6 ms·

People who work in or near genetic engineering: CRISPR seems to solve the problem of editing DNA to contain arbitrary human-designed data. Are there major cave

by gradys 7y ago

People who work in or near genetic engineering: CRISPR seems to solve the problem of editing DNA to contain arbitrary human-designed data. Are there major caveats or limitations to this that people outside the field don't recognize? e.g. is delivery to the right cells a major obstacle? If editing is solved, it seems like the major remaining problem is designing the right data to insert. I'd imagine this is a much vaster problem than editing. What are the major subproblems there? What does the frontier look like? Besides the mechanics of editing and designing the sequences to insert, what other major problems stand in the way of sci-fi level genetic engineering?

aaavl2821 7y ago
I'm not a scientist but have friends who do research in this field and I've worked at startups in this field. Delivery is a major obstacle. DNA, RNA and proteins don't typically get into cells on their own. Typically you use vectors, often viruses or lipid nanoparticles, to get DNA into cells, but these vectors have their own challenges. One of which is that they aren't typically very specific to particular cell types. It can also be tough to get vectors to the right tissue. So most early gene therapy efforts focus on blood disease, liver disease and eye disease because it is easier to deliver to those areas. CNS has also become a popular target because a particular viral vector, AAV9, tends to get into neurons pretty well But viral vectors are commonly "immunogenic", ie the immune system learns to reject them as foreign after one dose. These are viruses after all, and the body is designed to reject them. Some people have pre-existing antibodies to many popular vectors. And often you can only dose a viral vector once, so if you don't dose it right, or if the effect isn't permanent, you don't get another shot I can't speak to some of the other more technical problems, but one limitation is that you are limited as to how large your DNA payload can be. But it seems this paper is a big step towards removing that barrier. Which I believe is why this paper is such a big deal (others more knowledgable should correct me) Also CRISPR works best to cut DNA as of now (which renders a gene non-functional), but it is a bit harder to insert DNA reliably from what I understand. Again, a scientist in this field would much better understand the state of the art than I do, but this is my understanding Another major unknown is the degree to which gene editing tech causes off-target edits. If your DNA editing tool accidentally snips a gene that protects you from getting cancer, it can lead to cancer. I'm not close enough to the science to know what current thinking is on this risk or how to best mitigate it, but it is very real. More primitive gene therapies did in fact cause cancer in patients (often children) In many ways sci-fi level genetic engineering is possible. We just don't do it because we don't know the risks, and the risks are huge. We can already genetically modify human embryos, and our toolkit for doing so grows every day. Editing human embryos is a very scary proposition. EDIT: I will also add that if you are interested in learning about this field, try reading / struggling through a couple scientific papers, really trying to understand every detail. Ideally with the help of a scientist friend. It is time consuming and daunting to get through all of the jargon, but the papers lay out the design, engineering and testing process in some detail. Often they provide the derivation of the mathematical models used and the specific DNA sequences used. You can begin to appreciate how amazing this work is when you get into these details
- cauthon 7y ago
  This is a good summary, and I can follow up on a couple points. (I don’t have direct experience with Crispr data, but I’m currently a PhD student in a computational genomics program, and have 4 years of experience in a lab that does crispr experiments in addition to my projects.) Related to both the idea of inserting DNA and off target effects - the issue is that scientists can engineer a cut, but then rely on the cell’s repair mechanisms to “stitch” the cut DNA back together. These mechanisms are inherently stochastic and error prone and are how many somatic mutations, such as those in cancer cells, arise. So for off target effects, even if you make the cut at the right place, you can still end up with new unintended SNPs and indels nearby. And for inserting new DNA, it’s not a guarantee. You can provide the template that you want to incorporate, but it might get missed, or get copied in more than once or in the wrong orientation.
  - subroutine 7y ago
    IMO the biggest concern is not whether CRISPR-Cas9, and DNA ligase, will do their job reliably (given appropriate guide RNA, and methodology)... the biggest concern to me is that we don't know what else the target SNP tends to do, aside from whatever we've found from GWAS. Here's an example... APOE e4 is associated with Alzheimer's disease risk. People in Guatemala have a high prevalence of this allele. Some researchers from the US might decide it's a good idea to fly down to Guatemala and launch a CRISPR clinical trial to 'protect' newborn Guatemalans from this increased risk of Alzheimer's. So they do; come back to the US, have a toast to longevity for these children. It has recently come to light however, that APOE e4 confers protection against Malaria parasites. Not a big deal if you live in Norway - huge deal if you live in Guatemala. Suddenly, the risk of developing an age related dementia doesn't seem all that pertinent. So, my 2 cents is that, if we're really going to start CRISPRing babies, we'd better be doing our due diligence, and for now limit to diseases that significantly and immediately impair wellbeing.
- gradys 7y ago
  Thanks for the excellent and detailed reply! Are there any papers you recommend im particular? One of the harder parts of getting into a new domain of science is that it's hard to tell the good/important papers from the others.
COGlory 7y ago
CRISPR biologist here, albeit in far more interested in CRISPR as a microbial immune system, and the bulk of my work has been examining CRISPR-associated Rossman fold proteins, not Type 2/Cas9 that is used for genetic engineering. However, I still have a fairly large working knowledge of the field and I guess I could be considered something of a tangential expert, at least compared to journalists and lay people. In short, there are huge limitations. The first issue is how to edit people reliably. Cas9 simply isn't accurate nearly enough of the time. It's not a system that could be employed in a living human being without risking mistarget effects. How severe those would be is something that still isn't known, and would be difficult to model because where you miss would likely depend on the target gene. There's a few thought processes on how to do it more safely. One is to make the edits in lab, verify the changes, propagate them. This is by far the safest route, but leaves other issues. How do you propagate the changes? We don't know how to propagate the changes to an adult human whatsoever. Stem cells get floated but that's still an underexplored field so it's difficult to say what's going to happen, or if it's even possible. Regarding that to put in, that's the easy part. We've been designing sequences for artificial proteins for decades. Researchers at CalTech just published a paper last November where they designed a system of viral proteases (proteins that chop up other proteins) in such a way that they could perform logic circuits in a living cell, and do things like kill the cell if it developed cancer. The first things to come will be targeted drug delivery, biomarkers, and cancer prevention. However we are still a long way from propagating changes throughout an adult human, if it's even even possible with CRISPR.

by gradys 7y ago

aaavl2821 7y ago

I'm not a scientist but have friends who do research in this field and I've worked at startups in this field. Delivery is a major obstacle. DNA, RNA and proteins don't typically get into cells on their own. Typically you use vectors, often viruses or lipid nanoparticles, to get DNA into cells, but these vectors have their own challenges. One of which is that they aren't typically very specific to particular cell types. It can also be tough to get vectors to the right tissue. So most early gene therapy efforts focus on blood disease, liver disease and eye disease because it is easier to deliver to those areas. CNS has also become a popular target because a particular viral vector, AAV9, tends to get into neurons pretty well But viral vectors are commonly "immunogenic", ie the immune system learns to reject them as foreign after one dose. These are viruses after all, and the body is designed to reject them. Some people have pre-existing antibodies to many popular vectors. And often you can only dose a viral vector once, so if you don't dose it right, or if the effect isn't permanent, you don't get another shot I can't speak to some of the other more technical problems, but one limitation is that you are limited as to how large your DNA payload can be. But it seems this paper is a big step towards removing that barrier. Which I believe is why this paper is such a big deal (others more knowledgable should correct me) Also CRISPR works best to cut DNA as of now (which renders a gene non-functional), but it is a bit harder to insert DNA reliably from what I understand. Again, a scientist in this field would much better understand the state of the art than I do, but this is my understanding Another major unknown is the degree to which gene editing tech causes off-target edits. If your DNA editing tool accidentally snips a gene that protects you from getting cancer, it can lead to cancer. I'm not close enough to the science to know what current thinking is on this risk or how to best mitigate it, but it is very real. More primitive gene therapies did in fact cause cancer in patients (often children) In many ways sci-fi level genetic engineering is possible. We just don't do it because we don't know the risks, and the risks are huge. We can already genetically modify human embryos, and our toolkit for doing so grows every day. Editing human embryos is a very scary proposition. EDIT: I will also add that if you are interested in learning about this field, try reading / struggling through a couple scientific papers, really trying to understand every detail. Ideally with the help of a scientist friend. It is time consuming and daunting to get through all of the jargon, but the papers lay out the design, engineering and testing process in some detail. Often they provide the derivation of the mathematical models used and the specific DNA sequences used. You can begin to appreciate how amazing this work is when you get into these details

cauthon 7y ago
This is a good summary, and I can follow up on a couple points. (I don’t have direct experience with Crispr data, but I’m currently a PhD student in a computational genomics program, and have 4 years of experience in a lab that does crispr experiments in addition to my projects.) Related to both the idea of inserting DNA and off target effects - the issue is that scientists can engineer a cut, but then rely on the cell’s repair mechanisms to “stitch” the cut DNA back together. These mechanisms are inherently stochastic and error prone and are how many somatic mutations, such as those in cancer cells, arise. So for off target effects, even if you make the cut at the right place, you can still end up with new unintended SNPs and indels nearby. And for inserting new DNA, it’s not a guarantee. You can provide the template that you want to incorporate, but it might get missed, or get copied in more than once or in the wrong orientation.
- subroutine 7y ago
  IMO the biggest concern is not whether CRISPR-Cas9, and DNA ligase, will do their job reliably (given appropriate guide RNA, and methodology)... the biggest concern to me is that we don't know what else the target SNP tends to do, aside from whatever we've found from GWAS. Here's an example... APOE e4 is associated with Alzheimer's disease risk. People in Guatemala have a high prevalence of this allele. Some researchers from the US might decide it's a good idea to fly down to Guatemala and launch a CRISPR clinical trial to 'protect' newborn Guatemalans from this increased risk of Alzheimer's. So they do; come back to the US, have a toast to longevity for these children. It has recently come to light however, that APOE e4 confers protection against Malaria parasites. Not a big deal if you live in Norway - huge deal if you live in Guatemala. Suddenly, the risk of developing an age related dementia doesn't seem all that pertinent. So, my 2 cents is that, if we're really going to start CRISPRing babies, we'd better be doing our due diligence, and for now limit to diseases that significantly and immediately impair wellbeing.
gradys 7y ago
Thanks for the excellent and detailed reply! Are there any papers you recommend im particular? One of the harder parts of getting into a new domain of science is that it's hard to tell the good/important papers from the others.

COGlory 7y ago

CRISPR biologist here, albeit in far more interested in CRISPR as a microbial immune system, and the bulk of my work has been examining CRISPR-associated Rossman fold proteins, not Type 2/Cas9 that is used for genetic engineering. However, I still have a fairly large working knowledge of the field and I guess I could be considered something of a tangential expert, at least compared to journalists and lay people. In short, there are huge limitations. The first issue is how to edit people reliably. Cas9 simply isn't accurate nearly enough of the time. It's not a system that could be employed in a living human being without risking mistarget effects. How severe those would be is something that still isn't known, and would be difficult to model because where you miss would likely depend on the target gene. There's a few thought processes on how to do it more safely. One is to make the edits in lab, verify the changes, propagate them. This is by far the safest route, but leaves other issues. How do you propagate the changes? We don't know how to propagate the changes to an adult human whatsoever. Stem cells get floated but that's still an underexplored field so it's difficult to say what's going to happen, or if it's even possible. Regarding that to put in, that's the easy part. We've been designing sequences for artificial proteins for decades. Researchers at CalTech just published a paper last November where they designed a system of viral proteases (proteins that chop up other proteins) in such a way that they could perform logic circuits in a living cell, and do things like kill the cell if it developed cancer. The first things to come will be targeted drug delivery, biomarkers, and cancer prevention. However we are still a long way from propagating changes throughout an adult human, if it's even even possible with CRISPR.