Repurposing anti-mesothelin CAR-NK immunotherapy against colorectal cancer

Abstract

Background: Colorectal cancer (CRC) is the third most common cancer worldwide, with highly variable prognosis and response to treatment. A large subset of patients does not respond to standard treatments or develops resistance. As an alternative, adoptive immunotherapy based on chimeric antigen receptor (CAR)-transduced immune cells has been proposed, however with significant adverse events. We therefore evaluated alternative CAR targets already tested in other tumour types and employed the natural killer cell line NK-92 for CAR transduction because of its more favourable toxicity profile.

Methods: As an alternative antigen, we considered mesothelin (MSLN), the most represented target in CAR-based clinical studies for solid tumours. MSLN RNA expression was analysed in large series of CRC tumours (n = 640) and cell lines (n = 150), to evaluate its distribution and to identify MSLN-overexpressing models. NK-92 cells were transduced with anti-MSLN CAR, and subsequently sorted and cloned. Activity of CAR-NK-92 cells against target-expressing ovarian and CRC cells was assessed in vitro and in vivo. Statistical significance of efficacy was evaluated by t-test and log-rank test.

Results: Large-scale expression analysis highlighted that about 10% of CRCs overexpress MSLN at levels comparable to those of ovarian cancer, a typical target of MSLN-CAR-based therapy. Intriguingly, MSLN overexpression is more frequent in poor prognosis and KRAS/BRAF-mutant CRC. Lentiviral transduction of NK-92 cells with the MSLN-CAR, followed by sorting and cloning, led to the identification of one clone, MSLN.CAR.NK-92.cl45, stably expressing the CAR and retaining the NK phenotype. As expected, the clone demonstrated significant in vitro and in vivo activity against ovarian cancer cells. When repurposed against models of CRC expressing high MSLN levels, it displayed comparable efficacy, both in vitro and in vivo. Specificity of the clone was confirmed by the absence of activity on control models with low or absent MSLN.

Conclusions: Our results provide preclinical evidence that a subset of colorectal cancers expressing high mesothelin levels can be effectively targeted by MSLN-CAR-based immunotherapy. The potential therapeutic impact of these findings is enhanced by the fact that frequently MSLN-overexpressing CRCs display worse prognosis and resistance to standard care.

Fig. 1

MSLN mRNA expression in CRC tumours and cell lines. A Dot plot showing MSLN mRNA expression in CRC and Ovarian cancer data from the TCGA PanCancer Atlas. The red line indicates the threshold for CRC MSLN expression to be considered similar to ovarian cancer. B, C Dot plots showing MSLN mRNA expression in CRC TCGA data (B) and cell lines (C), stratified by CRIS subtypes. Kruskal–Wallis test for higher MSLN expression in CRIS-B vs non-B cases yielded p-values < 0.004–0.0001. D Dot plot showing in CRC cell lines higher MSLN expression in KRAS/BRAF mutant CRC

Fig. 2

Flow cytometry analysis of MSLN protein expression in selected CRC cell lines. A MSLN expression on the surface of control cell lines known to be MSLN-positive (HeLa, purple) and negative (293 T, grey). B MSLN expression on the surface of CRC cell lines (COLO320, MSLN-, grey; HCT116 MSLN ±, red; LIM2099 MSLN +, green; SW837 MSLN + + , blue)

Fig. 3

MSLN-CAR transduction, sorting and cloning. A Schematic representation of the lentiviral vector for expression of the MSLN CAR under the human EF1A promoter. The car is composed of the MSLN-specific scFv antibody fragment (VH + VL), a modified CD8α hinge region (CD8α) and transmembrane domain (CD8TM), followed by the intracellular domains of 41BB and CD3ζ. B Flow cytometry analysis of MSLN-CAR surface expression on NK-92 cells, before and after CAR transduction and sorting. C Flow cytometry analysis of MSLN CAR expression (mean fluorescence intensity) in 68 clones obtained after sorting transduced cells; yellow squares show a second MFI measurement on selected clones after expansion. The red circle indicates the selected clone 45. D Flow cytometry analysis of MSLN CAR expression in NK-92 WT (grey) vs the sorted population (blue) and clone 45 (red)

Fig. 4

In vitro efficacy of MSLN.CAR.NK-92.cl45 against CRC cells. A–C Assessment by flow cytometry of cancer cell viability (%) after 24 h of coculture with NK-92 WT (white bars) or MSLN.CAR.NK-92.cl45 (black bars) at different E:T ratio, as indicated. D–F Assessment by luciferase chemiluminescence of cancer cell viability (%) after 24 h of coculture with NK-92 WT (white bars) or MSLN.CAR.NK-92.cl45 (black bars) at different E:T ratio, as indicated. G Quantification by ELISA of INFγ release in the supernatant of different cell lines after coculture (E:T = 1:1) with NK-92 WT (white bars) or MSLN.CAR.NK-92cl45 (black bars). (H) Flow cytometry analysis of the degranulation marker CD107a in different cell lines after coculture (E:T = 1:1) with NK-92 WT (white bars) or MSLN.CAR.NK-92cl45 (black bars). For all plots, data are expressed as the mean ± SD, with n ≥ 3. Stars indicate t-test p-values: *p < 0.05, **p < 0.01, ***p < 0.001

Fig. 5

In vivo efficacy of MSLN.CAR.NK-92.cl45 against CRC cells. A–C–E–G In vivo growth of CRC cell line xenografts, either untreated (green line), or treated with NK-92 WT cells (blue line) or MSLN.CAR.NK-92.cl45 cells (red line): HCT116 silenced for MSLN (A), HCT116 overexpressing MSLN (C), LIM2099 cells (E), and SW837 cells (G). Data are expressed as median ± SEM (cohort size = 6). Stars indicate T-test p-values comparing tumour size of the MSLN.NK-92.cl45 treated cohorts vs the CTRL cohorts at the endpoint: *p < 0.05; **p < 0.01. B-D-F–H Survival curves of the same cell lines, as indicated. Stars indicate the p-value of Mantel-Cox tests comparing CTRL and MSLN.NK-92.cl45: **p < 0.01; *** ≤ 0.001