Article Text

Download PDFPDF

319 Transcriptome-guided design of logic-gated CAR T cells for treatment of acute myeloid leukemia
  1. Tess Woodring1,
  2. Katherine A Freitas2,
  3. Elena Sotillo2,
  4. Crystal L Mackall2 and
  5. Rebecca M Richards1
  1. 1University of Wisconsin-Madison, Madison, WI, USA
  2. 2Stanford University School of Medicine, Stanford, CA, USA
  • Journal for ImmunoTherapy of Cancer (JITC) preprint. The copyright holder for this preprint are the authors/funders, who have granted JITC permission to display the preprint. All rights reserved. No reuse allowed without permission.


Background In acute myelogenous leukemia (AML), CD93 is an appealing target for chimeric antigen receptor (CAR) T-cell therapy, offering leukemic killing without myelosuppression in preclinical studies.1 However, CD93 CAR T-cell development is hampered by CD93 expression on healthy endothelium. Our group previously demonstrated the efficacy of an inhibitory CAR (iCAR) to counteract off-tumor CD93-directed activating CAR (aCAR) signaling in a model system.1 Here, we identify endothelial-specific iCAR targets in resting and cytokine-activated states to minimize off-tumor toxicity for logic-gated (a NOT b) CD93 CAR T cells.

Methods We analyzed differentially expressed genes (DEGs) from RNA-Seq of 3 AML (Kasumi-1, THP-1, NOMO-1) and 2 endothelial (iHUVEC, TIME) cell lines, incubated with and without proinflammatory cytokines (IFNγ, TNFα). We applied a statistical filter (FDR<0.05 & log2FC>10) to identify genes specific to endothelial cells and selected surface-expressed genes using annotations from a surfaceome database (figure 1).2 DEGs present in both resting and cytokine-activated conditions were reviewed as potential iCAR targets and cross-referenced with CD93 in a healthy transcriptome atlas (Tabula Sapiens) and 3,225 pediatric AML samples (TARGET-AML).3 4

Results From the thousands of DEGs in resting and cytokine-activated conditions, statistical and surfaceome filters reduced DEGs by 99%, with only half of remaining genes (N=16) upregulated in both conditions. Among these, we identified 3 potential iCAR targets (CHD5, KDR, TEK) encoding well characterized, single-pass receptors with strong endothelial cell expression (figure 2A). Validating our cell line data, we found iCAR targets expressed along with CD93 in all 21 organ-specific endothelial cell subpopulations in Tabula Sapiens (figure 2B). By contrast, median expression of iCAR targets was extremely low in TARGET-AML samples (<0.2 transcripts per million [TPM] for all targets), nearly 200 times less than that of aCAR target CD93 (figure 3). Notably, CD93 was detected in nearly all AML samples (97.2%, 3134/3225, TPM >1), with median expression twice that of existing AML immunotherapy target CD33 (CD93, 36.7 TPM; CD33, 17.2 TPM).

Conclusions Our findings suggest that VE-cadherin (CDH5), VEGFR-2 (KDR), and Tie2 (TEK) could function as iCAR targets in NOT-gated CD93 CAR T cells for AML, predicted to protect endothelium without sheltering leukemic cells. Limited overlap of endothelial-specific DEGs in resting and cytokine-activated conditions reinforces the importance of screening iCAR targets that remain stably expressed on healthy tissue during inflammation anticipated with immunotherapy. Overall, our work demonstrates an efficient and generalizable transcriptome-guided approach to design logic-gated CAR T cells.


  1. Richards RM, et al. NOT-Gated CD93 CAR T Cells Effectively Target AML with Minimized Endothelial Cross-Reactivity. Blood Cancer Discov, 2021;2(6):648–665.

  2. Bausch-Fluck D, et al. The in silico human surfaceome. Proc Natl Acad Sci U S A, 2018;115(46):E10988-E10997.

  3. Tabula Sapiens Consortium, et al. The Tabula Sapiens: A multiple-organ, single-cell transcriptomic atlas of humans. Science, 2022;376(6594):eabl4896.

  4. TARGET-AML, National Cancer Institute GDC Data Portal. [].

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.