The present study addresses an unmet pharmacological need in TNBC, a subtype characterised by the absence of targetable hormone receptors and the overexpression of EGFR, which collectively limits therapeutic options and drives poor clinical outcomes. Using MDA-MB-231 cells as an established model of highly metastatic TNBC, we systematically characterised the cytotoxic and anti-proliferative activity of three phytocannabinoids (CBD, CBG, and CBD-A) and the sesquiterpene β-caryophyllene (BCP), both as single agents and in combination.

The antiproliferative activity of CBD in MDA-MB-231 cells is consistent with prior reports implicating the EGF/EGFR axis as a relevant target in this context [33] and extends earlier work by providing GR-normalised potency data suitable for quantitative combination analysis. CBG, by contrast, remains poorly characterised in breast cancer models; the present data provide quantitative evidence of its antiproliferative and cytotoxic potential, adding to the limited pharmacological profile of this cannabinoid in oncology. CBD-A exhibited modest cytotoxicity only at the highest concentrations tested and was excluded from combination studies, both for its limited efficacy and its well-documented chemical instability: thermal decarboxylation to CBD during standard extraction and handling procedures compromises reproducibility. Although CBD-A may retain pharmacological relevance in other contexts, including anti-migratory activity in MDA-MB-231 cells [36], its physicochemical liabilities preclude its inclusion in a rigorous combination design.

Three binary and ternary combinations of CBD, CBG, and BCP emerged as the most potent growth-inhibiting treatments. Because conventional IC₅₀-based metrics do not account for differences in cell division rates across experimental conditions, drug responses were quantified using growth rate (GR) inhibition metrics, which normalise drug effect to the proliferation rate of untreated controls and thus provide a more informative measure of compound potency [17]. A critical observation, only accessible through GR-based quantification, was that these three combinations produced fundamentally different biological outcomes despite appearing equally effective in endpoint metabolic assays. Combinations C1 and C3 drove net cytotoxicity (GR < 0), whereas C2 produced predominantly cytostatic effects (GR near 0). This distinction has direct therapeutic implications: cytotoxic combinations eliminate cells, while cytostatic ones arrest their proliferation without reducing viable cell number, a difference that is masked entirely by endpoint viability readouts calibrated at a single time point.

Synergy quantification was performed using SynergyFinder+, which integrates four reference models (HSA, Bliss, ZIP, and Loewe) across full dose-response matrices [19]. Combinations lacking CBD consistently showed reduced efficacy and lower synergy scores across all models, confirming CBD as the indispensable pharmacological anchor of the active combinations. Among the four models, Loewe additivity consistently yielded the lowest synergy scores, an outcome that reflects a conceptual limitation of this model in the present context. Loewe additivity assumes mutual dose-substitutability between agents, which is only valid when compounds share identical mechanisms of action. CBD, CBG, and BCP engage partially overlapping but pharmacologically distinct targets: both CBD and CBG modulate TRPV1 channels and PPARγ signalling, while BCP acts as a selective CB₂ receptor agonist [34–35]. Given this partial, rather than complete, target overlap, Bliss independence and the related ZIP model, both of which are grounded in the assumption of probabilistic independence between agents, provide a more appropriate framework for interpreting multi-drug interactions in this setting [19].

The partial additivity observed for CBD and CBG combinations aligns mechanistically with their shared engagement of TRPV1 and PPARγ: when two compounds modulate the same pathway, their effects tend toward reinforcement rather than synergistic amplification. In contrast, CBD–BCP combinations displayed marked synergy, consistent with the non-redundant character of their respective mechanisms. BCP, as a selective CB₂ agonist, inhibits cancer cell proliferation through a pathway that is largely distinct from CBD’s multi-target pro-apoptotic activity, which operates through PPARγ upregulation and the concurrent suppression of mTOR and cyclin D1 [37–39]. Although CBD has negligible intrinsic affinity for CB₂, its combination with a selective CB₂ agonist enables concurrent engagement of CB₂-dependent and CB₂-independent death pathways, a pharmacological complementarity consistent with previously reported synergistic interactions between these two compounds in analgesic models [40].

At the cellular level, high-resolution label-free holotomographic imaging provided a morphodynamic characterisation of the death process triggered by the most efficacious combinations. Combinations containing BCP (C1 and C3) induced extensive cytoplasmic vacuolisation as an early and prominent feature, consistent with the effects reported for THC, CBD, and CBG on breast cancer cell lines, where vacuolisation has been attributed to ER stress and dysregulation of autophagic flux [13, 41]. The subsequent appearance of caspase-3 activation, cell shrinkage, cytoplasmic blebbing, and apoptopodia confirmed a terminal apoptotic phase, consistent with the pro-apoptotic profiles established for CBD [13], CBG [42] and BCP [43]. However, the ordered morphological sequence observed here: vacuolisation associated with focal swelling of the perinuclear space preceding plasma membrane blebbing, is not consistent with canonical apoptosis. In contrast, the morphological intermediary phase, characterised by focal swelling of the perinuclear space and progressive nuclear compression occurring before membrane fragmentation, corresponds to the hallmarks of autosis, a regulated form of cell death mechanistically distinct from apoptosis, necroptosis, and ferroptosis. Despite its emerging recognition in several pathological contexts, autosis remains poorly characterised in breast cancer, and only one prior study has documented its occurrence in TNBC specifically. Zhang et al. demonstrated that a Beclin 1-targeting peptide, Tat-SP4, triggered autotic cell death in MDA-MB-231 and two additional TNBC cell lines, exploiting an intrinsic vulnerability rooted in the haploinsufficiency of Beclin 1, which is monoallelically deleted in 40–75% of breast cancers and is expressed at particularly low mRNA levels in TNBC relative to other breast cancer subtypes [32]. The resulting autophagy deficiency, rather than conferring resistance, paradoxically renders TNBC cells susceptible to autosis: when residual autophagic capacity is overwhelmed, even a moderate pharmacological increase in autophagy induction can exceed the cell’s homeostatic threshold and precipitate autotic death [32]. The present findings extend this paradigm to a pharmacologically distinct class of inducers, demonstrating that phytocannabinoid–terpene combinations engage the same vulnerability through a completely different mechanistic entry point (without direct Beclin 1 targeting) and through entirely distinct upstream receptor pharmacology.

Autosis is defined by two absolute requirements: (i) functional autophagy machinery, since blockade of Beclin 1, ATG5, ATG7, ATG13, or ATG14 abolishes the phenotype, and (ii) activity of the Na⁺, K⁺-ATPase pump [25]. Its biochemical definition is further refined by its resistance to caspase inhibitors, necroptosis inhibitors, and ferroptosis inhibitors, which collectively exclude it from all major previously defined regulated death pathways. The morphological sequence of autosis follows a staged ultrastructural programme: an early phase of autophagosome accumulation with convoluted nuclear membranes and dilated ER; a mid-phase of inner/outer nuclear membrane separation; and a late phase of focal swelling of the perinuclear space, which exerts mechanical pressure on the nucleus, inducing its characteristic concave deformation [25,30,32]. Autotic cells maintain substrate adhesion throughout, in direct contrast to the detachment that typifies apoptosis, a feature directly observable in live-cell imaging.

The molecular axis governing autosis centres on a direct physical interaction between Beclin 1 and the α-subunit of Na⁺,K⁺-ATPase. This interaction is dynamically regulated: it increases substantially under autosis-inducing conditions such as starvation and ischaemia, and is disrupted by cardiac glycosides, whose protective effect against autosis appears to operate at least partly through blocking this protein–protein interaction rather than solely through enzymatic inhibition of the pump [25].

The trigger for autosis is the net accumulation of autophagosomes, which can arise from excessive induction of autophagosome biogenesis, from a blockade of autophagosome-lysosome fusion, or, as is particularly relevant in TNBC, from an insufficient baseline lysosomal clearance capacity that is rapidly saturated upon pharmacological stimulation. In the context of cannabinoid action, CBD, CBG, and BCP individually induce autophagic flux in breast cancer cells [13, 41, 42, 43]; the present data suggest that at the concentrations deployed in C1 and C3, their combination generates a net surplus of autophagosomes that exceeds the degradative capacity of MDA-MB-231 cells, precipitating autotic engagement. This interpretation is fully consistent with the model established by Zhang et al., in which TNBC cells with low Beclin 1 expression tolerate basal autophagy yet lack the spare capacity to absorb a pharmacologically imposed increase in autophagic load [32].

The morphodynamic analysis further identified a temporally invariant feature of the death programme: although C1 induced a higher rate of entry into the autotic phase than C3 (92.3% vs. 82.4% of cells, respectively), the interval between perinuclear space swelling and the transition to apoptotic blebbing was fixed at a median of 11.3 hours for both combinations. This kinetic invariance across pharmacologically distinct stimuli suggests that the autosis-to-apoptosis transition is governed not by upstream induction conditions, but by an intrinsic cellular threshold, likely reflecting the time required for irreversible depletion of bioenergetic reserves, disruption of ionic homeostasis beyond a critical point, or the accumulation of sufficient organellar damage to engage caspase-dependent membrane fragmentation. The conservation of this temporal window is consistent with the classification of the autosis-to-apoptosis transition as a non-stochastic, hard-wired programme in which the upstream stimulus modulates commitment probability, but not the kinetics of execution once commitment has occurred. To the best of our knowledge, this sequential engagement of autosis followed by a transition to apoptotic execution, as a single, temporally ordered death programme, has not been previously documented in any cellular model. This point deserves emphasis: Zhang et al., in the only prior report of autosis in TNBC, described autosis as the terminal event itself, culminating in necrotic plasma membrane rupture; rescue by the Na⁺,K⁺-ATPase inhibitor digoxin was complete, and no subsequent caspase-dependent phase was identified [32]. The present data indicate a fundamentally different trajectory: perinuclear space swelling systematically precedes, and is kinetically decoupled from, subsequent apoptotic blebbing, providing evidence that autosis and apoptosis can operate as consecutive modules within a single integrated death cascade rather than as independent or parallel processes. Critically, prior studies reporting the co-occurrence of autophagic and apoptotic features in cannabinoid-treated cancer cells established co-occurrence as a morphological observation, not mechanistic succession [13, 41]. This distinction implies that the upstream autotic phase represents a discrete, pharmacologically accessible commitment point that precedes, and may be required for, the subsequent apoptotic collapse.

From a therapeutic standpoint, the present findings carry two notable implications. First, the cannabinoid–terpene combinations characterised here achieve strong cytotoxic efficacy at sub-IC₅₀ concentrations of each component, a dose-sparing effect that reduces the likelihood of off-target toxicity while potentially broadening the therapeutic window. Second, the involvement of autosis as an initiating mechanism diversifies the death pathways engaged: because autosis is independent of caspases and is not suppressed by the resistance mechanisms that typically attenuate apoptosis (e.g., Bcl-2 overexpression, p53 loss), a combination strategy that triggers autotic entry before transitioning to apoptotic execution may be less susceptible to the resistance acquisition that limits single-agent apoptosis-based approaches. The polypharmacological nature of these compounds, engaging CB₂, TRPV1, PPARγ, mTOR, and the autophagy–Na⁺,K⁺-ATPase axis concurrently, may further reduce the selective pressure driving resistance emergence. That this cytotoxic programme can be elicited in MDA-MB-231 cells from two mechanistically unrelated upstream stimuli (Beclin 1-targeting peptides [32] and phytocannabinoid–terpene combinations, as shown here) reinforces the view that autotic susceptibility is a tractable and reproducible pharmacological vulnerability in TNBC.

In summary, CBD, CBG, and BCP act synergistically in MDA-MB-231 cells to induce cytotoxicity through a temporally ordered death programme consistent with autotic initiation followed by apoptotic execution. This sequence, demonstrable only through kinetic live-cell imaging, illustrates a mechanistic complexity that is invisible to endpoint assays and underscores the value of morphodynamic characterisation in preclinical drug combination studies. Building on the single prior report of autosis in TNBC, the present data extend this emerging paradigm to phytocannabinoid–terpene combinations and reveal, for the first time, that autosis can serve as an upstream commitment module within a sequential death programme rather than as a terminal event per se. The identification of Na⁺,K⁺-ATPase-regulated autotic death as a mechanism accessible to pharmacologically diverse inducers opens a therapeutic avenue that is orthogonal to conventional apoptosis-targeted strategies in TNBC.

The use of high-purity reference compounds in the present study was a deliberate methodological choice, enabling unambiguous attribution of observed effects to defined pharmacological entities rather than to batch-dependent compositional variation. This reductionist approach provides a rational, mechanism-grounded framework directly transposable to complex formulations: hemp chemotypes naturally co-expressing CBD, CBG, and BCP, or standardised phytopharmaceutical preparations enriched in these constituents, represent immediate translational vehicles for the combinations characterised here. This perspective aligns with the observations of Blasco-Benito et al., who demonstrated that full-spectrum cannabis extracts produced markedly greater anticancer effects than purified THC alone, an outcome now interpretable considering the cooperative pharmacodynamic interactions documented in the present work [44].