Building Disease Models With CRISPR The Zhang Lab Divisional That Claims Modelling Before Medicine

This application sits inside the foundational CRISPR gene editing patent family filed out of Feng Zhang’s laboratory at the end of 2013, and it is worth being precise about what this particular member of the family asks for. Claim 1 is not a treatment. It is a method of modelling a disease that is associated with a genomic locus, carried out in a eukaryotic organism or a non-human organism, by delivering an engineered composition that assembles a CRISPR complex at a chosen target sequence. Therapy claims do appear further down the set, tied to viral vector delivery, but the independent claim that opens the document is about deliberately creating the mutation rather than correcting one. The applicants are the Massachusetts Institute of Technology and the Broad Institute.

The Problem

The specification frames the difficulty in the language of the early 2010s, before CRISPR had become a household word. Genome sequencing had advanced far enough to catalogue and map genetic factors across a wide range of biological functions and diseases, but reading a genome and rewriting it are different problems. What was missing, the background section says, was a way to selectively perturb individual genetic elements so that causal genetic variations could be reverse engineered systematically.

The tools that existed were not good enough for the job. The specification names designer zinc fingers, transcription activator-like effectors and homing meganucleases, and concedes that they can produce targeted genome perturbations. Its complaint is practical rather than scientific. There remained a need, it says, for genome engineering technologies that are affordable, easy to set up, scalable, and amenable to targeting multiple positions within a eukaryotic genome. Every one of those four adjectives is a criticism of protein-guided editing, where each new target requires a new protein to be designed and built.

The summary adds a second and more commercial problem. Despite valid therapeutic hypotheses and strong development effort, the specification notes, there have only been a limited number of successes using small molecules to treat diseases with strong genetic contributions. Diseases with a clear genetic cause are exactly the ones a small molecule is least suited to, because the defect is in the instruction set rather than in a protein that can be blocked. Adding CRISPR to the therapeutic repertoire, the applicants argue, also accelerates the ability to develop animal models for genetic diseases, which is the thread claim 1 picks up.

The third problem is the one that occupies most of the document. Knowing that a nuclease works in a dish tells you very little about whether it can be delivered to the right cells in a living animal, expressed there at a useful level, and kept away from everything else. The specification states plainly that to use the system effectively without deleterious effects it is critical to understand engineering, optimisation and cell-type, tissue and organ specific delivery, and that those aspects are aspects of the claimed invention.

What This Invention Does

Claim 1 is a method of modelling a disease associated with a genomic locus, in a eukaryotic organism or a non-human organism, by manipulating a target sequence within a coding, non-coding or regulatory element of that locus. The manipulation is achieved by delivering a non-naturally occurring or engineered composition, and the claim then offers two alternative compositions.

In the first alternative, the composition contains a CRISPR-Cas system RNA polynucleotide sequence made up of three parts arranged in a 5′ to 3′ orientation: a guide sequence able to hybridise to the target, a tracr mate sequence, and a tracr sequence. Alongside it sits a polynucleotide sequence encoding Cas9, optionally carrying one or more nuclear localisation sequences so the enzyme reaches the nucleus. When transcribed, the tracr mate sequence pairs with the tracr sequence and the guide sequence steers the assembled complex to the target. The claim is explicit that the Cas9-encoding polynucleotide may be either DNA or RNA.

The second alternative breaks the same machinery into separate pieces. The guide sequence and one or more tracr mate sequences sit on one polynucleotide, the Cas9 coding sequence on another, and the tracr sequence on a third. The functional requirement is identical. What differs is the packaging, and packaging is the practical constraint the rest of the specification wrestles with.

The dependent claims then move outward from that core. Claim 2 narrows the enzyme to SaCas9, the Cas9 of Staphylococcus aureus, which matters because it is small enough to be carried where the more familiar Streptococcus pyogenes enzyme is not. Claim 3 covers delivery of the RNA components by liposomes, nanoparticles, cell penetrating peptides, exosomes, microvesicles or a gene gun. Claim 5 sets out a parallel modelling method delivered by a viral vector system.

The therapeutic claims arrive at claim 7 and again at claim 41, and both are tied specifically to an adeno-associated virus or lentivirus vector system carrying the composition under first, second and third regulatory elements. Claims 10, 11, 42, 45 and 46 repeatedly narrow the condition to ocular disease and then to retinitis pigmentosa or achromatopsia. Claims 15 to 21 cover the manufacture of those vectors. Claims 13 and 14 cover delivering the enzyme as mRNA rather than DNA. Claim 44 closes the loop between the modelling claims and the treatment claims: introduce the disease-correlated mutations into cells or a transgenic animal, test candidate treatments on those cells, then treat the patient based on what the testing showed.

Key Features

• A claim written for models, not patients. Claim 1 covers modelling a disease in a eukaryotic organism or a non-human organism, so the protected act is creating a faithful genetic model rather than administering a therapy.
• SaCas9 as the preferred enzyme. Claim 2 narrows Cas9 to SaCas9 and several of the later independent claims state that the Cas9 is preferably SaCas9, reflecting a size advantage that matters when the payload has to fit inside a viral capsid.
• Two ways to package the same complex. Every independent claim offers the choice between a single RNA polynucleotide carrying guide, tracr mate and tracr sequences in order, or the same elements split across separate polynucleotides under separate regulatory elements.
• Delivery written into the claims. Liposomes, nanoparticles, cell penetrating peptides, exosomes, microvesicles and a gene gun appear in claim 3, while adeno-associated virus and lentivirus vectors carry the therapeutic claims at 7, 12, 41 and 43.
• The eye as the worked indication. Ocular disease, and specifically retinitis pigmentosa and achromatopsia, is the condition the dependent claims return to five separate times, which is consistent with the eye being small, enclosed and reachable by direct injection.
• Enzyme delivered as messenger RNA. Claims 13 and 14 cover delivering mRNA encoding Cas9 to the cell instead of DNA, which limits how long the nuclease persists and therefore how long it has to make an unintended cut.

Who Is Behind It

The applicants are the Massachusetts Institute of Technology and the Broad Institute, the Cambridge genomics institute founded jointly by MIT and Harvard. Six inventors are named: David Benjamin Turitz Cox, Le Cong, Matthias Heidenreich, Randall Jeffrey Platt, Lukasz Swiech and Feng Zhang. Zhang is a core institute member at the Broad and a professor at MIT, and the laboratory he runs published the first demonstrations of CRISPR editing in human cells. Cong was a graduate student on that work and now runs his own laboratory at Stanford. Platt and Swiech are the authors of the in vivo brain editing work that the examples section of this specification draws on, and Platt now leads a laboratory at ETH Zurich. The specification records that the work was made with government support under an NIH Director’s Pioneer Award and a second National Institutes of Health grant, and states that the government has certain rights in the invention.

The chain of applications is long and the specification sets it out precisely. AU 2026202160 is a divisional of Australian patent application 2023219828, which is itself a divisional of Australian patent application 2021204395, which is a divisional of Australian patent application 2014361781. That earliest Australian application claims priority to seven United States provisional applications, all filed on 12 December 2013: 61/915,176, 61/915,192, 61/915,215, 61/915,107, 61/915,145, 61/915,148 and 61/915,153. The sequence listing at the back of the document records the international application as PCT/US2014/070127, filed 12 December 2014. Everything in that paragraph comes from the specification itself. Three generations of Australian divisional filings from a single 2013 priority date is a fair measure of how much commercial ground the family is understood to cover.

Why It Matters

The examples in this specification are the reason the modelling claim is the one at the front. The worked example is the delivery of Cas9 into the mouse brain using a two-vector adeno-associated virus system, one carrying the enzyme under a neuron-specific Mecp2 promoter and the other carrying the guide cassette, in order to disrupt the Mecp2 gene in post-mitotic neurons. Mecp2 is the gene behind Rett syndrome. The point of the experiment was not to fix anything. It was to produce the mutation in a living animal quickly, then read out the consequences by sequencing, imaging, electrophysiology and behaviour, without waiting for generations of breeding to produce a transgenic line.

That is the quiet argument running underneath the whole family. Before CRISPR, building an animal model of a suspected disease gene was a multi-year project, which meant that only genes with strong prior evidence were ever modelled. When making the model becomes a matter of designing a short guide sequence, the economics of asking the question change, and the bottleneck moves from making the animal to deciding which question to ask.

The size problem the claims keep circling is also real rather than academic. An adeno-associated virus genome holds roughly 4.7 kilobases and the Streptococcus pyogenes enzyme alone takes up about 4.2 of them, which leaves almost nothing for a promoter, a guide cassette and a polyadenylation signal. The specification’s answer is either to split the system across two vectors, as the brain example does, or to use the smaller Staphylococcus aureus enzyme that claim 2 names. Both answers show up in the claim structure, and both are still the two live options for anyone trying to deliver an editor in a single injection today.

Finally, the repeated narrowing to eye disease is worth reading as a statement about where delivery is solvable rather than where the need is greatest. The eye is small, immune-privileged, physically enclosed and can be injected directly, so a dose that would be impossible to achieve systemically is routine there. That is why retinal indications led the gene therapy field generally, and why they appear again in a document whose central preoccupation is getting the tool to the tissue.

Related Concepts

• Cas9 – the RNA-guided nuclease that every claim in this application is built around.
• Guide RNA – the short targeting sequence that the claims describe as a guide sequence, a tracr mate sequence and a tracr sequence.
• Genetically modified mouse – the kind of disease model claim 1 is aimed at producing.
• Retinitis pigmentosa – the inherited retinal condition named in five of the dependent claims.
• Zinc finger nuclease – one of the older editing tools the background section measures CRISPR against.

---

AU 2026202160 was published in the Australian Official Journal of Patents on 9 April 2026 and is open for public inspection. Patent applications represent inventions that are sought to be protected and do not necessarily reflect commercially available products.