Hupehenine targets an inflammatory pathway linked to diabetes and colorectal cancer

Hupehenine, a steroidal alkaloid found in the bulbs of Hubei fritillary, bound to all three proteins in an inflammatory signaling module linked to diabetes-associated colorectal cancer in laboratory experiments, according to a study published in Frontiers in Pharmacology. The compound showed its strongest measured affinity for the chemokine receptor CXCR2, interfered with CXCL8 binding to that receptor, and reduced colorectal cancer cell viability, migration and invasion under conditions designed to mimic selected features of diabetes. In parallel, it lowered markers associated with an M2-like macrophage profile and increased markers associated with an M1-like profile. These findings remain preclinical: the researchers used cell cultures, computational models and purified-protein assays rather than animals or people with cancer.

The study began with an effort to identify molecular features shared by type 2 diabetes and colorectal cancer. Researchers combined 1,043 diabetes-associated genes with 3,632 colorectal cancer-associated genes gathered from DisGeNET, the Therapeutic Target Database, GeneCards and OMIM, identifying 531 genes linked to both conditions. Gene Ontology and pathway analyses pointed toward inflammatory responses, PI3K/AKT signaling, cancer-related microRNAs and the AGE-RAGE pathway associated with diabetic complications. A protein-protein interaction network then nominated IL1B, TNF, NFKB1, TLR4, IL6, CXCL8, IFNG and IL10 as a consensus set of candidate hubs. Because the upstream cytokine IL1β can induce CXCL8, and CXCL8 signals through CXCR2, the researchers focused on the IL1β-CXCL8-CXCR2 axis as a possible connection between metabolic inflammation and tumor biology.

Analysis of The Cancer Genome Atlas colorectal cancer data added observational context but also revealed a complicated pattern. IL1B and CXCL8 were significantly more highly expressed in tumors than in adjacent normal mucosa, and CXCR2 was also upregulated and positively co-expressed with the other two components. Unexpectedly, higher expression of IL1B or CXCL8 was associated with longer overall survival, with the longest observed survival among patients expressing both at high levels. Immune deconvolution associated high axis expression with greater CD8-positive T-cell infiltration and a higher ratio of M2- to M1-associated macrophage signatures. These are expression-based associations rather than evidence that the pathway improves survival, and the researchers said the findings could reflect an immune-infiltrated tumor environment that simultaneously contains regulatory or inhibitory immune programs.

The study then examined 371,223 cells from a published single-cell colorectal cancer atlas spanning 62 patients. IL1B and CXCL8 were expressed almost exclusively by myeloid populations, including monocytes, granulocytes and macrophages, while CXCR2 was concentrated mainly in granulocytes. Across patients, the proportion of CXCL8-expressing myeloid cells correlated positively with the M2-associated transcriptional score at r = 0.667 and with the M1-associated score at r = 0.557, both with P values below 0.001. The difference between those correlations was not formally tested, so their relative sizes do not establish preferential M2 polarization. Mismatch-repair-deficient tumors had higher T-cell exhaustion-associated and M2-associated expression scores than mismatch-repair-proficient tumors, but the authors cautioned that such scores do not demonstrate functional exhaustion or suppressive macrophage activity. Computational ligand-receptor analysis inferred possible myeloid communication, yet it likewise did not demonstrate signaling in intact tissue.

To find compounds capable of engaging several nodes in the proposed pathway, the researchers combined deep-learning prediction with molecular docking. Screening a natural-product library with the DeepPurpose tool and AutoDock Vina produced four consensus candidates, among which hupehenine had the highest predicted multi-target scores for IL1β, CXCL8 and CXCR2. Docking estimates were −5.75 ± 0.06 kcal/mol for IL1β, −6.44 ± 0.20 kcal/mol for CXCL8 and −7.49 ± 0.32 kcal/mol for CXCR2. In 100-nanosecond molecular-dynamics simulations, hupehenine remained in the CXCR2 complex for every recorded frame but detached from CXCL8 after about 91 nanoseconds and repeatedly separated from IL1β from about 53 nanoseconds onward. These trajectories were single simulations rather than independently replicated ensembles, and the authors warned that trajectory-wide binding-energy estimates for the two complexes affected by dissociation could not be treated as reliable bound-state affinities.

Surface plasmon resonance supplied direct biophysical measurements. Hupehenine bound in a concentration-dependent manner to each purified target, with dissociation constants of 505.3 ± 4.6 nanomolar for CXCR2, 2.2 ± 0.1 micromolar for CXCL8 and 12.7 ± 1.0 micromolar for IL1β. The micromolar values indicate substantially weaker measured binding to CXCL8 and IL1β than to CXCR2. In reciprocal assays, IL1β bound immobilized CXCR2 with a dissociation constant of 16.2 ± 0.76 nanomolar, while CXCL8 bound it at 1.5 ± 0.13 nanomolar. Pre-saturating the receptor surface with 10 micromolar hupehenine reduced subsequent binding by both proteins by more than tenfold. The experiment supports interference with binding to the sensor-bound receptor, but it does not by itself establish the precise molecular mechanism, such as direct competition at an orthosteric site.

Cell experiments used a “three-factor” diabetic-mimetic environment consisting of 25 millimolar glucose, 100 nanomolar insulin and 200 micromolar palmitic acid. In HCT116 colorectal cancer cells, prolonged exposure to this mixture increased viability, migration and invasion in a time-dependent manner, with changes emerging after one to three weeks of conditioning. Hupehenine reduced those tumor-associated readouts. The half-maximal inhibitory concentration in chronically conditioned HCT116 cells was 26.7 ± 1.3 micromolar, compared with 72.2 ± 3.6 micromolar under standard conditions, indicating roughly 2.7-fold greater sensitivity in the model. SW480 cells likewise became more sensitive. By contrast, conditioning did not increase sensitivity to 5-fluorouracil or oxaliplatin, and hupehenine did not significantly reduce viability of the non-tumorigenic HCEC-1CT colon epithelial cell line at concentrations up to 200 micromolar. This suggests a possible selectivity window in vitro, not evidence of systemic safety or diabetes-specific efficacy.

At the signaling level, the diabetic-mimetic conditions increased expression and secretion of axis components and the measured activity of the PI3K/AKT/mTOR pathway. Hupehenine selectively reduced CXCL8 messenger RNA and secretion without producing a parallel decline in the total intracellular CXCL8 protein pool. It also attenuated PI3K/AKT/mTOR readouts while leaving NF-κB p65, IL1β and CXCR2 largely unchanged. Knockdown experiments produced changes resembling those following hupehenine treatment, while PI3K/AKT overexpression largely bypassed its inhibitory effects, supporting action upstream of that pathway in the experimental system. Co-immunoprecipitation also showed that hupehenine disrupted the physical interaction between purified CXCL8 and CXCR2. At a single 30-micromolar concentration, hupehenine reduced HCT116 viability to 26% ± 1.5%, compared with 54% ± 2.2% for the CXCR2 antagonist navarixin, 62% ± 1.9% for the IL-1 receptor antagonist anakinra and 31% ± 2.3% for the two antagonists combined.

Hupehenine also shifted measured macrophage markers. Conditioned medium from HCT116 cells exposed to the three-factor environment increased M2-associated markers and decreased M1-associated markers in THP-1-derived macrophages. Adding hupehenine either to the cancer-cell culture or directly to the macrophage culture reduced the M2-associated profile and increased M1-associated markers. It also counteracted M2-associated marker increases induced by exogenous IL1β plus CXCL8. These findings show that a selected marker profile can be altered, but they do not establish functional repolarization. The study did not directly test macrophage nitric oxide production, phagocytic activity or suppression of T-cell responses, all of which would be required to characterize functional behavior. Nor did it determine which of hupehenine’s targets caused the marker changes.

Several limitations sharply define the study’s scope. The three-factor cell model reproduces selected metabolic stresses but cannot capture the full physiology of diabetes, including systemic endocrine, immune, vascular and microbiome effects. The proposed diabetes-related expression subgroup used a 14-gene signature and was not equivalent to clinically confirmed diabetes status. Survival associations require multivariable adjustment and independent clinical validation, while inferred cell communication and expression-based exhaustion scores cannot substitute for functional measurements. Hupehenine’s pharmacokinetics, oral bioavailability and broader target profile—including possible activity at nuclear hormone receptors, hERG channels and other off-targets—remain uncharacterized. The authors placed the compound at the lead-discovery stage rather than treating it as a clinical candidate. The next steps they identified include immunocompetent diabetic colorectal cancer models, direct macrophage-function assays, broader safety and pharmacokinetic profiling, medicinal-chemistry optimization and experiments designed to establish whether multi-target engagement is the principal cause of hupehenine’s observed effects.

Subject of Research: In vitro characterization of hupehenine targeting the IL1β-CXCL8-CXCR2 axis in a diabetic-mimetic colorectal cancer model

Article Title: Hupehenine suppresses tumor cell phenotypes and modulates macrophage marker profiles in a diabeticmimetic colorectal cancer model

Article References: Zhu, M. Q., Shi, Z. T., Hu, H. M., Xue, Q. D., Gu, X. Y., Li, X. C., Wang, Q. Y., Zhao, Y., Zhang, Y. W., & Zhao, J. Z. (2026). Hupehenine suppresses tumor cell phenotypes and modulates macrophage marker profiles in a diabeticmimetic colorectal cancer model. Frontiers in Pharmacology, 17, Article 1962620. https://doi.org/10.3389/fphar.2026.1962620

Image Credits: AI Generated

DOI: 10.3389/fphar.2026.1962620

Keywords: colorectal cancer, diabetes, hupehenine, IL1β, CXCL8, CXCR2, macrophage markers, inflammation, drug discovery, cell model, suppresses, tumor