Novel Humanised ROR1 Chimeric Antigen Receptors for the Treatment of Haematological Malignancies

Introduction:

Receptor Tyrosine Kinase Like Orphan Receptor 1 (ROR1) is a surface antigen expressed on a range of haematological and solid malignancies including Chronic Lymphocytic Leukaemia (CLL). Although expressed during embryogenesis, its virtual absence on normal adult tissues makes it an attractive target for immunotherapy, especially with Chimeric Antigen Receptor modified T-cells (CAR T-cells). We have generated novel fully humanised ROR1CAR constructs for the treatment of CLL and other ROR1 positive malignancies.

Results:

Following a rat immunisation programme 38 oligloclonal hybridoma clones were single cell sorted and subjected to 5'RACE. Of 13 novel anti-ROR1 antibodies isolated, 10 retained specific binding when cloned into a heavy-linker-light single chain variable fragment (scFv) format. These scFvs in combination with a second generation CAR architecture comprising CD3zeta and 4-1BB demonstrated specific toxicity against ROR1 positive cell lines after T-cell transduction using lentiviral vectors.

We found cytotoxicity with ROR1CAR T-cells was dependent on target cell ROR1 density. In order to ensure our screening assays allowed us to select which of the 10 binders was most suitable for targeting primary CLL, we assessed the antigen density of ROR1 and CD19 on CLL cells. Median expression of ROR1 was 2304 molecules/cell (Range 800-4828), compared to CD19, which had a much higher density of 12,583 (Range 5894-23,652). In view of this, subsequent functional assessment was focused on SKW and Jeko1 cell lines with constitutive ROR1 expression at levels similar to CLL cells, as opposed to those transduced to express supra-physiological levels.

Our initial optimisation focused on modifying the CAR extracellular spacer region. We demonstrated a reciprocal relationship between cytotoxicity and the distance between T-cells and target cells. This was assessed by using clones that bound either the membrane-distal immunoglobulin domain or a more membrane-proximal frizzled domain of ROR1. The use of an optimum spacer enhanced cytotoxicity of all scFv constructs but yielded two lead candidates: Clones A & F. These showed consistently superior cytotoxicity against target cell lines compared to the other isolated clones. In addition epitope mapping revealed binding sites unique from the previously described rabbit R12 and murine 4A5 anti-ROR1 CAR T-cells.

One of the advantages of targeting ROR1 as opposed to CD19 is sparing the normal B-cell compartment from CAR mediated eradication. However this comes with the consequent risk of B-cell mediated immune responses against rat-derived scFvs. To minimise immunogenicity we undertook a humanisation programme and grafted the complementary determining regions (CDR) of the heavy and light chains of Clone A and F into 5 acceptor human germline VH and VL sequences, generating 25 potential scFvs for each. Binding assessment showed seventeen successfully humanised binders for Clone A and three for Clone F. Of these, 5/17 and 3/3 showed activity in a CAR format against target cells. A final selection was made based on specific cytotoxicity, enhanced cytokine secretion (Interleukin-2 and Interferon gamma) and proliferation compared to the parental clones resulting in 2 unique constructs targeting different extracellular domains of ROR1. In addition, we have demonstrated cytotoxicity against a panel of ROR1 positive solid cancer cell lines to demonstrate their wider applicability.

Conclusion:

ROR1CAR T-cells have the potential to be an effective therapeutic not just for CLL but also Acute Lymphoblastic Leukaemia, Mantle Cell Lymphoma and solid malignancies. We have described the first humanised ROR1 CARs, which target novel epitopes and have proved effective in relevant pre-clinical assays. Although other ROR1 CARs have been described, we believe the unique properties of these constructs merits further investigation and comparison in the preclinical and clinical setting.