A couple of weeks ago we had a paper published in the journal Scientific Reports. This was an important paper for me personally for a couple of reasons, but it’s also a hugely important paper for the field of biology (I know – I would say that). The project also ended up as quite a nice example of what working in research can be like – first it was exciting, then it was soul-destroyingly horrible, then it got exciting again and we made a discovery (which is the whole point after all). I should add here that this was done with a huge amount of help from our wonderful transgenics facility at the University of Manchester, namely Antony Adamson and Neil Humphreys. I must also admit that I stole the title from Antony too!
So what did we do? In short we showed that CRISPR, a fairly new technique which is being hailed as ‘the best thing since sliced bread’ and even now being used in actual humans (which is scarily rapid development), can lead to some crazy unexpected effects and should be used with more care in the future.
That’s the concise story. The real story is far more complex and far more interesting, but it’ll need a bit of background. Most of that background can be summarised in this blog I wrote on a previous paper. There, we designed new compounds to limit the potentially dangerous release of a cytokine called interleukin-1 beta. But this time we looked slightly differently…
Twins but different
The IL-1 family, including IL-1β, is a group of immune proteins that are involved in diseases such as stroke, Alzheimer ’s disease and atherosclerosis. IL-1β is not alone in the IL-1 family; it has a twin brother/sister sitting right next door on chromosome 2 called IL-1α. IL-1α is extremely similar to IL-1β. They are both critically involved in inflammation and fever and they both get produced as a large form then chopped down to an active ‘mature’ form. Surprisingly, however, they seem to have quite different functions inside the body. This is particularly unexpected to us because both proteins signal through the same receptor and they most likely arose as a result of what’s called a gene duplication (essentially IL-1β cloned itself around 200 million years ago – watch this space for more info on that!).
If you glanced briefly at the IL-1 twins alpha and beta you’d think they are basically the same, but you’d be very wrong. Being a twin myself I can certainly attest to this misconception! One of the best ways to understand the differences between twins is to look closer. If you looked closer at my sister and I you’d see that… well I think there’s a pretty obvious difference. Concomitantly, if you look closer at IL-1α and IL-1β you also begin to see a difference. Looking at a molecular level (approx. a millionth of a millimetre) you’d see that, whilst IL-1β has very little of much significance in its front half, IL-1α possesses a very interesting structure called a nuclear localisation sequence (NLS). This is important because it means that IL-1α lives in the nuclear compartment of the cell (along with all the DNA etc.) whilst IL-1β lives mainly in the cytoplasm of the cell (the bit that isn’t the nucleus or the major other compartments). WE DON’T KNOW WHY THIS IS! Why would two proteins that seem so similar live in completely different parts of the cell? And why would a protein whose main job is to be secreted from cells to stimulate an immune response hide away in the nucleus?? It just doesn’t make any sense! There have been some attempts to find an explanation but this is has all been done in simple models – we need to find out why this nuclear localisation is important for IL-1α in a more complex model. We need to make a version of IL-1α that is identical to every way to the normal version but without the NLS. For that, we need to use CRSIPR.
CRISPR than a winter’s morning
Historically, genetically modifying proteins in an organism has been really challenging. Previous techniques were hampered by long wait times, regular instances of off-target effects and high failure rates. But that all changed in 2012/13 when labs in Lithuania, California and Massachusetts showed that a bacterial immune system can be harnessed to perform quick, clean and relatively cheap genome edits in almost any cell type you desire. The story behind the discovery of CRISPR (which stands for ‘clustered regularly interspaced palindromic repeats’) is more fascinating than I have time for and features a Spanish salt marsh, Saddam Hussein & the French dairy industry (yeah, seriously) so for now I’ll have to make do with the basics.
CRISPR was initially discovered as a type of adaptive immunity employed by some bacterial strains to allow them to fight infection from viruses called phage. I say adaptive immunity because these bacterial strains developed immunity to certain viruses by storing information about them, similar to how human T and B-cells recognise and remember signatures from invading pathogens. It’s not dissimilar from your local police storing the fingerprint of a previous criminal. CRISPR works because the host bacteria can store short sequences of the DNA of an invading virus within their own DNA. Now, when the virus returns, it can recognise this DNA and chop it up with a pair of molecular scissors called Cas9. This process has developed over millions of years of evolution and is highly specific and efficient. For that reason it has been harnessed by labs across the world to chop up and inactivate the sequence of any DNA you desire. All you have to do is cook up your own version of the CRISPR-Cas9 system. Here’s the recipe:
- Look up the gene you want to cut and find a part of it with a ‘protospacer adjacent motif’ or PAM site. This will help Cas9 do its thing and is easier than you think to find!
- Design and build guideRNA sequences (sg-RNAs) specific to the DNA of the gene you want to cut. This needs to contain a ‘scaffold’ region which will recruit the molecular scissors Cas9 and a ‘spacer’ region specific to your gene. This sounds hard but in reality you can do it all online relatively cheaply and easily!
- Add the Cas9 enzyme either by injecting it directly into your cell or by giving the cell the information to make its own!
And voila! You gone and done a CRISPR! This recipe will lead to a big old chop in your DNA sequence of choice. When this happens the cell will try and repair it but usually makes a right old pig’s ear of it and you get complete loss of function of your gene! Bingo!
“But Mike you didn’t want to lose your whole gene you just wanted to make a small edit of the NLS!”. Ah yes avid reader you’re quite right! The whole gene delete wasn’t what we were after, so we employed a slightly different approach. Added into the recipe above we also included a section of DNA that we made ourselves. This ‘repair template’ contained the normal version of IL-1α but with two changes made in the NLS. Now all we had to do was cut the IL-1α gene in two places either side of the NLS, add in our predesigned ‘repair template’ and we should have our mutant IL-1α with a non-functional NLS!
But there’s one problem. We are only making changes to NLS but to do this we are making two cuts either side of the NLS with sg-RNAs that recognise the normal, unedited sequence. That means that, once our edited repair template gets incorportated into the DNA, we’re going to run the risk of chopping up our new mutant version just as fast as we can we chop up the old! One way round this is including what’s called ‘shield mutations’ in our repair template. These should prevent the CRIPR Cas9 system from chopping up our new version and make sure it’s only the old version that is cut. So after doing all this, our CRISPR ingredients were injected into a single fertilised mouse egg and we generated a line of mice with this edited form of IL-1α. Approximately a year later – it’s experiment time and I could not have been more excited…
The first thing I needed to test was whether these mice were largely normal. To do this I took samples of the parts of their body that classically hold immune cells (the spleen and bone marrow) and tested to see if they have normal levels of each type of immune cell. Sure enough there was nothing unusual there.
I next wanted to see if cells from these mice responded differently to stimulus. To do this I took cells from multiple different compartments such as the bone marrow and essentially gave them a fake infection. Once again all the cells responded as I would expect and we even saw no difference in the secretion of IL-1α’s twin, IL-1β. Looks like our mutation has had no effect – does that mean the NLS on IL-1α does nothing??
So the obvious question now was, what about the levels of IL-1α itself? I ran my previous tests just to check I hadn’t messed around with anything other than the localisation of IL-1α (it should no longer be in the nucleus). Now it’s safe to say this result ruined my day… IT. WAS. GONE. Completely disappeared, nothing there. What on earth had happened?!
Now we were at panic stations. We needed to find out what had gone wrong. The first thing we did was check a cell type that should produce IL-1α without being stimulated by an infection. It could be that my ‘fake infection’ from before hadn’t worked for IL-1α. Alas, when I isolated unstimulated platelet cells from the blood of my mice I found they also had no IL-1α. Hmm… The next question was whether this disappearance of IL-1α occurred before or after the DNA was made into RNA (All proteins are made: DNA > RNA > protein. I needed to find where in that pathway the issue has taken place!). I found that the RNA of IL-1α was also lost meaning that my issue has occurred between DNA and RNA – a process called transcription.
At a loss for how our tiny changes in the DNA of IL-1α had caused this complete loss of transcription (I’m pretty inconsolable right now) we decided to head to the literature. Amazingly, a lab in Massachusetts had just published a paper showing that IL-1α relies on something called a long noncoding RNA (lncRNA) to be produced. I won’t go into too much detail about lncRNAs but the important part is that this one (called ‘AS-IL1a’) sits on the opposite stand of the DNA to the protein-coding IL-1α (DNA has a double-helix structure which means there are two stands coiled up. For a long time we thought that only one strand was important but we now know that the opposite strand has loads of vital functions!). Right! We had a hypothesis – what if our mutations in the IL-1α gene (which will have effects on both strands) had affected the lncRNA ‘AS-IL1a’ and led to a loss of IL-1α transcription therefore explaining why I see no IL-1α being produced?
I first tested whether there were different amounts of the lncRNA and found no difference. Still no luck. This was all going rather badly so I decided to swap my mammalian mouse for a computer mouse and do some bioinformatic work to see if our changes in the IL-1α gene had affected the structure of the lncRNA. Using some fancy prediction models I found that this was exactly the case! The small changes we made in the IL-1α gene (especially the ‘shield mutations’) had changed the predicted structure of the lncRNA suggesting that it can no longer work to switch on IL-1α. We had ourselves a discovery!
So why should you care? CRISPR went wrong – whatever. We’re not going to be doing this sort of stuff to humans any time soon right?? Well scarily that’s actually not the case. CRISPR technology has absolutely exploded and is currently be used around the world for a huge number of applications such as directly editing the genome of humans in order to cure disease.
In 2015 a Chinese group began working with human embryos from IVF treatments – how crazy is that? This was just 2 years from when we discovered that this bacterial immune system can be harnessed for other cells and already groups are using this technology on human embryos ?! (Disclaimer – these are non-viable embryos aquired from IVF procedures and are not used past 15 days). Since then groups have shown CRISPR can be used to fix mutations for multiple hereditary diseases in embryos and even in humans. There’s a good chance that 2018 will be ‘the year of the CRISPR trial’ with numerous companies requesting permission to undertake clinical trials.
Our paper shows that, even if you make all the right changes in your gene without any ‘off-target effects’, you can still get an unexpected result because, well, biology is complicated. CRISPR is truly an incredible technology that will no doubt change the way we treat disease but is accelerating at an alarming rate and we still do not fully understand it. That’s why we need to be careful what we CRISPR and do more work to better understand this wonderful tech and start kicking disease’s butt!