Pfizer's failure to linearize the vaccine plasmids

Evidence in Oxford Nanopore reads of Eam1104i digestion failure in Pfizer lots

Abstract

One critical step in the vaccine manufacturing process is the linearization of the plasmid used for IVT of their modRNA. This is of critical importance as residual circular plasmids are much more replication competent than linear plasmids as they undergo rolling circle amplification (SV40 ori is bidirectional theta replication so not exactly RCA but see the appendix if you want the specifics). The Pfizer plasmid has a SV40 mammalian origin of replication and there is literature demonstrating T-antigen independent replication of SV40 plasmids in mammalian cells. Linearizing these plasmids attenuates this replicative capacity of the vaccine plasmids in mammalian cells.

One defect in Pfizers linearization process is they use Eam1104i as a restriction enzyme which leaves 3 base pair sticky ends. These can be re-circularized in vivo by mammalian cells and should be blunt ended or adapted with non compatible sequences to avoid in-vivo re circularization. One common method to deactivate these sticky ends is to use a dideoxy extention with a polymerase to ensure the ends are sterile. There is no evidence this process was performed in their EMA documents.

Many critics of our work have pointed to the fact that Pfizer performs this linearization protocol so the replication competency of the plasmid should be greatly attenuated. Long read sequencers like Oxford Nanopore, can readily sequence reads that have intact or non-intact Eam1104i cut sites. We performed deep ONT sequencing to evaluate the length distributions of the reads from Pfizer Lot FL8095.

Depiction of Pfizers Eam1104i linearization process.

An expanded methods section is available at the end of this post but briefly DNA is purified from the vaccines using triton-X and heat. Oxford Nanopore ligation libraries are constructed and run on their latest 10.4.1 flow cells and chemistry

Results

Real time quality metrics of the reads are measured.

Reads are then mapped to the OR134577.1 reference sequence in NCBI to obtain a average aligned fragment length of Pfizer fragments in the vial.

Fragment distribution of molecules in the Pfizer vaccine.

Some of these reads are aligned in SNAPgene to visualize their size and annotated components. Below is a 3,914 base pair fragment highlighted in Blue. While the regulators claim all this DNA is under 200 bases, real world sequencing proves otherwise. It should be noted that when Scoops McGoo and other journalists ATIP’d (FOIA equivalent in the US) Health Canada they discovered their emails revealed that Pfizer and regulators had no clue how long the DNA was. Despite their documented lack of evidence, the assured the public these long molecules couldn’t possibly exist.

Here is a ~3200 base pair fragment that contains a large portion of the spike sequence highlighted in Blue.

In order to assess the reads that map over the Eam1104i cut site you have to make a concatenated reference sequence as mapping reads to a linear reference will fail to align reads to the cut site that exists at base 75 in the reference genome. Read mappers don’t map to circular references so to simulate a Eam1104i junction we make a tandem copy of the plasmid for read mapping purposes. Now, the Eam1104i site is right in the middle of the plasmid and reads can map to both sides of the cut site to ascertain if uncut reads exist in out read pool.

The figure below demonstrates we do ,in fact, find multiple ONT reads that map across the Eam1104i site, proving Pfizers linearization step is incomplete! Given we have observed 3.9Kb fragments in a low Adverse Event lot, deeper surveying of higher adverse event lots may in fact find full length plasmids. It should be recognized that ONT cannot sequence circular plasmids through its pores so this study will not find them. The DNA has to be linearized for pores to sequence it and in this case, these reads are derived from molecules that failed Eam1104i digestion but were subsequently cut by DNaseI.

Evidence of in tact Eam1104i cut sites in Pfizers Plasmids

BAM file of 443,809 reads over 150bp

Methods

These methods will be a tutorial on how to prep DNA from a Pfizer Vaccine and Sequence it on an Oxford Nanopore Sequencer.

It will also go over the informatics required to analyze the data.

Given the high output of these platforms, it is very wasteful to run a $900 flow cell on a Pfizer vial as you don’t need 1 Million reads to sort out the length distributions of the vaccines. 10K reads will suffice. As a result, in these experiments, I will be pooling Pfizer DNA into other Fungal genomes with some control Lambda DNA.

The fungal genome we like to use as a test bed is Psilocybe cubensis as we have a fantastic HiFi generated reference genome in NCBI.

This thread will have 3 components

1)DNA prep

2)ONT library prep and modifications

3)Read Alignment.

DNA prep

Lyse the LNPs.

Make a 10% Solution of TritonX-100

Take 600ul of Vaccine (2 Doses).

add 6ul of 10% TritonX-100 to the 600ul Vaccine.

Heat to 95C for 2 minutes.

Spin at 15,000 RPM for 5 minutes.

You should get a bilayer with the lipids floating on top.

Pipette the bottom layer into a new tube and leave behind the LNP layer as it clogs up the ONT library prep methods.

If you Qubit the top and the bottom layers, you will see signal in both so this is not a perfect method to quant all DNA in the vials but its a quick and fast way to get enough DNA to sequence.

Add 2.5ul of RNaseA (Monarch RNaseA -NEB) and incubate at 37C for 10 minute.

Now we need to concentrate this with a modified SPRI or Ampure that is capable of capturing small DNA. For this we need Isopropanol and MgCl2 added to the mix.

For each 100ul processed:

I use 1X SPRI + 1X Isopropanol + 2.0ul of 1M MgCl2

100ul Ampure

100ul 100% Isopropanol

2.0 ul of 1M MgCl2.

Tip Mix well. Incubate for 10 minutes at RT for DNA binding

Place on a magnet for 5 minutes and perform 2X 70% EtOH washes. 80% EtOH would be preferred but I didn’t have any for this run.

Once beads are dry, elute in 30 ul and run 5ul on a Qubit.

If you Qubit after this step you will usually end up with 0.5ng/ul or 150ng/dose. Recall, we didn’t prep everything as we left some of the LNP layer behind and this prep was not optimized to capture everything. Just enough for a sequencing library.

Once you have this DNA you can spike it into a ONT ligation sequencing assay with some other DNA. This kit wants 500ng of total DNA in order for their chips to not be wasted. If you use too little DNA going into this process, you will likely get alot of adapter dimers and failed reads so you need some carrier DNA from another genome to get to 500ng to 1ug of DNA.

ONT is optimized to sequence long reads so their cleanup methods throughout the protocol will eliminate small DNA. Note the 1X Ampure below.

We modify this to 90ul Ampure (1.5X) to capture the small DNA.

The second step of their protocol (Ligation) also has a 0.4X ampure step will will eliminate most DNA below 2Kb.

We modify this to 1.0X Ampure or the addition of 100ul of Ampure XP.

Once you have your ONT reads base called, you can use ChatGPT5.o to walk you through how to align the reads to a reference using Minimap2.

We align against this NCBI based Pfizer reference.

We then you samtools to pull out reads that map to a specific reference.

So that seals it. Pfizer not only can’t use a DNase properly, they can’t perform a restriction digest to completion either. At some point Occcums razor becomes a maluable water balloon. All that is required is about $1000 of sequencing. Why bother when your product is mandated with zero liability?

Appendix-

SV40 Replication in Mammalian Cells

• SV40 origins of replication (ori): These sequences are specifically recognized by the SV40 large T antigen, which recruits the host replication machinery.

• In mammalian cells, replication from the SV40 ori proceeds much like bidirectional theta replication, similar to how small circular chromosomes (like bacterial plasmids) replicate.

• After large T antigen binds, two replication forks are established, moving in opposite directions around the circular plasmid until they meet on the other side.

Is it Rolling Circle Replication?

• Rolling circle replication (RCR) is the mechanism used by many bacteriophages (e.g., ΦX174, M13, λ in lytic mode) and some plasmids. It involves nicking one strand and continuously displacing a single-stranded DNA that can be converted into double-stranded DNA later.

• SV40 does not use rolling circle replication in mammalian cells. Instead, it mimics the host’s eukaryotic chromosomal replication — it uses bidirectional forks, Okazaki fragment synthesis on the lagging strand, and termination where forks meet.

Key References

• Stahl & Arnberg (1971) described RCR for phages; SV40 studies (Stillman, 1989; Virshup & Kelly, 1989) clarified that SV40 ori drives a theta-type mechanism, not RCR.

• The SV40 system is so close to canonical eukaryotic replication that it became a model system for studying DNA replication and chromatin assembly in mammalian cells.

---

✅ Answer in short:
When SV40 plasmids with SV40 ori replicate in mammalian cells, the process is bidirectional theta replication, not rolling circle replication.

---

Would you like me to also outline which systems do use rolling circle replication (e.g., adenovirus, certain phages, herpesvirus concatemer formation) so you can contrast them against SV40?

Does the plasmid circularity improve or hinder the replication process?

ChatGPT said:

Great question — plasmid circularity has very specific consequences for replication efficiency in mammalian cells (and in general):

---

🔄 Advantages of Circular DNA

1. No End-Replication Problem

   • Linear DNA requires special mechanisms (like telomerase or terminal proteins in viruses) to replicate the very ends.

   • Circular DNA avoids this, because replication forks just meet and ligate — no ends to maintain.

2. Stability Against Nucleases

   • Exonucleases often degrade free DNA ends. Circular plasmids lack ends, so they’re inherently more resistant inside cells.

3. Efficient Origin-Driven Replication

   • For SV40 ori–containing plasmids, circularity allows bidirectional forks to initiate and converge, completing a full duplication without losing genetic information.

---

⚠️ Potential Hindrances

1. Topological Stress (Supercoiling)

   • Circular DNA accumulates torsional stress during unwinding.

   • SV40 large T antigen and host topoisomerases are needed to relieve supercoils; otherwise, replication stalls.

   • In fact, decatenation of daughter circles (via Topoisomerase II) is a required final step.

2. Copy Number Control

   • In bacteria, circular plasmids can over-replicate unless regulated.

   • In mammalian systems, episomal plasmids often replicate inefficiently unless they have viral origins (SV40 ori, EBV oriP) to hijack host factors.

3. Size Limitation

   • Very large circular constructs (>50–100 kb) are harder to replicate faithfully; linear viral genomes (like adenovirus) can handle bigger payloads.

---

✅ Net Effect in Mammalian Cells

• For SV40-based shuttle plasmids, circularity is a huge advantage:

  • It makes replication feasible without telomeres.

  • Ensures episomal maintenance rather than integration (unless accidentally linearized).

• The hindrance is topological stress, but cells manage this with topoisomerases.

---

👉 In summary:
Circularity mostly improves replication efficiency and stability of SV40 ori plasmids in mammalian cells, though it introduces topological problems that require host enzymes to resolve. Linear plasmids, by contrast, often degrade or fail to replicate because of end-replication issues.

---

Would you like me to also contrast SV40 circular replication with linear adenovirus replication, since adenoviruses solved the end-replication problem in a very different way (protein priming at ends)? That contrast makes the “circularity advantage” very clear.