Abiraterone-based HDAC inhibitor targets glioblastoma stemness and redox adaptation via filaggrin

Glioblastoma, the most aggressive primary brain cancer in adults, has long defeated the standard chemotherapeutic temozolomide through a frustrating combination of molecular defenses: the DNA repair enzyme MGMT, a persistent population of self-renewing glioma stem cells, and a finely tuned redox system that neutralizes the oxidative damage meant to kill the tumor. Now, a team of medicinal chemists and neuroscientists at Taipei Medical University reports a compound that attacks all three of these resistance mechanisms at once, while resurrecting an unexpected molecular ally from an unlikely corner of biology—the skin barrier protein filaggrin.

The compound, designated cp8, is a first-in-class hybrid molecule built on the scaffold of abiraterone, the CYP17A1 inhibitor approved for prostate cancer, onto which the team installed a hydroxamic acid group acting as a zinc-binding head for histone deacetylase inhibition. In the traditional three-part architecture of HDAC inhibitors—a surface recognition part, a linker, and a zinc-binding group—abiraterone serves as the surface recognition unit, tethered through an N-benzyl acrylamide linker in a critical beta configuration at the 3-position of the sterol core. That stereochemical detail proved decisive: among ten derivatives screened, only cp8, with the correct beta orientation, drove cell viability in temozolomide-resistant Pt#3-R glioblastoma cells down to 12.8 percent at 20 micromolar, while analogs bearing inverted configurations or alternative sulfonyl, benzoyl, and benzamide linkers fell short.

The potency numbers are striking. Against a panel of human and murine glioblastoma lines, including temozolomide-resistant and MGMT-positive cells such as T98G and P1S, cp8 achieved half-maximal inhibitory concentrations of 3 micromolar or less. The clinically validated pan-HDAC inhibitor vorinostat, known as SAHA, required concentrations of 30 micromolar or higher to reach comparable effects—a roughly tenfold gap. Mechanistically, cp8’s dependence on HDAC6 was confirmed using CRISPR-Cas9 knockout cells: when HDAC6 was deleted, cp8 treatment at 4 micromolar left nearly 75 percent of U87MG cells alive, whereas wild-type cells were reduced to about 10 percent viability. Treatment with cp8 also produced a striking accumulation of acetylated histone H3 and H4, including acetylation at residues H3K18, H3K23, H3K27, H4K5, H4K12, and H4K20, confirming functional blockade of multiple HDAC isoforms. In colony formation assays, cp8 reduced colony counts from 660 in the control group to just 6 at 0.8 micromolar, outperforming SAHA at identical doses.

The most surprising discovery emerged from RNA sequencing of drug-resistant Pt#3-R cells. Cp8’s transcriptional fingerprint included a distinctive upregulation of the filaggrin gene, FLG—a structural protein famous for aggregating keratin filaments in the epidermis and maintaining the skin’s hydration and barrier function, but essentially unstudied in brain tumors. Mining of The Cancer Genome Atlas and the Chinese Glioma Genome Atlas revealed that higher FLG expression correlated significantly with better survival in glioblastoma patients (p = 0.001), while FLG mutations, present in 8 to 14 percent of tumors in different datasets, were associated with worse outcomes. Immunohistochemistry on patient tissue microarrays and experimental mouse tumors showed FLG expression markedly depleted in glioblastoma compared with adjacent normal tissue.

Functional experiments established FLG as a genuine suppressor rather than a bystander. When researchers knocked down FLG with small interfering RNA in T98G cells, cell survival rose, active caspase-3 fell, and the stem cell markers Oct4 and SOX2 climbed 2.1-fold and 1.8-fold respectively, accompanied by enhanced temozolomide resistance. Pathway analysis of FLG-silenced cells revealed activation of pro-inflammatory and growth-factor signaling networks centered on IL1B, TNF, IL6, FGF2, JUN, PTGER2, and CREB1—transcriptional programs that sustain glioma stem-like states, phenotypic plasticity, and therapy resistance. Conversely, overexpressing FLG through CRISPR-Cas9–mediated promoter insertion reduced glioblastoma cell viability, raised caspase-3 activity, and, critically, sensitized MGMT-positive T98G cells to temozolomide at 600 micromolar. Because the filaggrin precursor protein spans roughly 4,061 amino acids, conventional cloning proved impractical, making the gene-editing approach essential. Notably, the FLG-boosting effect was unique to cp8; SAHA failed to induce FLG expression, and individual silencing of HDAC1, HDAC2, or HDAC6 each raised FLG while lowering SOX2, linking the epigenetic target to the filaggrin effect.

The compound simultaneously dismantled the other pillars of resistance. In MGMT-positive T98G cells, cp8 at 3 micromolar cut MGMT protein expression by 68 percent, directly undermining the primary enzymatic defense against temozolomide-induced DNA alkylation. In glioma sphere assays modeling glioma stem cells, cp8 shrank neurosphere size by 54 percent and suppressed Oct4 and SOX2 in both monolayer and spheroid cultures, an effect mirrored by accumulation of acetylated tubulin, the classic readout of HDAC6 inhibition. Immunofluorescence confirmed visibly weakened Oct4 and SOX2 staining after cp8 exposure. When MGMT was experimentally overexpressed in Pt#3 cells, the antiproliferative effect of cp8 was partially blunted, but its induction of reactive oxygen species was untouched, indicating the compound’s oxidative assault proceeds independently of MGMT status.

That oxidative assault is central to cp8’s mechanism. Using MitoSOX and CellROX probes, the team documented dose-dependent surges in mitochondrial superoxide and total cellular ROS after 48 hours of treatment, with the ROS signal colocalizing precisely with active caspase-3–positive apoptotic cells. Staining for 4-hydroxynonenal, a marker of toxic lipid peroxidation, intensified in cp8-treated glioma spheres, and TUNEL staining of mouse tumor tissue showed a clear increase in apoptotic cells. The picture is one of a compound that floods resistant cells with mitochondrial ROS they can no longer clear—overwhelming the redox adaptation that ordinarily lets glioma stem cells evade apoptosis and sustain temozolomide resistance.

The in vivo results were decisive. In a CT-2A allograft model, intraperitoneal cp8 at 10 milligrams per kilogram twice weekly extended median survival to 59 days, compared with 34 days in vehicle controls (p < 0.001) and 49 days for SAHA-treated animals, while reducing tumor volume by 72 percent. Tumors from cp8-treated mice showed reduced Oct4 and SOX2 and elevated FLG by immunohistochemistry, confirming the mechanism operated inside living brain tissue. In an orthotopic xenograft of temozolomide-resistant Pt#3-R cells, cp8 alone extended median survival to 55.5 days versus 24 days for controls, and interestingly, adding temozolomide conferred no statistically significant additional benefit—likely because cp8’s single-agent efficacy on resistant cells left little room for improvement. Against SAHA head-to-head in a luciferase-tagged U87MG xenograft, cp8 more powerfully suppressed tumor growth as measured by IVIS imaging and nearly doubled survival advantage (59 days versus 49 days in one model; 49 versus 30 in another). Tolerability testing in healthy C57BL/6 mice at doses up to 80 milligrams per kilogram revealed stable body weights and unremarkable liver and kidney biochemistry, with no histological signs of toxicity in hepatic or renal tissue.

Pharmacokinetic analysis in Sprague–Dawley rats showed cp8 penetrates the blood–brain barrier with a brain-to-plasma exposure ratio of roughly 21.4 percent, achieving rapid equilibrium between circulation and brain parenchyma. The caveats are real: a short plasma half-life of about 0.48 hours and high systemic clearance of 12.6 liters per hour per kilogram mean the compound will need medicinal chemistry optimization or specialized delivery strategies to widen its therapeutic window. The authors also note that distinguishing mutated from functional filaggrin protein was not possible with available antibodies, leaving questions about which FLG species matters most in tumors.

Even with those caveats, the study delivers two significant contributions at once. It validates filaggrin—long known as a dermatology gene—as a previously unrecognized tumor-suppressive factor and therapeutic target in glioblastoma, and it demonstrates that a single rationally designed molecule can strike stemness, MGMT expression, and redox homeostasis simultaneously, precisely the triad of mechanisms that has made temozolomide resistance so difficult to defeat. For a disease where more than half of patients develop drug resistance and median survival remains measured in months, a multitargeted compound that reached 59-day survival in aggressive mouse models without systemic toxicity offers a genuinely new template for the next generation of anti-glioblastoma drug discovery.