The authors’ answer is that there is no single answer. mTOR assembles into two distinct complexes, mTORC1 and mTORC2, which share some upstream inputs but control largely non-identical downstream programmes. mTORC1 integrates amino-acid availability, growth-factor signalling and cellular energy status to regulate protein and lipid synthesis, metabolic adaptation and the repression of autophagy. mTORC2 adds a separate layer of control through AKT-dependent cell survival, cytoskeletal organization and stress adaptation. Because the biological output of either complex depends on the cell state, the spatial niche within the plaque, the substrate being measured and the magnitude and duration of pathway perturbation, the same nominal target can produce opposite consequences in different settings. The review’s central claim is that the question is not simply whether mTOR should be inhibited, but how a defined perturbation is translated into distinct cellular and lesion-level responses under specific biological contexts.

The pharmacological model used throughout the review is the rapalog class — rapamycin and its analogues such as sirolimus and everolimus. These agents bind the protein FKBP12 and inhibit mTOR allosterically. Acute exposure is weighted toward rapamycin-sensitive mTORC1 outputs, whereas prolonged exposure can gradually impair mTORC2 assembly and AKT signalling in susceptible cell types. This duration-dependent behaviour distinguishes rapalogs from ATP-competitive mTOR kinase inhibitors, which engage the catalytic site and can acutely suppress both complexes regardless of exposure time. The authors caution that conclusions about duration dependence, complex selectivity and pathway recovery should not be generalized automatically across all mTOR-directed compounds, because the same nominal target does not imply equivalent complex selectivity, substrate coverage or temporal pharmacology.

The strongest genetic evidence comes from plaque macrophages, and even there the picture is internally discordant. In a macrophage-specific Raptor-disruption model, which disables mTORC1, chemokine expression and experimental atherosclerosis were both reduced, indicating that selected macrophage mTORC1 programmes can promote disease in that context. mTORC2 tells a more complicated story. In lethally irradiated LDLR-deficient mice reconstituted with bone marrow carrying myeloid Rictor deficiency, loss of mTORC2 reduced lesion formation by impairing monocyte and macrophage proliferation and survival. Yet in an independent, non-transplanted LysM-Cre/Rictorflox/flox/ApoE−/− model, mTORC2 loss drove larger and more complex lesions, with expanded necrotic cores, heightened macrophage apoptosis and defective efferocytosis, driven by derepression of the transcription factor FoxO1 and IL-1β-dependent inflammatory signalling. The review stresses that these studies differed in host genotype, irradiation, bone-marrow transplantation and the context of Rictor deletion, so the opposing outcomes cannot be attributed to disease stage alone; stage dependence remains a hypothesis requiring direct testing in stage-matched models.

Smooth muscle cells illustrate why cell lineage alone is an inadequate guide to mTOR function. SMCs provide the structural integrity of the fibrous cap, yet they can lose canonical contractile markers, adopt alternative phenotypes and contribute to foam-cell populations. In advanced atherosclerosis, P2RY12-driven PI3K–AKT–mTOR signalling suppresses autophagy in SMCs, promoting SMC-derived foam-cell formation and lesion progression in the studied model — a disease-specific pathological programme that should not be generalized to imply that mTOR activity is uniformly detrimental across all SMC-derived states. Matrix-producing SMC-derived states support cap integrity, and the review proposes, as a hypothesis rather than a demonstrated requirement, that mTOR activity may support matrix synthesis in selected cap-associated SMC states. To keep these distinctions explicit, the authors introduce a three-tier evidence framework: tier one denotes direct genetic perturbation of mTOR or a complex component with lesion outcomes, tier two denotes disease-relevant mechanistic evidence without cell-specific genetic disruption, and tier three denotes supportive evidence from non-atherosclerotic or reductionist systems. Conflating these levels, they argue, would create false symmetry between directly demonstrated plaque mechanisms and biologically plausible but unproven homeostatic functions.

Amplitude and duration add a quantitative dimension to this context dependence. Rapamycin-sensitive mTORC1 substrates do not respond uniformly: S6K activity can remain strongly suppressed even at low drug exposure, whereas 4E-BP1 phosphorylation may be less completely or less durably inhibited, with the divergence varying substantially among cellular contexts. A drug concentration that virtually abolishes S6K phosphorylation may therefore leave 4E-BP1-dependent translation control largely intact, which means a single readout such as p-S6 cannot represent displacement of all mTORC1-regulated programmes. Earlier work showed that leucine-mediated macrophage mTORC1 activation suppresses mitophagy, promotes accumulation of dysfunctional mitochondria and increases macrophage apoptosis in experimental atherosclerosis, and a subsequent study identified a leucine-dependent threshold effect on mTORC1 activation and downstream responses in human monocyte-derived macrophages. The review emphasizes, however, that this threshold is specific to leucine stimulation, the macrophage context and the experimental conditions used, and cannot be directly extrapolated to define a rapalog dose–response curve or therapeutic window.

The pharmacological literature, though methodologically heterogeneous, does establish that mTOR perturbation can modify experimental atherosclerosis. Systemic rapamycin reduced plaque progression and monocyte and macrophage-associated responses in ApoE-deficient mice, and low-dose oral sirolimus reduced atherogenesis and vascular inflammation in LDLR-deficient mice. In cholesterol-fed New Zealand rabbits subjected to balloon-induced aortic injury, oral rapamycin initiated after lesion induction reduced plaque burden and inflammatory-marker expression, increased fibrous-cap thickness and lowered a histological vulnerability index — although the reported reduction in plaque rupture should be interpreted cautiously because rupture was pharmacologically provoked rather than observed spontaneously. In ApoE−/−Fbn1C1039G+/− mice, everolimus initiated only after 12 weeks of Western diet, when advanced lesions were already present, reduced circulating Ly6Chigh monocytes and plaque macrophage content, abolished intraplaque neovascularization and was associated with improved cardiac function and survival. Drug-eluting stents demonstrate that local arterial mTOR inhibition is feasible, but pathological studies also document delayed healing and incomplete re-endothelialization, and restenosis after percutaneous intervention follows a mechanical-injury programme that is not equivalent to spontaneous lipid-rich plaque evolution.

Lesion-directed delivery systems push the evidence further toward spatial control. Platelet-membrane-coated rapamycin nanoparticles preferentially accumulated in atherosclerotic lesions and reduced plaque progression while improving indices of plaque stability in ApoE-deficient mice. LOX-1-targeted rapamycin spherical nucleic acids enhanced plaque-directed delivery and reduced atheroma progression with improved plaque-stability measures. Biomimetic leukosome-based delivery reduced aortic macrophage proliferation and pro-inflammatory signalling after a short treatment schedule, and a 2026 study evaluated a PP1-peptide-targeted, ultrasound-responsive rapamycin nanosystem in ApoE−/− mice after 12 weeks of high-fat feeding had already established plaques; subsequent 21-day treatment enhanced plaque localization and, combined with low-intensity pulsed ultrasound, reduced plaque and lipid area together with macrophage- and inflammation-associated readouts. The review concludes that these platforms demonstrate the feasibility of localized delivery and lesion remodelling, but that drug accumulation does not establish cell-specific target engagement or pharmacodynamic selectivity, and histological stability indices are not equivalent to reduced spontaneous plaque rupture.

To close these gaps, the authors propose an exposure-guided framework that separates quantities often treated as interchangeable: administered dose, systemic exposure, plaque drug content, local pharmacologically available drug, cellular availability, FKBP12-dependent target engagement, substrate-specific pharmacodynamic response and pathway recovery. Total plaque drug, total intracellular drug and mTOR target occupancy are non-equivalent, because rapamycin must first bind FKBP12 before the complex engages the FRB domain of mTOR, and lesion architecture — extracellular matrix, lipid content, cellular density and local transport — can each alter arterial drug distribution. A therapeutic window for plaque-directed rapalog modulation is presented as putative rather than established: its lower boundary would correspond to exposure insufficient to modify the intended maladaptive output, and its upper boundary to exposure deep or prolonged enough to perturb unintended homeostatic programmes. Intermittent or pulse exposure is framed as a testable strategy, contingent on measurable suppression of the pathological output, subsequent decline of active exposure, recovery of homeostatic signalling and no accumulation into quasi-continuous occupancy — none of which has yet been demonstrated in plaques.

Translational testing, the review argues, should stratify by lesion phenotype rather than propose universal dosing. In inflammation-dominant lipid-rich plaques, the candidate objective is suppression of maladaptive inflammatory outputs while minimizing prolonged engagement of non-target programmes. In repair-fragile thin-cap plaques, the priority shifts to safety: whether sustained pathway suppression perturbs cap integrity, endothelial continuity and SMC-derived matrix-producing states. Pharmacodynamic readouts should combine p-S6 or p-S6K and p-4E-BP1 for mTORC1 displacement, and AKT Ser473 together with AKT Thr308, upstream feedback measurements and Rictor–mTOR complex integrity for mTORC2, since AKT Ser473 alone cannot establish rapalog-induced mTORC2 disruption when relief of mTORC1/S6K negative feedback can independently raise AKT phosphorylation. Serial imaging with OCT, IVUS, NIRS, CTA and PET could then link measured exposure and cell-resolved pharmacodynamics to structural plaque change. Four safety boundaries remain: systemic risks including immunosuppression, glucose dysregulation and dyslipidaemia; the improbability of absolute spatial selectivity; the inability of mouse pharmacology to establish human lesion penetration and recovery kinetics, requiring large-animal validation; and the fact that target engagement plus favourable imaging does not establish clinical event reduction. The near-term objective, the authors conclude, is not a precision mTOR dosing algorithm but the experimental definition of the exposure–engagement–response relationships that determine whether rapalog-mediated mTOR modulation produces benefit, neutrality or harm within the evolving plaque.

Subject of Research: Context-dependent mTOR signalling and exposure-guided rapalog pharmacology in atherosclerotic plaque stability

Article Title: Context-dependent mTOR pharmacology in atherosclerotic plaque stability and exposure-guided modulation

Article References: Han, R., Peng, M., Song, Z., Tang, Y., Yao, Z., She, Y., Cheng, S., Li, X., & He, P. (2026). Context-dependent mTOR pharmacology in atherosclerotic plaque stability and exposure-guided modulation. Frontiers in Pharmacology, 17, Article 1994231. https://doi.org/10.3389/fphar.2026.1994231

Image Credits: AI Generated

DOI: 10.3389/fphar.2026.1994231

Keywords: mTOR, atherosclerosis, rapamycin, rapalogs, plaque stability, macrophages, smooth muscle cells, mTORC1, mTORC2, drug delivery, pharmacokinetics, target engagement