(W1230-02-09) LC-MS FFPE Tissue Analysis Across Multiple Murine Tissues Supports Mouse Relevance for On-Target/Off-Tumor Toxicity Assessment for Anti-STEAP2 CAR-T Therapeutic Development

Purpose: Six-transmembrane epithelial antigen of the prostate-2 (STEAP2) is a novel therapeutic target that has established high cell surface expression in prostate cancer.¹ AZD0754 is a first-in-class CAR-T candidate therapeutic targeting STEAP2 for prostate cancer treatment. The therapeutic development and safety assessment of this CAR-T cell included the use of an immunocompromised mouse xenograft model to demonstrate activity against tumor and to detect activity towards healthy tissue, also expressing STEAP2. Confirmation of STEAP2 protein expression across multiple healthy tissues in the pre-clinical species underpins the safety assessment and justifies the relevance of the in vivo model. No STEAP2 antibody was available for detecting murine STEAP2 by immunohistochemistry, so we established a high sensitivity nano-liquid chromatography high-resolution mass spectrometry (nanoLC-HRMS) approach to quantify endogenous STEAP2 levels in murine FFPE tissues.

Methods: Sample preparation: Murine formalin-fixed paraffin-embedded (FFPE) tissues were deparaffinized with xylene, followed by protein de-crosslinking, reduction, alkylation, and trypsin digestion.² Tryptic peptides were desalted, resuspended, and spiked with a constant level of internal standard (IS). The bioanalytical method employed STEAP2 peptide IPIEIVNK calibration and QC standards with C-terminally labeled ¹³C₆ ¹⁵N₂ IPIEIVNK as IS. Murine spleen FFPE was used as control matrix to quantify the endogenous STEAP2 in various murine FFPE tissues due to its undetectable expression of STEAP2. Data acquisition: The sample material was analyzed on a Thermo Q Exactive HF-X mass spectrometer coupled with a Waters nanoAcquity M-Class UPLC system. Parallel reaction monitoring (PRM) data were analyzed by Skyline software for STEAP2 targeted quantitation. Data independent acquisition (DIA) data were processed using Spectronaut.

Results: A major challenge in developing a STEAP2 FFPE method is the lack of representative standard material. While peptide standards are available, even a full-length recombinant protein is not representative of the membrane-bound analyte preserved in FFPE. We implemented a global proteomics approach to assess the effect of extraction conditions for potential protein targets. DIA consistently identified > 4000 proteins. However, approximately 15 % of the total proteins did not achieve consistent recovery from the FFPE tissue across the examined various extraction conditions. STEAP2 was not detectable in the DIA approach. Thus, when we moved to a more sensitive targeted quantitation approach for STEAP2 detection, all tissue samples were analyzed with PRM approach using optimized sample extraction conditions. The established bioanalytical method (nanoLC-HRMS PRM) employed eight non-zero calibration standards over the concentration range of 15.2 – 5000 amol/μg of peptide, and the correlation coefficients for individual runs were ≥0.985. Intra-assay accuracy and precision (A&P) were evaluated for each quality control sample (LLOQ, LQC, MQC, and HQC) by multiple analyses (n=6) of each level during one qualification run. All quality control samples passed ±20 % accuracy and precision acceptance criteria (except for LLOQ at ±25 %). The established reinjection reproducibility was determined to 92 hours at 2-8 °C. Additional qualification parameters included recovery (analyte spiked pre-extraction and post-extraction), chromatographic carryover, internal standard variability, and chromatographic interference. In-situ hybridization data for STEAP2 mRNA expression indicated high expression in murine kidney and low expression in the heart and liver tissues. However, confirmation at the protein level is necessary to establish NSG MHC I/II DKO murine model relevance for on-target/off-tumor toxicity assessments. Thus, the qualified methodology was applied to multiple murine FFPE tissues (kidney, liver, and heart) from five NSG MHC I/II DKO mice. The results demonstrate that STEAP2 protein was detectable the murine kidney but not in the heart or liver (see Figure 1).

Conclusion: The detectable STEAP2 level in the murine kidney supports the animal model suitability for the on-target/off-tumor toxicity assessment. This work demonstrates that nanoLC-HRMS is an effective tool for evaluating therapeutic target protein expression in FFPE tissue to advance current anti-cancer biologics and cell therapies. This work also highlights that extraction efficiency is a critical consideration and emphasizes the need to assess this factor individually for each protein during method development of targeted protein quantitation from FFPE tissue.

References: 1. Zanvit, P., van Dyk, D., Fazenbaker, C. et al. Antitumor activity of AZD0754, a dnTGFβRII-armored, STEAP2-targeted CAR-T cell therapy, in prostate cancer J. Clin. Invest. 2023, 133, e169655.
2. Sweet, S., Chain, D., Yu, W. et al. The addition of FAIMS increases targeted proteomics sensitivity from FFPE tumor biopsies. Sci. Rep. 2022, 12, 13876.

Acknowledgements: The authors thank Steve Sweet, David Chain, and Yeoun Jin Kim from AstraZeneca for helpful discussions. The authors declare the following competing financial interest(s): Liu Yang, David R. Spiciarich, Kévin Contrepois, Georgina Cornish, Clare Hoover, Alexandra Kathenes, Michael Lehmann, Deborah L. Berry, Anton I. Rosenbaum, and John K. Meissen are employees of AstraZeneca and may have stock ownership and/or stock interests or options in the company.

Figure 1: Chromatograms of STEAP2 (peptide: IPIEIVNK, product ion: y7++) from (A) MC38/STEAP2*, (B) LLOQ, (C) kidney, (D) liver, and (E) heart. *MC38 murine STEAP2 transduced cell line was used as a positive control.