Was writing this reply, but the old topic was closed before I could submit.
I went over in great detail in 2 previous replies (here and here) for why the data doesn’t support their conclusions.
First off, I would like to let everyone know that the U.S. Right To Know is an anti-GMO organization that is known for harassing 40 scientists by abusing the FOIA, which was also done against climate scientists by climate deniers in the “climate gate” debacle. A classic anti-science move. Can’t refute the science, attack the scientists.
Moving on, nowhere do they actually address these problems. The only rebuttal they make is against the fact that using such type IIS like BsaI and BsmBI are specifically used in such a way that would automatically remove the recognition sites from the final assembly, for which they say that these recognition sites nevertheless could’ve been retained in the final product, but I have already addressed this previously:
Although the USRTK article does make a point that I want to address next.
Please pay close attention. Remember that the specific recognition sites in question are those of BsaI and BsmBI which are type IIS recognition enzymes, and recall that the problem with the claim that these enzymes were used to construct the SARS-CoV-w genome is that the recognition sites of such type IIS enzymes are removed during the genome assembly process (unless oriented in a specific way, which they are not).
In their response to this objection, they point out among 10 synthetic coronavirus only 2 were constructed using the Golden Gate assembly where all recognition sites are removed, while the recognition sites of unnamed enzymes are retained in the other 8 synthetic coronaviruses.
The main concern voiced by experts is that a classic approach used to assemble coronavirus genomes — so called “Golden Gate Assembly” — usually removes restriction sites or “scars” from the final product.
“I’ve even joked … in the future we will easily be able to tell real vs. synthetic DNA just by seeing if there are BsaI and BsmBI sites present, because most of the synthetic methods to make DNA will leave the final product absent of these sites,” said Tom Ellis, a professor of synthetic genome engineering at Imperial College London.
Sylvestre Marillonnet, who pioneered Golden Gate Assembly, told The Economist that the pattern of restriction sites indeed appeared to be typical of engineering, but that he was struck by the fact that the restriction sites remained in the genome.
Stanley Perlman, a University of Iowa coronavirologist with expertise in reverse genetic systems, also thought it was unusual that sites used to generate fragments for in vitro assembly would be left in the assembled genome. As a result, he found the preprint to be “not very credible.”
However, the preprint authors said that they analyzed synthetic coronaviruses assembled before the pandemic, and eight out of ten retained their restriction sites.
Golden Gate Assembly is also not the only way to clone viral genomes. Other methods keep the restriction sites in the final genome.
Does anyone else notice this bizarre switcheroo? The objection is NOT that all/any recognition sites are removed during genome assembly, the objection is that the sites of type IIS enzymes like BsaI and BsmBI are removed. And we aren’t specifically talking about Golden Gate Assembly (this is just one example where type IIS recognition sites are removed). We are talking about any genome assembly process (Golden gate or not) that would remove recognition sites of type IIS enzymes.
Clearly, they are obscuring the relevant details. Oh, I wonder why?? So, I looked up the 10 synthetic coronavirus genomes and see which restriction enzymes they used (type IIS or not) and whether they are retained in the final product.
1: TGEV (2000)
Here they used recognition enzymes BglI, BstXI, NotI, and ApaI. The assembled genome product contains recognition sites for all enzymes, except ApaI, and the recognition sites for BglI were artificially inserted by using primer-mediated PCR mutagenesis. BglI and Bst*XI in particular are convenient to use since the sticky ends they create are often unique, so they aren’t compatible with other sticky ends produced by the same enzyme. This allows for fragments to be stitched together in a specific order (A to B to C, as opposed to C to A to B). This is similar to type IIS which also produce unique sticky ends. However, the significant differences is that these enzymes recognize palindromic (symmetrical) sequences and they cleave the DNA within the recognition site (SEE: left image below). On the other hand, as I have explained before, type IIS have non-palindromic (asymmetrical) recognition sites and they cleave the DNA at a defined distance away from the recognition site in a specific direction (SEE: right image below).
Since type IIS recognition sites have a specific orientation (due to their asymmetry) and since the cleavage site is outside the recognition site, this means the recognition sites are easily removed during genome assembly. This is NOT the case for the enzymes that were used to assemble assembled the TGEV genome. So, it’s not a surprise that the recognition sites were retained… Thus, the fac that these recognition sites are retained is irrelevant to the objection that the theoretical use of type IIS enzymes like BsaI and BsmBI to assemble the SARS-CoV-2 genome would’ve removed the very recognition sites in the assembled genome.
2: SARS-CoV-1 Urbani (2003)
Here they also used BglI to connect the subclones (A to F) of the genome. So, the fact that here BglI restriction enzyme sites were retained is also irrelevant just like in the previous example. However, here they also used an type IIS enzyme called AarI to make each of the individual A-F subclones. It’s not BsaI nor BsmBI specifically, but at least it’s a type IIS enzyme. So, were the recognition sites retained in the final product?? NOPE!! You can see that from the figure 1C that these recognition sites are removed after ssembly. The paper even spells this out in the text:
To rapidly assemble consensus clones, we used class IIS restriction endonucleases [AarI] that cut at asymmetric sites and leave asymmetric ends. These enzymes generate strand-specific unique overhangs that allow the seamless ligation of two cDNAs with the concomitant loss of the restriction site.
The only recognition sites that were retained are not of the type IIS enzymes DITTO #1.
And the recognition sites that are of the type IIS were NOT retained, thus only given more strength to the objection.
3. SARS-CoV-1 Urbani (2006)
Here they used the following restriction enzymes: ClaI, MluI, PmeI, BamHI, and NheI.
NONE of these are type IIS enzymes. DITTO #1
4. MHV (2008)
This is one of the two where Golden Gate Cloning was used, and all type IIS enzyme recognition sites were removed.
To ensure unidirectional ligation of all fragments, native or engineered type IIs restrictions sites were used to form junctions at the ends of each fragment, which allowed the restriction site to be removed by restriction digestion
5. SARS-CoV (2008) and 6. A bat SARS-CoV (Same 2008 paper)
Restriction enzymes used: BglI, AflII, NotI
Once again, NONE of these are type IIS enzymes. DITTO #1
7. HCoV-NL63 (2008)
Restriction enzymes used: BstAPI and BsmBI
- BstAPI is NOT a type IIS restriction enzyme, DITTO #1
- BsmBI is indeed a type IIS restriction enzyme, and it is one of the two that the preprint paper was focusing on. However, the recognition sites for BsmBI were NOT retained in the final assembly DITTO #2
The strategy employed by our laboratory has been to divide the genome into stable cDNA fragments flanked by native or engineered type IIS [BsmBI] restriction endonuclease sites that form unique junctions at the ends of each fragment. […] For assembly, the fragments are cleaved by restriction digestion, which removes the nonnative portion of the restriction site and sequence, leaving unique ends that allow for a seamless, unidirectional ligation of the full-length cDNA clone.
8. MERS-CoV (2013)
Restriction enzyme used: BglI
Again, NOT a type IIS restriction enzyme. DITTO #1
9. WIV1 (2016)
Restriction enzymes: SacII, AscI, and BglI.
Again, NOT a type IIS restriction enzyme. DITTO #1
10. PEDV (2018)
This is one of the two where Golden Gate Cloning was used, and all type IIS enzyme recognition sites were removed. DITTO #4
TL;DR:
The USRTK article said the following:
The main concern voiced by experts is that a classic approach used to assemble coronavirus genomes — so called “Golden Gate Assembly” — usually removes restriction sites or “scars” from the final product.
“I’ve even joked … in the future we will easily be able to tell real vs. synthetic DNA just by seeing if there are BsaI and BsmBI sites present, because most of the synthetic methods to make DNA will leave the final product absent of these sites,” said Tom Ellis, a professor of synthetic genome engineering at Imperial College London.
Sylvestre Marillonnet, who pioneered Golden Gate Assembly, told The Economist that the pattern of restriction sites indeed appeared to be typical of engineering, but that he was struck by the fact that the restriction sites remained in the genome.
However, the preprint authors said that they analyzed synthetic coronaviruses assembled before the pandemic, and eight out of ten retained their restriction sites.
Golden Gate Assembly is also not the only way to clone viral genomes. Other methods keep the restriction sites in the final genome.
Instead, it should say that the restriction sites of restriction enzymes that are not of the type IIS class are indeed often retained, but this is irrelevant to the objection that was made. On the other hand, whenever they are used, the restriction sites of type IIS restriction enzymes (such as BsaI and BsmBI) are NOT retained in the final assembly in any of these synthetic coronaviruses.
FIN