Mitotic stress-induced secretome primes cancer cells to apoptosis and maximizes paclitaxel response in breast tumors when combined with BCL-xL-targeting BH3 mimetics

ABSTRACT

We recently identified a previously unappreciated ability of antimitotics to propagate apoptotic priming across cancer cell populations. The underlying paracrine cytotoxic signal, fueled by undead cells activating the cGAS/STING pathway, is required for in vivo antitumor response and it can be further exploited by delayed, but not synchronous, BCL-xL inhibition.

Upon exposure to antimitotic drugs, cancer cells experience highly heterogeneous fates due not only to their individual proliferative status but also, when cycling, to an intrinsic competitive molecular race between cell death and mitotic slippage signaling networks. Indeed, in antimitotic- responsive cycling cancer cells, drug-induced mitotic arrest leads either to regulated cell death (mainly based on apoptosis during mitotic arrest) or to post-slippage multinucleation followed by resumption of transcriptional programs and/or apoptosis.^1,2 Stochastic variations in key actors of instrumental pathways contribute to the various observed cell fates and finally to tumor fractional response to antimitotic therapy. This may significantly participate in the maintenance of antimitotic resistant residual cancer cells, potentially responsible for later cancer relapse.

Apoptosis is executed by caspases activated downstream of mitochondrial outer membrane permeabilization (MOMP), itself regulated by a network of intracellular interactions between anti-apoptotic BCL-2 homologs (BCL-xL, MCL-1) and proapoptotic counterparts (multidomain BAX/BAK and BH3-only proteins such as BID, BIM, or NOXA).³ We and others showed that, among anti-apoptotic proteins, BCL-xL is strongly required for survival maintenance either during mitotic arrest or in multinucleated cells. This underlines that BH3 (BCL2 homology 3) mimetic inhibitors of BCL-2 homologs, of BCL-xL in particular, may improve chemotherapeutic outcome but, even though this was somehow confirmed in preclinical studies, a clear understanding of how exactly antimitotic drugs may promote BCL-xL dependency in heterogeneously cycling cancer cell populations is required.

We explored this notion and recently established that antimitotic treatment enhances BCL-xL-dependent apoptotic priming in cancer cells not only by intrinsic signals (in cycling cancer cells) but also by extrinsic ones (in nonproliferating cells). Using human primary breast tumors maintained in organotypic or organoid cultures, manipulated ex vivo or grafted in in vivo models, in addition to genetically engineered cell lines, we indeed observed that upon antimitotic treatment of sensitive breast tumors, proliferating cancer cells activate the cytosolic DNA sensor cGAS/STING pathway to produce a type I interferon (IFN-I) and TNFα proapoptotic secretome. Most importantly this effect was critical for paclitaxel tumor response since genetic ablation of either cGAS or STING strongly impaired antitumor efficacy in in vivo immunodeficient models. Mechanistically, we identified antimitotic-induced micronuclei as providers of danger-associated molecular patterns (DAMP) prone to recruit cGAS upon nuclear membrane collapse and cytosolic exposure of unprotected DNA (experimentally prevented by Lamin-B2 surexpression).

Apoptotic signaling across the cancer cell populations exposed to antimitotics relies in great part on transcriptional induction of NOXA coding gene PMAIP1, both in mitotically stressed secretory cells and in cells in receipt of paracrine signals. The exact mechanisms involved in NOXA induction in either case remain to be detailed. Since NOXA essentially functions as an inhibitor of MCL-1, this expression enhances apoptotic load on BCL-xL and sensitizes both donor and recipient cells to BH3 mimetic inhibitors of the latter. Consistent with this, we observed that the administration of a BH3 mimetic targeting BCL-xL after paclitaxel improved in vivo response (see Figure 1). Enhancement of the pro-apoptotic effects paclitaxel-induced secretome by BCL-xL inhibition seems all the more critical as this secretome can also contribute to the expansion of tumor-initiating cells in the long term.⁴ Importantly, we found that synchronous treatments with paclitaxel and BCL-xL inhibitors were not as efficient as delayed ones, most likely because they kill mitotically stressed (BCL-xL dependent) cells and thereby prevent them from producing a critical secretome. Thus, timing in the use of BCL-xL inhibitors needs to be finely tuned. As this implies that a significant number of cells capable of providing a paracrine cytotoxic signal are required for in vivo efficiency, it brings further support to the notion that communications between cells in receipt of death stimuli are critical for populational responses.⁵ More precisely here, the presence of slowly dying (or even resistant) cancer cells is expected to foster collective tumor response, advocating that transiently attenuating apoptotic rates might be paradoxically useful for cancer treatments.⁶ A contrario, this implies that residual tumors may present a heterogenous landscape wherein cells resistant to paracriny coexist with cells whose high sensitivity to death induction by antimitotics prevents them from producing it. This substantially differs from the classically envisioned view of residual diseases as homogenously composed of intrinsically resistant cells.

Figure 1.

A mitotic stress-induced secretome maximizes the sequential combination of paclitaxel and a BCL-xL-targeting BH3 mimetic.

Paclitaxel induces a mitotic stress in the subset of cycling cancer cells that leads to the production of an inflammatory secretome including TNFα and type I Interferons through cGAS/STING signaling pathway. This secretome spreads apoptotic priming in the whole cancer cell population, which is revealed by the delayed use of BCL-xL targeting BH3 (BCL2 Homology 3) mimetics. How this impacts immune tumor response and potential long-term tumor relapse, remains to be studied.

Our work raises the possibility that modifications in the cGAS/STING and TNF/IFN-I signaling pathways (such as silencing STING, rewiring of cytokine signaling …) might contribute to antimitotic resistance. Alterations in restricted subsets of cancer cells (were they to exist prior selection or to adapt to it) might suffice. This work evokes others, which ascribed to STING tumor-intrinsic ambivalent effects, ranging from anti- (cell death) to pro- (metastasis) tumoral functions.^7,8 In addition, given the critical roles played by the cGAS/STING and TNF/IFN-I signaling pathways in antitumor immune response our work also suggests that antimitotic treatments might indirectly impact on the immune system through the induction of an inflammatory secretome. Even though IFN-I signatures were correlated to apoptosis/antiproliferative responses in tumors and to favorable clinical outcome in the context of anthracycline-based chemotherapy,^9,10 chronic IFN-I signaling may have also pro-tumor effects including immune suppression and metastasis promotion.¹¹

Finally, how antimitotic-induced secretome influences long-term tumor evolution and how would BCL-xL inhibition interfere with immune control need now to be deciphered and deserve careful consideration.

Funding Statement

Our work is funded by the Ligue contre le Cancer Grand-Ouest, the Canceropole Grand-Ouest, the Region Pays de la Loire and the Agence Nationale de la Recherche and the SIRIC ILIAD program.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.