Thread by @stevenemassey

Some further insights into the SARS2 spike sequences found in Pseudomonas aeruginosa datasets, recorded as being sampled in 2019 🧵

2/ Complete SARS2 spike gene sequences were found in contigs generated from Pseudomonas aeruginosa cultures sampled in 2019, by @iximeno

The spike sequence displayed codon optimization and lacked the furin cleavage site

3/ The spike sequences are found in four contigs 👇, inserted into the pcDNA3.1 plasmid h/t @raqueltobes , with a t-PA (tissue plasminogen activator) leader h/t @Daoyu15

4/ I re-examined the following 3 features of the sequences:

i) the C-terminus extension of the spike sequence
ii) the N-terminus t-PA (tissue plasminogen activator) extension
iii) Reasons for variation between the contigs that contain the spike and plasmid sequences

5/ i) The C-terminus of the spike sequences has a peptide extension not found in SARS2

@VBruttel proposed that this is a trimerization domain, but it does not match anything in the database

https://x.com/VBruttel/status/1617580234941026304

6/ The C-terminus extension can be traced to a publication from Hong Ling's group, Harbin Medical University

The sequence is termed an 'MTQ' domain and is designed to promote trimerization of overexpressed virus glycoproteins
link.springer.com/article/10.100…

7/ The construct described by Ling's paper also uses the pcDNA3.1 expression vector, with a t-PA leader, as well as the MTQ domain, a close match therefore to the contig sequences

8/ The MTQ domain has had limited published use for the overexpression and trimerization of HIV, RSV and SARS2 surface glycoproteins (spike protein is classed as a glycoprotein)

These are all immunogenic as they are exposed to the host immune system, so of medical interest

9/ The MTQ extension starts with a flexible linker GGGSGGS, then followed by a trimerization domain designed de novo using PSIPRED and MARCOIL1.0, incorporating elements that produce coiled coil alpha helices (including heptad repeats)

10/ Trimerization of the spike protein means that it is in its native state and reflects how it is orientated in the SARS2 membrane surface (as a trimer).

This native state conformation can be important for boosting the immunogenicity of subunit vaccines

11/ The MTQ domain has been patented by Harbin Medical University, which means that if used commercially there should be royalty payments

patents.google.com/patent/CN10172…

12/ The MTQ domain is mentioned in a patent for a RSV subunit vaccine by Jiangsu Ruike Biotechnology Co Ltd (however the plasmid used is not pcDNA3.1)

patents.google.com/patent/CN11758…

13/ The identification of the MTQ domain verifies that the spike proteins in the 4 contigs are likely intended for subunit vaccines either for humans, or for testing in mouse/bat (as commonly conducted by the WIV, and described in DEFUSE h/t @VBruttel )

14/ Using 'MTQ' as a search term on Addgene () yields no results (C-terminal tags are sometimes provided in plasmid descriptions), so this indicates it is only rarely used (hence no matches on Genbank) and constitutes a useful identifieraddgene.org

15/ There are 14 Chinese spike protein subunit vaccines (2 approved, 12 undergoing trials) listed at COVID19 vaccine tracker (however note that site is not updated after Dec 2022)

covid19.trackvaccines.org/country/china/

16/ Not all of these use the entire spike protein: several use the NTD/RBD instead

Of those that use the spike protein none of those that report which trimerization domain was used report using MTQ

17/ They report using alternative trimerization domains such as:
collagen (SCB-2019)
foldon (SCTV01C)
fibritin (202-CoV9)

18/ Consequently, the spike constructs in the 4 contigs do not appear to have been used in official vaccine trials

This may be because the project was abandoned, never openly subjected to trials, or in a trial after Dec 2022

19/ b) The N-terminus of the spike sequences in the contigs possesses a t-PA leader sequence

The t-PA leader is immunogenic, but is also a secretion signal in mammalian protein expression systems (this was Ling's purpose for adding the t-PA leader)

20/ It is notable that P22 of t-PA has been replaced by asparagine (N) 👇

In the publication by Hong Ling's group, they test two modifications of the t-PA leader:

P22A and P22G

21/ The insertion of small, neutral amino acids was used to enhance cleavage of the t-PA signal sequence from the overexpressed protein after secretion

SignalP was used to predict enhanced cleavage of the tPA signal peptide due to the P22A and P22G substitutions

22/ Oddly, however, using SignalP-6.0, the P22N mutation reduces cleavage efficiency of t-PA, compared to P22A and P22G (probability 0.30 vs 0.92 and 0.81, respectively)

23/ In addition, if they had added the tPA signal sequence 1 amino acid upstream of the current join site, resulting in inclusion of an additional valine, this would result in much better cleavage (probability 0.97)

24/ This indicates that the P22N was suboptimal for cleavage, whether deliberate or not is unclear

Given suboptimal cleavage, these exact constructs are likely to have been unsuccessful as subunit vaccines or in experiments

25/ This emphasizes that designers are perfectly capable of designing suboptimal cleavage sites, with relevance to the supposedly suboptimal FCS of the SARS2 spike protein

26/ Of note, the P22A substitution has been used before Ling's 2011 paper, in 2004 👇 This indicates that the effect of mutating P22 on cleavage was known before Ling's paper, and so it is not a unique marker

ncbi.nlm.nih.gov/pmc/articles/P…

27/ iii) Lastly, I looked for differences between the contig sequences

When aligned I found that they are identical, but differ in size. I attribute this to the assembly of few reads may produce contigs of variable sizes (their sequence depth is low h/t @Kevin_McKernan )

28/ This substack by Kevin contains a lot of interesting ideas and background information on the plasmids

anandamide.substack.com/p/chocolate-in…

29/ A 2023 paper by Dongsheng Zhou and others describes the P.aeruginosa genome sequences, but unfortunately does not provide the sequencing date or location, which could help trace the provenance of the plasmid sequences

ann-clinmicrob.biomedcentral.com/articles/10.11…

30/ Can the observations overall tell us more about the date the spike sequences were generated ?

Not directly, but the identity of the MTQ domain verifies the purpose of the sequences (subunit vaccine), and indicates that there should be documentation if a post-2019 project

31/ (documentation is likely deleted if it is pre-2020)

32/ The observations are consistent with contamination of the P.aeruginosa sequence datasets either during sample / library prep or sequencing

This is because the construct is almost identical to Ling's, which was used for mammalian expression

33/ Lastly, in a striking coincidence the P.aeruginosa contigs match experiments described in DEFUSE

https://x.com/VBruttel/status/1618616822278029317

34/ Of interest also, pcDNA3.1 was used extensively by Zhengli Shi for a variety of purposes, including cloning codon optimized SARS1 spike h/t @VBruttel





nature.com/articles/natur…
journals.plos.org/plosone/articl…
nature.com/articles/s4142…
journals.plos.org/plospathogens/…
link.springer.com/article/10.100…

35/ ZLS has also used the t-PA signal sequence with pcDNA3.1




Unfortunately, plasmid sequences are not provided in her papers, so use of the MTQ domain is unclear

The use of MTQ would be a red flag sciencedirect.com/science/articl…
journals.asm.org/doi/10.1128/jv…

36/ The pcDNA3.1 expression system is quite widely used; entering the search terms "pcDNA3.1" "spike" "SARS-CoV-2" "S" into Addgene brings back 55 entries from over 7 groups

However, the MTQ domain is rare and key for tracing provenance
addgene.org/search/catalog…

37/ Strictly speaking, these plasmids cannot be ruled out as coming from the WIV, or that they were generated pre-pandemic

This is due to the lack of an audit of the WIV

@threadreaderapp unroll