An Ovarian Cancer CAR With a Protease Gate How Pfizer and Allogene Borrow CD45 to Keep a T Cell Switched Off Outside a Tumour

This specification describes an engineered immune cell receptor with a built-in safety catch, and a separate but related receptor aimed at an ovarian cancer marker. The safety catch is the part that gives the document its title: a chimeric antigen receptor carrying an extra domain that drags the cell’s own CD45 phosphatase alongside it and damps the receptor down, joined to the working receptor by a linker that a tumour enzyme can cut. The claims actually filed with this Australian divisional, however, are directed at the other half of the disclosure, the anti-MUC16 receptor that the gated design uses as its business end. It was filed jointly by Pfizer and Allogene Therapeutics.

The Problem

The background section is short and it names one obstacle. Tumour specificity, it says, is one of the major hurdles for the development of CAR T cell therapy in solid tumour indications.

The specification explains why blood cancers have been easier. Targets such as CD19, CD20 and BCMA are heavily restricted to the haematologic lineage. A CAR T cell aimed at CD19 will destroy healthy B cells along with malignant ones, but because normal blood cells repopulate from progenitor stem cells, the collateral damage is survivable and therefore acceptable. That tolerance is what made the first generation of CAR T therapies possible.

Solid tumours do not offer the same bargain. Many solid tumour targets, the specification says, show an appreciable level of expression in some healthy organs or tissues, and the collateral destruction of those tissues by CAR T cells may lead to severe adverse side effects and even death. The document uses the field’s standard name for this, on-target off-tumour toxicity, and it is worth stressing that the receptor in that scenario is working exactly as designed. It has found its antigen. The antigen was simply in the wrong place.

That leaves engineers with an uncomfortable choice. Pick a target expressed only on tumour, and the list of candidates is very short. Pick a target that is merely enriched on tumour, and the therapy risks attacking the healthy tissue that also carries it. The specification states the need directly: a cell-based cancer therapy with reduced on-target off-tumour toxicity.

What This Invention Does

The disclosed answer is to stop looking for a perfectly specific antigen and instead make the receptor conditional on where it is. The invention adds what the specification calls a CD45 gate.

CD45 is a phosphatase found in large quantities on the surface of T cells, and the specification summarises its double-edged role. It can activate T cell receptor signalling by dephosphorylating kinases such as LCK, and it can also shut signalling down by dephosphorylating CD3 zeta directly. The design exploits the second behaviour. A CD45 recruiting domain is added to the outside of the receptor, so that the receptor is physically tethered next to a CD45 molecule on its own cell surface. Held in that company, the receptor can still bind its antigen but cannot properly transmit the resulting signal, because CD45 keeps stripping the phosphate groups the signal depends on.

The release mechanism is the linker. The CD45 recruiting domain is joined to the antigen binding domain by a linker containing at least one protease cleavage site. When the cell reaches an environment rich in the relevant protease, the linker is cut, the recruiting domain floats away, and what remains is an ordinary functional CAR that is no longer tied to its own off switch. The specification frames this as reversible colocalisation, and the proteases it names, matrix metalloproteinases 2 and 9, urokinase plasminogen activator and matriptase, are the ones characteristically abundant in a tumour microenvironment and comparatively scarce in healthy tissue. In effect the tumour supplies the key that arms the cell that is attacking it.

The claims of this particular Australian application, though, do not cover the gate. They cover the receptor that the gate was built around. Claim 1 reads on a MUC16-specific chimeric antigen receptor comprising an extracellular ligand-binding domain, a first transmembrane domain and an intracellular signalling domain, where the extracellular ligand-binding domain comprises a single chain variable fragment having binding specificity for the extracellular domain of MUC16. MUC16 is better known by the name of the protein fragment it sheds into blood, CA-125, the marker used to monitor ovarian cancer for decades.

There are twenty claims in total and they build outward in the conventional way. Claims 2 to 6 fix the antibody fragment by its six complementarity determining regions, then by its heavy and light chain variable region sequences, then by the full receptor sequence. Claims 7 and 8 cover the encoding polynucleotide and an expression vector. Claims 9 to 13 cover engineered immune cells carrying that vector or receptor, optionally including a suicide polypeptide, with claim 13 naming RQR8. Claim 14 covers a population of between about one thousand and about ten billion such cells, and claim 15 a pharmaceutical composition. Claims 16 to 19 cover treatment, including a long list of cancers, and claim 20 covers the method of making the cell.

Key Features

• A receptor tethered to its own off switch. The CD45 recruiting domain holds the receptor beside a CD45 phosphatase on the same cell surface, which at least partially inhibits activation even when the receptor has bound its antigen.
• A linker that acts as the trigger. The recruiting domain and the antigen binding domain are joined by a linker carrying one or more protease cleavage sites, and cutting that linker is what converts a damped receptor into a working one.
• Tumour enzymes as the key. The named cleavage sites are recognised by matrix metalloproteinases 2 and 9, urokinase plasminogen activator and matriptase, and some linkers carry two different sites so that either enzyme can release the gate.
• Recruiting domains borrowed from elsewhere. The specification builds the CD45 recruiter from truncated forms of UL11, sec49K and BTN3A1, proteins already known to engage CD45, as an alternative to an anti-CD45 antibody fragment.
• The claimed receptor is the MUC16 one. Claim 1 of this divisional is an anti-MUC16 CAR defined by a single chain variable fragment binding the extracellular domain of MUC16, with the sequence-level definitions following in claims 2 to 6.
• An allogeneic route written into the claims. Claim 19 covers cells derived from allogeneic immune cells taken from a donor other than the patient, which is the off-the-shelf manufacturing model rather than the patient-specific one.

Who Is Behind It

The two applicants are Pfizer and Allogene Therapeutics, and their joint appearance on the front page is a direct consequence of how Allogene came to exist. Allogene was formed in 2018 by former leadership of Kite Pharma, and Pfizer contributed its allogeneic CAR T immuno-oncology portfolio to the new company under an asset contribution agreement in exchange for an equity stake. Work that began inside Pfizer and continued at Allogene therefore carries both names, which is what has happened here.

Eight inventors are named: Siler Panowski, Barbra Johnson Sasu, Yi Zhang, Nguyen Tan, Zhe Li, Michael Thomas Bethune, Shanshan Lang and Thomas John Van Blarcom. Sasu served as Allogene’s chief scientific officer from the company’s formation in 2018, and several of the others are long-standing members of the same protein engineering and cell therapy group.

The priority position is set out clearly in the specification and does not need to be inferred. AU 2026202161 is a divisional of Australian patent application 2021409801, which is the Australian national phase of PCT/US2021/064615. That international application claims the benefit of two United States provisional applications, 63/128,667 filed on 21 December 2020 and 63/289,984 filed on 15 December 2021.

One quirk is worth recording for anyone reading the document itself. In preparing the Australian description the drafters replaced the word claims with the word paragraphs throughout the body text, which is ordinary practice for a divisional but has been applied a little too thoroughly. The cross reference sentence now says the application paragraphs the benefit of priority, and the eighty-four numbered embodiments at the end of the description are all called paragraphs. Paragraphs 1 to 64 describe the protease-activating CD45-gate receptor and its methods of use. Paragraphs 65 to 84 describe the anti-MUC16 receptor. Only the second group has been carried through into the actual twenty claims of this divisional.

Why It Matters

The gated design is an attempt to solve a problem that has held solid tumour cell therapy back for a decade, and it does it by changing the question. Every earlier approach has tried to find an antigen that appears on cancer and nowhere else, which for most solid tumours does not exist. This design accepts an imperfect antigen and adds a second condition that has nothing to do with antigen at all, namely whether the cell is standing in a protease-rich environment. Two weak signals combined can be more specific than one strong one, which is the same logic behind logic-gated and masked receptor designs generally.

Using CD45 as the brake is the elegant part. The inhibitor is not something that has to be dosed, manufactured or kept stable in circulation. It is already sitting on the T cell surface in abundance, so the gate ships with the cell and cannot run out. The recruiting domain does not need to inhibit anything itself. It only needs to hold the receptor in the wrong neighbourhood.

The choice of MUC16 for the claims of this divisional is also readable. MUC16 is heavily expressed on ovarian and some other gynaecological cancers and is the antigen behind the CA-125 blood test, but it is not absent from healthy tissue, and it sheds a soluble fragment that can soak up receptors before they reach a tumour. The specification’s experiments address that directly, reporting that binding to soluble MUC16 was not significantly affected by the presence or absence of the gate, and showing anti-tumour activity for the 53B6 receptor in two orthotopic ovarian tumour models in mice. An antigen that is real, abundant and slightly too widely distributed is exactly the kind that a conditional receptor is meant for.

The allogeneic angle in claim 19 rounds it out. Making cell therapy from a healthy donor rather than from each patient is what turns a bespoke procedure into a product, and it is the premise Allogene was built on. It also raises the stakes on safety, since an off-the-shelf cell cannot be tailored to the individual it is given to. A receptor that stays quiet until the tumour switches it on is a reasonable thing to want in that setting.

Related Concepts

• Tumor microenvironment – the protease-rich local environment that supplies the signal releasing the gate.
• Matrix metalloproteinase – one of the enzyme families whose cleavage sites are written into the linker.
• Lck – the kinase whose regulation by CD45 sits at the centre of the gating mechanism.
• Ovarian cancer – the disease most associated with the MUC16 antigen that claim 1 targets.
• Allotransplantation – the donor-to-patient model behind the allogeneic cells of claim 19.

---

AU 2026202161 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.