top of page

A Novel MHC-Independent Mechanism of Tumor Cell Killing by CD8+ T Cells

Lerner*, Woroniecka* et al. (BioRxiv). DOI: 10.1101/2023.02.02.526713

A Novel MHC-Independent Mechanism of Tumor Cell Killing by CD8+ T Cells


●  CD8+ T lymphocytes

●  Major histocompatibility complex class I

●  NKG2D

Main Findings

Loss of major histocompatibility complex class I (MHC-I) expression is a common strategy of tumour cells to evade antigen-specific CD8+ T-cell-mediated immune surveillance and restricts the use of immune-checkpoint blockade (ICB). In this preprint, the authors provide experimental evidence that – surprisingly – a functional CD8+ T cell compartment is still required for immunotherapeutic efficacy in αPD-1/α4-lBB-treated mice implanted with tumours cells in which beta-2-microglobulin was genetically ablated (b2MKO). MHC-I-independent killing of b2MKO ­tumour cells by antigen-specific T cells in vitro could be achieved when they were co-cultured with antigen-loaded macrophages or b2MWT tumour cells expressing the cognate antigen. Similarly, antigen-specific T cells were found capable of killing antigen-mismatched (b2MKO) tumours in vivo upon adoptive transfer of macrophages loaded with the cognate antigen, suggesting that antecedent TCR is both necessary and sufficient to drive MHC-I-independent and antigen-agnostic killing of tumour cells by conventional CD8+ T cells. Mechanistically, this was found to depend on ligation of the activating C-type lectin-like receptor NKG2D on the surface of CD8+ T cells.


  • Figure 2A and S2A show that more antigen-specific (tetramer-positive) CD8+ T cells accumulate intratumorally in αPD-1/α4-lBB-treated mice with MHC-I-negative tumours compared to those with MHC-I-positive tumours. The authors do not discuss the mechanisms or implications of this finding, nor follow-up on it experimentally.

  • The bulk RNAseq data presented in Figure 5B also show differential upregulation of KLRC1 (which encodes the C-type lectin-like inhibitory receptor NKG2A), which might affect CD8+ T-cell activity upon co-culture with b2MKO cells (that lack surface expression of the NKG2A-ligand Qa-1, the murine orthologue of HLA-E). Therefore, any involvement of the NKG2A/Qa-1-axis in MHC-I-independent tumour cell killing should be ruled out by the authors.

  • The authors do not validate the bulk RNAseq data (Figure 5B) by flow-cytometric analysis of NKG2D expression in T cells co-cultured with TRP2-loaded macrophages and CT2A-TRP2-b2MKO cells. On a side note, others (Vries et al., 2023. Nature) have observed downregulation of NKG2D on effector cells in response to co-culture with MHC-Inull target cells, which was interpreted as involvement of NKG2D in its tumoricidal activity. How do the authors explain that NKG2D was instead upregulated on CD8+ T cells co-cultured with MHC-Inull tumours cells?

  • In Figure 5D and S4EFG the authors show NKG2D ligand expression of b2MKO tumours cells but not b2MWTcells. They write in the discussion that “only tumor expression of NKG2D ligand was required, and tumor cells lacking MHC-I appeared to have a reflexive increase in these ligands, leaving them more subject to this mode of cell kill”. As the authors do not provide a reference to any other primary research work demonstrating evidence for this claim, it would be necessary to verify if there is any difference in (ex vivo) NKG2D ligand expression between b2MKO and b2MWT tumours to support this claim.

  • The authors conclude that Fas/Fas-L interactions are not involved in the proposed MHC-I-independent CD8+T-cell-mediated killing mechanism. However, we believe that this claim should be interpreted with caution. Figure S4J shows an increase in Fas-L mRNA expression in antigen-cognate TCR-activated CD8+ T cells upon co-culture with MHC-Inull tumours. Figure S4K shows a small increase in Fas-L protein expression in αPD-1/α4-lBB-treated mice with b2MKO tumours versus those with b2MWT tumours, although this difference is not statistically significant. However, while the sample size (n = 4) in the b2MKO ­tumour/αPD-1/α4-lBB-treated mice group might just satisfy the minimum number of animals required according to a ‘resource equation’ calculation, it could as well be underpowered because the expected effect size might be very small, as observed in Figure S4J. The Fas knock-out experiment presented in Figure S4L does not include Yummer-FasWT + CT2A-OVA and Yummer-FasWT only controls, which are necessary to draw sound conclusions about Fas/Fas-L interactions being involved in the MHC-I-independent NKG2D-dependent killing mechanism.

  • Overall, the diversity of methods and assays used in the paper is relatively low. The authors only present flow-cytometry-based cytotoxicity assays using CellTrace dyes to determine tumour cell killing in vitro, which could be complemented by other assay modalities (e.g., 51Cr-release assay, bioluminescence or colorimetric assays or real-time impedance-based cytotoxicity assays). It would also be of considerable interest to provide a more detailed (ex vivo) phenotypic characterisation of the CD8+ T cells involved in the MHC-I-independent killing (including other effector functions, such as cytokine production).

  • Raw flowcytometry data (including gating strategies) are not available for the flow-based cytotoxicity assays.

  • Addressing the recent findings of De Vries et al. (2023, Nature) – i.e., that γδ T cells (in particular Vδ1+) are involved in killing of MHC-I-negative MMR-defective gastro-intestinal tumours in a manner partially dependent on NKG2D – either experimentally or in the main text, would further strengthen the manuscript.


The authors of this preprint have uncovered a novel mechanism of CD8+ T-cell-mediated tumour killing, which is independent of both MHC-I and tumour antigen expression. Herein, they identified a key role for the NKG2D receptor as an important cytotoxic receptor (similar to NK cells) rather than just being a (weak) costimulatory on CD8+ T cells. In this regard, the results presented in the preprint complement a recent study by De Vries et al. (Nature, Jan 2023) which demonstrates that γδ T cells preferentially kill tumours with HLA-I defects in a manner partially dependent on NKG2D/NKG2DL interactions.

The findings of this study challenge our traditional notions of anti-tumour immunity and ICB immunotherapy efficacy and, as such, might inform novel immunotherapeutic approaches, in particular for MHC-Inull/low tumours. At a more fundamental level, the findings make us rethink our current – and perhaps our most basic – understanding of CD8+T cell function, blurring the boundaries between the adaptive and innate, showing that TCR-activated CD8+ T cells can kill cells that do not present the cognate antigen through a bystander-like effect. (On a side note, this might mechanistically explain why CMV- and EBV-specific CD8+ T cells have been associated with enhanced tumour control and favourable patient outcomes (despite the tumour not expressing CMV- or EBV-associated antigens), which has been previously attributed to a bystander-like effect). The results of the preprint might have implications beyond immuno-oncology, for instance in the field of infectious diseases (e.g., viral infections are often associated with MHC-I downregulation.


Reviewed by Bas W.A. Peeters as part of a cross-institutional journal club between the Icahn School of Medicine at Mount Sinai, the University of Oxford, the Karolinska Institute and the University of Toronto.

The author declares no conflict of interests in relation to their involvement in the review.

bottom of page