Artificial Intelligence-Guided Cancer Engineering for Tumor Normalization Executed by Tumor Lysosomal-Triggered Supramolecular Chiral Peptide (2025)

Identifying Key Aberrant Signaling Pathways in LUAD Using AI

Researchers initially downloaded mRNA expression data from TCGA for 59 normal lung tissues and 515 LUAD lung cancer tissues. These data were analyzed using ssGSEA to evaluate critical cancer pathways and immune cell infiltration pathways (Figure S1). The results were then fed into ChatGPT 4OT, which analyzed the data and identified the gene sets contributing most significantly to cancer behavior in malignant tissues compared to normal tissues (Figure 1A). Further analysis using ChatGPT 4OT identified the most frequently activated signaling pathways in the cancerous tissues. Two AI assessments revealed that the key aberrantly activated signaling pathways in tumor tissues were the Wnt signaling, PD-L1 signaling, and macropinocytosis pathways (Figure 1A).

Furthermore, the mRNA expression level of CTNNB1 (β-catenin), a pivotal protein in the Wnt signaling pathway, exhibited significantly elevated levels in tumor tissues compared to healthy tissue in LUAD (Figure 1B), thereby corroborating the findings derived from AI analysis. Additionally, there was a statistically significant positive correlation observed between CD274 (PD-L1) mRNA expression and CTNNB1 levels (Figure 1C). Similar results were confirmed in macropinocytosis, where the mRNA expression levels of key proteins involved in its regulation were significantly elevated in cancerous tissues compared to healthy tissues (Figure S2A). Moreover, there was a significant positive correlation between the mRNA expression levels of these key regulatory proteins (particularly RAC1, CDC42, RAB5A, PIK3CA, EEA1, PAK1 and KRAS) and the expression level of CTNNB1 (Figure S2B). Additionally, it was discovered that lithium chloride, an activator of the Wnt signaling pathway, effectively augmented dextran uptake in LLC, PC-9 and erlotinib-resistant PC-9 lung adenocarcinoma cells, thus indicating further activation of the macropinocytosis pathway (Figures 1D and S2C,D). Finally, overexpression of β-catenin in LLC lung adenocarcinoma cells confirmed that activation of the Wnt/β-catenin pathway significantly enhances cellular macropinocytosis activity (Figure S2E,F). This observation was further substantiated by demonstrating that inhibitors of macropinocytosis such as Amiloride and Cytochalasin D impeded dextran uptake (Figure 1E). These findings indicate that macropinocytosis is activated in tumor cells, and this activation is dependent on the Wnt pathway, potentially facilitating nutrient acquisition for cancer cells. Collectively, these results further substantiate the AI driven discoveries that the key signaling pathways aberrantly activated in tumor tissues include Wnt signaling, PD-1 signaling, and macropinocytosis pathways.

Design, Synthesis, and Characterization of CCtor

Inspired by these clinical data analysis results, researchers hypothesized that concurrently correcting the aberrant activation of the Wnt and PD-1/PD-L1 pathways in LUAD patients could normalize tumor cells, thereby achieving tumor treatment. Additionally, the tumor-specific increase in macropinocytosis could be exploited for tumor-specific targeting of therapeutic molecules (Figure 2A). Given that macropinocytosis is an effective pathway for the internalization of soluble proteins, it was proposed to coassemble a previously developed Wnt pathway inhibitor CA with a D-type PD-L1 antagonist peptide (PPI) into a protein-like particle, named CCtor. For this assembly, the CA-induced polymerizing H1 sequence was linked to the C-terminus of PPI, leveraging the CA-induced polymerization to construct the protein-like particle CCtor (Figure 2A). Once internalized by tumor cells, the lysosomal degradation of the L-type polypeptide part of CCtor would release CA and the D-type dodecapeptide PPI, allowing CA to enter the cytoplasm and PPI to function extracellularly by binding to PD-L1 (Figure 2A).

To achieve this design, the polypeptide Ctor linking H1 and PPI was synthesized using solid-phase synthesis, purified using reverse-phase chromatography, and characterized for purity and molecular structure by LC-MS (Figure S3A). The supramolecular assembly of CCtor was constructed in a one-pot process induced by CA, with the particle size tunable by varying the Ctor ratio (Figure 2B). Transmission electron microscopy (TEM) observations showed uniform particle size with a diameter of about 218.2 nm at a Ctor mass ratio of 1:1. To ensure better penetration in tumor tissues, an equal ratio (Ctor = 1:1) was adopted for subsequent work. Furthermore, TEM also observations revealed that in the absence of the L-type H1 peptide (Sequence: VNLINYQDDAEL), the particles formed through the self-assembly of D-type PPI peptide and CA were sparse and irregular (Figure S4A). This finding was consistent with previous studies, suggesting that the L-type H1 peptide facilitates CA self-assembly. To further investigate the effect of immune checkpoint inhibitors on blocking the PD-1/PD-L1 interaction, we performed surface sealing of the PD-L1 antigen on tumor cells. As a result, the uptake of these particles by tumor cells was significantly reduced in the absence of the D-type PD-L1 antagonist peptide (PPI) (Figure S4B–C). These results support our initial hypothesis that PPI exerts its extracellular effects by binding to PD-L1. Subsequently, the ultraviolet–visible (UV–vis) absorption and FTIR spectra confirmed the formation of CCtor, showing higher typical absorption peaks of peptides and the characteristic peaks of both Ctor and CA (Figure 2C–D). HPLC analysis of the supernatant after centrifugation of the prepared CCtor solution indicated that 100% of Ctor and 92.1% of CA were assembled into the CCtor supramolecule (Figure S3B–C). Dynamic light scattering (DLS) and ζ-potential measurements under simulated tumor acidic microenvironment conditions showed a hydrated particle size of 280.2 nm and a ζ-potential of −20.1 mV for CCtor at pH 6.5 (Figure 2E,F). High-resolution-TEM (HRTEM) and energy-dispersive X-ray spectroscopy (EDS) analysis confirmed the uniform distribution of N, C, and O in the observed CCtor particles, consistent with their components (Figure S3D,E). These data indicate the successful construction of the protein-like polypeptide condensate CCtor. Furthermore, to evaluate the kinetic stability of CCtor upon entering the bloodstream, the particle size of CCtor was measured using dynamic light scattering (DLS) at 0, 24, 72, 96, 120, and 144 h in a PBS solution containing 20% serum (to mimic the in vivo environment) (Figures S5). Over the course of 1 week, the particle size remained approximately 280 nm, demonstrating excellent colloidal stability.

Pharmaceutical Characteristics of CCtor

To investigate the internalization feature of CCtor, FITC-labeled CCtor and Ctor (CA-free control) were employed to assess uptake by Jurkat (T cells), RAW264.7 (macrophage), and LLC cells (LUAD cells). As anticipated, both confocal microscopy and flow cytometry consistently demonstrated a greater uptake of CCtor in LLC cells compared to Jurkat and RAW264.7 cells (Figure 3A–C). In order to confirm the micropinocytosis-mediated internalization of CCtor into LLC, macropinocytosis inhibitors Amiloride and Cytochalasin D were preincubated with LLC, resulting in significant suppression of CCtor internalizations (Figure 3D).

Following internalization, FITC-labeled CCtor exhibited significant colocalization with the lysosome probe Lysotracker (Pearson coefficient 0.86), indicating that CCtor undergoes degradation within the lysosome and subsequently releases D-type PPI into the extracellular space (Figure 3E–H). Subsequent analysis confirmed the release of biotin-labeled PPI from LLC cells, with the highest biotin content observed at 12 h postinternalization (Figure 3I). Further investigation into PPI’s role in PD-L1 degradation on the cell membrane revealed a substantial reduction in PD-L1 fluorescence on LLC cells treated with CCtor compared to controls (Figures S6A and S7A,B). These findings suggest that CCtor is internalized via macropinocytosis, undergoes lysosomal degradation, and releases PPI to degrade PD-L1, thereby restoring normalcy to the tumor immune microenvironment (Figure 3J).

Furthermore, due to the distinctive activation of micropinocytosis in tumors, CCtor is anticipated to exhibit remarkable tumor targeting capabilities. To validate this hypothesis, Cy5-labeled CCtor was synthesized and intravenously administered to C57BL/6 mice with subcutaneous LLC transplantation tumors. As anticipated, CCtor demonstrated maximum accumulation at the tumor sites compared to various other organs such as the heart, liver, spleen, lungs, and kidneys (Figure 3K–M), indicating a preferential profile for tumor accumulation.

CCtor Effectively Inhibits Wnt/β-Catenin Signaling and Enhances Immune Sensitization

CCtor significantly inhibited tumor cell proliferation at the cellular level, as evidenced by marked cell cycle arrest (Figure S8A–C). GSEA analysis of CCtor-treated cells showed significant downregulation of nuclear β-catenin and Wnt signaling pathways compared to controls (Figure 4A–D). Western blot analysis confirmed the significant downregulation of β-catenin and its downstream molecules C-myc and Cyclin D1 in CCtor-treated cells, with the most pronounced downregulation observed compared to CA-treated and control groups (Figures 4E,F and S9). The β-catenin was a critical component of the Wnt signaling pathway, and clinical data analysis revealed a positive correlation between β-catenin and PD-L1 expression. To further investigate whether CCtor regulates PD-L1 expression at the protein level, we conducted experiments on two distinct lung adenocarcinoma cell lines (Figure S10A,B). Western blot results demonstrated that CCtor downregulated PD-L1 expression at various time points. Notably, CA-induced downregulation of PD-L1 was significantly delayed compared to CCtor, suggesting that CA indirectly modulates PD-L1 expression via the Wnt/β-catenin pathway. In contrast, PPI, a D-type PD-L1 antagonist peptide within CCtor, directly interacts with PD-L1. In conclusion, the CCtor effectively suppressed the activation of the canonical Wnt/β-catenin signaling pathway and downregulated PD-L1 expression in tumor cells.

In vivo single-cell sequencing of LLC subcutaneous tumors identified various immune cells and a distinct tumor cell population (Figure 4G,H). GO and KEGG clustering of significantly downregulated genes in CCtor-treated tumor cells revealed suppression of aberrantly elevated Wnt signaling, energy metabolism, and cell cycle pathways (Figure 4I). T cell subpopulation analysis revealed a significant reduction in Treg cell markers and an elevation in cytotoxic T cell markers in CCtor-treated tumors, suggesting the normalization of the tumor-suppressive immune microenvironment (Figure 4J).

In Vivo Antitumor Efficacy of CCtor in C57 Mice Bearing LLC Tumors

Given CCtor’s significant tumor cell enrichment and Wnt/β-catenin signaling inhibition at the cellular level, its antitumor efficacy was evaluated in vivo using a C57 mouse subcutaneous LLC tumor model. Mice were treated with 2.5 mg/kg CCtor, PD-1 mAb, or CA for six sessions via tail vein injection, with tumor volumes measured daily. Tumor growth curves showed that CCtor significantly inhibited tumor growth compared to controls, with a TGI of 71.53%, outperforming PD-1 mAb (TGI < 50%) and CA (TGI < 30%) treatments (Figure 5A). Mice in the CCtor group-maintained weight gain throughout the treatment period, indicating no significant toxicity (Figure 5B). Photographs of mice and excised tumors post-treatment showed a 62.24% reduction in tumor weight in the CCtor group compared to the PD-1 mAb group and a 78.58% reduction compared to controls, corroborating the tumor growth inhibition results (Figure 5C–D). CCtor also significantly inhibited proliferation and promoted apoptosis in tumors compared to other treatments (Figure 5E–G). Mechanistic analysis revealed significant inhibition of β-catenin, Cyclin D1, and C-Myc expression in CCtor-treated tumors (Figures 5H and S11A,B), linking CCtor’s antitumor effect to Wnt/β-catenin signaling inhibition. Analysis of PD-L1 expression in tumor tissues showed that CCtor significantly reduced PD-L1 expression, indicating enhanced antitumor immune sensitization compared to other treatments (Figure 5I). Furthermore, a pronounced decrease in TGF-β expression was detected in the CCtor treatment group (Figure SI2A). TGF-β and the Wnt signaling pathway are intricately interconnected, with each pathway modulating the activity of the other. CCtor likely disrupts the β-catenin/BCL9 interaction by interfering with the Wnt/β-catenin signaling pathway, thereby reducing TGF-β expression. Within the tumor microenvironment, TGF-β plays a crucial role in regulating the secretion of chemokines by both tumor cells and immune cells. CCL22 and CCL20, members of the CC chemokine subfamily, are key immunosuppressive chemokines involved in the recruitment of regulatory T cells to the tumor microenvironment. In the CCtor-treated tumors, the expression levels of CCL22 and CCL20 were markedly decreased, leading to reduced infiltration of immunosuppressive T cells into the tumor, thereby enhancing the therapeutic efficacy of CCtor. Mechanistically, CCtor inhibits the infiltration of regulatory Treg cells via the TGF-β-CCL20/CCL22 axis. Further analysis of Treg and CTL populations using multichannel FCM revealed a significant reduction in Treg cells and an increase in CD8T/Granzyme B and CD8T/IFN-γ populations in CCtor-treated tumors compared to PD-1 mAb-treated tumors (Figure 5J). These results demonstrate that CCtor inhibits tumor cell infiltration by Tregs, enhances CTL cytotoxicity, and exerts potent antitumor immune sensitization.

Potent Immunotherapy Efficacy of CCtor in a Mouse LUAD Orthotopic Model

The remarkable inhibition of subcutaneous tumors prompted the establishment of a more clinically relevant model for the tumor microenvironment. By establishing an orthotopic LUAD model in mice through tail vein injection, we evaluated the in vivo effects of CCtor (Figure 6A). Mice were treated with PBS (control), PD-1 mAb (2.5 mg/kg), or CCtor (2.5 mg/kg) every 3 days for 11 sessions (Figure 6A). Post-treatment, CCtor significantly reduced lung tumor nodules compared to controls and PD-1 mAb treatments, with significant differences in statistical analysis (Figure 6B–C). Histological analysis showed that CCtor alleviated lung fibrosis, reduced collagen fiber deposition, and decreased lesion areas compared to other treatments (Figure 6D,E). CCtor-treated mice exhibited a median survival time of 36.5 days, outperforming control (23 days) and PD-1 mAb (27.5 days) treatments (Figure 6F). Notably, PD-1 mAb treatment did not significantly differ from controls in tumor nodule number, tumor area, collagen deposition, or median survival, indicating resistance (Figure 6B–F). Tunel staining showed increased apoptosis in CCtor-treated tumors compared to controls (Figure S13A). Ki67 IHC staining indicated significant downregulation of proliferation in CCtor-treated tumors (Figure 6G). Overall, these results demonstrate that CCtor inhibits LUAD cell growth and effectively treats LUAD orthotopic tumors with superior antitumor efficacy compared to PD-1 mAb.

IHC analysis of lung tumor tissues revealed significant downregulation of β-catenin, Cyclin D1, C-Myc, and PD-L1 in CCtor-treated tumors compared to controls and PD-1 mAb-resistant groups, indicating effective inhibition of Wnt/β-catenin and PD-L1 pathways (Figure 6H–L). Consistent with the findings in subcutaneous tumor models, the CCtor treatment group exhibited markedly reduced expression of TGF-β, as well as significantly lower levels of the immunosuppressive chemokines CCL22 and CCL20 (Figure S14A). These observations suggest that CCtor inhibits the infiltration of regulatory Treg cells via the TGF-β-CCL20/CCL22 axis. Immune cell infiltration analysis showed a significant increase in CD3+/CD8+ T cells and a decrease in CD4+/CD25+ T cells in CCtor-treated tumors, confirming enhanced immune response (Figure 6M,N). These results suggest that CCtor effectively blocks Wnt/β-catenin and PD-1/PD-L1 pathways, enhancing tumor-specific immune responses and providing a promising antitumor strategy.

Biocompatibility and Safety of CCtor

To evaluate CCtor’s biocompatibility, healthy C57BL/6 mice were injected with PBS (control) or CCtor (high dose 7.5 mg/kg, low dose 2.5 mg/kg) every other day for seven sessions. Weight monitoring revealed no significant differences between CCtor-treated and control groups, with continuous weight gain throughout the treatment period (Figure 7A). Hematological analysis revealed no significant adverse effects on WBC, RBC, LYMPH, or PLT in CCtor-treated groups (Figure 7B). Histopathological examination of organ slices post-treatment showed no significant changes in the morphology of vital organs, and liver and kidney function levels remained unchanged (Figure 7C–G). Biochemical analysis of immune-related toxicity indicators (TNF-α, IFN-γ, IL-1β, IL-4) in serum and organ tissues showed no immune-related adverse reactions (Figure 7H–L). These results demonstrate the excellent biocompatibility and safety of CCtor in vivo.

General Remarks

All chemical reagents for this research were supplied by Sigma-Aldrich. All commercial products were directly used without additional purification.

Synthesis of Ctor

The Ctor peptide was synthesized on a CS bio 336X automated peptide synthesizer using an HBTU activation/DIEA in situ neutralization procedure optimized on a suitable resin based on Fmoc chemical SPPS. After cleavage and deprotection in a reagent cocktail containing 88% TFA, 5% phenol, 5% H₂O, and 2% TIPS, it was precipitated with cold ether and purified to homogeneity by preparative C18 reversed-phase HPLC. The relative molecular masses of synthetic Ctor were detected by electrospray ionization mass spectrometry (ESI-MS).

Preparation of CCtor

The previously lyophilized Ctor powders were dispersed into distilled water at room temperature at the final concentration of (1 mg/mL). Subsequently, 0.5, 1, 2 mg CA was respectively dissolved in DMSO, which were dropwise added to the above Ctor solution and ultrasonicated for reaction 10 min, resulting in a milk white solution, namely CCtor.

Physicochemical Characterization of CCtor

The prepared series of CCtor solutions (the mass ratios of Ctor/CA respectively were 1:0.5, 1:1, and 1:2) were dripped onto the carbon-coated copper, and then the shape and size of CCtor particles were recorded at an acceleration voltage of 120 kV by Transmission Electron Microscope (Talos L 120C G2, Thermo Fisher Scientific). The morphology and energy-dispersive spectroscopy (EDS) analysis were observed by Lorenz transmission electron microscopy (Talos F200X, Thermo Fisher Scientific). The hydrodynamic size distribution and ζ-potential of the CCtor (with a mass ratio of Ctor/CA = 1:1, 1 mL) were characterized using dynamic light scattering (DLS). The modified surface chemical structure of CCtor was determined by Fourier Transform Infrared Spectroscopy (FTIR) and UV–vis Spectroscopy. In addition, the proportion of Ctor and CA loaded on CCtor was measured by high performance liquid chromatography (HPLC). Specifically, CCtor solution was centrifuged at 5000 rpm for 5 min to obtain supernatant, which then assayed for the amount of free Ctor and CA by HPLC.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.4c14264.

• Detailed descriptions of reagents, rigorous protocols for evaluating the in vitro stability and in vivo activity of CCtor, systematic procedures for constructing animal-based LLC tumor xenograft models and orthotopic lung adenocarcinoma mouse models, and essential experimental guidelines, and statistical analyses, along with 14 figures (PDF)

• Corresponding Authors

  • Wangxiao He - Departmentof Medical Oncology and Department of Talent Highland, The First Affiliated Hospital of Xi’an JiaotongUniversity, Xi’an 710061, P. R. China; https://orcid.org/0000-0002-2054-6022; Email: [emailprotected]

  • Jin Yan - Departmentof Infectious Diseases and Hepatology, TheSecond Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710004, P. R. China; National& Local Joint Engineering Research Center of Biodiagnosis andBiotherapy, The Second Affiliated Hospitalof Xi’an Jiaotong University, Xi’an 710004, P. R. China; https://orcid.org/0000-0001-6468-6003; Email: [emailprotected]

  • Yanmin Zhang - Departmentof Infectious Diseases and Hepatology, TheSecond Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710004, P. R. China; StateKey Laboratory of Shaanxi for Natural Medicines Research and Engineering, Xi’an 710061, P. R. China; School ofPharmacy, Health Science Center, Xi’anJiaotong University, Xi’an 710061, P. R. China; https://orcid.org/0000-0001-7307-9408; Email: [emailprotected]

• Authors

  • Jingmei Wang - Departmentof Infectious Diseases and Hepatology, TheSecond Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710004, P. R. China; Institutefor Stem Cell & Regenerative Medicine, The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710004, P. R. China

  • Weiming You - Departmentof Infectious Diseases and Hepatology, TheSecond Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710004, P. R. China; StateKey Laboratory of Shaanxi for Natural Medicines Research and Engineering, Xi’an 710061, P. R. China

• Author Contributions

  W.H., J.Y., and Y.Z. contributed equally to this work.

• Notes

  The authors declare no competing financial interest.
  

This work was supported by The National Key Research and Development Program of China (No. 2022YFE0133500), The National Natural Science Foundation of China (Nos. 22007076, 82272782, 32171256), Thousand Talents Plan of Shaanxi Province (For W.H.), The Key R&D plan of Shaanxi Province (No. 2023-YBSF-098), “The Young Talent Support Plan” of Xi’an Jiaotong University (W.H.), “Young doctor training program” of Xi Jing Hospital (J.Y.). We thank Instrument Analysis Center of Xi’an Jiaotong University for their assistance with TEM analysis. We extend our heartfelt gratitude to the Home for Researchers (www.home-for-researchers.com) for their meticulous work in color-matching the Figure 2A and Toc graphic.

Download PDF