Polygonatum sibiricum, known in Chinese as huangjing, is an edible botanical material with a long history in traditional food culture. Like many botanical products, it is not simply sold as harvested. Producers apply traditional food-processing methods that can involve repeated steaming and drying cycles, soaking in honey, or treatment with wine. Each of these treatments is believed to alter the chemical composition of the plant material, changing its flavor, its spectral fingerprint, and potentially its functional properties. That variability is precisely what makes authentication hard: a dried rhizome that has been steamed nine times and sun-dried nine times looks, to the naked eye, broadly similar to one that has merely been dried, yet the market treats the two as distinct products with distinct values.

The core analytical tool in the new study is FTIR spectroscopy, a technique that shines broadband infrared light through or onto a sample and records which frequencies are absorbed. Because chemical bonds absorb infrared radiation at characteristic frequencies, the resulting spectrum acts as a molecular fingerprint. O–H stretches associated with hydroxyl groups, C–H vibrations from carbon-hydrogen bonds, and C–O and C–O–C vibrations typical of ethers and carbohydrates all leave signatures in the mid-infrared region. For carbohydrate-rich plant materials such as Polygonatum rhizomes, these bands carry a great deal of information, since processing methods such as steaming are known to transform sugars and polysaccharides. Crucially, FTIR requires minimal sample preparation, no reagents, and only minutes of measurement time, which makes it attractive for routine quality control compared with slower chromatographic methods.

A raw spectrum, however, is far too complex for a human inspector to decode into a processing category. The bands overlap, baselines drift, and the differences between processing types may be subtle shifts in relative intensity rather than the appearance of new peaks. This is where machine learning enters. The researchers treated each spectrum as a high-dimensional data vector and trained classification algorithms to recognize the patterns associated with each labeled processing type. But they went a step further than many previous studies by insisting on interpretability: rather than accepting a black-box verdict, they wanted the model to show which spectral features drove its decisions.

A central methodological concern in the study is information leakage, a subtle but serious pitfall in spectroscopy-based machine learning. When spectra are preprocessed, features selected, and models evaluated using randomly split measurements, spectra recorded from the same physical sample can end up on both sides of the train-test divide. The model then partly memorizes sample-specific quirks rather than learning genuine chemical differences, inflating reported accuracy. The team avoided this by performing model training, feature selection, and evaluation at the physical-sample level, ensuring that all measurements from a given product were kept together during splitting. This design choice makes the reported performance a more honest estimate of how the system would behave on genuinely new products.

Before classification, the researchers confronted the dimensionality problem. An FTIR spectrum contains thousands of data points, many of them redundant or noisy, and feeding all of them into a classifier invites overfitting. The team compared three dimensionality-reduction strategies: principal component analysis, the classic statistical technique that compresses data into orthogonal axes of maximum variance; the successive projections algorithm, a variable-selection method that seeks subsets of columns with minimal collinearity; and simulated annealing, an optimization procedure inspired by the metallurgical process of slowly cooling material to reach a low-energy state. Simulated annealing explores the space of possible feature subsets by accepting not only improvements but occasionally worse solutions, allowing it to escape local optima that greedy search methods would get stuck in.

The standout configuration paired simulated annealing feature selection with a Random Forest classifier, an ensemble method that builds many decision trees on random subsets of the data and averages their votes. The resulting SA–Random Forest model showed stable discrimination performance across the four labeled processing categories. Stability, in this context, matters as much as raw accuracy: a model that performs brilliantly on one data split but collapses on another is of little use in a regulatory or commercial setting. By anchoring both feature selection and evaluation to physical samples, the authors could demonstrate that the model’s discrimination reflected reproducible chemical differences rather than statistical artifacts.

The interpretability layer came from SHapley Additive exPlanations, or SHAP, a framework borrowed from cooperative game theory that assigns each input feature a contribution value for every individual prediction. Applied to the spectral data, SHAP analysis linked the model’s outputs to specific interpretable spectral features, identifying variables related to O–H, C–H, and C–O/C–O–C vibrations as the key discriminators. These assignments make chemical sense. Repeated steaming is known to promote transformations in carbohydrate-rich components, honey processing introduces sugars with their own hydroxyl-rich signatures, and wine processing can alter the profile of soluble carbohydrates and other constituents. The fact that the model latched onto hydroxyl and glycosidic-bond regions suggests it was reading genuine processing chemistry rather than exploiting confounders such as particle size or instrument drift.

The broader significance of the work lies in what it says about the future of food and botanical authentication. FTIR-based chemical fingerprinting coupled with machine learning has already been applied to saffron adulteration, edible oil verification, sea cucumber origin tracing, and the geographic identification of medicinal herbs, and the new study extends this toolkit to processing-type discrimination, a dimension of authenticity that is often overlooked. Because the approach works at the laboratory level with rapid measurements and no chemical reagents, it could in principle be adapted to screening incoming batches, auditing supplier claims, or supporting market surveillance of traditionally processed botanical foods. The authors are careful to frame the results as demonstrating potential at the lab level rather than a deployable field method, an appropriate caution given that the study examined commercially available products whose labels cannot be independently verified.

There are also instructive lessons here for anyone building spectroscopy-machine learning systems. The comparison of dimensionality-reduction methods shows that feature selection strategy can matter as much as classifier choice, and the emphasis on sample-level data handling sets a standard that many published chemometric studies still fail to meet. The pairing of a powerful ensemble classifier with SHAP explanations demonstrates that accuracy and transparency need not be traded off against each other. For Polygonatum sibiricum specifically, the study establishes that the chemical consequences of traditional processing are large enough to be captured in an infrared fingerprint and robust enough to survive honest validation. As demand for traditionally processed botanical foods grows across Asia and beyond, tools like this one may become the difference between a market built on trust and one built on verifiable chemistry. The work was supported by Yunnan Fundamental Research Projects, and the authors report no competing financial interests.

Subject of Research: FTIR spectroscopy combined with interpretable machine learning to classify processed Polygonatum sibiricum products

Article Title: Interpretable FTIR spectroscopy and machine learning for discriminating commercially available Polygonatum sibiricum products labeled as different processing types

Article References: Guo, Z., Jiang, Y., Cheng, Y., Liu, G., Xie, J., & Ou, Q. (2026). Interpretable FTIR spectroscopy and machine learning for discriminating commercially available Polygonatum sibiricum products labeled as different processing types. Food Science and Biotechnology. https://doi.org/10.1007/s10068-026-02326-2

Image Credits: AI Generated

DOI: 10.1007/s10068-026-02326-2

Keywords: Polygonatum sibiricum, FTIR spectroscopy, machine learning, Random Forest, simulated annealing, SHAP, food authentication, chemical fingerprinting, traditional food processing, chemometrics, infrared spectroscopy, food analysis

The story begins with a chemical problem. Polygonati rhizoma polysaccharide, or PRP, is the most abundant bioactive component of Huangjing and has previously been shown to reduce dopaminergic neuron loss, suppress oxidative stress, and temper microglial inflammation in models of Parkinson’s disease. But PRP is built mostly of glucose and fructose monosaccharides whose abundant hydroxyl groups form strong intramolecular hydrogen bonds. That makes the molecule poorly reactive in water and difficult to combine with drugs or biomaterials, limiting its use in drug delivery or tissue engineering. To unlock its potential, the team turned to controlled periodate oxidation, a strategy widely used to convert the vicinal diols of polysaccharides such as dextran, alginate, and hyaluronic acid into reactive aldehyde groups.

Treating PRP with sodium periodate yielded oxidized PRP, or OPRP, with a degree of oxidation of roughly 16.33 percent. Structural elucidation was exhaustive: proton and carbon NMR, two-dimensional NMR experiments including COSY, HSQC, HMBC, and NOESY, FTIR spectroscopy, methylation analysis, and monosaccharide composition together identified OPRP as an inulin neoseries-type fructan with a weight-average molecular weight of 4.6 kilodaltons. Fructose made up more than 93 percent of the monosaccharide content, the backbone was dominated by 1,2-linked fructose residues exceeding 55 percent, and a low dispersity ratio of 1.04 confirmed the purity of the isolated polysaccharide. FTIR revealed the characteristic carbonyl absorption at 1730 per centimeter, while furanose-specific peaks at 1016, 1115, 939, and 816 per centimeter confirmed the β-D-furanose fructose architecture.

With reactive aldehydes in hand, the researchers crosslinked OPRP with carboxymethyl chitosan through Schiff base reactions, the reversible condensation between aldehyde and amino groups that underlies many self-healing injectable hydrogels. The optimized formulation, containing 2.5 percent OPRP and 5 percent carboxymethyl chitosan, gelled within ten minutes and displayed a storage modulus of about 230 pascals, a stiffness squarely within the soft 0.1 to 1 kilopascal range considered favorable for neural tissue. Scanning electron microscopy revealed a uniform porous network with an average pore diameter near 45.4 micrometers and porosity of roughly 95.4 percent. Rheological testing showed the gel could withstand strains up to 370 percent before yielding, then fully recover its modulus after repeated damage-healing cycles. It flowed smoothly through a 30-gauge needle fine enough to write letters, and degraded slowly in saline, retaining about 34 percent of its mass after fourteen days.

The material’s drug delivery credentials were demonstrated with Fast Green FCF, a hydrophilic stand-in for levodopa, which diffused steadily from the gel over a fourteen-day dialysis experiment before reaching a plateau around day ten. But the more provocative finding concerns what the gel does without any drug at all. In free radical scavenging assays, drug-free COPRP clearly outperformed a bioinert control hydrogel made from carboxymethyl chitosan and a dibenzaldehyde-terminated polyethylene glycol crosslinker, and adding levodopa more than doubled the scavenging capacity, an effect attributed to the polyphenolic chemistry of the drug itself. The authors are careful to frame the levodopa-loaded version, COPRPL, not as a simple carrier-drug relationship but as a deliberate dual-mechanism design in which the OPRP backbone remodels the diseased brain microenvironment while levodopa replenishes dopaminergic signaling.

In cellular experiments, the team challenged BV2 microglia with lipopolysaccharide to provoke inflammation and reactive oxygen species. Hydrogels containing OPRP significantly reduced ROS fluorescence, cut the M1-type pro-inflammatory markers IL-1β, iNOS, and TNF-α by more than 40 percent, and raised the M2-type anti-inflammatory marker CD206 by over 1.8-fold, with the levodopa-loaded gel pushing that upregulation past threefold. In a human neuroblastoma SH-SY5Y model of Parkinsonian injury induced by 6-hydroxydopamine, COPRP restored cell viability, reduced early and late apoptosis, rescued mitochondrial membrane potential as measured by JC-1 staining, and preserved expression of tyrosine hydroxylase, the rate-limiting enzyme of dopamine synthesis. The bioinert control gel did none of this, isolating the benefit to the OPRP component.

The in vivo results were the most consequential. The researchers lesioned the medial forebrain bundle of Sprague Dawley rats with 6-hydroxydopamine, confirmed successful Parkinsonian induction in about 94 percent of animals by apomorphine-evoked rotation, then injected four microliters of saline, inert hydrogel, COPRP, or COPRPL stereotaxically into the lesion site. By day seven, COPRP-treated rats were already rotating significantly slower than controls, and by day fourteen their circling speed had fallen to roughly 8.2 seconds per round. In the cylinder test, forelimb contact rates on the impaired side recovered by more than 15 percent in OPRP-gel groups and over 22 percent with levodopa loading, while open-field testing showed increased locomotion and exploration. Histology confirmed the pattern: more tyrosine hydroxylase-positive dopaminergic neurons survived in the substantia nigra, activated Iba1-positive microglia and GFAP-positive astrocytes were markedly reduced, and cerebral superoxide dismutase activity rose to approximately 53 units per milligram of tissue in the levodopa-loaded group.

To find the mechanism, the team compared brain proteomes of inert-hydrogel and COPRP-treated rats using data-independent acquisition mass spectrometry on a timsTOF HT platform. Principal component analysis cleanly separated the groups, and differential expression identified 161 upregulated and 38 downregulated proteins. KEGG enrichment pointed to focal adhesion, sphingolipid signaling, axon guidance, Rap1 signaling, and neurotrophin pathways, with selective remodeling of extracellular matrix components such as the Col6a collagens, paxillin, and talin. Critically, gene set enrichment analysis confirmed significant upregulation of the entire TGF-β signaling pathway. Western blotting then verified that COPRP increased TGF-β1 and TGF-β2 protein abundance, raised total Smad2/3, and markedly enhanced its phosphorylated form, the signature of pathway activation. Unexpectedly, GFAP expression fell rather than rose, leading the authors to propose that COPRP guides astrocytes away from a chronically hyperactivated, potentially neurotoxic phenotype toward a protective, repair-oriented state.

The causal test came from pirfenidone, an antifibrotic agent that attenuates TGF-β-associated signaling. Rats pretreated with pirfenidone before lesioning and COPRP treatment showed substantially reversed TGF-β2 expression and Smad2/3 phosphorylation, and their behavioral recovery across rotational, cylinder, and open-field tests was significantly blunted, though still better than untreated Parkinsonian rats. The authors acknowledge the limits of this design: systemic pirfenidone given before modeling cannot isolate striatal TGF-β signaling, so the data support the pathway as an important mediator without establishing pathway-specific causality. They also note that the study could not establish clinical or pharmacoeconomic advantages over established levodopa regimens, since the current approach requires stereotactic intracranial administration and specialized biomaterial preparation. Long-term biosafety, biodistribution of degradation products, and cell-type-specific knockout models to pinpoint the cellular targets of COPRP-mediated neuroprotection remain future work. Still, the framework is clear: a structurally defined fructan from a traditional medicinal herb, chemically functionalized into an injectable self-healing matrix, can couple local drug release with active immunomodulation and neuroprotection, positioning polysaccharide hydrogels as bioactive therapeutic agents rather than passive depots in the fight against neurodegeneration.

Subject of Research: A bioactive oxidized Polygonati rhizoma polysaccharide crosslinked chitosan hydrogel that modulates TGF-β/Smad signaling for Parkinson's disease treatment

Article Title: Modulation of the TGF‐β/Smad Signaling Pathway by a Bioactive Oxidized Polygonati Rhizoma Polysaccharide Crosslinked Chitosan Hydrogel as Therapeutic Carrier for Parkinson's Disease Treatment

Article References: Dai, P., Zhao, J., Xue, C., Xu, K., Yong, L., Zhang, Z., Huang, Y., Shuai, J., Chen, X., Nie, S., & Xu, J. (2026). Modulation of the TGF‐β/Smad Signaling Pathway by a Bioactive Oxidized Polygonati Rhizoma Polysaccharide Crosslinked Chitosan Hydrogel as Therapeutic Carrier for Parkinson's Disease Treatment. Advanced Science, Article e77842. https://doi.org/10.1002/advs.77842

Image Credits: AI Generated

DOI: 10.1002/advs.77842

Keywords: Parkinson's disease, hydrogel, Polygonatum sibiricum, polysaccharide, TGF-β/Smad signaling, levodopa, neuroinflammation, oxidative stress, Schiff base reaction, dopaminergic neurons, proteomics, drug delivery

The protein at the center of the new work is the retinoic acid receptor alpha, or RARα, which is highly active in colorectal tumors and drives both proliferation and metastatic invasion. Traditional RARα modulators, built on retinoic acid scaffolds, suffer from chemical instability, excessive lipophilicity, and poor oral bioavailability, and they can provoke toxicity in the liver, skin, and bones. To escape these limitations, the researchers turned to proteolysis-targeting chimera, or PROTAC, technology. PROTACs are bifunctional molecules that simultaneously grip a target protein and an E3 ubiquitin ligase, forming a ternary complex that tags the target for destruction by the cell’s own ubiquitin–proteasome system. Because each PROTAC molecule dissociates after every degradation event and can engage targets repeatedly, this catalytic mechanism holds the promise of overcoming the drug resistance that plagues conventional inhibitors.

The team began with AR7, an atypical RARα antagonist, and selected its optimized derivative CA77.1 as the target-binding ligand. Molecular docking against the RARα structure (PDB:5K13) revealed that CA77.1 binds with a calculated energy of −7.379 kcal/mol, anchored by a key hydrogen bond to Arg276 and an extensive hydrophobic network. Crucially, a terminal methyl group on CA77.1 protrudes into the solvent-exposed region outside the ligand-binding pocket, providing an ideal attachment point for a linker without disturbing the core binding interactions. By covalently joining CA77.1 to thalidomide, a ligand for the E3 ubiquitin ligase CRBN, the researchers created a degrader that remains active in both oxygen-rich and hypoxic tumor regions, overcoming the hypoxia-dependent activation that limited the parent compounds.

Through systematic synthesis of six candidate molecules with varying carbon-chain and piperazine linkers, the team identified Z1 as the standout. Z1 degraded RARα with a DC50 of 6.02 ± 1.05 μM and a maximum degradation of 85.4 percent, an activity roughly 43.6 times greater than that of CA77.1 itself. Western blotting confirmed that Z1 left the related receptors RARβ and RARγ untouched, demonstrating high subtype selectivity. Mechanistic experiments sealed the case: cycloheximide chase assays showed accelerated decay of pre-existing RARα protein, while pre-treatment with the competitive ligand ATRA, the E1 enzyme inhibitor MLN4924, the proteasome inhibitor MG132, or excess thalidomide all blocked degradation, confirming the classical PROTAC pathway of direct binding and ubiquitin-proteasome-mediated clearance.

Molecular dynamics simulations running 100 nanoseconds provided atomistic support for the design. Z1 formed stable hydrogen bonds with both RARα and CRBN, with calculated binding free energies of −139.394 kJ/mol and −111.138 kJ/mol respectively, and per-residue energy decomposition pinpointed Pro407, Val395, and Ser225 as the thermodynamic core of RARα recognition. Functionally, Z1-induced degradation suppressed phosphorylation of AKT at Ser473 in a concentration-dependent manner while leaving total AKT unchanged, and it markedly reduced expression of MMP2, an enzyme that tumors use to invade surrounding tissue. This AKT–MMP2 axis, the authors note, had not been systematically characterized in earlier RARα degradation studies and offers a fresh mechanistic window into how acute protein elimination restrains cancer progression.

In vitro assays in HCT116 colorectal cancer cells painted a striking picture of Z1’s antitumor potency. Flow cytometry showed concentration-dependent arrest of cells at the G2/M phase, with the arrested population rising 3.7-fold at 100 μM. At 50 μM, Z1 completely abolished colony formation, whereas the same concentration of CA77.1 allowed more than 300 colonies to grow. Annexin V staining revealed that Z1 drove 34.6 to 55.2 percent of cells into late apoptosis, well above the 24.6 percent seen with CA77.1. Scratch assays demonstrated up to 90.4 percent inhibition of cell migration, and Transwell experiments showed that 50 μM Z1 cut the invasion rate to 10.8 percent, an 81.2 percent reduction relative to the parent compound.

Yet like all PROTACs, Z1 carries the field’s chronic handicaps: high molecular weight, poor water solubility, and weak membrane permeability. The team’s solution came from an unexpected source, the roots of Polygonatum sibiricum, a plant long valued in traditional medicine. Using a juicing, filtration, and ultracentrifugation workflow, the researchers isolated exosome-like nanoparticles, or PsELNs, that measure roughly 116 nanometers across, carry a mildly negative zeta potential of −5.8 mV, and display the characteristic disk-like morphology of plant vesicles. Their cargo of triglycerides forms a hydrophobic core that stabilizes hydrophobic drugs, while their natural tropism for colonic tissue and their ability to accumulate in tumors through the enhanced permeability and retention effect make them unusually well suited to colorectal cancer applications.

Encapsulation of Z1 within PsELNs achieved a drug-loading efficiency of 43.1 percent, exceeding values reported for goji berry and ginger vesicles, and the loaded particles grew only modestly to 143 nm while retaining an intact bilayer. Stability testing was impressive: at acidic pH 5.5 mimicking the tumor microenvironment, 77.38 percent of Z1 remained intact after 24 hours versus just 40.65 percent for the free drug, and the formulation resisted thermal stress at 44 °C, stayed stable in serum for 72 hours, and showed minimal leakage over 60 days of storage. Confocal microscopy and flow cytometry confirmed that Z1/PsELNs entered HCT116 cells far more efficiently than free Z1, and the encapsulated drug degraded RARα with a DC50 of 2.95 μM, a 2.2-fold improvement that translated into stronger apoptosis induction and greater cytotoxicity in vitro.

The decisive test came in living animals. Near-infrared imaging of tumor-bearing nude mice showed that DiR-labeled PsELNs accumulated in tumors at 2.13 times the level of the free dye, peaking at 24 hours after injection, while hemolysis rates below 5 percent confirmed blood compatibility. Over 16 days of tail-vein treatment, the Z1/PsELNs group showed significantly stronger tumor growth inhibition than mice receiving free Z1, with in vivo antitumor efficacy increasing 1.8-fold. Histopathology revealed extensive tumor cell damage, Ki67 staining showed the fewest proliferating cells, and both immunohistochemistry and Western blotting confirmed the lowest RARα levels in the nanoparticle-treated tumors, all without body-weight loss or organ damage. The authors propose that the intestinal stability and colon-targeting behavior of PsELNs could eventually support an oral PROTAC formulation, a goal that would address both gastrointestinal degradation and site-specific accumulation. By fusing event-driven protein degradation chemistry with a biocompatible, naturally derived delivery vehicle, the study charts a credible translational path for PROTAC therapeutics in colorectal cancer and illustrates how plant nanotechnology may help the next generation of degrader drugs reach their targets intact.

Subject of Research: A RARα-targeting PROTAC delivered by Polygonatum sibiricum exosome-like nanoparticles for colorectal cancer therapy

Article Title: Synthesis of RARα–PROTAC and its delivery via Polygonatum sibiricum exosome-like nanoparticles for the treatment of colorectal cancer

Article References: Zhu, G., Zhao, Y., Chen, M., Lu, S., Xu, L., Chen, G., Zeng, L., & Chen, J. (2026). Synthesis of RARα–PROTAC and its delivery via Polygonatum sibiricum exosome-like nanoparticles for the treatment of colorectal cancer. Materials Today Bio, 40, Article 103651. https://doi.org/10.1016/j.mtbio.2026.103651

Image Credits: AI Generated

DOI: 10.1016/j.mtbio.2026.103651

Keywords: colorectal cancer, PROTAC, RARα, protein degradation, Polygonatum sibiricum, exosome-like nanoparticles, drug delivery, nanomedicine, ubiquitin-proteasome system, AKT signaling, MMP2, targeted therapy