The 2013 PLoS One article Complete Genes May Pass from Food to Human Blood is often used as evidence that genes from GMO can “transfer” into our bodies (such as in this article from Collective Evolution). In this post, I’d like to review the paper with you and discuss this nightmare-inducing scenario.
The authors of the paper examined the content of DNA outside the human cell, known as “cell free DNA” or cfDNA. As a reminder, the DNA we inherit from both our parents is packed up nicely and tucked away within the nucleus of the cell. The paper outlines that the source of DNA in our plasma (i.e. the stuff that’s in the space between our cells) is thought to originate from cells that have died. However, there are also foreign sources of DNA in plasma from bacteria, viruses, and from our food. Fetal DNA can also be detected in maternal plasma and is the basis for non-invasive prenatal testing (NIPT).
The authors of the paper took 200 blood samples from 4 different types of patients who had different intestinal diagnoses, and included patients with no symptoms as their control. They separated the blood cells from the plasma, extracted the DNA and pooled the DNA from each group. So, for example, if there were 50 patients with irritable bowel syndrome and 50 control patients, each group of 50 was pooled into a single tube so that there were only 2 samples at the end: one sample representing the irritable bowel syndrome patients and another representing controls. They sequenced the DNA in the pools.
The authors threw out all the DNA sequences from vertabraes because a) they weren’t interested in human DNA sequences and b) it would be difficult to tell what organism the DNA came from due to similarities in DNA sequences (after all, we’re more similar to chickens than we’d like to believe). They took the remaining DNA samples and compared them to a database of sequences of chloroplast DNA.
Chloroplast DNA is unique because it is separate from the DNA found in the nucleus of plant cells. It is circular and there are multiple copies of chloroplast DNA in each plant cell (sounds a bit like mitochondrial DNA, if you’re familiar with that from 23&Me or other ancestry DNA sequencing services). The authors found that there were quite a few chloroplast DNA sequences, particularly sequences for potato and tomato chloroplast.
Then, they wanted to determine the original size of the DNA fragment. It is generally thought that most DNA gets fragmented during digestion, so if the authors could demonstrate that the DNA sequenced was long, then they might be able to make a case that entire genes could be floating around. THAT would be a pretty interesting finding. However, this is pretty difficult to demonstrate because during the process of preparing a sample for sequencing, you generally chop up the DNA into bits and pieces.
Here’s an analogy. Imagine that the preparation of a sample for sequencing is like baking cookies and the DNA are walnut pieces that go into the batter. You get a bag of walnut pieces, which may contain a few whole walnuts. The recipe calls for throwing the walnut pieces into the food processor before you add them to the cookie batter. So, how do you figure out how many whole walnut pieces were in the original bag, if any at all, based on the walnut pieces in the final cookies??
To get around this conundrum, the authors physically filtered the DNA according to size. They had 3 filtration sizes which became 3 different samples. Each sample was then chopped up. When it was sequenced, they could infer that the DNA’s original size was larger than the filtration cutoff. If we go back to our walnut analogy, imagine that you take the bag of walnut pieces and pass it through a 1/2 inch sieve. Everything that gets caught goes in one bowl. Then you take the stuff that went through and you pass it through a 1/4 inch sieve. You repeat the process with a 1/8 inch sieve. Then you take the 3 bowls of walnuts and you put each one of them through the food processor, make the cookie batter, and end up with 3 batches of cookies. All 3 batches will have roughly the same walnut size, but you can infer that the original starting size of the walnut pieces was >1/2″, 1/2-1/4″, and 1/4″-1/8″.
The authors infer that a lot of DNA sequence came from the largest filter size in patients diagnosed with irritable bowel syndrome (IBS). The filter size that they used was 10 kilobases. If you consider that the average size of a human gene is 10-15 kilobases, it implies that most of the cell-free DNA in patients with IBS is large enough to have a gene in it.
The authors then wanted to confirm their findings. They searched publicly available DNA databases and found 909 samples of cell-free DNA, representing 907 individuals. They also found non-human DNA in the electronic data, but noted that the amount that was present had “large variations” from person to person. They followed the same data analysis workflow as before. The DNA in the public databases came from 2 projects: one project was studying patients with an autoimmune disorder and the second was trying to detect fetal DNA in pregnant women. Here’s the breakdown of the DNA from the two studies.
- Autoimmune disorder: The most common matches were to chloroplast DNA from Brassica rapa, as well as orange. The authors state that there’s a lot of plant DNA in these samples when compared to control. Since this is the same observation noted in the patients with irritable bowel syndrome, the authors state that high levels of plant DNA circulating in plasma may be associated with inflammation.
- Pregnant women: The authors were able to determine that the most common match to chloroplast DNA were from soybean. Additionally, since these samples weren’t actually pooled together (i.e., each sample was sequenced independently), the authors were able to identify differences in the abundance of plant DNA in these samples, which represents differences in the diets of the pregnant women. This finding suggests that the plant DNA detected in these samples are not actually contaminants.
The authors conclude that the presence of foreign DNA in the plasma is not unusual, that its concentration is highest in patients with inflammation, and that these findings should lead us to revisit our views on the degradation and absorption of DNA/RNA in our bodies.
I think that the finding that there is plant DNA circulating in our bodies isn’t a big deal. The paper provides several references for studies that have examined this issue and have found DNA from our food in our organs and tissues (see here and here). However, what’s novel and unique in this paper is the suggestion that it’s whole genes, not gene fragments, that are circulating in our plasma and the suggestion that increased levels of circulating plant DNA may be associated with inflammation. As such, I’m going to focus on these unique findings from the paper.
I have several issues with the experiment the authors performed in their lab (i.e not the data analysis work on the plasma samples from autoimmune disorder patients or pregnant women):
1) Contamination. As I stated at the beginning of this piece, the authors are sequencing the DNA in the space between our cells. There’s very little DNA in there so the risk of sequencing a contaminant is high. It’s a matter of abundance: if you had actual cellular material, all that plant DNA would get drowned out by the vast amount of human DNA that you’d end up sequencing. The authors probably had very little DNA when they started, so any DNA from the environment or from their equipment could be mistaken for DNA from their samples.
Since the risk of contamination is higher, the authors should have included a negative control yet the authors failed to do this simple test. More recent studies have noted the importance of a negative control in experiments that use new sequencing technologies, particularly those with low biomass (full disclosure: I work for companies that develop sequencing technologies). Another paper’s finding suggest that contaminant sequencing “are ubiquitous” and that cross contamination between samples probably goes unnoticed. The absence of a negative control in this study, particularly given the little amount of DNA that they’re working with, is a glaring omission and should have been an important consideration in the experimental design.
2) The authors find high levels of tomato and potato DNA in all their samples. This doesn’t make much sense to me. Why would the authors find the same two DNA samples to be of highest abundance in all the different patient types and filtration sizes? As seen in the study with pregnant women, there should be variation between the different groups. I know that tomatoes definitely don’t make up the biggest part of my veggie/fruit diet, so this strikes me as odd.
3) The authors find abnormally high levels of plant DNA in the irritable bowel syndrome patients, but only for the largest filtration size. The authors conclude that foreign DNA in plasma is elevated in patients with inflammation. As such, you’d expect to see increased levels of foreign DNA in every filtration size. However, the medium and small filtration sizes have plant DNA levels equivalent to the patients with no symptoms. There’s one thing that I think you can agree with: concluding that “plant DNA is elevated in patients with inflammation” is a HUGE conclusion to draw from a single sequencing run.
4) Filtration controls? Where are you? The authors infer DNA size based on physical separation of DNA. However, they have no controls. It would be fairly simple to just spike in DNA of different, but known, sizes (the use of a “ladder” in DNA size separation is very common, so it would have been trivial to do). This size control would have also helped determine contamination: if you find some of the large DNA control in the small DNA results, then you know that some sort of contamination may have occurred during the filtration process. It would be similar to placing a brazil nut, a hazelnut, and a peanut whose sizes you’ve measured into the walnut size separation. The brazil nut should filter out with the large walnut chunks, the hazelnut with the medium chunks and the peanut should end up in the small bits and pieces. If any pieces of these nuts appear in the “wrong” cookie batch, then you could conclude that there was contamination. Maybe you didn’t wash the blade on your food processor well enough. Or maybe you got carried away by the music you were playing in the kitchen and made an inadvertent mistake. Seriously. Anything is possible, and if you don’t have controls, you’ll never know.
5) Why chloroplast DNA? I think it’s odd that they focused exclusively on the analysis of DNA from the chloroplast, and not the DNA from the nucleus of the plant cell. Is this truly reflective of all the DNA in the cell? Is it possible that due to the circular nature of chloroplast DNA, it can avoid degradation more readily? Since there are more copies of chloroplast DNA in each cell, how does this affect their findings?
But, let’s imagine that the findings of the paper are not an error and that someone else actually replicates the results. What does it mean?
- This has little to do with GMOs. If a transgene is floating in our system, so is a full gene from a traditionally bred crop as well as any other cellular material we eat. It doesn’t matter if the DNA came from fried bacon, pesticide-laden spinach or organic blueberries: your body doesn’t know the difference and can’t pick/choose what DNA to absorb. This fact alone should debunk titles of articles such as “Genetically Modified DNA transfers from food to blood“.
- The most important question: then what? There are two scenarios that I can think of (but please feel free to comment below if you can think of something else):
- Somehow these whole genes that are floating about have to make their way through the outermost layer of the cell (cell membrane), avoid getting degraded by proteins that chop up foreign DNA, make their way into the nucleus, and then somehow get integrated into the cell’s DNA (i.e. act like a virus even though it doesn’t have any of the viral proteins/genes). But let’s say that somehow one of these scenarios were to play out, and the gene that was floating about was the transgenic gene from a GM corn (the odds of this alone are 1 or 2 in 32000, since there are only 1-2 transgenic genes added to corn, which has 32000 genes). The DNA somehow manages to defy all odds and get made into RNA. The RNA will then be made into a protein. And let’s pretend that this happens stably: meaning that this protein keeps getting made. That’s 1 cell out of the 46-68 trillion in our body that is making a foreign protein. The two most likely fates for this protein produced by this single cell in your body is a) your immune system will take care of matters or b) the protein will just fade away (all proteins have a half life; they don’t just float around forever).
- The second scenario is that the gene gets integrated into the DNA from the bacteria in our gut. Again: that’s one cell out of the 100 trillion bacteria in our gut. In order for it to proliferate and “take over”, the gene that gets integrated would somehow have to confer the bacteria some form of selective advantage. Why would that happen specifically with a transgene and not with anything else that we eat? Additionally, what sort of selective advantage would Bt-resistance, for example, give to the bacteria in our gut? With the microbiome sequencing projects that are currently underway, there’s no evidence to date that “gene integration from our food” happens.
However, if you want to lose sleep over these scenarios, go right ahead. I’m more worried about the zombie apocalypse, and the CDC thinks you should be too.
Great breakdown of bad science. Lack of controls should have sent this paper to the circular file immediately. So much for proper peer review in this case.
This was a really good and detailed review. Like you I wondered why chloroplast DNA from only potato and tomato. It really does look like an artifact.
There are lots of copies of the chloroplast DNA in each cell, so lots of copies of each element. That might be why they were found more commonly than nuclear DNA.
The lack of controls is indeed a problem. You can get all sorts of spurious sequence with PCR if you don’t run the appropriate controls (says he from experience).
« So much for proper peer review in this case »?
Only in this case?
I am afraid the scientific community – and indeed society as a whole – has a major problem with peer review in areas where science is polluted by opinion.
Thanks for this. I’m not a scientist and have no experience whatsoever in the use of the sequencing technology or study protocols, but you did a good job of bringing the hay down to the cows so to speak in explaining the design and conclusions so that a layman like myself could have a working understanding.
I also wonder how even if the findings are correct, that the conclusions implicate genes in foods from ge varieties pose an elevated health risk by the mechanism advanced here. After all, lots of foods we eat that have no ge versions have genetic sequences that produce harmful substances — in fact much of the point of genetic manipulation of food crops over the centuries has been to reduce the level of harmful substances food crops produce closer to their wild state so that we can safely add them to our diet.
The finding of chloroplast DNA from tomatoes and potatoes is interesting. Both are are a nightshade descendant and even today, domesticated varieties still produce low levels of solanine. According to Wikopedia “Solanine is a glycoalkaloid poison found in species of the nightshade family (Solanaceae), such as the potato (Solanum tuberosum) and the tomato (Solanum lycopersicum). It can occur naturally in any part of the plant, including the leaves, fruit, and tubers. . . . Solanine poisoning is primarily displayed by gastrointestinal and neurological disorders. Symptoms include nausea, diarrhea, vomiting, stomach cramps, burning of the throat, cardiac dysrhythmia, nightmare, headache and dizziness. In more severe cases, hallucinations, loss of sensation, paralysis, fever, jaundice, dilated pupils, hypothermia and death have been reported. . . Ingestion of solanine in moderate amounts can cause death. One study suggests that doses of 2 to 5 mg per kilogram of body weight can cause toxic symptoms, and doses of 3 to 6 mg per kilogram of body weight can be fatal.”
I certainly would not want the right genetic sequences from potatoes and tomatoes I eat to somehow be incorporated into my cells or those of my stomach flora and start producing solanine with any abundance. Taking your thought experiments and assuming that gene sequences floating around in our blood somehow breach all the barriers to survival, let alone being incorporated and active in human cells, I am skeptical that acquiring this naturally occurring genetic code of tomatoes and potatoes to produce solanine is more palatable than somehow acquiring traits to date added by ge methods (e.g. HT or BT). Before I would consider this a concern limited to or heightened by ge manipulation, how is it a given that the one or two or handful of genes of a plant acquired by biotech mediated process out of the entire genome are more likely to be susceptible to incorporation than the sequence for producing say solanine or chlorphyl, or carotene, and so forth. There would have to be some reason why the ge process somehow made the ge traits or the thousands of non ge traits more likely to overcome all the obstacles.
Here is one more point. If you screened people’s blood for solanine with methods able to detect at extremely minute levels, I wouldn’t be surprised you’d find it floating around in there, perhaps in higher concentrations than the ge traits, or glysophate, or whatever in all potato and tomato eaters. It might even be found in breast milk, umbilical chords, etc. I’m waiting for the day this is done and we have a health panic that we need to avoid tomatoes and potatoes.
It’s definitely a problem. According to Wikipedia, PLoS One was launched with the perspective that “PLOS ONE only verifies whether experiments and data analysis were conducted rigorously, and leaves it to the scientific community to ascertain importance, post publication, through debate and comment.” I understand why this is important: we need journals where we can publish negative results, results that aren’t groundbreaking, etc. which can be important but rarely get published by high impact journals. However, I think that the way that most of the public receives “scientific” information today does not allow for this model, particularly on topics that are deemed controversial. PLoS One was launched in 2006, and since then, the amount of misinformation surrounding scientific/medical topics online has drastically increased. I’ve seen this particular paper in dozens of anti-GMO publications/websites. Does the average reader go through the comments in PLoS One to note that critics have deemed the absence of a control to be a fatal flaw? They don’t. They only get the blurb from a website stating that a peered reviewed article in PLoS One found that DNA from a GMO is in your blood. The Genetic Literacy Project had an article about this problem: http://www.geneticliteracyproject.org/2014/10/13/peer-review-on-controversial-science-issues-letting-all-of-us-down/
“This has little to do with GMOs. If a transgene is floating in our system, so is a full gene from a traditionally bred crop as well as any other cellular material we eat. It doesn’t matter if the DNA came from fried bacon, pesticide-laden spinach or organic blueberries: your body doesn’t know the difference and can’t pick/choose what DNA to absorb.”
Of course is matters!! I’m sure Monsanto engineered it to rapidly incorporate into the human genome, facilitating world domination;-o
Next story: Food Shown to have Genes–Should be Avoided
I think the usual argument is something along the lines of “transgenes are special because they were inserted with scary genetic modification techniques”.
I actually wrote a manuscript to rebut this paper that just came out today:
Just read it! Thanks for all the work that went into the study! Another recent paper, which still hasn’t been published, suggests setting guidelines for low biomass samples (http://biorxiv.org/content/early/2014/07/16/007187). I agree that the whole field should make negative controls mandatory for these samples.
I appreciate that you broke this study down into concepts understandable by non-scientific laypeople. But, at best the conclusions you draw raise questions about the reported study but don’t disprove them. A leap of faith is required in both scenarios as the truth is that science simply does not have the full answer about the possible implications of GMO foods on human health.
As someone who chooses to eat a very healthy diet, I think the pro GMO crowd entirely misses the point. That is, I think GMO products are absolutely perfect for you and your families. I think you should embrace them and eat them to your heart’s content. But, I also think that for myself and my family, I should have the product information necessary to make an informed choice.
The issue at this juncture isn’t which ideological camp has raised the best pseudo-scientific conclusions, but instead about the choices people want to make about the foods they consume.
Yes, thanks for the analysis. I didn’t notice any mention of bacteria DNA. Since bacteria are so adaptable, should there be any apprehension at all that they use this? I know that the compatibility rate is low between humans and bacteria, but is there a possibility that this manipulated bacteria will alter the bacteria that we are currently able to cope with?
The bacterial DNA Richard Lusk found in the data set is from skin bacteria. This is clearly evidence of contamination of the original sample. Given the methods the authors of the original study used, it is little surprise that they were able to identify contaminant DNA. This is one of the risks of modern molecular biology techniques that people who are using them need to be aware of.
We need a label!
“Warning! This product may contain DNA.”
Please correct me if I’ve misunderstood, but I believe you’re asking if genes from our food have transferred into the bacteria in our gut?
If that’s the case, I’ve looked over several studies that have sequenced bacteria from the gut of animals fed GM-feed and controls. The studies have found no or little difference in the number and representation of bacteria. Granted, they aren’t sequencing the entire bacteria: they’re sequencing just enough to be able to identify them (here’s an example paper, which cites several other previous studies: http://aem.asm.org/content/78/12/4217.full).
I think there are a few important things to keep in mind 1) in order for a meaningful transfer to happen, the entire gene has to survive its way into the gut so the resulting transfer can encode a protein. 2) Even if the transfer happens, that’s one out of trillions of bacteria in our gut. Recent studies indicate that the distribution of these bacteria is constantly in flux (there’s even a significant shift in their distribution in day/night http://www.medscape.com/viewarticle/833344). It suggests that they’re constantly dying off and replicating. 3) In order for the bacteria that has taken up a gene to “spread” and to be of concern, the gene/protein that it takes up has to give that bacteria a significant selective advantage. I don’t see how any of the genes we’ve added to GM-crops would give bacteria in our gut a selective advantage. How would a bacteria in our gut benefit from making Bt? Or from developing resistance to the papaya-ringspot virus? It doesn’t really make sense.
Thanks for your review, its really good.
I must notice that the pro and anti anything is always quick to label things such as ‘bad science’ etc. as the first commenter shows. This is NOT the bad science, its just the usual science. You can find flaws in every paper and the reasons are not only due to the bad thinking but can be because of anything – resource limitations for example. That is why other people have to repeat the work.
This has nothing to do with GMO too and show again that people are prone to invalid interpretations when it fits their agenda. The question was simple: can or can’t the full gene cross the intestinal barrier (avoid digestion). It appears here that it can. The work needs to be repeated ofc and to make sure that there isn’t any contamination, but that doesn’t make this work a bad science, only a first step (remember, there were no studies like this one before and it is the usual scenario that first studies about the subject have many places to be improved upon).
Only after there is a clear evidence that gene can cross into circulation, there is a need to discuss a potential effects of that.
To be honest, before reading this study I was sure that this thing must happen in individuals with impaired intestinal barrier (i.e. in gluten sensitive people, IBD etc.) so this is not something that was unexpected.
The last part of this review on the likelyhood of this becoming integrated into cells doesn’t make much sense IMO. There are numerous examples where single mutant cell can cause problems, particularly if it is the stem cell (cancer being the most known one) and examples where highly unlikely situations definitely happen given enough time.
As you point out most studies will have minor flaws because there are always things that you could have done differently, or could have improved if you had more resources. But there are major flaws that can invalidate the findings of a study, which is why the peer review process is so important.
I completely agree with your perspective that even a paper with a major flaw can improve our knowledge of a field: with each attempt to replicate a finding, we learn more about a system. But unfortunately, when papers are on topics that are controversial (such as the safety of GMOs or vaccines), flawed papers do more harm than good. Take for example this paper: http://www.ncbi.nlm.nih.gov/pubmed/21931358 The findings suggest that miRNA from rice can regulate genes in humans. There have been several attempts at replicating the findings and no one has been able to do so. A recent paper suggests that the findings are most likely due to contamination. It is due to such papers and the one I’ve reviewed here (and probably many others) that recent guidelines/recommendations on sequencing low abundant DNA have arisen. But if you read websites that stand against GMOs, you’ll read about the dangers of miRNA and only the original paper will be cited, none of the follow-ups.
To add an additional layer of review to papers on topics of controversy would not be fair to the authors and adds additional burden to an already overloaded review system. So I’m not sure what the solution could be.
You must also acknowledge that in highly controversial landscape everybody has its own agenda and its very hard to follow the truth. Examples are common trough the history (h. pylori discovery for example is a good proof of that, but there are many others including Paulings work on vit. C and cancer where we know now that rebutals were actually flawed and that he was more right then wrong).
There is no good way or rule or politic here IMO. Time will simply tell, as always, and until then, we will have to accept some collateral damage. Democratization of the science is IMO a good thing and journals like PLOS one.
I value that you recognize that even the highly flawed studies can contribute. This is not specific for medicine only, it exists in virtually any field. For instance, some wrong ideas in math are responsible for the chain of events that finished with some crucial discoveries.
Comments are closed.