Structure Guided Design of CCL27 CAR T-Cells Against CCR10 for Multiple Myeloma

Background:

Even in the BCMA CAR-T era, many multiple myeloma (MM) patients often relapse and need additional treatment options. Our group identified CCR10 as a therapeutic target for MM and showed it is correlated with significantly worse outcomes and relapse. Additionally, we developed anti-CCR10 CAR-Ts that utilized the natural ligand CCL27 as the binder (Ferguson et al., Nat Comm 2022). However, these proof-of-concept CAR-Ts only had moderate efficacy in in vitro experiments. Here, we further validated CCR10 as a therapeutic target for MM and utilized structure-guided and computational modeling approaches to improve the performance of these CCL27 CAR-Ts to the level of current BCMA CAR-Ts.

Methods:

CAR-Ts were generated by lentiviral transduction of CD3+ human T cells. In vitro and in vivo performance of CAR-Ts was assessed against cell lines engineered to express luciferase. Statistical analysis on cytotoxicity was performed by two-way ANOVA. We used the Seurat and Harmony packages in R for data analysis and integration of single-cell RNAseq data. AlphaFold3 was utilized for generating predicted structural models, visualized by ChimeraX. Murine studies were performed with 1e6 MM.1S intravenously implanted in NSG mice, with 3e6 CAR-Ts administered 8 days after tumor inoculation (n=3 mice/arm).

Results:

Flow cytometry of 16 primary MM samples, 10 previously published and 6 new cases, showed that all samples were positive for CCR10. We also confirmed the expression of CCR10 on 7 MM cell lines by flow cytometry. Analyzing single-cell RNAseq data from 9 MM patients and 7 healthy donors, we found CCR10 was highly expressed in myeloma patient plasma cells, with lower expression in healthy donor plasma cells.

To improve on our proof-of-concept CAR-Ts, we used homologous chemokine-chemokine receptor structures, the limited literature on CCL27, and AlphaFold models to identify the importance of the N-terminus of CCL27 in its interaction with CCR10. This structure-guided approach suggested that there was scope to improve the binding of CCL27 by modifying N-terminal residues and that our N-terminal MYC tag may have prevented CCR10 binding. From this, we created a small mutational library of 10 unique CCL27 CAR-T constructs with modified N-terminal residues, while moving the MYC tag to the C-terminus. Screening this library for in vitro cytotoxicity against the MM cell lines MM.1S and AMO1, we identified two mutants that performed similarly to BCMA CAR-Ts in one or both cell lines. These were both N-terminal additions of a single aromatic amino acid, tryptophan (W-CCL27) or phenylalanine (F-CCL27). Further validation in two new T-cell donors indicated that these mutants performed significantly better than the wild-type natural ligand (WT-CCL27) in MM.1S, AMO1, and LP-1 (p<0.05). Efficacy was correlated with CCR10 expression and equivalent to BCMA CAR in cell lines with high levels of CCR10. In these experiments, WT-CCL27 had minimal efficacy. Knocking out CCR10 in the AMO1 cell line, we observed no cytotoxicity, showing the specificity of these mutants. In an ongoing murine study with an orthotopic MM.1S model, we have observed no signs of toxicity and an increased efficacy of CCL27 mutants compared to WT-CCL27, similar to BCMA CAR.

We further investigated these mutant CCL27s using AlphaFold3 to model their interaction with CCR10. The confident models indicated that the additional aromatic N-terminal amino acid on the mutants is interacting with a region of CCR10 not accessed by the WT-CCL27. We identified four amino acids in this binding pocket of CCR10 (V47, S48, Y120, and L300) that are predicted to interact with the mutant CCL27s.

Finally, we evaluated on-target off-tumor toxicity of these CAR-Ts. Profiling GM-CSF-mobilized peripheral blood, we found no CCR10 on CD34+ HSPCs (n=3). Despite no signal in single-cell and bulk RNAseq data, we surprisingly observed expression of CCR10 by flow cytometry on monocytes in both healthy donors and MM primary samples. However, we saw minimal cytotoxicity of the mutant CCL27 CAR-Ts against the THP-1 cell line, a model of monocytes that expresses high levels of CCR10. We are now working to validate the potential MM cell-type specificity of these CCL27-based CAR-Ts.

Conclusion:

These anti-CCR10 mutant CCL27 CAR-Ts represent a promising therapeutic candidate for MM that indicates the potential of using a structure-guided design to optimize natural ligand-based binders.