling pathways of: TGF-beta, thrombin, paxilin, ephrin, PAK, and CXCR4) were also downregulated in COVID-19 thrombi.Lastly, five proteins related to the coronavirus replication pathway were downregulated in COVID-19 thrombi.These were: COPI coat complex subunits alpha and gamma 1 (COPA and COPG1) and tubulin proteins beta 1 class VI, beta 6 class V, and beta 2a class IIa (TUBB1, TUBB6 and TUBB2A) (Figures 3B and  6A).Analysis of proteins related to specific diseases and functions revealed a significant increase in functions associated with thrombocytopenia, bleeding time, and lung inflammation.A decrease in proteins associated with viral infection was found in COVID-19 thrombi.
Between in situ thrombi and thromboemboli, two pathways showed significant differential expression, Increase in venous thromboembolism in SARS-CoV-2 975 that being upregulation of EIF2 signalling and downregulation of the 'deactivation of coronavirus pathogenesis' pathway.This pathway consisted of nucleophosmin 1 (NPM1) and six ribosomal proteins (RPS3/8/14/18/20/21) (Figures 3C and 6B), which were all significantly upregulated in the in situ  thrombi group compared to the thromboemboli group (P-values varying between 0.007 and 0.029).However, these proteins showed no significant differential expression between influenza and COVID-19 thrombi, independent of thrombus type.In situ thrombi also showed a significant increase in functions related to viral infection.

L A S E R C A P T U R E M I C R O D I S S E C T I O N A N A L Y S I S : E N D O T H E L I U M
Protein profiling of isolated endothelium identified 2366 different proteins among all samples.On average, 1130 proteins and peptides were identified, ranging between 891 to 1484 per sample.LCM analysis of endothelium based on underlying disease showed 149 differentially expressed proteins between endothelium from COVID-19 and influenza cases.With LCM analysis of endothelium surrounding thrombi of different histomorphology, differentially 38 expressed proteins were identified between embolism and in situ cases.
Pathway analysis of endothelium stratified by disease identified differential expression of two pathways.The DHCR24 signalling pathway, involved in cholesterol synthesis, was upregulated in COVID-19 endothelium; the CSDE1 signalling pathway was downregulated.The activity of viral infection and replication was decreased in COVID-19 endothelium.There were no differentially expressed pathways regarding coagulation, thrombocytopenia, or haemostasis.Regarding analysis of protein expression between endothelium of vessels with either in situ thrombi or thromboemboli, no significant differentially expressed pathways were identified (Figure 3E).Similarly, no proteins related to specific diseases or functions showed significant differential expression.

Discussion
In this study we analysed proteins in postmortem lung tissue from severe COVID-19 and severe influenza patients, as well as isolated endothelium and Increase in venous thromboembolism in SARS-CoV-2 977 thrombi from pulmonary blood vessels through LC-MS.Although whole-tissue analysis predominantly showed upregulation of liver proteins, analysis of isolated COVID-19 thrombi showed downregulation of multiple platelet activity pathways that may indicate a relative increase in thromboembolism in COVID-19.Analysis of thrombi based on histomorphology showed upregulation of the coronavirus pathogenesis pathway in in situ thrombi, emphasizing the increase of in situ pulmonary thrombosis in COVID-19 as well.
Proteomics analysis of COVID-19 thrombi identified downregulation of various proteins related to coagulation compared to influenza thrombi (Intergrin, TGFbeta, thrombin, paxilin, ephrin, PAK, and CXCR4 signalling).The integrin signalling pathway was the most downregulated, which plays a vital role in platelet adhesion and aggregation at sites of endothelial damage. 10,11A decrease of integrin and other platelet-related proteins in COVID-19 thrombi compared to influenza thrombi may indicate a thrombus origin.Integrin signalling has been described to have a key role in thrombosis in arteries 12,13 ; downregulation of the integrin signalling pathway could thus mean a relative decrease of arterial in situ thrombosis.Additionally, the decrease of platelet transmembrane proteins integrin and ephrin in thrombi suggests an overall reduction of platelets in COVID-19 thrombi, as arterial thrombi contain more platelets compared to venous thrombi. 14,15These findings imply that, in general, thrombi in COVID-19 are more often of venous origin compared to other diseases with pulmonary thrombi.A meta-analysis of thrombosis in COVID-19 supports our findings, reporting a high incidence of venous thromboembolic events in COVID-19 compared to similar lung diseases. 16astly, ephrin signalling was upregulated in the whole-tissue analysis of COVID-19 lung tissue, suggesting that COVID-19 lung tissue taken as a whole has increased platelet activity.Thus, our data of these proteomics analyses strengthen the hypothesis that there is a relative increased risk of increased venous thromboembolism in COVID-19 compared to other diseases with pulmonary thrombi.
Multiple pathways related to inflammatory activity were also downregulated in thrombi of COVID-19 cases.8][19][20][21][22][23][24][25][26][27] We are not sure that measurement in serum can be directly compared to tissue.However, in general a reduction in platelet activity may occur in severe COVID-19.Recent studies found impaired integrin activity in platelet of COVID-19 patients versus healthy controls, with a more pronounced decline of integrin alpha IIb in severe COVID-19 patients. 28,291][32] A similar phenomenon regarding certain cytokine signalling pathways may also occur.Neutrophil exhaustion was seen in patients dying of COVID-19, which may explain the decrease of interleukin-8 signalling activity. 33,34hus, a state of leukocyte and platelet depletion after cytokine storms and a hypercoagulation phase cannot be excluded.

C O R O N A V I R U S P A T H W A Y S I N I S O L A T E D T H R O M B I
Analysis of thrombi based on histomorphology, in situ thrombus, or thromboembolus found significant downregulation of coronavirus pathogenesis pathway in the latter, which IPA was based on upregulation of the following seven proteins: NPM1 and ribosomal proteins RPS3/8/14/18/20/21.These ribosomal proteins are all part of ribosome subunit 40s, which monitors tRNA and mRNA to protein translation.Nonstructural proteins of SARS-CoV-2 (NSP1) binds to ribosome subunit 40, which inhibits its function.Thus, upregulation of these ribosomal proteins may imply intact function.However, several counterarguments can be made against the predicted downregulation of coronavirus pathogenesis.First, ribosomal proteins 3, 8, 14, 18, 20, and 21 were also found to be potential biomarkers for SARS-CoV-2 infection. 35- 38Second, NPM1 (nucleolar phosphoprotein B23) is involved in key steps of replication of RNA and DNA viruses. 39Therefore, upregulation of NPM1 and ribosomal proteins RPS3/8/14/18/20/21 instead suggests coronavirus pathogenesis activity in in situ thrombi.
Analysis of proteins in thrombi sorted by disease also revealed downregulation of the coronavirus replication pathway in COVID-19 thrombi compared to influenza thrombi based on the decrease of four proteins: COPI coat complex subunits alpha and gamma 1 (COPA and COPG1), as well as tubulin proteins (TUBB1, TUBB6, and TUBB2A).SARS-CoV-2 hijacks coatomer protein for replication, 40 while microtubules allow intracellular transport for SARS-CoV-2 to happen. 41,424][45] Thus, downregulation of these proteins in COVID-19 may be a result of more extensive viral load in the endothelium of pulmonary blood vessels.
The high expression of proteins and pathways related to the liver in COVID-19 tissue and isolated COVID-19 endothelium may be a result of multiple organ failure.Liver failure in hospitalized COVID-19 patients is seen more often compared to influenza patients. 46,47lthough liver failure was clinically present in both groups (two out of eight COVID-19 cases and 3/11 influenza cases), no significant differential expression of liver proteins was found between patients with or without liver failure.Second, aldehyde dehydrogenase is also highly expressed in fat 48 ; obesity was found to be a risk factor for COVID-19 and associated with increased severity of COVID-19, 49,50 which was not uncommon in COVID-19 patients included for this study: three out of eight COVID-19 cases had a BMI >30 compared to 1 out of 11 influenza patients.
Still, some of the upregulated pathways give insight into the pathophysiology of COVID-19 despite the presence liver proteins.A histamine degradation pathway was among the top three upregulated pathways in COVID-19 tissue, with seven corresponding differentially expressed proteins including (Figure 4) histamine N-methyltransferase (HNMT ).][53] Our results are in line with these findings and imply increased mast cell activity, which may contribute to pulmonary fibrosis and thrombosis in COVID-19. 54,55pregulation of arginase (ARG1) in COVID-19 patients may also contribute to organ failure and pulmonary fibrosis.0][61] Thus, the influx of arginase in lung tissue may also play a role in fibrosis of endstage COVID-19.

L I M I T A T I O N S
There were some limitations to this study.First, criteria for differentiating in situ thrombi and thromboemboli were based on expert opinion and not validated clinically.Differentiation between these phenotypes of thrombi based on these histomorphological thrombus were also hard to appreciate in microthrombi >100 lm.Additionally, since HE slides represent a three-dimensional structure by a twodimensional image, classification of thrombi based on histology are prone to sampling error.Thus, the results of this study should be interpreted with caution.Future studies should evaluate the clinical significance of this classification.Second, there was some selection bias.A small fraction of lung tissue was sampled and analysed with pathway analysis tools that were not all validated with orthogonal techniques.In addition, only lung tissue samples of severe COVID-19 and severe influenza in their lethal stage were used for this study.Inherent to the design of this study, performed analyses could only display protein expression at one instance in time.No correlation could be made between alterations in patient health and biomarker levels during admission.Data from COVID cases in this study would most likely be infected with the alpha variant of SARS-CoV-2.Since it is unknown if similar results would apply for other variants or milder phenotypes, the results of this study should be interpreted with caution.The difference in age between influenza and COVID-19 patients may have also influenced some difference in the identified proteins.Third, while laser microdissection was able to accurately isolate thrombi, inclusion of smooth muscle tissue and occasional intravascular erythrocytes for endothelium analysis was unavoidable.Lastly, influenza type from one case could not be retrieved from available patient data.

Conclusion
Deep venous thrombosis and pulmonary thromboembolism play a vital role in COVID-19 pathophysiology and clinical outcome.While proteomics profiling of SARS-CoV-2 infected lung tissue and isolated endothelium predominantly shows upregulation of liver metabolism pathways in severe COVID-19, which may play a role in pulmonary fibrosis, analysis of isolated thrombi show a decrease of platelet activation proteins and pathways in COVID-19 thrombi compared to influenza thrombi.This may suggest a relative increase in classic venous thromboembolism in Increase in venous thromboembolism in SARS-CoV-2 979 increase of venous thromboembolism without undercutting the involvement of in situ thrombosis in severe COVID-19 at a molecular level.

Figure 1 .
Figure 1.Examples of mild to severe histopathological characteristics of diffuse alveolar damage.These include hyaline membranes (A,B), alveolar histiocytes (C,D), acute fibrinous, and organizing pneumonia (AFOP) (E,F), haemorrhage (G,H), pneumocyte hyperplasia (I,J), and fibrosis (K,L).For each displayed feature, left examples portray the mild presence of said feature (score 1) and right examples portray the severe presence (score 3).The severity of each feature varied between COVID-19 and influenza cases; the displayed examples originated from both COVID-19 (A,B,C,E,F,I,L) and influenza cases (D,G,H,J,K).

Figure 2 .
Figure 2. Examples of pulmonary thrombi.In situ thrombi (A-C) show inward growth of thrombus from the vascular wall towards the lumen.Proliferation of fibroblasts (A,B) and vascular infiltration of lymphocytes (C) were seen in in situ thrombi.Thromboemboli (D-F) are virtually unconnected to the vascular wall.Thromboemboli consisted of a considerable amount of erythrocytes, occasionally forming layers of fibrin and erythrocytes; lines of Zahn (F).

Figure 3 .
Figure 3. Volcano plots showing differentially expressed proteins in whole-tissue analysis and LCM analysis.Each dot represents a unique protein or peptide.Green dots represent proteins and peptides with a P-value <0.05 and a fold change <À1 or >1 and were considered differentially expressed.Highlighted proteins are displayed with corresponding gene name.(A) Whole-tissue analysis of influenza versus COVID-19.Blue dots represent upregulated proteins participating in the urea cycle, magenta represents proteins involved with histamine degradation.(B) LCM analysis of influenza and COVID-19 thrombi.Blue dots represent proteins related to COVID-19 replication.Magenta dots represent proteins related to integrin signalling.The majority of these proteins were also related to thrombin, CXCR4, PAK, paxilin, and ephrin signalling pathways.(C) LCM analysis of thromboemboli and in situ thrombi.Blue dots represent proteins related to the coronavirus pathogenesis pathway, upregulated in in situ thrombi.(D) LCM analysis of endothelium between influenza versus COVID-19.Magenta dots represent upregulated proteins in COVID-19 involved with DHCR24 signalling; a key pathway of cholesterol synthesis.(E) LCM analysis of endothelium from cases with thromboemboli and in situ thrombi.No differentially expressed pathways were identified.

Figure 4 .
Figure 4. Whole-tissue analysis pathway examples.(A) Urea cycle pathway.Four proteins were found to be upregulated in COVID-19 lung tissue, represented by blue framed diamonds.Intensity of the orange colour represents predicted activation.Orange ovals represent predicted upregulated metabolites.(B) The histamine degradation pathway.Upregulated proteins related to this pathway are represented by magenta framed diamonds.Note the fading of orange in aldehyde dehydrogenase; three out of six subunits were not considered differentially expressed (P < 0.05 and/or fold change between À1 and 1).Blue and magenta diamonds represent dots of the same colour in Figure 3A.

Figure 5 .
Figure 5. LCM thrombi analysis: Integrin signalling pathway.This pathway was significantly downregulated in COVID-19 thrombi.Magenta framed icons depict proteins and peptides identified in influenza and COVID-19 thrombi (see Figure2B).Green icons portray identified downregulated proteins in COVID-19 thrombi; blue and orange icons portray respectively down-and upregulated proteins or functions.

Figure 6 .
Figure 6.Coronavirus replication pathway (A) and coronavirus pathogenesis pathway (B) identified by IPA.(A) Downregulation of coronavirus replication pathway is based on virus assembly, although involved proteins are not pathognomonic for SARS coronavirus.Blue framed icons represent dots of the same colour in Figure 3B.(B) Downregulation of SARS CoV pathogenesis is based on increased levels of multiple Ribosomal 40s subunits and NPM1 by SARS coronavirus, ultimately leading to ARDS.Magenta framed circles represent identified upregulated proteins in COVID-19 thrombi.
severe COVID-19 compared to other diseases with pulmonary thrombi.Based on histomorphology, in situ thrombi show upregulation of SARS-CoV-2 pathogenesis proteins compared to thromboemboli, which may indicate increased in situ pulmonary thrombosis in COVID-19.Therefore, this study supports the Ó 2024 The Authors.Histopathology published by John Wiley & Sons Ltd., Histopathology, 84, 967-982.

Table 1 .
Patient characteristics

Table 3 .
Identified proteins and pathways in performed analyses Proteins were considered differentially expressed if P < 0.050.Top five upregulated pathways were based on highest and lowest Z-scores per analysis.Pathways were considered differentially expressed if Z-scores were >2 or <À2.INF: influenza cases.COV, COVID-19 cases; EMB, thromboembolism cases; INS, in situ thrombi cases.a Down-and upregulated proteins or pathways per analysis are displayed in parentheses.