Genome editing is an immensely powerful tool which has the potential to revolutionise the life sciences, including medicine and agriculture. In recent years, the technology has moved on apace heralding myriad scientific opportunities, but can the law keep pace with the science? 

The science

DNA comprises a collection of genes (the 'genome'), which provide the code for sequences of amino acids that are the building blocks for making proteins. Proteins are essential molecules in all living systems, playing a wide variety of roles that are crucial for life – there are around 20,000 different types of protein in humans. They include enzymes, which catalyse biochemical reactions, such as the breakdown of food into absorbable nutrients by amylases and proteases; hormones, which play a crucial role in cell signalling and communication to trigger specific responses, such as the regulation of blood sugar by the hormone insulin signalling cells to take in glucose from the bloodstream; and antibodies, which help to protect organisms from infection. 

By modifying DNA, we are able to instruct cells to make modified proteins, or even new proteins which have not been made before. The ability to use genome editing tools to make precise changes to the DNA of living organisms may therefore hold the key to correcting genetic defects, and to treating cancer and infectious diseases; to addressing the pressure of global food supplies by improving crops and enhancing livestock; and to reducing the impact of climate change on agriculture and natural ecosystems. 

Genetic modification is not new. However, its application was for many years limited by the fact that it was a complex and labour-intensive process. That has changed significantly with the advent of CRISPR, which is a revolutionary gene editing technology developed from a natural defence mechanism found in bacteria. But first, let's take a look at what went before. 

Gene editing B.C. (Before CRISPR)

A number of years ago, during a discovery exercise for a patent infringement and revocation action, I had the privilege of discussing the genesis of a genetically modified crop with one of the scientists who created a highly successful herbicide resistant strain.  It was with some excitement that he shared stories of the work that he and his colleagues carried out in their lab in the 1980s and the many avenues that they explored. The technique that was ultimately successful was something of a surprise to me.  They used DNA containing a gene that provides resistance to the herbicide to coat tiny ballistic gold particles. These particles were then physically shot into plant tissue using a 'gene gun' that they had invented in their spare time from various bits of kit – in the earliest prototypes this included pieces of crisp packet! Somewhat incredibly, the DNA integrated into the plant's genome, the gene was expressed and a herbicide resistant strain was born. 

The prototype 'gene gun' is now a museum piece at the Smithsonian, and times have moved on.  Other techniques have been successful – most using nuclease enzymes (proteins that cleave DNA) that are able, to some extent, to recognise and cut specific DNA sequences. These include enzymes with names that sound straight out of science fiction: Meganucleases, Zinc Finger Nucleases and TALENs. But it has been the advent of CRISPR (in around 2015) that has marked a new era in gene editing, providing a more simple, efficient and precise way to find a specific sequence of DNA inside a cell, and to cut it so that it can be edited. CRISPR has revolutionised the field of genetic modification. 

CRISPR

CRISPR stands for Clustered Regularly Interspaced Short Palindromic Repeats. These repetitive DNA sequences were first observed in bacteria where they play a role in defending a bacterium against infection by viruses called bacteriophages. When a bacterium is invaded by a bacteriophage, the first stage of the bacterium's immune response is to capture bacteriophage DNA and insert it into its own genome as a 'spacer' sequence between the CRISPR repeats. CRISPR 'arrays' are thereby formed which comprise the repeats interspersed with spacer sequences derived from DNA fragments of bacteriophages that have previously infected the bacterium.

CRISPR arrays can then be used by the bacterium to detect and destroy DNA from similar bacteriophages during subsequent infections – a form of acquired immunity. In order to do so, RNA is transcribed from the CRISPR array and processed into smaller CRISPR RNAs (CrRNAs) containing the spacer sequences that are specific for the target bacteriophage DNA.  The crRNAs then guide nucleases known as CRISPR-associated proteins (or "Cas") to cut the target sequences of bacteriophage DNA, thereby destroying the virus. 

Based on this naturally occurring system, researchers demonstrated that RNAs could be constructed to guide a Cas nuclease (a commonly used one is known as Cas9) to a specific DNA sequence, improving the chances that DNA will be cut at that site and not elsewhere in the genome. This is the basis of the technology known as CRISPR-Cas9 that can be used to edit genes within living organisms – it is a two-component system comprising a guide RNA (gRNA) that contains the code designed to match with a particular target DNA sequence, and an enzyme (Cas9) that cuts the DNA.  These two components form a complex that binds to the DNA in the cell and moves along it looking for the gRNA's complementary sequence, cutting the DNA when it finds a sufficiently similar sequence. What used to be costly and laborious can now be achieved relatively quickly and cheaply. The development of CRISPR-Cas9 for genome editing was recognised by the Nobel Prize in Chemistry in 2020, awarded to Emmanuelle Charpentier and Jennifer Doudna. Its potential is phenomenal. 

Using CRISPR it is now possible to find a specific gene and change or delete it, or add a completely new gene to a genome. So it is now theoretically possible to insert DNA to correct mutations that cause human diseases, or to specifically engineer cells (for example, those from the immune system) for use as a cancer therapy, enabling the treatment of patients using a modified version of their own immune cells. Last year, the first CRISPR-based therapy obtained regulatory approval. It uses CRISPR to edit genes in bone marrow stem cells to enable the production of fetal hemoglobin. The stem cells are then transplanted back into the patient where they help to replace defective red blood cells.

CRISPR-based therapies have also been developed which do not cut DNA or alter it in any way, but instead make a cell express more or less of the protein that is encoded by a gene, or switch a gene off completely. These include CRISPRa (where a is for 'activation') and CRISPRi (where i is for 'interference')[1].

Another powerful technique is using CRISPR to genetically modify bacteria to introduce genes encoding a product of interest in order to create mini factories which can churn out specific proteins, such as enzymes and hormones. 

More recently, another type of CRISPR which targets RNA has been developed, using an enzyme called Cas13. This too has huge potential in biomedical applications.  When a gene is expressed in human cells, one of the first steps is the creation of RNA so being able to target RNA has huge potential, for example, in blocking the expression of a particular gene which is implicated in a disease.  Further, RNA is the main genetic material in viruses (including flu and Covid), so CRISPR-Cas13 has potential uses in preventing or treating viral infection.   

What is next for CRISPR?

Like many other areas of science, attention has been turned to Artificial Intelligence and what it might be able to do to improve, augment and enhance the use of CRISPR. How might these two powerful technologies be used together? 

DNA by design

Generative AI can be used to design proteins and compose the DNA that encodes them. This opens up the possibility of designing and making bespoke proteins, even those not known to exist in nature.  Couple this with CRISPR technology and it is now possible to use bacteria to generate proteins that may have endless uses in addressing the world's challenges.  

By way of example, the Institute for Protein Design (IPD) at the University of Washington has a mission to create new proteins that solve a diverse array of challenges in medicine, technology and sustainability[2].  For example, in pursuit of drug development, the IPD is crafting proteins that selectively bind to cancer cells and others that inhibit viral infection. In response to environmental challenges, the IPD is pursuing novel proteins which can be used to create new biodegradable molecules and break down environmental pollutants. In order to achieve this, the IPD has developed a number of AI tools.   One, based on generative AI, can be used to create novel protein structures in seconds, with the proteins designed in a specific way, for example to bind with a particular target receptor.  Another of IPD's AI tools uses multiple neural networks to create sequences of amino acids for those protein structures, which can then be used to code the relevant DNA. The resulting DNA construct can then be introduced into bacterial cells by gene editing, creating a microscopic source of a bespoke protein which has been designed and produced by a combination of AI and CRISPR.   

Improved gene editing tools

And what about the potential of AI to improve CRISPR technology itself? Whilst the currently available CRISPR technology has revolutionised gene editing, it is not a perfect system. With all technology, there is risk. With CRISPR the risk is that it will cut DNA where it is not supposed to, potentially affecting off-target genes, leading to potentially harmful mutations and side effects. In addition to the risk of making edits in the wrong places, some CRISPR-based tools may lack efficiency, editing too few cells to be effective. One of the goals in developing CRISPR-based therapy is therefore maximising the activity on the target gene and minimising off-target activity. 

Deep learning models have been developed to accurately predict the activity and specificity of different Cas9 variants. For example, the Flatiron Institute's Center for Computation Biology (CCN) has developed an AI-based model which could lead to safer genome editing by predicting the chances that a given Cas9 enzyme will make off-target cuts[3]. Profluent has an AI-based protein design platform that is being used to effect improvements in gene editing technology, from precisely tuning existing CRISPR systems to creating entirely novel gene editing systems[4]. Their AI-generated gene editor, which they call OpenCRISPR-1, displayed a 95% reduction in editing at certain off-target sites, with unwanted genetic insertions and deletions at a rate of less than 1%[5].

Similarly, in CRISPR-Cas13 technology, specificity is key. With this in mind, researchers at New York University, Columbia University, and the New York Genome Center have engineered a deep learning model called TIGER (Targeting Inhibition of Gene Expression via guide RNA design) using data from CRISPR screens to develop an AI tool for predicting off-target activity of RNA-targeting CRISPRs[6].  By using TIGER it is possible to design Cas13 guides with improved specificity, avoiding undesirable off-target activity and therefore lowering the risk of side effects. 

Fine-tuning therapy

In addition to its use as a tool for predicting off-target activity, TIGER can be used to precisely modulate the amount of a particular gene that is expressed, by mismatching the guide so that it does not knock-out the RNA completely. This may be useful for diseases in which there are too many copies of genetic material (such as Down's syndrome) or where there is aberrant gene expression (which, for example, may lead to uncontrolled tumour growth in some cancers).

The law – patentability 

As a patent specialist, these exciting developments lead me to ponder the extent to which patents may be available in this field. The idea of scientists assembling a 'gene gun' and firing DNA-coated ballistic particles into plant tissue seems to me to be the epitome of what the patent system is designed to reward (and, indeed, patents were granted). So too the development of CRISPR itself (that is the subject of another article in this series). But what impact might the use of AI in combination with CRISPR have on patentability? 

Patent law in the UK as we know it today dates back to the 1970s, when the idea of AI was the stuff of science fiction. As the legal framework for protecting innovation was established without AI in mind, it is inevitable that the law is having to play catch-up with the commercial reality of using AI to assist in innovation (such as the AI systems developed by IPD to assist in protein design and coding of the related DNA sequences) and with the development of AI technology itself (like the TIGER tool). The increasing prevalence of AI may well demand a reassessment of fundamental patent law concepts and their application to this new and fast-moving technology. 

Conversations and consultations have been going on worldwide. The UK Intellectual Property Office (UKIPO) sought views on AI and IP in 2020, following up with a further consultation specifically on whether the patent system is equipped to deal with AI in 2022. It concluded that there was no evidence that UK patent law is currently inappropriate to protect inventions made using AI, but it remains open to the possibility that change may be needed in the future when there is a stronger technological case. The UKIPO has published, and subsequently updated, guidelines for examination of patent applications relating to AI. 

So, the position remains that patents may be granted in the UK for AI-related inventions, provided that the application satisfies the legal requirements set out in the Patents Act 1977. Specifically, the application must claim something that is novel, involves an inventive step, is capable of industrial application and does not fall within one of the statutory exclusions from patentability, which include computer programs, mathematical methods, business methods and presentation of information. The invention must also be disclosed clearly and completely enough for it to be performed by the skilled person. 

Many questions and challenges are likely to arise with the use of AI. Inventiveness may become an increasingly difficult hurdle as AI platforms become ever more sophisticated and widely used, making many elements of research and development routine and leading to an increase in the threshold for obviousness. It is still early days for the CRISPR-AI combination. But what happens when AI becomes so sophisticated that it can routinely churn out bespoke designed proteins that are reliably fit for a defined purpose, and compose the DNA that encodes them, with that DNA then routinely being inserted into cells using CRISPR-based gene editing techniques that have been optimised and verified using other AI systems?  Where is the invention that justifies a 20-year monopoly? Will we see a move away from protecting developments by traditional intellectual property rights because AI makes solutions to problems easy to come by? 

Further, there is no obligation to disclose in a patent specification that AI was involved in the inventive process that resulted in, say, a DNA sequence for a protein designed for a specific purpose, so will all patent applications have to be examined on the assumption that it was?  Will companies that do not have cutting edge AI at their disposal then be at a disadvantage because the inventive merit that used to flow from more traditional research is now presumed to have been replaced by AI?  Inventiveness also has to be assessed by considering AI as it was at the priority date, ignoring capabilities developed later.  However, trying to pinpoint the state of the art some years down the line in an area as fast-moving as AI is likely to be a challenge. And are we going to need the concept of 'the skilled AI' alongside 'the person skilled in the art' in order to consider what is obvious?

And what of the AI systems themselves? Can AI tools that provide DNA to be used in the CRISPR system, or that assess or improve the gene-editing technology itself, be patented? One of the main issues faced when trying to patent AI inventions in the UK is the law relating to exclusions from patentability. UK legislation declares that anything that consists of certain categories of excluded matter, such as "a program for a computer … as such" or a mathematical method, is not an invention and therefore cannot be protected by a patent. AI by its nature is based on computational models and mathematical algorithms, so these statutory exclusions raise tricky questions as to whether it is possible to patent AI-related technology. A recent decision of the UK Court of Appeal in the Emotional Perception case[7] has provided some much needed guidance on this. As things stand currently (subject to any appeal to the UK Supreme Court), AI- implemented inventions are in no better and no worse position than other computer-implemented inventions – they are patentable if there is sufficient 'technical contribution' – and the UKIPO has updated its guidelines for examining patent applications relating to AI in light of the case.  My more detailed analysis of the Emotional Perception judgment can be found here: Patentability of AI – A New Perspective from the UK on Emotional Perception | DLA Piper. 

In summary

It is an exciting time in the field of genomics. CRISPR has been a game-changer in gene editing and AI's ability to analyse vast datasets and make predictions makes it a tremendously powerful partner.  The application of AI to genomics and protein science will undoubtedly play a significant role in shaping the future, potentially transforming it in ways that we cannot currently predict.  

Technological advances will no doubt challenge the legal system that is the bedrock of protection for innovation and which facilitates a return on investment in research. With the rise of digital biology, patentability of inventions comprising or derived from AI is likely to be replete with questions and challenges for many years to come. We need to ensure that the answers that develop do not stifle innovation, so that the life sciences industry can reap the full benefit of modern technology. 

Next week in our DLA Piper Genomics Series. 

Please tune in for next week’s feature, where Marion Abecassis and colleagues will examine the hurdles that innovative gene therapies face in obtaining regulatory approvals across the UK and EU, balancing the need for safety and efficacy with the urgency of bringing life-saving treatments to market.
 

[1] What is CRISPRa vs. CRISPRi? (horizondiscovery.com)

[2] Institute for Protein Design (uw.edu)

[3] Physics-Informed AI Method Could Help Make CRISPR Safer (simonsfoundation.org)

[4] Profluent

[5] Design of highly functional genome editors by modeling the universe of CRISPR-Cas sequences | bioRxiv

[6] Prediction of on-target and off-target activity of CRISPR–Cas13d guide RNAs using deep learning | Nature Biotechnology

[7] Drug discovery and development in the age of AI