Validity of report of reverse-transcription of Covid-19 vaccine mRNA in cultured human liver tumor-derived cells

Question

The following paper reports results suggesting that when Huh7 cells (severely karyotypically abnormal immortalized cells derived from a human liver tumor) were incubated with the Pfizer/BioNTech COVID-19 mRNA vaccine BNT162b2 the lipid nanoparticles were taken up. The paper argues that the vaccine mRNA was reverse-transcribed and suggests that the LINE 1-encoded reverse transcriptase activity might be involved.

Aldén et al. (2022) Curr.Issues Mol. Biol. 44, 1115–1126 “Intracellular Reverse Transcription of Pfizer BioNTech COVID-19 mRNA Vaccine BNT162b2 In Vitro in Human Liver Cell Line”

The clinical relevance of these results, if correct, relates to the possibility that the integration of viral cDNA may occur in human liver tissue following administration of the vaccine, that the integrated gene for the virus spike protein may be subsequently expressed and could therefore be responsible for some of the rare side-effects of the vaccine.

Question: Do the results really indicate reverse-transcription of viral mRNA and integration of the cDNA into the genome of this cell line? If so, to what extent is this relevant to side-effects of the vaccine administered in the clinic.

Footnote
The publisher of this journal, MDPI, has been classified as a borderline predatory publisher — hence my scepticism — although the subject is far from my field.

Answer 1

Re: Intracellular Reverse Transcription of Pfizer BioNTech COVID-19 mRNA Vaccine BNT162b2 In Vitro in Human Liver Cell Line by Alden et al. ( Curr. Issues Mol. Biol. 2022, 44(3), 1115-1126; https://doi.org/10.3390/cimb44030073 )

Regarding the question: "Question: Do the results really indicate reverse-transcription of viral mRNA...?" They provide results suggesting this, but missing key additional data (see Figure 5 comments below).

Regarding the question: "Do the results really indicate ... integration of the cDNA into the genome of this cell line?" No, they provide no evidence of that (see Figure 5 comments below).

Regarding the question: " If so, to what extent is this relevant to side-effects of the vaccine administered in the clinic." As they fail to convincingly show the above points, and are working in a cell line system of questionable relevance to normal human cells in the body, this paper seems to have no real relevance to side-effects of the vaccine administered in the clinic. See below comments.

This paper is NOT reliable. A non-exhaustive list of problems in no particular order appears below, but first let's consider their Results:

Figure 2: Yes, they show that BNT162b2 vaccine mRNA enters the cultured cells. However, all their statistical comparisons except for those against the controls are invalid -- see pt. 2 below.

Figure 3: They fail to show that BNT162b2 affects LINE1 mRNA levels. First, the control plots are visually not different from those of the vaccine-treated samples. Second, essentially all their statistical comparisons seem to be invalid -- see pt. 2 below.

Figure 4: Fluorescence always makes pretty pictures, but this figure again seems to fail to show anything useful. First, the LINE1 protein antibody used binds to the LINE1 ORF1 protein, which does not have reverse transcriptase activity, rather than the ORF2 protein, which does contain reverse transcriptase activity. An anti-ORF2 Ab is available, but they did not use it. Also, from Fig. 3 we see no significant difference in LINE1 RNA levels, so LINE1 protein levels would be unlikely to vary. Finally, the statistical comparisons shown are likely to be invalid -- again, see pt. 2 below.

Figure 5: This section claims to detect reverse-transcribed BNT162b2 in DNA preps from buffer-washed cells previously exposed to BNT162b2. Maybe they have, but key controls are missing.

The one piece of evidence shown is successful PCR amplification of a segment of the vaccine sequence. They do show some useful controls, such as lanes for cells not treated with the vaccine, but miss other essential controls such as cells treated with a different mRNA vaccine not containing that amplicon, and DNA from relatively karyotypically and morphologically normal cultured cells treated with BNT162b2, another mRNA vaccine, or nothing. It's all too easy to have a few stray molecules of a DNA fragment present in the lab (like the BNT162b2 PCR amplicon) contaminate other samples and lead to false positives after lots of cycles of amplification. They also provide no evidence any reverse-transcribed BNT162b2 is integrated into the cells' genomic DNA.

As for the paper's significance to BNT162b2 influence on actual normal human cells, there really isn't any because of how abnormal this immortalized liver tumor-derived cell line is in terms of of genomic stability and of gene expression. It's sort of like using Godzilla as a model organism for humans.

The promised non-exhaustive list of problems:

1. It's clear they don't understand some of the features they discuss, such as LINE1 elements, calling LINE1 "endogenous reverse transcriptase" when they are actually transposons. They also seem to think there is just one LINE1 gene, not understanding that they are present in thousands of copies throughout the genome (having "jumped" into many genes).

2. It's clear they don't understand how to do statistics properly. If you just do more and more comparisons between different experiments, eventually you will find a difference just by chance. To correct for this, you must reduce your significance threshold to take into account the number of comparisons you make. An easy way of doing this is called "Bonferroni correction" and there is no evidence they used it or any other such correction. As far as I can tell, after the Bonferroni correction is applied, essentially none of their meaningful comparisons end up being statistically significant.

3. Their main "control" consists of no vaccine, rather than a dummy/different vaccine. Effects associated with vaccine presence could for example be due to the lipid nanoparticles or modified RNA -- that is, they could be non-specific responses.

4. The fundamental flaw: as a stand-in for human cells, they use only a highly chromosomally abonormal cultered liver cancer cell line Huh-7. According to a description of the line, "The majority of Huh-7 cells show a chromosome number between 55 and 63 (mode 60) and are highly heterogeneous." (normal cell count is 46) In addition to "...containing many mutations and INDELS, it is worthy to note the Huh7 cells have a point mutation in the p53 gene."

5. The LINE1 protein they image likely appears in tumor cell lines (like the Huh-7 line they use), but not in normal cells. From "Methods Mol Biol. 2016; 1400: 261–280. doi: 10.1007/978-1-4939-3372-3_17 ""Using immunohistochemistry, we found nearly half of human cancers stain positively for ORF1p, with immunoreactivity in some common cancers approaching 100% of cases. No staining was observed in the cognate normal tissues"

The Zhang et al. (PNAS May 25, 2021 118 (21) e2105968118; https://doi.org/10.1073/pnas.2105968118 ) paper mentioned in this paper is discussed in my answer to a Medical Sciences StackExchange question: MedicalSciencesQuestionAboutZhangPaper The summary of my answer to "can the SARS-COV-2 RNA modify our DNA[based on this paper]" is "The short answer is maybe, but rarely, and the whole Covid virus has never been seen to integrate into the cell's DNA intact. Any integration requires "helper" molecules not found in the Covid virus, and only very rarely found in a normal cell."

Answer 2

The paper may be reliable in its finding that

"BNT162b2 mRNA is reverse transcribed intracellularly into DNA in as fast as 6 h upon BNT162b2 exposure",

however, the authors apparently refrained from trying to find integration into the genome of the hepatitic cells. They do not claim to have found integration into the genome. They say,

"(f)urther studies are needed to demonstrate the effect of BNT162b2 on genomic integrity, including whole genome sequencing of cells exposed to BNT162b2, as well as tissues from human subjects who received BNT162b2 vaccination."

In that context they do not refer to a previous study which - before the question had been edited anew - had been linked to: Zhang et al., Reverse-transcribed SARS-CoV-2 RNA can integrate into the genome of cultured human cells and can be expressed in patient-derived tissues. Interestingly, this study did not use tissues from human subjects who received vaccine (or had been infected), however in their setup used whole genome sequencing to detect recombinded virus-host-DNA. This study did not.

Isn't whole genome sequencing easy to do these days? Was there a lack of "tissues from human subjects"? In fact, mentioned previous study by Zhang et al. (on virally infected cells) recurred to existing RNA-analysis (existing data) when trying to find out about normal cells, after they had examined artificially LINE1-upregulated cells in vitro. That may hint to what does not go without saying, which neither the study challenged by question nor its co-study mentioned above explains and elaborates on, and only to experts seems obvious and apparent:

The study questioned did not undertake to prove whether retrotranscribed DNA - which was undebatedly found - is of host cell genome origin (post integration) or of viral origin (post retro-transcription).

Both studies do not speak of "free viral cellular DNA". That may be some blunt way to describe the issue, however, first, the technique of recombinant DNA vaccine shows very cleary (and very unexpectedly) that DNA in cells may exist without being integrated into the genome of the host cell. Adenovirus vector technique is based on viral DNA in cells not being integrated into the genome (There is no finding of vaccine recombinant DNA being retro-transcibed into the genome (after transcription to m-RNA)). Second, it is accepted knowledge that several virus species persist latently in host cells in so called episomal form. E.g. herpes virus DNA is known to persist episomally in the nucleus without integrating into the host cell's genome. The following quote refers to adeno associated virus (different from Adenovirus used in vaccine): "Although AAV vector genomes can persist within cells as episomes, vector integration has been observed in various experimental settings, either at non-homologous sites where DNA damage may have occurred or by homologous recombination." Deyle and Russell, Adeno-associated virus vector integration; see also, e.g., Chong et al.Transfection types (...): "Stable transfection refers to (...) integrating foreign DNA into the host nuclear genome or maintaining an episomal vector in the host nucleus as an extra-chromosomal element (...). The transgene may then be constitutively expressed even with the replication of cells (...). In contrast, transient transfection does not require integrating nucleic acids into the host cell genome (...)"

This shows that retrotranscription as a first step may very well be distinguished and proved independently from integration into the genome.

The study in question here proved retro-transcription of vaccine m-RNA. Different from Zhang et al. they did not do any "whole genome sequencing" to find out about integration. The authors do not give reasons. In case of positive findings, we still wouldn't know if CoV-DNA integrates in normal tissue.

Whereas the first study searched for

"chimeric transcripts"

i.e. DNA sequences that comprise quote "partial (...) subgenomic RNA sequence that was flanked by a duplicated host cell DNA target sequence" to lay proof of integration into the genome, the study challenged here did not. You will not find the word "chimeric transcript" being mentioned in the description of the methods employed:

"Genomic DNA was extracted from cell pellets ... PCR was then performed using primers targeting BNT162b2 (sequences are shown in Table 1) (Remark:no chimeric sequences among) ... Bands corresponding to the amplicons of the expected size (444 bps) were cut out and DNA was extracted ... The sequence of the DNA amplicon was verified by Sanger sequencing."

However, it seems unanimously accepted that chimeric transcripts are valid proof of integration - if validly found. The latter had been questioned by peer review in respect of the study by Zhang et al.

To sum up: There is no reason given for not using "long sequencing" technique that might have found chimeric transcripts. Alternatively, the study might have given reasons why chimeric transcripts cannot be accepted as proof of integration. The study questioned here seems to self-restrict itself in significance and reach.

It may be some biased opinion to think of the previous study having been reviewed as technically not valid in respect of its findings of chimeric transcripts as a reason why the study challenged here refrained from long sequence technique (Remark: mentioned previous study did not either in respect of tissue from patients, in its non in vitro follow up, as it then used pre-existing data that resulted from RNA, not DNA search and sequencing which in my opinion might explain the artifacts they were blamed for).

Final remarks: Merits of this study may lie in the use of nano lipids in combination with cancer cells. It had been shown before that any cell can possibly uptake the nanolipids; however, in respect of the previous study by Zhang et al., this study results in some strong argument for LINE1-expression not being dependent either on artifically induced over-expression nor - as it used nano lipids - on viral infection (host-virus membrane fusion or receptor mediated endocytosis). Natural infection may very well appear as a prerequisite for upregulation of LINE1.

Still, any study might be challenged for not being "reliable" or "serious" (quote pre-edited version of this question), in my opinion, as long as the mechanism of LINE1 activation, of inducing LINE1, remains unknown.

There are no known side-effects of LINE1 integrating DNA, and they have not been searched on by the study questioned nor are there any other studies up to date on effects of CoV-DNA integration. It is beyond the scope of this study to offer cues that hint at chimeric transcripts (integration) coresponding (not to side-effects but) to severity of disease (Zhang et al.: "...in some patients-derived tissues ... a large fraction of the viral transcripts could have been transcribed from SARS-CoV-2 sequences integrated into the host genome.")