The COVID-19 pandemic, starting in the last months of 2019 in Wuhan (China) and caused by the RNA virus SARS-CoV-2, so far (January 4, 2022) has resulted in more than 293.0 million (Mio) infections and more than 5.46 Mio deaths (data source: https://www.worldometers.info/coronavirus/). The world has faced the fastest development and production of vaccines ever, resulting in a total of four vaccines that have received conditional market authorizations by the regulatory bodies: the mRNA vaccines from BioNTech/Pfizer (BTN162b2/Comirnaty; EMA: 21.12.20) and Moderna (mRNA-1273/Spikevax; EMA: 6.1.21), as well as the adenoviral vector-based vaccines from AstraZeneca (AZD1222/ChAdOx1-S/Vaxzevria; EMA: 29.1.21) and Janssen (Ad26.COV2.S; EMA: 11.3.21). All four vaccines encode slightly different forms of the SARS-CoV-2 Spike glycoprotein, which mediates virus binding to the host cell membrane and entry via ACE2 and TMPRSS2, respectively (Hoffmann et al., 2020).

While vaccination with BioNTech and Moderna mostly causes only mild and typical immediate vaccination side effects, severe side effects thromboses (ITP) or thrombosis with thrombocytopenia syndrome (TTS) were first observed with Vaxzevria in Europe (in March 2021). The most severe side effects entailed rare events of thrombocytopenia combined with cerebral venous sinus thrombosis (CVST). The higher than expected occurrence of CVST led to a rapid suspension of the Vaxzevria vaccination campaign in several European countries (March 11–14, 2021: Denmark, several northern European countries, Thailand, Ireland; March 15, 2021: Germany, Italy, France, and Spain).

The latest data from the security surveillance of the Paul-Ehrlich-Institute in Germany (report dated from December 23, 2021; https://www.pei.de/DE/newsroom/dossier/coronavirus/sicherheitsbericht-covid-19-impfstoffe-aktuell.html) revealed still a significant risk increase for the AstraZeneca vaccine compared to the other vaccines (see Table 1). Thrombocytopenia and immunothrombocytopenia (ITP) have been observed in 314 cases after 96.6 Mio vaccination doses with Comirnaty (3.3 cases per Mio injections), in 28 cases with Spikevax after 10.5 Mio vaccination doses with Comirnaty (2.6 cases per Mio injections), in 269 with Vaxzevria after 12.7 Mio vaccination doses cases (21.2 cases per Mio injections), and in 23 cases with Ad26.COV2.S after 3.4 Mio vaccination doses cases (6.6 cases per Mio injections). Thrombosis-related death cases for these four vaccines were 74, 6, 81, and 12, respectively. This indicates that thromboembolic events associated with case mortality were in the range of 21.4–52% (see Table 1).

TTS has been observed for these four vaccines in 36, 5, 200, and 24 patients, indicating a relative risk of 0.4, 0.5, 15.7, and 6.9 cases per Mio injections. These thromboembolic events in combination with thrombocytopenia share similarities with another condition, ‘heparin-induced thrombocytopenia’ (HIT). The term ‘vaccine-induced immune thrombotic thrombocytopenia’ (VITT) was coined for this vaccine-induced condition, after finding out that autoantibodies against the platelet factor 4 (PF4) could be causally linked to it (Greinacher et al., 2021b; Greinacher et al., 2021a). However, despite intensive research activities by many laboratories, the underlying initiating mechanisms for the observed side events were not yet fully understood.

At the molecular delivery mechanism of the approved vaccines, there are fundamental differences between mRNA- and adenoviral vector-based vaccines. The mRNA vaccines encode, as the naming implies, the Spike sequence as a modified RNA sequence encapsulated in lipid nanoparticles. Upon intramuscular injection, the nanoparticles are taken up by the cells, where the mRNA cargo is released into the cytosol. Since both mRNA vaccines encode a signal peptide, the process of protein translation occurs at the rough endoplasmatic reticulum (ER) and results in membrane-anchored Spike proteins. Further post-translational modifications occur in the ER and Golgi apparatus (glycosylation) from where the Spike protein trimer is transported in vesicles to the outer cell membrane.

In contrast, adenoviral vector-based vaccines deliver the Spike sequence as a codon-optimized DNA sequence instead of a RNA sequence. Vaxzevria is based on a genetically modified chimpanzee adenovirus type Y25, while the Janssen vaccine is based on a human adenovirus type 26 (HAdV-D26), both of which are replication incompetent. Upon transduction of the host cell and disassembly of the vector particle, the linear double-stranded DNA is released and enters the nucleus. Subsequently, viral DNA-encoded genes become transcribed using the host cell transcription machinery (Doerfler and Böhm, 2003). In the nucleus and before the mRNA is exported to the cytosol for translation (see above), the resulting RNA molecule is subject to the same post-transcriptional RNA processing machinery as endogenous primary transcripts, including the possibility of splicing.

SARS-CoV-2 is a single-stranded RNA virus of the betacoronavirus family that naturally has a purely cytoplasmic replication cycle. Therefore, coronaviruses have not been exposed to any evolutionary pressure that would have an impact on biological processes taking place in the nucleus. Almost all nuclear-encoded genes contain introns that become eliminated from precursor RNA transcripts by the splice machinery (Lee and Rio, 2015). Defined RNA consensus sequences are recognized by RNA/protein complexes (spliceosome), which consist of nearly 100 proteins and specific U-RNAs. Noteworthy, splice events are influenced by exonic splice enhancer sequences and exonic splice silencer sequences, which bind splice-activating or -suppressing proteins. During this evolutionary conserved splice process, a so-called ‘branch A’ nucleotide is performing a nucleophilic attack at a G•G-dinucleotide of the 5'-splice consensus sequence G•GUNNGU. The resulting 3'-OH of the final exonic G nucleotide then makes a second nucleophilic attack at the G•G-dinucleotide of the 3'-splice consensus sequence YYYYYCAG•G. To this end, two exons become fused, and the intronic sequence is being eliminated. For each gene, this process guarantees that all introns flanked by SD and SA sites are eliminated in order to produce the final mRNA that encodes a specific protein-coding sequence.

When using cloned cDNA in molecular biology or biotechnology, no intronic sequences are usually present, because cDNA synthesis is made from already spliced mRNA molecules. However, in case of the SARS-CoV-2 virus, the cloned cDNA for the Spike open reading frame was not coming from a nuclear-encoded gene, rather from an RNA virus that replicates solely in the cytosol of infected cells. Therefore, such a cDNA exhibits potential SD and SA sites just by chance (~130 cryptic splice signal sequences in the total SARS-CoV-2 virus sequence), because no negative selection has ever taken place during the evolution of coronavirus to eliminate such sequences. Therefore, we hypothesized that nuclear splicing of DNA-encoded Spike RNA sequences might result in spliced, and thus, truncated Spike protein variants with unknown fate and function. As mentioned above, this type of splicing would be quite artificial, because an RNA virus does per se not contain any intronic sequences in its various open reading frames. Thus, all such splicing events are derived from the so-called cryptic splice sites. This is unrelated to the mechanism of alternative splicing, which could be functionally attributed to certain genes of our genome. Cryptic splicing occurs only if strong splice sites are present, and thus, these cryptic splice sites will be used equally in different cell types or tissues.

To address this question of cryptic splicing, we analyzed nuclearly transcribed wildtype and codon-optimized Spike genes in a specially designed splice reporter vector system. We also investigated splicing of Spike RNAs following transduction of representative cells with a Spike-expressing human adenovirus type 5 vector and the two approved adenoviral vector-based vaccines ChAdOx1-S and Ad26.COV2.S. Here, we show that splicing of DNA-encoded SARS-CoV-2 Spike RNA molecules takes place at positions as they were predicted by in silico analysis. These splice events resulted in Spike protein variants that lack the transmembrane (TM) anchor, thus may become secreted due to the nature of the secretory pathway. Codon optimization performed to enhance expression levels further augmented the abundance of splicing and was most prominently observed for the ChAdOx1-S vector. In addition, we analyzed the production of Spike protein and its secretion to the outside medium. Implications of our findings for the safety of vector-based vaccines and future consequences for the design of vector-based vaccines for RNA viruses will be discussed.

Here, we provide experimental evidence that adenovirus-based vaccines designed to express the SARS-CoV-2 Spike protein from a cloned, codon-optimized cDNA may result in unwanted splicing events, if the presence of functional splicing sites in the coding sequence is not avoided during vector design.

SARS-CoV-2 is an RNA virus with an infectious cycle that takes place entirely in the cytoplasm. Thus, there has never been an evolutionary selection against presence of SD and SA sites in its gene sequences. While mRNA-based vaccines that are functionally only active in the cytoplasm, transcripts from adenoviral vectors are generated in the cell nucleus, where also post-transcriptional RNA processing steps – including splicing – is taking place. Thus, the findings shown and discussed here have implications for DNA-encoded vaccine development not only against SARS-COV-2 but also against other RNA viruses.

In this study, we experimentally validated our hypothesis that Spike protein variants lacking the natural membrane anchor – thereby being subjected to secretion – are generated from adenoviral vaccine vectors as a consequence of nuclear RNA splicing events. We showed (1) that in a Spike/splice reporter system use of in silico predicted SD sites within the wildtype Spike gene resulted in Spike-Luciferase protein fusion events, (2) that in this system splice reactions lead to secreted protein variants due to the splicing-dependent loss of the TM membrane domain, (3) that two of the three evaluated adenoviral vectors that encode a Spike gene that is codon-modified for improved expression exhibited more and stronger SD sites than the viral wildtype sequence, and (4) that following transcriptional read-through, splicing occurred to a strong SA in the adenoviral pIX transcript, going also along with loss of the TM anchor of Spike.

We wish to point out that in our experimental studies we have investigated a distinct spectrum of splice events, since we used primers binding to either Luciferase or the adenoviral pIX-coding sequence for PCR amplification. Alternative splice events, for example, internal splice events (in-frame, out-of-frame) were mostly cloned and sequenced by chance. The analysis of provided RNA-Seq data after cells were infected with the AstraZeneca vaccine gave insight into additional trans-splicing events with host cell genes. Viral RNA genes evolutionary contain splice site consensus sequences due to a missing counter selection. Thus, we conclude that nuclear expression of such RNA virus-derived genes in mammalian cells may result in unwanted splicing events with host cell genes that should be considered when developing new vaccines in the future.

Our results clearly indicate that Spike splice reactions occur and that secreted soluble Spike protein variants are generated. Western blot analyses from total cells and supernatant thereof revealed a series of Spike protein variants that vary in their molecular weight, indicative for the previously analyzed splicing events. Soluble Spike protein has been described to cause adverse effects, for example, a strong inflammatory response on endothelial cells (Lei et al., 2020; Nuovo et al., 2021; Patra et al., 2020; Amraei and Rahimi, 2020). Obviously, severe life-threatening thromboembolic events occur due to infection of recipient cells by wildtype SARS-CoV-2. Even pseudoviruses with Spike protein located at their surface may cause strong inflammatory reactions in tissues and endothelial cells, indicating a potential risk of soluble Spike protein to cause severe side effects when systemically present in the vascular system (Lei et al., 2021).

According to our results, soluble Spike variants could potentially explain the occurrence of rare but severe events observed in particular after vaccination with Vaxzevria. Noteworthy, the Ad26.COV2.S vaccine carries fewer SD sequences, especially SD₅₀₆ and SD₃₆₁₄ are absent (see Table 2B–D), which are the strongest predicted SD sites in the Ad5.S- and ChAdOx1-S-encoded Spike sequences (see Figure 1). This was further confirmed by the observation that in contrast to Ad5.S and ChAdOx1-S, no significant amounts of spliced RNA molecules upon cell transduction with Ad26.COV2.S were generated. This may explain the ~3-fold lower incidence for thromboembolic side effects with the Janssen vaccine when compared to the Vaxzevria vaccine (see Table 1). It may also explain the ~2-fold lower risk for TTS with the Janssen vaccine.

Based on our data we pose the hypothesis that soluble Spike via its receptor-binding domain will bind to ACE2 and anti-Spike antibodies generated during immunization will recognize Spike variants, bound to the endothelial cell surface. Inflammatory reactions could locally occur through different mechanisms including antibody-dependent cell-mediated cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC), both of which could serve as starting point for thrombus formation. In ADCC, mainly NK cells are recruited to surface-bound antibodies via Fc-gamma receptors (CD16 or CD32) and cause cell lysis after their activation. In CDC, anti-Spike antibody-coated endothelial cells may recruit and activate complement that may result in platelet activation via binding of C3 and terminal complement complex on the surface of platelets and thrombin/serotonin secretion (Luo et al., 2020). By this means, thromboses may occur in any site of the human body where endothelial cells express ACE2.

Why should such events occur in the cerebral dural venous sinus system? It is conceivable that due to the absence of venous valves bi-directional and posture-dependent blood flow in the dural venous sinus, the residence time of soluble Spike protein is increased and the possibility to bind to endothelial cells expressing ACE2 enhanced. This could increase the probability of adverse immune reactions at places of slow blood flow.

Recently the VITT syndrome has been described, suggesting a complex inflammatory process. Related to HIT it has been reported that the induction of anti-PF4 autoantibodies and platelet activation in a subset of patients may result in the observed combination of thrombocytopenia and thromboses mainly in large vessels including cerebral venous sinus (Figure 11, left side). Additionally, it is possible that protein impurities from the vaccine production process may be involved in the generation of a pro-inflammatory status (Krutzke et al., 2021). Here, we propose a third pathomechanism as it is depicted in Figure 1, right side. Based on our splicing data, membrane-anchored and secreted soluble Spike protein variants are produced after vaccination. Secreted Spike protein variants are disseminated throughout the vascular system and may concentrate at endothelial cells expressing the ACE2 at their surface. These ACE2-bound Spike will then become a target of anti-Spike antibodies, generated after vaccination with the potential to result in ADCC- and/or CDC-mediated inflammatory reactions. It may also cause the already described NET reaction of neutrophils that has been causally linked to thromboembolic events (Schönrich et al., 2020; Veras et al., 2020; Middleton et al., 2020). Also the rarely observed capillary leak syndrome (Clarkson disease; Kawabe et al., 2002) could potentially be explained by such a mechanism.

Figure 11

Download asset Open asset

The distinguishing features of the three different adenovirus vectors Ad5.S, ChAdOx1-S, and Ad26.COV2.S have been summarized in Table 3. Based on the analysis of the design of the vaccines and their splice behavior of the encoded transcripts as described in this report, we have identified several aspects that should be considered for future adenovirus vaccine design and development.

We have two different data sets. One data set has been created by large-scale sequencing of splice events of all three adenoviral vector systems after transduction in HeLa and HepG2 cell lines (Figures 5—7), where the different splice events of all three vectors have been investigated by simply analyzing a large number of PCR fragments through direct sequencing or subcloning and sequencing. The second data set is based on experiments, where all three adenoviral vectors were analyzed for their splice activity after adjusting the MOIs in the HeLa, HepG2, and C2C12 cell lines (Figure 8). In the latter, read-through transcripts have been investigated for the activity of the different poly(A) signal sequences (BGH vs. SV40), but also for the amount of splice events in that particular transcript (Figure 9). Transcriptional read-through obviously is a precondition for the generation of splice events between the Spike-coding sequence and the pIX gene encoded by the adenoviral vector genome. Very little transcriptional read-through was observed with ChAdOx1-S that can be linked to the BGH poly(A) signal used in this construct. Apparently, the SV40 poly(A) signal is not as efficient in termination of the transcript as we observed with Ad5.S. In these cases, differences were observed for splice events which may point to the fact that tissue-specific splicing may account as well as for some differences in the creation of soluble Spike protein variants.

Based on our findings, we suggest that the Spike open reading frames – wildtype or codon-optimized for expression – in DNA-based vaccines should always be analyzed for the presence of strong SD sites. If possible, strong SD should be avoided to prevent unintended splice. We note that the Janssen Ad26.COV2.S vaccine has been modified to reduce the number of strong SDs, while the Ad5.S and ChAdOx1-S vectors still contain strong SDs, which is reflected in a large number of detectable splicing events, while only very few splicing events were detected in the Ad26.COV2.S vaccine.

Very little is known about fate and tissue distribution of adenovirus-based vaccine in humans. In mice, it has been observed that following intramuscular injection significant amounts of adenovirus vectors end up in hepatocytes, where expression of the transgene occurs (Kron et al., 2011). It is likely that the chimpanzee ChAdOx1-S is also a hepatotropic adenovirus, since it is based on the chimpanzee adenovirus Y25 (Dicks et al., 2012), which belonged to a group of isolates from chimpanzees associated with outbreaks of viral hepatitis (Hillis and Goodman, 1969). If transduction of hepatocytes would occur in (some of the) vaccines after intramuscular injection in a significant amount, it is conceivable that larger amounts of Spike proteins, having lost the transmembrane domain by splicing, will be secreted into the systemic circulation via the hepatic veins.

Taken together, from our work it appears that for DNA-encoded sequences (especially when derived from RNA viruses), most importantly, strong SDs should be avoided and strong poly(A) signals should be used for RNA termination in expression constructs. This is especially important when adenoviral vectors are used as transport vehicles. Usually, transgenes sequences are embedded in the adenoviral genome in the gap of the deleted E1-region, located directly upstream of the adenoviral pIX with a strong SA site. Alternatively, while most transgenes in adenoviral vectors have a ‘left-to-right’ orientation, also an inversion of the expression cassette would avoid splicing events to occur between the transgenic-coding sequence and the viral pIX transcript. Taking this into account, chances of splicing will be strongly reduced or null.

We strongly emphasize the consideration of these aspects for the rational design of safer future adenoviral vaccine vectors and also for adenoviral vector-based gene transfer approaches in general.

The original WUHAN SARS-CoV-2 sequence is available in the NCBI database (NCBI Reference Sequence: NC_045512.2); the adenoviral and codon-optimized Spike sequence data have a protected intellectual property by the companies. The primary sequence of Ad5.S, designed and used by the colleagues in Ulm, can be retrieved upon request (contact Prof. Stefan Kochanek).

1. 1. Wu F
   2. Zhao S
   3. Yu B
   4. Chen YM
   5. Wang W
   6. Song ZG
   7. Hu Y
   8. Tao ZW
   9. Tian JH
   10. Pei YY
   11. Yuan ML
   12. Zhang YL
   13. Dai FH
   14. Liu Y
   15. Wang QM
   16. Zheng JJ
   17. Xu L
   18. Holmes EC
   19. Zhang YZ

   (2020) NCBI GenBank

   ID NC_045512.2. A new coronavirus associated with human respiratory disease in China.

   https://www.ncbi.nlm.nih.gov/nuccore/1798174254

1. 1. Almuqrin A
   2. Davidson AD
   3. Williamson MK
   4. Lewis PA
   5. Heesom KJ
   6. Morris S
   7. Gilbert SC
   8. Matthews DA

   (2021) SARS-CoV-2 vaccine ChAdOx1 nCoV-19 infection of human cell lines reveals low levels of viral backbone gene transcription alongside very high levels of SARS-CoV-2 S glycoprotein gene transcription

   Genome Medicine 13:43.

   https://doi.org/10.1186/s13073-021-00859-1

   • PubMed
   • Google Scholar
2. 1. Amraei R
   2. Rahimi N

   (2020) COVID-19, Renin-Angiotensin System and Endothelial Dysfunction

   Cells 9:E1652.

   https://doi.org/10.3390/cells9071652

   • PubMed
   • Google Scholar
3. 1. Dicks MDJ
   2. Spencer AJ
   3. Edwards NJ
   4. Wadell G
   5. Bojang K
   6. Gilbert SC
   7. Hill AVS
   8. Cottingham MG

   (2012) A novel chimpanzee adenovirus vector with low human seroprevalence: improved systems for vector derivation and comparative immunogenicity

   PLOS ONE 7:e40385.

   https://doi.org/10.1371/journal.pone.0040385

   • PubMed
   • Google Scholar
4. Book

   1. Doerfler W
   2. Böhm P

   (2003) Adenoviruses: Model and Vectors in Virus-Host Interactions

   Berlin, Heidelberg: Springer-Verlag.

   https://doi.org/10.1007/978-3-662-05597-7

   • Google Scholar
5. 1. Greinacher A
   2. Selleng K
   3. Mayerle J
   4. Palankar R
   5. Wesche J
   6. Reiche S
   7. Aebischer A
   8. Warkentin TE
   9. Muenchhoff M
   10. Hellmuth JC
   11. Keppler OT
   12. Duerschmied D
   13. Lother A
   14. Rieg S
   15. Gawaz MP
   16. Mueller KAL
   17. Scheer CS
   18. Napp M
   19. Hahnenkamp K
   20. Lucchese G
   21. Vogelgesang A
   22. Flöel A
   23. Lovreglio P
   24. Stufano A
   25. Marschalek R
   26. Thiele T
   27. Immune-Response in COVID-19 Vaccination Study Group

   (2021a) Anti-platelet factor 4 antibodies causing VITT do not cross-react with SARS-CoV-2 spike protein

   Blood 138:1269–1277.

   https://doi.org/10.1182/blood.2021012938

   • PubMed
   • Google Scholar
6. 1. Greinacher A
   2. Thiele T
   3. Warkentin TE
   4. Weisser K
   5. Kyrle P
   6. Eichinger S

   (2021b) A Prothrombotic Thrombocytopenia disorder resembling Heparin-induced Thrombocytpenia following Coronavirus-19 vaccination

   Researchsquare 10:362354.

   https://doi.org/10.21203/rs.3.rs-/v2

   • Google Scholar
7. 1. Hillis WD
   2. Goodman R

   (1969)

   Serologic classification of chimpanzee adenoviruses by hemagglutination and hemagglutination inhibition

   Journal of Immunology (Baltimore, Md 103:1089–1095.

   • PubMed
   • Google Scholar
8. 1. Hoffmann M
   2. Kleine-Weber H
   3. Schroeder S
   4. Krüger N
   5. Herrler T
   6. Erichsen S
   7. Schiergens TS
   8. Herrler G
   9. Wu NH
   10. Nitsche A
   11. Müller MA
   12. Drosten C
   13. Pöhlmann S

   (2020) SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

   Cell 181:271–280.

   https://doi.org/10.1016/j.cell.2020.02.052

   • PubMed
   • Google Scholar
9. 1. Kawabe S
   2. Saeki T
   3. Yamazaki H
   4. Nagai M
   5. Aoyagi R
   6. Miyamura S

   (2002) Systemic capillary leak syndrome

   Internal Medicine (Tokyo, Japan) 41:211–215.

   https://doi.org/10.2169/internalmedicine.41.211

   • PubMed
   • Google Scholar
10. 1. Kowarz E
    2. Merkens J
    3. Karas M
    4. Dingermann T
    5. Marschalek R

    (2011)

    Premature transcript termination, trans-splicing and DNA repair: a vicious path to cancer

    American Journal of Blood Research 1:1–12.

    • PubMed
    • Google Scholar
11. 1. Kowarz E
    2. Löscher D
    3. Marschalek R

    (2015) Optimized Sleeping Beauty transposons rapidly generate stable transgenic cell lines

    Biotechnology Journal 10:647–653.

    https://doi.org/10.1002/biot.201400821

    • PubMed
    • Google Scholar
12. 1. Kron MW
    2. Espenlaub S
    3. Engler T
    4. Schirmbeck R
    5. Kochanek S
    6. Kreppel F

    (2011) miRNA-mediated silencing in hepatocytes can increase adaptive immune responses to adenovirus vector-delivered transgenic antigens

    Molecular Therapy 19:1547–1557.

    https://doi.org/10.1038/mt.2011.83

    • PubMed
    • Google Scholar
13. 1. Krutzke L
    2. Allmendinger E
    3. Hirt K
    4. Kochanek S

    (2020) Chorioallantoic Membrane Tumor Model for Evaluating Oncolytic Viruses

    Human Gene Therapy 31:1100–1113.

    https://doi.org/10.1089/hum.2020.045

    • PubMed
    • Google Scholar
14. 1. Krutzke L
    2. Roesler R
    3. Wiese S
    4. Kochanek S

    (2021) Process-related impurities in the ChAdOx1 nCov-19 vaccine

    Researchsquare 10:477964.

    https://doi.org/10.21203/rs.3.rs-477964/v1

    • Google Scholar
15. 1. Lee Y
    2. Rio DC

    (2015) Mechanisms and Regulation of Alternative Pre-mRNA Splicing

    Annual Review of Biochemistry 84:291–323.

    https://doi.org/10.1146/annurev-biochem-060614-034316

    • PubMed
    • Google Scholar
16. Preprint

    1. Lei Y
    2. Zhang J
    3. Schiavon CR
    4. He M
    5. Chen L
    6. Shen H
    7. Zhang Y
    8. Yin Q
    9. Cho Y
    10. Andrade L
    11. Shadel GS
    12. Hepokoski M
    13. Lei T
    14. Wang H
    15. Zhang J
    16. Yuan JXJ
    17. Malhotra A
    18. Manor U
    19. Wang S
    20. Yuan ZY
    21. Shyy JYJ

    (2020) SARS-CoV-2 Spike Protein Impairs Endothelial Function via Downregulation of ACE2

    bioRxiv.

    https://doi.org/10.1101/2020.12.04.409144

    • Google Scholar
17. 1. Lei Y
    2. Zhang J
    3. Schiavon CR
    4. He M
    5. Chen L
    6. Shen H
    7. Zhang Y
    8. Yin Q
    9. Cho Y
    10. Andrade L
    11. Shadel GS
    12. Hepokoski M
    13. Lei T
    14. Wang H
    15. Zhang J
    16. Yuan JX-J
    17. Malhotra A
    18. Manor U
    19. Wang S
    20. Yuan Z-Y
    21. Shyy JY-J

    (2021) SARS-CoV-2 Spike Protein Impairs Endothelial Function via Downregulation of ACE 2

    Circulation Research 128:1323–1326.

    https://doi.org/10.1161/CIRCRESAHA.121.318902

    • PubMed
    • Google Scholar
18. 1. Luo S
    2. Hu D
    3. Wang M
    4. Zipfel PF
    5. Hu Y

    (2020) Complement in Hemolysis- and Thrombosis- Related Diseases

    Frontiers in Immunology 11:1212.

    https://doi.org/10.3389/fimmu.2020.01212

    • PubMed
    • Google Scholar
19. 1. Middleton EA
    2. He XY
    3. Denorme F
    4. Campbell RA
    5. Ng D
    6. Salvatore SP
    7. Mostyka M
    8. Baxter-Stoltzfus A
    9. Borczuk AC
    10. Loda M
    11. Cody MJ
    12. Manne BK
    13. Portier I
    14. Harris ES
    15. Petrey AC
    16. Beswick EJ
    17. Caulin AF
    18. Iovino A
    19. Abegglen LM
    20. Weyrich AS
    21. Rondina MT
    22. Egeblad M
    23. Schiffman JD
    24. Yost CC

    (2020) Neutrophil extracellular traps contribute to immunothrombosis in COVID-19 acute respiratory distress syndrome

    Blood 136:1169–1179.

    https://doi.org/10.1182/blood.2020007008

    • PubMed
    • Google Scholar
20. 1. Mittereder N
    2. March KL
    3. Trapnell BC

    (1996) Evaluation of the concentration and bioactivity of adenovirus vectors for gene therapy

    Journal of Virology 70:7498–7509.

    https://doi.org/10.1128/JVI.70.11.7498-7509.1996

    • PubMed
    • Google Scholar
21. 1. Nuovo GJ
    2. Magro C
    3. Shaffer T
    4. Awad H
    5. Suster D
    6. Mikhail S
    7. He B
    8. Michaille JJ
    9. Liechty B
    10. Tili E

    (2021) Endothelial cell damage is the central part of COVID-19 and a mouse model induced by injection of the S1 subunit of the spike protein

    Annals of Diagnostic Pathology 51:151682.

    https://doi.org/10.1016/j.anndiagpath.2020.151682

    • PubMed
    • Google Scholar
22. 1. Parks RJ

    (2005) Adenovirus protein IX: a new look at an old protein

    Molecular Therapy 11:19–25.

    https://doi.org/10.1016/j.ymthe.2004.09.018

    • PubMed
    • Google Scholar
23. 1. Patra T
    2. Meyer K
    3. Geerling L
    4. Isbell TS
    5. Hoft DF
    6. Brien J
    7. Pinto AK
    8. Ray RB
    9. Ray R

    (2020) SARS-CoV-2 spike protein promotes IL-6 trans-signaling by activation of angiotensin II receptor signaling in epithelial cells

    PLOS Pathogens 16:12.

    https://doi.org/10.1371/journal.ppat.1009128

    • PubMed
    • Google Scholar
24. 1. Schiedner G
    2. Hertel S
    3. Kochanek S

    (2000) Efficient transformation of primary human amniocytes by E1 functions of Ad5: generation of new cell lines for adenoviral vector production

    Human Gene Therapy 11:2105–2116.

    https://doi.org/10.1089/104303400750001417

    • PubMed
    • Google Scholar
25. 1. Schönrich G
    2. Raftery MJ
    3. Samstag Y

    (2020) Devilishly radical NETwork in COVID-19: Oxidative stress, neutrophil extracellular traps (NETs), and T cell suppression

    Advances in Biological Regulation 77:100741.

    https://doi.org/10.1016/j.jbior.2020.100741

    • PubMed
    • Google Scholar
26. 1. Veras FP
    2. Pontelli MC
    3. Silva CM
    4. Toller-Kawahisa JE
    5. de Lima M
    6. Nascimento DC
    7. Schneider AH
    8. Caetité D
    9. Tavares LA
    10. Paiva IM
    11. Rosales R
    12. Colón D
    13. Martins R
    14. Castro IA
    15. Almeida GM
    16. Lopes MIF
    17. Benatti MN
    18. Bonjorno LP
    19. Giannini MC
    20. Luppino-Assad R
    21. Almeida SL
    22. Vilar F
    23. Santana R
    24. Bollela VR
    25. Auxiliadora-Martins M
    26. Borges M
    27. Miranda CH
    28. Pazin-Filho A
    29. da Silva LLP
    30. Cunha LD
    31. Zamboni DS
    32. Dal-Pizzol F
    33. Leiria LO
    34. Siyuan L
    35. Batah S
    36. Fabro A
    37. Mauad T
    38. Dolhnikoff M
    39. Duarte-Neto A
    40. Saldiva P
    41. Cunha TM
    42. Alves-Filho JC
    43. Arruda E
    44. Louzada-Junior P
    45. Oliveira RD
    46. Cunha FQ

    (2020) SARS-CoV-2-triggered neutrophil extracellular traps mediate COVID-19 pathology

    The Journal of Experimental Medicine 217:e20201129.

    https://doi.org/10.1084/jem.20201129

    • PubMed
    • Google Scholar
27. 1. Wang M
    2. Marín A

    (2006) Characterization and prediction of alternative splice sites

    Gene 366:219–227.

    https://doi.org/10.1016/j.gene.2005.07.015

    • PubMed
    • Google Scholar
28. 1. Zuallaert J
    2. Godin F
    3. Kim M
    4. Soete A
    5. Saeys Y
    6. De Neve W

    (2018) SpliceRover: interpretable convolutional neural networks for improved splice site prediction

    Bioinformatics (Oxford, England) 34:4180–4188.

    https://doi.org/10.1093/bioinformatics/bty497

    • PubMed
    • Google Scholar

Author details

1. Eric Kowarz

   Institute of Pharmaceutical Biology/DCAL, Goethe-University of Frankfurt, Frankfurt am Main, Germany

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Supervision, Writing - review and editing

   Contributed equally with

   Lea Krutzke

   Competing interests

   No competing interests declared

2. Lea Krutzke

   Department of Gene Therapy, Ulm University, Ulm, Germany

   Contribution

   Conceptualization, Data curation, Formal analysis, Methodology, Supervision, Writing - review and editing

   Contributed equally with

   Eric Kowarz

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-4092-4131

3. Marius Külp

   Institute of Pharmaceutical Biology/DCAL, Goethe-University of Frankfurt, Frankfurt am Main, Germany

   Contribution

   Data curation, Formal analysis, Investigation

   Competing interests

   No competing interests declared

4. Patrick Streb

   Institute of Pharmaceutical Biology/DCAL, Goethe-University of Frankfurt, Frankfurt am Main, Germany

   Contribution

   Data curation, Formal analysis, Investigation

   Competing interests

   No competing interests declared

5. Patrizia Larghero

   Institute of Pharmaceutical Biology/DCAL, Goethe-University of Frankfurt, Frankfurt am Main, Germany

   Contribution

   Data curation, Investigation, Software, Visualization

   Competing interests

   No competing interests declared

6. Jennifer Reis

   Institute of Pharmaceutical Biology/DCAL, Goethe-University of Frankfurt, Frankfurt am Main, Germany

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

7. Silvia Bracharz

   Institute of Pharmaceutical Biology/DCAL, Goethe-University of Frankfurt, Frankfurt am Main, Germany

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

8. Tatjana Engler

   Department of Gene Therapy, Ulm University, Ulm, Germany

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

9. Stefan Kochanek

   Department of Gene Therapy, Ulm University, Ulm, Germany

   Contribution

   Conceptualization, Formal analysis, Methodology, Supervision, Writing - review and editing

   Competing interests

   No competing interests declared

10. Rolf Marschalek

    Institute of Pharmaceutical Biology/DCAL, Goethe-University of Frankfurt, Frankfurt am Main, Germany

    Contribution

    Conceptualization, Funding acquisition, Project administration, Supervision, Visualization, Writing - original draft, Writing - review and editing

    For correspondence

    Rolf.Marschalek@em.uni-frankfurt.de

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0003-4870-3445

• 10,839

  views

• 783

  downloads

• 47

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.