PD-L1 and mTOR form a signaling loop that drives tumor invasion

Programmed death-ligand 1, the transmembrane glycoprotein that tumors exploit to switch off cytotoxic T lymphocytes, is now emerging as something far more subversive than an immune checkpoint. A narrative review published in Molecular Biology Reports pulls together experimental evidence that PD-L1 actively drives cancer cell migration and invasion through a reciprocal signaling relationship with mechanistic target of rapamycin, the central serine/threonine kinase that integrates growth, metabolic, and environmental cues. In this crosstalk, mTOR activity promotes PD-L1 expression, while PD-L1 itself appears to feed back into mTOR activation, forming a loop that sustains the mesenchymal, invasive phenotype that tumors need to metastasize.

The canonical story is well known. PD-L1, also called CD274 or B7-H1, is a B7-family glycoprotein of the immunoglobulin superfamily, expressed on T cells, B cells, macrophages, and dendritic cells as well as many tumor types. When it binds PD-1 on activated lymphocytes, phosphorylation of the immunoreceptor tyrosine-based inhibitory and switch motifs recruits the phosphatases SHP-1 and SHP-2, dampening T cell receptor signaling, proliferation, cytokine production, and cytotoxicity. Cancers including lung cancer, melanoma, glioma, and breast cancer co-opt this axis through PD-L1 upregulation, often associated with poorer prognosis. But the review emphasizes that PD-L1 expression in cancer cells is not always a marker of intrinsic tumor biology; it can be an adaptive response to cytokines such as interferon-gamma, tumor necrosis factor-alpha, and interleukin-6 secreted by a pre-active immune infiltrate.

The non-canonical functions are where the picture becomes provocative. Genetic or pharmacological suppression of PD-L1 reveals direct cell-intrinsic dependencies across cancer types. In gastric cancer cells, PD-L1 knockdown impaired proliferation, migration, invasion, cell-cycle progression, and tumorigenicity. In triple-negative breast cancer, PD-L1 depletion reduced proliferation and triggered both intrinsic and extrinsic apoptosis. In murine sarcoma models, tumor-specific PD-L1 blockade disrupted growth and attenuated glycolytic metabolism via mTOR suppression. PD-L1 also functions as an intracellular signaling molecule: in TNBC cells, cytoplasmic PD-L1 binds and stabilizes mRNA transcripts encoding DNA damage response proteins, while nuclear PD-L1 cooperates with DNA-dependent protein kinase to activate the MEK/ERK pathway. Its nuclear import is facilitated by vimentin, phosphorylated STAT3, and karyopherin beta 1, and once inside the nucleus PD-L1 transcriptionally activates pro-tumorigenic genes including GAS6, WIP, and beta-catenin by binding their promoters.

PD-L1 is also tightly linked to cancer stemness and epithelial-mesenchymal transition, the phenotypic reprogramming marked by increased N-cadherin and vimentin and decreased E-cadherin, orchestrated by transcription factors such as SNAIL1, SNAIL2, and ZEB1. RNA interference studies show PD-L1 directly regulates core stemness factors including OCT4, NANOG, and BMI1, and it induces OCT4 and NANOG expression through PI3K/AKT signaling, feeding into mTOR-dependent control of self-renewal. In colorectal carcinoma, PD-L1 induces a stem and mesenchymal state by interacting with HMGA1 and activating PI3K/AKT and MEK/ERK. In glioblastoma and other epithelial malignancies, PD-L1 overexpression upregulates N-cadherin, vimentin, SLUG, and beta-catenin while downregulating E-cadherin through ERK-dependent mechanisms. In hypopharyngeal squamous cell carcinoma, PD-L1 knockdown reduced N-cadherin and vimentin alongside diminished AKT phosphorylation, while enforced PD-L1 expression boosted migration and invasion in nasopharyngeal carcinoma through PI3K/AKT activation.

The relationship with EMT appears bidirectional. EMT-associated transcription factors can induce PD-L1 expression, while PD-L1 contributes to EMT maintenance. In breast cancer, EMT induction increases PD-L1, whereas PD-L1 silencing reverses EMT markers, with a positive correlation between EMT gene signatures and PD-L1 expression in patient samples. In lung cancer cells, ZEB1 induces PD-L1 via miRNA-200 and IFN-gamma signaling, and in NSCLC, TNF-alpha and TGF-beta co-stimulation elevates PD-L1 through ZEB1 and SNAIL1. In colorectal tumor cells, PD-L1 deletion or silencing impairs migration and invasion, an effect dependent on PD-L1 isoform c, whose overexpression increases EMT markers such as DDR1 and fibronectin while reducing E-cadherin. This shift reflects the hallmark EMT transition from cell-cell adhesion to cell-extracellular matrix interactions. The review concludes that PD-L1 may not be a passive biomarker of EMT but an active player sustaining the invasive phenotype while being induced by it.

On the mTOR side, the kinase operates through mTORC1, activated by amino acids, growth factors, and energy levels to promote protein synthesis, lipogenesis, and mitochondrial biogenesis; mTORC2, which regulates proliferation, survival, glucose metabolism, and cytoskeletal organization; and the more recently described mTORC3, identified in tumor contexts and able to sustain mTOR kinase activity even under rapamycin inhibition. Multiple lines of evidence show mTORC1 activation promotes EMT by upregulating Snail and Twist and downregulating E-cadherin, enhancing motility and invasion. In colorectal tumors, mTOR inhibition reduces EMT-associated gene expression and significantly decreases migration and invasion. mTORC2 reinforces stemness by sustaining AKT signaling that supports self-renewal and therapy-resistant subpopulations.

Direct evidence connects mTOR activity to PD-L1 regulation. In a mouse model of lung squamous carcinoma with biallelic deletion of Lkb1 and PTEN, both negative regulators of mTOR, PD-L1 levels were increased in lung tissue. NSCLC cell lines treated with PI3K and AKT inhibitors showed reduced PD-L1 expression, an effect mediated by mTORC1 rather than mTORC2, since RAPTOR but not RICTOR silencing prevented PD-L1 expression. mTOR activation through EGF and IFN-gamma also increased PD-L1 in NSCLC cell lines. Upstream regulators vary by tumor type: EGFR signaling correlates positively with PD-L1 in patient tumors, EGF stimulation upregulated PD-L1 in lung adenocarcinoma cells through PI3K/AKT and JAK/STAT pathways, and EML4-ALK rearrangements increased PD-L1 expression through MEK/ERK and PI3K cascades. In prostate cancer, PD-L1 expression depended on ATM kinase via JAK/STAT, and JAK inhibition was sufficient to downregulate PD-L1 and reverse EMT markers.

Strikingly, PD-L1 itself seems to activate mTOR. In a comprehensive glioblastoma study, PD-L1 overexpression increased cell migration and elevated phosphorylated ERK1/2, but not AKT or GSK-3beta; immunoprecipitation showed PD-L1 binding to RAS. MEK inhibition reduced the EMT markers SLUG and vimentin and increased E-cadherin, indicating the RAS/MEK/ERK pathway mediates PD-L1-driven migration. In vivo, GBM cells overexpressing PD-L1 showed greater tissue invasion with the same EMT markers. Given that the MEK/ERK pathway induces mTOR activation through TSC2 inhibition, PD-L1 binding to RAS may point to a direct link between PD-L1-expressing tumor cells and mTOR activation. Downstream of mTOR, transcription factors such as STAT3 and MYC regulate PD-L1 expression: gefitinib reduced IL-6, PD-L1, and p-STAT3 in EGFR-mutant NSCLC, while the MYC adaptor BIN1 downregulated PD-L1 when overexpressed and increased it when silenced.

Therapeutically, the review argues that targeting PD-L1 should involve inhibiting its intracellular tumor signaling alongside mTOR pathway inhibition to reduce metastatic potential. Direct preclinical evidence remains scarce: in SW480 colorectal cancer cells, combined PD-L1 and PI3K/mTOR inhibition by BMS-1166 and BEZ235 reduced cell migration, the only study to date evaluating both manipulations. Clinical trials mostly target tyrosine kinase receptors upstream of mTOR. In JAVELIN Renal 101, avelumab plus the VEGFR inhibitor axitinib significantly improved 5-year progression-free survival and objective response rate versus sunitinib, with particularly strong responses in sarcomatoid RCC, a subtype associated with EMT markers and higher PD-L1. In CALYPSO, the c-MET inhibitor savolitinib plus durvalumab achieved a 34% response rate in metastatic papillary renal cancer and 53% in c-MET-driven patients, though PD-L1 and tumor mutation burden did not correlate with response. In contrast, CONTACT-03 showed no benefit and increased toxicity for cabozantinib plus atezolizumab after prior immunotherapy, suggesting prior ICI treatment influences efficacy. The ICE-CAP trial of the AKT inhibitor ipatasertib with atezolizumab was well tolerated, with biopsy samples showing reduced regulatory T cells and increased CD8-positive lymphocytes independently of PI3K mutational status.

The review concludes that PD-L1 upregulation is closely associated with EMT in an active manner across cancer types, that the mTOR pathway plays a relevant role in PD-L1 expression and EMT induction and maintenance, and that PD-L1 itself may sustain mTOR signaling. The authors acknowledge limitations: as a narrative rather than systematic review, study selection followed reported experimental data on the proposed pathways rather than predefined criteria, so the conceptual framework may only partially represent the PD-L1/mTOR intersection. Key open questions include which PD-L1 domains mediate intracellular signaling, the mechanisms by which PD-L1 sustains EMT, and the contexts in which it promotes mTOR activation. Defining tumor-specific upstream regulators of mTOR will be critical for patient stratification, and direct preclinical evidence for combined PD-L1 and mTOR inhibition on migration, invasion, and EMT is still needed to establish a rational basis for combination strategies that target both the immune and tumor-intrinsic functions of PD-L1.

Subject of Research: PD-L1 and mTOR signaling in tumor migration and invasion

Article Title: The PD-L1–mTOR intersection in tumor migration and invasion

Article References: Palavro, R., Previato, T. T. R., Pazzin, D. B., Gonçalves, J. B., Boff, M. O., Marinowic, D., da Costa, J. C., & Scheffel, T. B. (2026). The PD-L1–mTOR intersection in tumor migration and invasion. Molecular Biology Reports, 53(1), Article 1793. https://doi.org/10.1007/s11033-026-12982-6

Image Credits: AI Generated

DOI: 10.1007/s11033-026-12982-6

Keywords: PD-L1, mTOR, EMT, cancer invasion, immune checkpoint, tumor microenvironment, PI3K/AKT, cancer stem cells, intersection, tumor, migration, invasion