Sunday, January 11, 2015

Horizontal Transfer of GMO DNA

To the critical science reader, the claims by the anti-GMO world are a goldmine of intellectual turds.  Over the last few weeks I've been blown away by the junk that is accepted and promoted by folks like (we don't know their actual name, just their fake sock puppet name) GMO USA.  Unfortunately I've been behind on a dozen writing assignments and can't very well put fun discussions like this into the public space without feeling a bit guilty.  However, I could not let this one go! 

To the rocket surgeons over at GMO USA, the work by Oraby et al (2014) is quite compelling. However, to anyone with even a high-school level understanding of the science, these data are just awful, and the conclusions unacceptable. The title says "Horizontal Gene Transfer" when the authors don't even test for it!

The manuscript was published in the African Journal of Biotechnology, a journal with an impact factor of 0.5 or so.  The lead author, Hanaa Oraby, has some publications here and there, so there's evidence of actual scientific training, but his work is rarely cited with a total of 109 citations since first publications in 1989.

This is important because the claim is that the authors have publishable evidence that the Cauliflower Mosaic Virus 35S promoter, the molecular switch that drives expression of a transgene at a high level in some transgenic constructs, is detectable in brain, blood and liver.  They claim that this is due to horizontal gene transfer, meaning a stable integration into genomes in these cell types.

They start with 19 experimental and 10 control rats, feed the former a GM diet from weaning and the latter a non-GM diet.  They sacrifice the animals at 30, 60 and 90 days,and prep DNA from the blood, liver and brain.  They then use PCR to detect evidence of DNA from the CaMV35S promoter using primer pairs designed against the DNA sequence.  PCR is the sensitive detection method used to amplify specific DNA fragments.  

The results show that after 40 cycles of PCR on 100 ng starting template that the authors can amplify evidence of the CaMV35S promoter in a fraction of their experimental rat parts, but not controls. 

Strengths of the work
1.  The authors actually sequenced the product to show that it is not just a spurious artifact. 
2.  The authors use non-GM fed rats as a control.

Major Weakness!
The two smaller primer pairs are "nested", meaning the sequences are found within the sequence of the large fragment.  However, the authors detect the large fragment, but in the same samples do not detect the small fragment!  They say that the short fragments are not taken up as well as the large ones.  That might be the case, but the small fragment is part of the large fragment-- so when the large one is detected, you'd have to detect the small one!   This point strongly suggests the authors are measuring false positives. 

1. Figure 1 does not show a negative control
2. The authors start with 100 ng of template!  This is a huge amount to start with. It increases the likelihood of contamination and false positives. 
3. The authors rely on 40 cycles of PCR to amplify their targets.  The use of this many cycles cannot be reliable, as a bona fide molecular target would amplify in 30.  Anyone that does quantitative real-time PCR knows that signals occurring after 35-40 cycles are evidence of contamination or artifacts.
4. The authors do not provide positive controls for PCR in their assays, which would be helpful.  A quantitative measurement would be good too.
5.  The "transfer efficiency" says that DNA was detected in 1-4% of samples, yet they authors present some level of detection in 100% of agarose gel samples. 
6. The values in Figure 6 depict "mean of GMtransfer" but shows no units and the values are based on 6, 6, and 7 rats.  Not sure what these numbers mean or how they were derived. 
7. When they report that positives were detected in "33, 37, or 52%" of samples from 30, 60, and 90 days, these are based on six rats, six rats and seven rats respectively.  They must be referring to positive samples, not positive rats (three samples per rat) to get these values, so rats with a "positive" in one sample were negative in others. Again, more evidence that they are measuring noise. 


When you have to rely on 40 cycles of PCR on 100 ng of starting template as your detection method, results are difficult to trust. Even with negative (non-GM consuming) controls, "detecting" a positive in 1-4% of samples after 40 cycles is not strong evidence of "horizontal gene transfer". 

Something concrete would go the extra step.  Prepare DNA from nuclei so that it eliminates fragments that might just be circulating in blood, as the authors cite that DNA can be found in the bloodstream (brains and livers have blood too).  Then amplify at fewer than 30 cycles using a nanogram of template, and show independent replicates and that multiple primer pairs targeted to the same DNA template amplify it as well.  That would be a start. 

The authors show some weak bands on a gel that represent only a fraction of the samples analyzed.   They apparently used 19 experimental animals and made three samples from each.  It was not clear if the samples showing positive for one primer set also showed confirmatory positives for separate set.  In fact, it appears that it is not the case, as there are only three positives for blood and more for others, when using different primer pairs.

The results were also from a single experiment, so any trace contamination would not be flagged by an inability to replicate in subsequent trials.


I get asked to review for the African Journal of Biotechnology all the time and never accept an assignment.  If they are willing to publish single replicates from cherry-picked data sets, then it is not a place that I need to invest my time.  These data are hardly compelling and my feeling is that if a comparable number of controls were used they'd see the same data there as well, even though the "data" are hardly reliable.  Size and sequence indicate they are amplifying and sequencing their target, but where there is zero evidence presented that it comes from horizontal gene transfer. To get that, they'd need to demonstrate that it is actually integrated, which is not impossible using similar techniques.

That would be a real story, and would be in Science or Nature, not the African Journal of Biotechnology