Method Special: Genome editing
Telling the Future from the Past
by Steven Buckingham, Labtimes 03/2017
Over the past few years, nuclease-based genome editing tools have taken life science labs by storm and given researchers unprecedented control over genomic DNA sequences. But it is not all sunshine in the brave new world of gene editing.
Foto: American Chemical Society
There is no surer way to understand where we are going than to recall where we have been. So come with us in the Lab Times time machine for a trip into the past.
The year is 1987. It is only about ten years since the Sanger Method for DNA sequencing was invented, and a little more than 30 years since the structure of DNA was solved. The human genome project has not yet started – that won’t happen for another three years or so. At the moment, there are no completed gene databases in existence, although preliminary work has begun, or will soon. There are, of course, the sequences of some genes available, but at the moment hardly anyone knows anything about any genome in its entirety.
Of course, even now in 1987, the more imaginative members of our species are at work speculating about what might be, now we have genomes almost within our grasp. “In a few years,” they say, “we will be able to chop and change genes at will, not waiting years for breeding programmes to yield their fruit.“
Even the more sober-minded scientists agree with a condescending nod. “And perhaps, just perhaps, one day we will be able to edit those very genes at any point we want, forcing them to make designer proteins”. Somewhat less enthusiastic agreement from the sober scientists. “And the day will come when this can be done quickly and cheaply, using inexpensive kits available on the marketplace”. Distinct disbelief from the sober scientists. “It can’t be done! Of course, it can’t! It is one thing to cut and paste genes, but the chemical precision needed to alter residues in situ is beyond anything imaginable! How are you going to target such changes? How will you make sure the changes are specific to one site? There are just too many difficulties to overcome. Cheap and convenient genome editing is just plain impossible!”
Now back to the present, and, of course, cheap and convenient DNA editing is not only possible, but it is becoming routine. How did we get here in so short a time? Well, that story has three clear chapters, and those chapters are headed “Zinc fingers”, “TALENs” and “CRISPR”. And like all good stories, each chapter builds upon the foregoing, and yet they all develop a common theme. In this story, that common theme is “cut and repair”. All three methods follow a similar pattern. First, find a way of cutting your DNA at a specific point. Then stand back while the cell repairs the DNA, while fooling the repair machinery into making a little “mistake” by introducing your letter into the code rather than the original one.
Let’s begin with the second part (a trick I learned with detective fiction – read the end first to avoid disappointment). How do these three genome editing methods trick the repair machinery into adding “our” nucleotide? Fortunately, the cell’s inbuilt editor does not have an eye for detail. When it tries to fix the break, it can be fooled into using a primer that does not perfectly match the template. So if we supply just such a primer, it gets incorporated into the DNA and we can then PCR away from there. Imagine helping someone repair a tapestry, and “helpfully” handing them the almost-correct strand of thread.
Now let’s tackle that first part – cutting the DNA at the right spot. Cutting DNA is easy, of course. Just use a nuclease. But the tricky part is making the cut just where you want it and nowhere else. What we really want here is a protein that, say, directs a nuclease to just one site, specified by the DNA sequence.
A lot to ask, but the lucky break came when people started looking closely at the way 5S RNA is stored in immature Xenopus oocytes. It was noticed that these unfertilised eggs contained a large amount of a factor (in fact the first eukaryotic transcription factor to be described) which bound both the RNA and the gene encoding the RNA, and which was required to control the rate of transcription.
This was just what we were looking for – a protein that binds a specific DNA sequence and that does so according to some predictably usable rules of sequence. By looking more closely at the structure, the eponymous “zinc finger” – so called because it holds on to the DNA – was identified. The importance of this was that a new DNA binding motif had been discovered, and that the sequence conferred specificity to a single nucleotide triplet in a predictable and therefore exploitable way. But there was something more: you could also string a number of fingers together, each with its own specificity, to create a binding peptide specific to a given sequence.
One of the first things done with this discovery was to fuse zinc fingers to activators or repressors. It was not long before scientists started fusing them with nucleases to create Zinc Finger Nucleases (ZFNs). By stringing together a number of fingers corresponding to the site you want to change, and fusing it with a nuclease, you can (in principle) combine this with homologous repair to introduce any change you want.
But it turned out that Zinc Finger Nucleases are not, as you may have guessed already, that easy to work with. For a start, there are issues with specificity. And there are no magic formulae for overcoming these problems. More often than not, it is a matter of tedious reagent optimisation, with no guarantee of success. ZFNs are also expensive, because you have to synthesise the finger.
Off-target effects are amongst the major issues of nuclease- based gene editing tools.
So let’s close the chapter on ZFNs, and open the one on Transcription activator-like effector nucleases (TALENs). In many ways, TALENs are a sister method to ZFNs, but with a different origin. It all came, like so many key discoveries, from an unexpected quarter. Researchers were interested in how a type of bacteria (Xanthomonas) was so effective in infecting their plant hosts. This was important because Xanthomonas infections affected the productivity of crop plants like rice, peppers and tomatoes.
It turned out that the bacteria were secreting a protein into the cytoplasm of the plant cells, and this protein was finding its way into the nucleus where it bound to the DNA. This binding was specific, and needed to be so in order to activate and suppress particular genes by copying the host’s transcription factors.
These TALE proteins (transcription activator-like effectors) came in three modular parts: a DNA-targetting domain, another domain that traffics the protein to the nucleus, and finally a transcription domain. OK, so swap the transcription domain for a nuclease and you have something very much like a ZFN – but with one difference: each repeat of the DNA-targeting domain in a TALE corresponds not to a triplet, but to a single nucleotide, giving an unprecedented flexibility in designing the recognition sequence. The rallying cry of “one monomer, one nucleotide” swept from one gene-editing bench to the next, pushing ZFNs to the side and winning Nature Methods’ coveted “method of the year” in 2011.
All is not sunshine for TALENs, though. Once again that old problem, off-site effects, reared its ugly head. Unfortunately, the binding efficiencies of the different DNA-recognising monomers are not the same, so you can sometimes get binding to sites that differ from the target by a few nucleotides. All the same, TALENs opened up a host of new opportunities for gene editing, as constant tweaks were made to protocols and sequence design algorithms.
However, TALENs’ predominance was short-lived, thanks to the rise just two years later of CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats). CRISPR complexes were found originally in very primitive organisms (Archaea), and are thought to be the adaptive immune system of bacteria. A CRISPR has two components: a “guide” RNA that does the job of guiding the molecule to the desired DNA target (like the Zinc Fingers or the TALEN monomers), and a remarkable enzyme called Cas-9 that does the cutting, under the direction of the guide RNA.
The edge that CRISPR gives over ZFNs and TALENs is simple – to change your target all you have to do is to make the appropriate guide RNA. You use the same Cas9 enzyme whatever your target. No rebuilding a new protein every time you want to try a new target. And you don’t have to stop at just one target. What is more, by adding several guide RNAs at the same time, any number of targets can be hit simultaneously. CRISPR isn’t restricted to single base changes either, as you can add quite long lengths of sequence (this is something you can do with TALENs, too, but not as easily).
Now I am sure the more mischievous of our readers will be wondering, “what would happen if the gene I inserted with CRISPR just happened to be a CRISPR construct as well?” The answer would be that the gene would insert itself into the complementary DNA strand and, over successive generations, would ensure its rapid propagation across the entire species. That is called a gene drive, and is considered enough of a viable proposition to support speculation about eliminating pest species and fixing disease-causing mutations throughout a species. Scared? We should be, and the threat is enough to cause the US National Academies of Science to publish a report containing a number of safety recommendations last summer.
Doc Brown's DeLorian could probably tell the future of genome editing. But it's a fair guess that the basic goals have already been reached.
So we can get back into our time machine and go back to 1987 to tell the dreamers they were right, and the sober scientists were wrong. But what if, instead, we went not into the past but into the future, would we indeed see genome editing changing our world?
I think I can imagine what today’s sober scientists will be saying right now. They would remind us that the hype over editing is greater than the reality. Didn’t we say earlier that ZFNs, TALENs and CRISPR all have off-site effects? Let’s look closer at our winning method – is CRISPR all it is made out to be (adjusting, of course, for the unfinished chore of fine-tuning the method and all the optimising that still has to be done)? In fact, one of CRISPR’s most serious drawbacks is an unintended consequence of one of its big selling points – specificity. If the sequence of your target differs from that expected by just one residue, CRISPR will probably fail. And even then, will the cells survive transfection? And have you got the best guide RNA? Or the best donor template (for homologous repair)? CRISPR is facile, but not trivial.
Sadly, CRISPR’s fame and the path it offers to glory has led to controversy. About a year ago, a researcher in China, HanChunyu, claimed he had a better alternative to CRISPR in a gene called NgAgo. More accurate, he said, and more flexible than CRISPR. The problem is that several attempts to reproduce these results by other laboratories have drawn a blank.
But CRISPR is too good a tool to throw away just because it is not perfect. For example, to overcome the limitation that it is too big to pack into a virus, labs are coming up with smaller versions of Cas9, such as Cpf1(Nature Biotechnol. http://dx.doi.org/10.1038/nbt.3609). Others are producing Cas9 variants that don’t just cut, but directly swap residues (Nature 533: 420-24).
Genome editing faces special challenges as it tries to enter the clinic. The biggest problem is getting it into the cells. To date, the payload is usually delivered by viral vectors, but these vectors are often limited in their load-carrying capacity and in their target-specificity, as well as having to negotiate the immune system. There are several impressive proof-of-concept demonstrations of gene editing therapies in animal models, but translation into the clinic is still largely out of reach.
But all that changed this January. London’s Great Ormond Street Hospital announced that a one-year old girl, Layla, while unresponsive to traditional treatment, was cured of acute lymphoblastomic leukaemia. Although clinical trials have been under way in which a patient’s T cells are removed and then reprogrammed using TALENs to express cancer-killing proteins, this approach was not open to Layla because she was so small – she simply didn’t have enough T cells to start with.
So the Great Ormond Street Hospital team took T cells from another donor and used TALENs not only to add the cancer-killing protein, but also to silence the genes that would have marked the donor cells as “foreign” and therefore have triggered Layla’s immune response.
Gene editing tools can also be used as a new way to fight infection. There are encouraging signs that you can not only target the host genome, but also the genomes of viruses that have infected the host. Bryan Cullen’s group at Duke University Medical Center, Durham, USA showed how CRISPR can knock out the dangerous papillomaviruses that cause cervical and other human cancers (J Virol doi: 10.1128/JVI.01879-14), while Kamel Khalili and Wenhui Hu of Temple University School of Medicine, Philadelphia have argued that there are compelling reasons to hope that CRISPR can provide the much sought-after ‘permanent or “sterile” cure’ of HIV-1/AIDS.
Sadly, the Lab Times time machine exists only in our imaginations (like genome editing once did . . . ?). But we at Lab Times are nonetheless prepared to hazard a guess at the future of genome editing. With continued fine-tuning of parameters, the basic goal of genome editing has more or less already been reached. The development of derivatives of Cas9, such as those designed to edit residues directly, will take the fundamental technique to its completion. Related gene editing methods, such as gamma peptide-nucleic acids, which bind to DNA and provoke the cell’s endogenous repair machinery, will add more colours to the ever-growing palette. Even that looming blank wall of a challenge – how to deliver it to the cells – is crumbling.
One of the two last challenges lies with us, the bench scientist. Given such manipulative power over the genome, what are the most informative questions to ask? The other lies with the regulatory community. With gene drives in mind, what dare we to do with it, and what should we forbid? Indeed, what can we forbid?
Last Changed: 28.06.2017