Hepatocellular carcinoma, the most common primary liver cancer, remains one of the world’s deadliest malignancies. It is the third leading cause of cancer-related death globally, and although viral hepatitis, alcohol consumption, aflatoxin exposure and poor diet all contribute to its development, the clinical picture is dominated by a single grim statistic: most patients are diagnosed at an advanced stage, when surgical options such as resection, transplantation or ablation are no longer viable. Approved drug therapies for advanced disease, including multikinase inhibitors such as sorafenib, lenvatinib and cabozantinib, and antibody combinations built around agents like bevacizumab and atezolizumab, have extended survival only modestly. The field’s persistent hunger for molecules with new structures and new mechanisms is what makes the KGA-1002 story notable.

The compound’s origin lies in the chemistry of Artemisia, the genus that gave the world artemisinin. Sesquiterpenoid dimers, molecules in which two sesquiterpene units are joined, have repeatedly shown stronger antitumor activity than their monomeric parents, but their scarcity in plants has historically blocked serious drug development. The Kunming group sidestepped that bottleneck through biomimetic synthesis: using a Diels-Alder reaction, they assembled KGA-1002 from dehydrocostus lactone, an abundant natural sesquiterpenoid, on a ten-gram scale. The resulting dimer inhibited the proliferation of three hepatocellular carcinoma cell lines with IC50 values of 5.9, 6.9 and 6.3 micromolar against HepG2, Huh7 and SK-Hep-1 cells respectively, outperforming sorafenib and beating the parent monomer dehydrocostus lactone by roughly five- to seven-fold.

Identifying what a small molecule actually touches inside a cell is the hardest part of natural product pharmacology, and the team attacked it with a battery of orthogonal techniques. A drug affinity responsive target stability assay, which exploits the fact that ligand-bound proteins resist enzymatic digestion, flagged a protein band that survived pronase treatment only in the presence of KGA-1002. Mass spectrometry of that band yielded 49 candidate proteins, which transcriptomic profiling and bioinformatics narrowed to six. The decisive experiment was genetic: when the researchers silenced each candidate in liver cancer cells, only knockdown of GRP94, the endoplasmic reticulum-resident member of the heat shock protein 90 family, blunted the drug’s antiproliferative and anti-migratory effects. Surface plasmon resonance then measured direct binding, with a dissociation constant of 454 nanomolar, and a cellular thermal shift assay confirmed that KGA-1002 stabilizes GRP94 inside cells.

The structural detail goes further than most target-identification studies dare. KGA-1002 carries an alpha, beta-unsaturated carbonyl motif, a well-known covalent warhead, and molecular dynamics simulations showed that this group is essential for stabilizing the GRP94-ligand complex. Mass spectrometric sequencing of the treated protein revealed a peptide whose mass had increased by exactly the molecular weight of the drug, and secondary fragmentation localized the covalent bond to serine 106, a hydroxyl-bearing residue ideally positioned for nucleophilic attack on the unsaturated carbonyl. When the team mutated serine 106 to alanine, binding affinity collapsed from 522 nanomolar to 461 micromolar, a nearly thousand-fold loss. Critically, KGA-1002 did not stabilize the cytosolic family members HSP90alpha or HSP90beta, whose indiscriminate inhibition has doomed pan-HSP90 drugs in clinical trials. That isoform selectivity, aided by an additional contact with GRP94-specific asparagine 276, is arguably the compound’s most clinically relevant property.

With the target pinned down, the downstream biology fell into place. GRP94’s day job is folding, assembling and trafficking client proteins inside the endoplasmic reticulum, and its inhibition floods the ER lumen with misfolded polypeptides. Transcriptomic analysis of treated cells showed enrichment of ER-related pathways, upregulation of stress genes such as ERN1, ATF4 and DDIT3, and downregulation of ER trafficking genes. The canonical unfolded protein response sensors PERK and IRE1, normally held inactive by the chaperone BiP, were activated, driving expression and nuclear translocation of the death-promoting transcription factors XBP1 and CHOP. Because GRP94 also buffers calcium in the ER, the drug produced a dose-dependent rise in intracellular calcium, climbing to nearly 98 percent calcium-positive cells at 7.5 micromolar in both SK-Hep-1 and Huh7 lines. The unfolded protein response, in other words, was not a side effect but the engine of tumor killing.

The most novel mechanistic finding concerns AKT, the kinase at the heart of the PI3K survival pathway. KGA-1002 lowered AKT protein levels without touching its mRNA, and cycloheximide chase experiments showed the drug accelerating AKT degradation; proteasome and lysosome inhibitors each partially rescued the protein, implicating both disposal routes. The explanation proved to be a chaperone-client relationship: GRP94 physically holds AKT, and pulldown assays demonstrated that KGA-1002 disrupts that interaction, both in cell lysates and with purified proteins. Destabilized AKT could no longer phosphorylate SKP2, the F-box substrate-recognition protein of an E3 ubiquitin ligase that normally tags the cell cycle brake P21 for destruction. Deprived of AKT’s stabilizing phosphorylation, SKP2 itself was degraded through the proteasome, P21 accumulated in the nucleus, and the cells arrested in G0/G1 and died. Clinical databases reinforced the story: both GRP94 and SKP2 are overexpressed in hepatocellular carcinoma tissues and correlate with poor survival.

The death program did not stop at apoptosis. With AKT signaling crippled, the antioxidant defenses of the cell, maintained through the AKT-mTOR axis, Nrf2 and GSK-3beta, eroded, and reactive oxygen species accumulated. Mitochondrial membrane potential collapsed, lipid peroxidation measured as malondialdehyde rose, the ferroptosis gatekeeper GPX4 fell, and labile iron climbed, all hallmarks of ferroptosis, the iron-dependent form of regulated cell death. The ROS scavenger N-acetylcysteine and the ferroptosis inhibitor liproxstatin-1 each blunted these effects, confirming causality. KGA-1002 thus kills liver cancer cells along two converging routes, ER stress-driven apoptosis and AKT-dependent ferroptosis, both traceable to a single upstream lesion at GRP94.

In vivo, the compound held up. In nude mice bearing subcutaneous SK-Hep-1 tumors, intraperitoneal KGA-1002 at 15, 30 and 60 milligrams per kilogram over 60 days shrank tumors by 44.1, 52.1 and 53.5 percent, with the top dose matching sorafenib; direct intratumoral injection performed even better, reaching 61.6 percent inhibition. Tumors from treated animals showed reduced Ki67 staining and elevated phosphorylated IRE1 and PERK, confirming that the ER stress mechanism operates in living tissue. The decisive target-validation experiment came from xenografts grown from GRP94-knockdown cells: with the target partially removed, the drug’s antitumor effect and its suppression of Ki67 were significantly weakened, establishing that KGA-1002’s efficacy is GRP94-dependent. Safety readings were reassuring, with no weight loss, behavioral changes, liver or kidney function abnormalities, and no histopathological organ damage even at 150 milligrams per kilogram in a separate subacute study.

The study’s authors are careful about scope, and so should readers be. The work rests on cell lines and subcutaneous mouse models, not orthotopic liver tumors or patients, and a selectivity index of roughly 1.5 to 1.7 over normal hepatocytes, while adequate, leaves room for optimization. Yet the conceptual payoff is substantial: GRP94 has been implicated in breast cancer metastasis and multiple myeloma, and the authors note it is also overexpressed in colon, esophageal and bladder cancers, so a validated, covalent, isoform-selective inhibitor scaffold is a tool the wider oncology community can build on. By showing exactly where the molecule docks, which residue it attacks, and how that single molecular event cascades through protein folding, calcium handling, AKT stability, SKP2 degradation and ferroptosis, the Kunming team has delivered something rarer than another cytotoxic natural product: a mechanistically coherent proof that GRP94 is a druggable vulnerability in liver cancer, and a lead compound engineered to exploit it.

Subject of Research: A selective GRP94 inhibitor derived from a guaianolide sesquiterpenoid dimer as a therapeutic agent against hepatocellular carcinoma

Article Title: Guaianolide dimer KGA-1002 targets GRP94 and triggers unfolded protein response-associated degradation of SKP2/AKT axis as a novel antihepatoma agent

Article References: Li, Q.-H., Li, T.-Z., Wang, Y.-C., Huang, X.-Y., Ma, W.-J., Li, F.-J., Huang, F.-D., Hu, M.-M., & Chen, J.-J. (2026). Guaianolide dimer KGA-1002 targets GRP94 and triggers unfolded protein response-associated degradation of SKP2/AKT axis as a novel antihepatoma agent. Journal of Advanced Research, 88, 1077-1093. https://doi.org/10.1016/j.jare.2026.01.010

Image Credits: AI Generated

DOI: 10.1016/j.jare.2026.01.010

Keywords: hepatocellular carcinoma, GRP94, KGA-1002, guaianolide dimer, unfolded protein response, AKT, SKP2, P21, ferroptosis, endoplasmic reticulum stress, natural products, drug discovery

Bortezomib is not a new molecule. The drug, best known for its role in treating multiple myeloma and mantle cell lymphoma, works by blocking the proteasome, the cellular machinery that degrades damaged or unwanted proteins. When proteasomal degradation grinds to a halt, misfolded proteins accumulate, stress pathways ignite and cells that cannot cope are pushed into programmed death. Earlier clinical and laboratory observations had suggested that bortezomib has activity against CTCL, but the molecular wiring behind that activity had never been fully mapped. The research team, led by Kirti S. Prabhu and Shahab Uddin, set out to answer two linked questions: does bortezomib modulate SKP2-associated cell-cycle regulation in CTCL cells, and can it be combined effectively with the platinum chemotherapy drug cisplatin?

To do this, the investigators used two well-characterized human CTCL cell lines, H9 and HH, exposing them to escalating concentrations of bortezomib and measuring viability with cell counting assays, then probing the consequences with fluorescence imaging and flow cytometry. The drug proved remarkably potent. The 24-hour concentration that killed half the cells, the IC50, was just 5.25 nanomolar in H9 cells and an even lower 1.82 nanomolar in HH cells. Values in the low nanomolar range are notable because they sit comfortably within concentrations achievable in clinical practice, lending translational weight to the laboratory findings. The team then asked what kind of death the cells were dying, and the answer was unambiguous: apoptosis, the orderly suicide program that cancer cells often manage to evade.

The apoptotic signature emerged at several levels simultaneously. Flow cytometry revealed a rising sub-G0/G1 population, a classic hallmark of cells with fragmented DNA, alongside an increase in Annexin V-positive cells, which flag the externalization of phosphatidylserine, an early molecular signal of apoptosis. Immunoblotting showed activation of the caspase cascade, the proteolytic machinery that executes cell death: caspase-8, caspase-9 and caspase-3 were all cleaved into their active forms, and PARP, a DNA repair enzyme that caspase-3 slices as a finishing blow, was cleaved as well. The intrinsic, or mitochondrial, arm of apoptosis was clearly engaged, because the ratio of the pro-death protein Bax to the pro-survival protein Bcl-2 increased, and the mitochondrial membrane potential collapsed, a depolarization event that releases apoptogenic factors from the intermembrane space into the cytosol.

To confirm that this death was truly caspase-dependent, the researchers pre-treated cells with z-VAD-FMK, a broad-spectrum inhibitor that shuts down all caspases. The inhibitor significantly blunted bortezomib-induced apoptosis, demonstrating that the caspase cascade is a functional component of the drug’s cytotoxicity rather than a bystander effect. That finding matters because it anchors bortezomib’s action in a defined, druggable death pathway, one that can be monitored and potentially amplified by rational drug combinations.

Perhaps the most mechanistically revealing part of the study concerns reactive oxygen species, or ROS. Bortezomib treatment increased both cellular and mitochondrial ROS and simultaneously depleted intracellular glutathione, the cell’s principal endogenous antioxidant buffer. This double hit, more oxidants and less antioxidant capacity, creates a state of oxidative stress that damages mitochondrial membranes and feeds the apoptotic circuit. The causal contribution of ROS was tested with N-acetyl-L-cysteine, a glutathione-replenishing antioxidant. When NAC was present, bortezomib’s disruption of the cell cycle, its induction of apoptosis and its activation of caspases were all partially reduced. In other words, oxidative stress is not the whole story, but it is a genuine upstream driver of the cytotoxic program, and the partial rescue suggests that bortezomib kills CTCL cells through converging ROS-dependent and ROS-independent routes.

The study also connected bortezomib’s effects to cell-cycle control, an area where CTCL cells are particularly vulnerable. The team observed a marked suppression of SKP2, S-phase kinase-associated protein 2, an E3 ubiquitin ligase component famous for tagging the cyclin-dependent kinase inhibitors p21 and p27 for destruction. When SKP2 falls, its targets accumulate, and that is exactly what the researchers documented: levels of p21 and p27 rose while the cyclin-dependent kinases CDK4 and CDK6, which drive cells through the G1 checkpoint, declined. This SKP2-p21/p27 axis functions as a brake on proliferation, and bortezomib effectively re-engages that brake. Because SKP2 is an oncogenic driver in many lymphoid malignancies, its suppression offers a candidate pharmacodynamic vulnerability, a measurable molecular endpoint that clinicians could track to confirm that the drug is hitting its intended target in patients.

Combination therapy was the final piece of the puzzle. Platinum-based agents such as cisplatin are mainstays of oncology, but their use in CTCL is limited by toxicity and modest durable benefit. Using the Chou-Talalay method, a widely accepted mathematical framework for quantifying drug interactions through the combination index, the researchers tested whether bortezomib and cisplatin act cooperatively. At selected dose pairings, bortezomib enhanced cisplatin-induced loss of viability and apoptosis, and the interaction was strongly synergistic at 5 nanomolar bortezomib combined with 10 micromolar cisplatin. Synergy means the combination kills more cells than the sum of the two drugs acting alone, a property that could, in principle, allow lower doses of each agent and a narrower toxicity profile. The mechanistic logic is plausible: bortezomib’s ROS generation and caspase priming may lower the apoptotic threshold that cisplatin-induced DNA damage must cross to trigger cell death.

The authors are careful to frame these results as a foundation rather than a clinical verdict. All of the experiments were performed in established cell lines, and the study’s own conclusions call for validation in primary CTCL samples from patients and in in vivo models before the strategy can move toward the clinic. That caveat is standard and appropriate, yet the work still carries real significance. It converts a clinical observation, that bortezomib has activity in CTCL, into a mechanistic account involving ROS accumulation, glutathione depletion, mitochondrial depolarization, caspase activation and SKP2 pathway suppression, and it nominates a specific synergistic drug pairing for further testing. For a disease with limited durable options in its advanced stages, that combination of mechanistic depth and translational direction is exactly the kind of progress that translational medicine aims to deliver, and it gives CTCL researchers a clear set of molecular signposts to follow in the next phase of preclinical and clinical evaluation.

Subject of Research: Proteasome inhibitor bortezomib-induced ROS-dependent mitochondrial apoptosis and SKP2 signaling suppression in cutaneous T-cell lymphoma cells

Article Title: Bortezomib induces ROS-dependent mitochondrial apoptosis and suppresses SKP2 signaling in cutaneous T-cell lymphoma cells

Article References: Bortezomib induces ROS-dependent mitochondrial apoptosis and suppresses SKP2 signaling in cutaneous T-cell lymphoma cells. (n.d.). https://doi.org/10.1186/s12967-026-09039-4

Image Credits: AI Generated

DOI: 10.1186/s12967-026-09039-4

Keywords: cutaneous T-cell lymphoma, bortezomib, SKP2, reactive oxygen species, mitochondrial apoptosis, caspases, p21, p27, cisplatin, proteasome inhibitor, glutathione, drug synergy