Knockout and tagged gene-editing can create cell lines ideal for antibody validation. Using CRISPR CAS technology, Horizon developed a streamlined process to cut production time. Read more in our interview with Lead Scientist Daniel Lackner
David Shifrin: Welcome to Horizon Discovery, I'm David Shifrin. Today I'm speaking with Dr. Daniel Lackner, a molecular biologist and team leader at Horizon Discovery. Dr. Lackner has an extensive background in molecular techniques and gene editing which he's using to develop some of the technologies that go into Horizon's cell lines.
In the fall of 2015 Dr. Lackner was the first author on a Nature Communications paper that described the really remarkable new application of CRISPR for gene tagging. That system, as I said, kind of underlies the new Horizon reporter cell lines, and we're going to be talking about both the science behind the system and how it fits in with a really important effort going on right now within the scientific community.
Dr. Lackner, how are you today? Thanks so much for your time.
Daniel Lackner: I'm fine. You're welcome.
David Shifrin: Today, as I said, I want to talk about a couple of things, including the technology behind the new Horizon reporter cell lines that you've helped develop. First I want to talk about how those efforts and those products fit in with current projects towards improving reproducibility and better validation of reagents that we're starting to see more and more within the scientific community.
I guess just to start, let's go big picture to make sure everyone's on the same page. If you could take a moment to describe the new cell lines that Horizon has developed for tagged protein expression.
Daniel Lackner: These new cell lines, basically, were discovered out of the need to have endogenous reporter genes. Basically there is many ways how you could basically generate the reporter gene. You can overexpress a gene of interest; however, all this comes with a shortcoming of overexpressing artificially your gene of interest. We wanted to create something that basically allows you to tag proteins, basically add the molecular tag to a gene at it's own genomic locations.
Out of this … and obviously in the company we also thought we want to make this in a way that's scalable and can be used at many genomic loci without a massive time effort and cost effort. We came up with this new technology where we introduce a tag just by a process called non-homologous end joining. Basically we cut in the gene of interest and we could describe it almost as the fact that this tag jumps in. There's no homology donors needed so basically this tag that you can insert can be used for any gene you would like to tag without the need to create a specific reagent for every gene/protein you want to tag.
That's how ... it was the need of this technology to be developed. The idea was that a big advantage of this is that you have basically, let's say the natural environment, the natural promoters, enhancers of the gene so the expression of the gene, so basically the amount that you have the protein in the cell in the end really reflects it's naturally occurring amount and is not driven higher artificially.
David Shifrin: That's really important for this quality control issue that we're seeing in the scientific community, right? That's where I want to go next before we get into the really deep scientific details. The ability to express a tagged protein of interest at endogenous levels is critical to be able to compare and make sure that what you're seeing is real, right? One of the hot topics right now, a lot of the company and organizations are highlighting the need for better antibody validation and so I think what Horizon is doing is really important for that project.
I guess by way of background for those listening, AbCam has a white paper in which they start out in the introduction by saying, "We're currently facing a problem with reproducibility in science. About 70% of scientists have failed to reproduce another scientist’s experiments and over 50% have failed to reproduce experiments of their own design." Of course that's not just because of antibodies or any one reagent or tool, but it's certainly part of it. I guess in the context of what Horizon is doing, could you talk about this problem and some of the efforts that are underway in the community at large to change things?
Daniel Lackner: Yes, and I think this is a very important problem, especially as you can see there's only, let's say a few thousand genes that are studied in depth. However, you can basically buy antibodies for almost all genes that occur in the human genome because they're somehow easy to generate. However the quality control behind that is not always up to par for the actual product. There's many antibodies out there that seem to be ... recognize a protein of the same size. We have a tool in hand which is our massive collection of knockout cell lines, especially the HAP1 cell lines which we can talk a bit about later, basically that really allows you to have this pair of cell lines, wild-type cell line, and a cell line that does not express your gene of interest anymore. Then basically you can use this pair to really test drive any antibody.
What you want to see, that in a normal screen scenario, you basically can detect a protein and as soon as you don't have protein anymore because it's genetically knocked out in our cell line, you don't see the corresponding bands by western blot.
David Shifrin: Great. Could you go into a little bit more detail about that? You mentioned things like knock down and using that to really get kind of a binary test on the quality of antibody and some other validation methods that are discussed in the white paper like knockout, mass spec, western blotting, protein arrays. Is there anything else as far as endogenous expression within transgenic cell lines that kind of ties into all of this?
Daniel Lackner: I think the main power, is really the high, the big portfolio, or let's say the big collection of knockout cell lines that we have. We have used CRISPR Cas-9 to modify our catalogue. I think we have more than 2,000 knockouts, or 2,000 genes in our freezer, and basically they can be used to validate antibodies because in this case you really go in with CRISPR Cas-9 and really delete the gene. There's no, not like previous methods where you basically can block an epitope or do a knock down with siRNA, but you really genetically modify the cell where it does not express the gene of interest anymore. Basically you can use the cell line then to show that an antibody does not recognize a protein anymore because it does not exist anymore in the cell. I think this is one of the most powerful tools when antibodies are developed to show that it really recognizes its intended target.
David Shifrin: Right, to dramatically reduce the number of false positives that we've all seen in cell biology.
Daniel Lackner: Yes. I agree and again, the problem was that there's unstudied cell lines. Obviously if you have a gene that's very well studied then you know where it localizes and you know the size, etc. Probably there's a lot of antibodies out there that have been used highly by the community and then basically the community itself has a kind of, let's say, correcting element to it. However now people find these screens or whatever feature or through literature searches new targets they want to explore, and you can be sure you can find antibodies for it, but if it's a good one you don't know. Basically having these knockout cell lines from Horizon really gives you the opportunity to really test this completely efficiently with the primary yes/no approach and see if this antibody is basically recognizing your target.
David Shifrin: All right, Dr. Lackner, before we get into the real depth of the science as far as the CRISPR system that you're using for these tagged cell lines, one thing is the cell line that you're using, and these are a proprietary cell line and have some unique feature so if you could, tell us a little bit about the HAP1 cells that you've been working with.
Attributes and characterization:
- Species: Human
- Growth: Adherent
- Morphology: Fibroblast-like
- Derived from: Male chronic myelogenous leukaemia cell line
Daniel Lackner: The HAP1 cells, basically HAP comes from haploid, so it's all in the name. The great advantage from a genome engineering point of view is that only have one copy of the whole genome information. Rather than having two chromosomes of each like a diploid cell they only have one chromosome. That basically allows us really in a very precise way generate mutations. If it's knockout or generate tagged cell lines and we can then also immediately have a very nice read out. We can easily genotype and find out where it's mutation is, what we have done and if we have done what we wanted to do, and on the other hand also it's very effective, right? Because in a diploid cell you might have a scenario where you generate a mutation and it's only in one allele in one chromosome whereas the second one is not harmed and you still have a functional protein, whereas in a haploid cell line and the HAP1s as soon as we generate a mutation we basically have, in most cases destroyed the gene in the protein.
The same goes for the tagged cell lines. I think the great advantage here is the fact that once you have integrated the tag and you have generated your tagged gene or in the end your tagged protein, 100% of the protein in the cell will have the tag. There's not a second allele that does not contain the tag and I think that it's a strong advantage for many applications.
David Shifrin: Okay. As researchers that are listening to this or reading up on the cell line on the Horizon website or thinking about possibly purchasing one of these, that's something to really be aware of. Any other considerations as to where the haploid cells are useful and where people may actually want to stay away from them or they might want to consider another option?
Daniel Lackner: I think the thing is always like, it's a new thing and people are always afraid of new things, even in the scientific community, which is sometimes of a surprise to us. But having said that, I think it's like any other cell line. It would be a great model for certain things and, for example we know for things like DNA damage and autophagy and imaging and many pathways that people generally work on, it's a really great model system where you can do affordable genome engineering and have a great collection of knockout cell lines. There would be, obviously, very specific applications like, obviously you would not be able to do immunology in the cell line.
I really think it depends on your approach and your question, but I think that there is nothing that is very specific a problem with the HAP1 cells. Like any other model cell system, it's very useful for a broad range of basic and academic research. It's also used by a lot of our biotech and large company partners to validate screens and validate knockout and you just have to know if it's available for you or for your kind of research. If it's something that really requires a specialized immune cell or neuronal cell, so obviously then it would be a problem, but generally it's just as any other immortalized cell that's used in academic labs.
David Shifrin: If people have questions about that, they can always get in touch with the Horizon team though the website. So, you said something a moment ago that I think is a perfect transition into kind of the scientific meat of what we're talking about here. You mention, I think it was 2,000 genes, you know, really kind of high throughput. You've been able to do this on number of different ... at a number of different loci using the CRISPR system. The basic idea, going back to that paper in Nature Communications that you published, and other work, is that the method you've developed is highly specific which is an obvious feature of CRISPR Cas-9 systems, but at the same time it does not require long specific adapter arms which add a lot of the time and cost. If you would, explain some of these problems, the existing problems with methods that are common on the marketplace right now for gene tagging, and then we'll get into how the Horizon solution you've developed solves that.
Daniel Lackner: Yes. Basically, as everybody knows these days, CRISPR Cas-9, it's a tool that basically hit the world of biology with, I would say, a big bang. It's a technology that really ... however they use it works. Basically, using it as the standard method to basically program these nucleases to cut anywhere in the genome allows you easily to make knockouts because these cuts are not always repaired perfectly and in this case insert insertions or deletions. However this technology has been used also for other approaches, for example, of tagging of genes. In this case what you usually would do is you would induce the break, the double strand break through CRISPR Cas-9, and then basically you would also rather than having non-homologous end joining, it's a repair pathway in the cell but basically the just-disrupted ends of the DNA are ligated together. With most cells, if you give it a repair donor, that means an uncut, let's say, healthy, non-destroyed version of the gene, then this will be used for repair.
If you do this, you basically can integrate other sequences, in this case, a tag, in your genome. This approach is a little bit cumbersome basically because, especially what you had to do, you had to make long homology arms. Basically if you wanted to do a GFP tag you had to basically clone your GFP tag and you have to clone a few ... up to kilobases or 100 base pairs of homology arms corresponding to a gene and hope that you get integration by homology accommodation.
David Shifrin: You have to have those long adapter arms simply to make sure that the target drops in where you want it to, correct?
Daniel Lackner: Exactly. Basically if they had homology arms it was a gene of interest.
David Shifrin: Okay, so you can't just do a handful of bases on either side, like you might for a PCR primer and make sure it works. You have to have a huge stretch to fit it in?
Daniel Lackner: That used to be the standard, yes. I mean, people try all kinds of variations now, but the technology that we have basically tried, and it's based on some reports, initially in zebrafish, was basically maybe we can try without any homology arms. The idea was, there's one pathway on how the cell basically repairs these cuts, so what happens if we introduce a very precise cut in the genome? In the gene of interest we wanted to tag with CRISPR Cas-9 and in the same way give it a lot of linear integration donor, whatever the sequences I want to integrate, let's say GFP or turbo GFP in our scenario, and see if by chance we get some of these integrated.
The real main advantage there is that you basically could try in parallel for hundreds of genes or thousands of genes without needing to change or remake an actual donor for each of these genes, but they can use the same donor for all of the genes. We had a lot of trial and error going into this because we tried PCR products or synthesized donors to find out that it works when you basically add your tag all of a sudden they form a plasmid that itself is cut by CRISPR Cas-9. You basically have a plasmid where CRISPR Cas-9 cuts out your tag and stays in a linear form, at the same time you cut your donor, you cut the genomic DNA, and then somehow this tag is basically integrated by this repair pathway called non-homologous end joining in a certain percentage of cases.
David Shifrin: It's really amazing that this works as well as it does, because you’re getting very specific insertion of a tag at a very precise location without having to make custom plasmids each time. You're also, what I found really remarkable, is that you're dropping a lot of exogenous material, in the context of the Cas and the different plasmids and there's a lot of cutting going on and yet it's such a precise system that it all comes together and doesn't overwhelm the cell. You actually get the integration that you want.
Daniel Lackner: Yes. I mean, obviously you have to basically play around a bit with how much Cas-9 and how much guide RNA, how much of the donor plasmid to basically put in, but yes it worked. I have to say this is one of the first experiments I did when I started in the company like two years ago and I would have said it basically falls under the category of slightly crazy ideas. We did it as a first test without optimizing a lot and over thinking it in a first kind of test, and I think it's actually even the first figure of the paper. I think in almost, I would say 80 or 90% of the loci really worked. It was really, really remarkable.
David Shifrin: Yeah, I'm looking at that figure now, the blot that you show, or I guess it's the PCR gels that you show.
Daniel Lackner: Yeah.
David Shifrin: You've got 11 out of 12 worked remarkably well.
Daniel Lackner: Exactly.
David Shifrin: The other thing that's quite impressive about this system is the efficiency of it. This is another thing where you kind of reference in the paper almost your amazement at a bit of luck where you kind of expected to have to screen a huge number of clones and remarkably it was quite efficient. You don't need to screen a large number of clones, which in addition to not having to take the time to design these long adapters, you know, anytime you can save time growing cells and screening them that's a huge benefit as well. What does that look like? How does the system work to be so efficient?
Daniel Lackner: I mean, I have to say I think we were actually a little bit on the lucky side and since then we have done many, many, many additional experiments at additional loci and this efficiency is a little bit loci specific and guide RNA specific. However, I think that the most important thing that helped us really to get great efficiency, especially for the Turbo GFP tagged cell lines is obviously the fact that we can sort and enrich by FACS. Basically you can transfect your cells with the cassette and because the plasmid and the cassette itself will not be expressed or translated, you will only get a GFP signal once you have the correct integration and the right reading frame in your cell. That basically allows you to select cells that are green and then you basically only take these cells and screen these for the integration events. The efficiency overall basically can be really improved using this method.
David Shifrin: Great. With all of this that you've learned over the last couple of years working on this system ... I mean, I guess this is a bit of an aside, but what are some of the bigger lessons that you've learned that scientists outside of the company could use to apply to their own experiments? Obviously, you've got a lot of proprietary things going on there within Horizon, but just these experimental setups that you've used to make an effective system, anything that's worth taking home for other scientists listening?
Daniel Lackner: I think an important lesson could be the fact like, we've tried this initially on a few loci and showed proof of principle to see if it works, and obviously then the robustness or the fine tuning of the system basically happens once you tried for 50 different or 100 different genes, right? Because then you will also find not only the easy parts, but also you will find some of the more trickier proteins or genes that are out there. Basically that will give you ... which obviously is not going to happen in academic research right? Nobody that shows that something works for five genes will go out and then test it for another hundred genes. Whereas we obviously, trying to produce a collection of cell lines, we'd want to do this. I think that that's an important lesson, that basically things can change and you still keep learning and improving the procedure by doing that.
David Shifrin: Okay. Along kind of those same lines of continually learning and improving, and you mentioned a moment ago that you've been using this system to insert turbo GFP and that's working quite well in terms of the Horizon cells. One question I have is how consistent is expression of turbo GFP? You mentioned that you're sorting to kind of pick expression levels that you want, but are you seeing a lot of variability that could confound imaging assays or quantification or whatever and if so, what are you doing to mitigate that?
Daniel Lackner: I think the turbo GFP is not a problem there. I mean, this is a very well ... stable and established fluorophore. I think the problem that you have there is not a problem with the biology, the truth is that it depends on which protein you want to tag, right? I mean, we have made a big case about the fact that we are tagging genes and they're on an endogenous loci, which basically gives it the really strong advantage of not seeing artefacts of over expression.
Of course the down side there is that some proteins now would only be expressed at low levels. Obviously, but this is basically the truth that is reflected within a cell. For certain screening procedures, etc., I mean, you would have to basically be aware of this fact. You can also take genes that are not expressed and can only be induced in a certain scenario, but obviously you will have a GFP-tagged cell line and you will look at it and in a steady state it's not going to be green. You have to be also aware of this for what you want to look for and what you're going to need the cell line for.
Obviously most highly expressed genes would give you a very nice signal, but then obviously you have certain genes are lowly expressed, then you will have only one GFP tag in the endogenous location and you will only have as much protein as is naturally occurring in a cell. Which is great for being close to the real life scenario, if you could call it so, but it's something people have to be aware of when using the cells.
David Shifrin: That goes back to the discussion around antibody validation as well, where you want something that's going to be as close to endogenous as possible, but also just be aware of variability in expression, whether that protein is naturally expressed at high levels or not, because you want to be able to compare the signal coming off the antibody with whatever you're seeing from the GFP tag or whatever it happens to be that you've inserted in order to get that good comparison.
Daniel Lackner: Yes, I mean, because we're talking about the antibody validation with the GFP expression, just to make this clear again, most of the antibody validation really happens with the knockout cell lines. I think the GFP tagged cell lines are really, let's say, an additional tool you could use to basically have a second line of testing that an antibody recognizes something and it goes to the right location. I would say this. But most of the current efforts to validate antibodies really is happening with the knockout cells where you really take out the protein of interest.
David Shifrin: Dr. Lackner, you and Horizon have been using this method to, as we've been discussing, putting the NanoLuc and the turbo GFP cassettes into cells, but because it could be used for any sequence, are there any other applications that either you're looking at or that other cell biologists are kind of playing around with that you think are exciting?
Daniel Lackner: In theory you can put any sequence of interest wherever you want. This is the idea. I mean, we haven't tested if you can put even larger… there's probably a certain amount where you go to a size that we'll find it harder to integrate because there's some indications ... not from our results, but from other tests that the shorter the better. But in theory, you could put enhancers, you could start, you know, taking out enhancers or promoters, switching them around. We had a customer who was interested in this paper and was thinking about markers for RNA or aptamers. Anything you want, in terms of adding a sequence anywhere in a genome where you have a good guide RNA and I think, you know, it's possible.
David Shifrin: Excellent. As we wrap up here, from your standpoint as a scientist, just the attitude of discovery and obviously you're working in a company that has a direction and goals and limited resources, but you know, just kind of a perfect world, where would you as a scientist want to see this tagged protein technology and these new adaptations of the CRISPR Cas-9 system go in the next few years?
Daniel Lackner: I mean, one of the things actually I'm personally, as a scientist, I'm very interested in is unknown genes. The unknown genes I think, with the CRISPR Cas-9 and also with the gene tagging, you have a really great tool to systematically integrate this set of genes that is not very well studied. Basically there still, you know, if you look at the amount of genes that are out there and the amount of genes that are very well placed in terms of function and localization, there is still a lot that are basically ... it is unknown what to do. I think the CRISPR Cas-9 system by generating knockout cell line allows you to basically look at the essentiality of the gene in the system. On the other hand, with the tagging approach, you could basically tag a whole lot of unknown genes and find where they go in the cell which also, together, this information could really give you a clue about a lot of these unknown genes and how to they tie in and how they function in the context of the cell. I think, for me as a scientist this is something that is very interesting, because it really allows you to systematically interrogate a proportion of genes that is not very often and very well studied.
David Shifrin: Excellent. Where can people go to learn more about everything that we've talked about today?
Daniel Lackner: I think a very good portal from our website is basically if you go to Horizon Discovery, that's from our company, and I think there you can basically, in this case it would be click through the Horizon cell lines, the news and literature section, and then basically you can find out about these cell lines. Also I think today it just came out in the top 10 innovations in The Scientist, The Turbo GFP tagged cell lines. You can also go there and have a read if you're interested in what's going on. Then obviously go to the original white paper describing our letter.
David Shifrin: All right, congratulations on it. That's an impressive accolade for you.
Daniel Lackner: Thanks.
David Shifrin: Daniel Lackner, thanks so much for your time. Again, congratulations and we look forward to seeing what work you continue to put out with Horizon in the future.
Daniel Lackner: You're welcome. Thanks a lot for taking the time to talk with me.
For further information about Horizon's tagged cell lines click the link below: