Thiadiazolidinone (TDZD) Analogs Inhibit Aggregation-Mediated Pathology in Diverse Neurodegeneration Models, and Extend C. elegans Life- and Healthspan

Abstract

Chronic, low-grade inflammation has been implicated in aging and age-dependent conditions, including Alzheimer's disease, cardiomyopathy, and cancer. One of the age-associated processes underlying chronic inflammation is protein aggregation, which is implicated in neuroinflammation and a broad spectrum of neurodegenerative diseases such as Alzheimer's, Huntington's, and Parkinson's diseases. We screened a panel of bioactive thiadiazolidinones (TDZDs) from our in-house library for rescue of protein aggregation in human-cell and C. elegans models of neurodegeneration. Among the tested TDZD analogs, PNR886 and PNR962 were most effective, significantly reducing both the number and intensity of Alzheimer-like tau and amyloid aggregates in human cell-culture models of pathogenic aggregation. A C. elegans strain expressing human Aβ_1-42 in muscle, leading to AD-like amyloidopathy, developed fewer and smaller aggregates after PNR886 or PNR962 treatment. Moreover, age-progressive paralysis was reduced 90% by PNR886 and 75% by PNR962, and "healthspan" (the median duration of spontaneous motility) was extended 29% and 62%, respectively. These TDZD analogs also extended wild-type C. elegans lifespan by 15-30% (p < 0.001), placing them among the most effective life-extension drugs. Because the lead drug in this family, TDZD-8, inhibits GSK3β, we used molecular-dynamic tools to assess whether these analogs may also target GSK3β. In silico modeling predicted that PNR886 or PNR962 would bind to the same allosteric pocket of inactive GSK3β as TDZD-8, employing the same pharmacophore but attaching with greater avidity. PNR886 and PNR962 are thus compelling candidate drugs for treatment of tau- and amyloid-associated neurodegenerative diseases such as AD, potentially also reducing all-cause mortality.

Keywords: Alzheimer’s disease; aggregation; glycogen synthase kinase 3β (GSK3β); molecular modeling; neurodegeneration; neurodegenerative disease; protein aggregation; proteostasis; thiadiazolidinones (TDZDs).

Figure 1

Structures of TDZD-8 and two novel anti-inflammatory drugs, PNR886 and PNR962. TDZD analogs were synthesized with structures shown: (A) TDZD-8; (B) PNR886; and (C) PNR962.

Figure 2

TDZD analogs PNR886 and PNR962 are more potent than TDZD-8 in reducing HEK-tau aggregation. HEK293-tau cells were cultured 48 h in the presence of the indicated doses of TDZD-8 (A), PNR962 (B), or PNR886 (C), displayed on a log(10) scale, comprising zero (controls), 1 nM, 10 nM, 1 µM, and (for TDZD-8) 10 µM. Cells were then stained with thioflavin T, and aggregate signal per cell was quantified from fluorescence images (see Materials and Methods). Significance of differences from controls (combined from two to four repeats) was assessed by two-tailed t tests, with nominal adjustment of the alpha level (α < 0.01) to compensate for multiple comparisons. ** p < 0.01, **** p < 0.0001. Based on fluorescence per cell at 1 nM (10⁻³ µM) of each drug, both PNR962 and PNR886 were significantly more effective than TDZD-8 at reducing tau-mediated aggregation (p < 10⁻⁴).

Figure 3

TDZD analogues reduce aggregation and paralysis in C. elegans models of HD. (A,B) Adult worms (C. elegans strain AM141, expressing Q40::YFP) were exposed to PNR886 or PNR962 from the end of larval development (late L4) through adult day 3, and imaged on adult day 3. Images were processed to quantify (A) the number of aggregate foci and (B) total worm fluorescence of foci as a measure of amyloid/worm. (C) The percent of paralyzed worms is shown for C. elegans strain CL4176, 48 h after inducing Aβ_1–42 expression in muscle by upshift to 25 °C at the L3/L4 transition. Worms were exposed to PNR886 or PNR962 from larval L3/L4 stages through day 3.5 post-hatch, i.e., 2 days prior to image capture. Significance (as shown) was assessed by two-tailed heteroscedastic t tests, with adjustment of the alpha level (to α = 0.01) to compensate for multiple comparisons. (D) Typical AM141 images.

Figure 4

PNR886 and PNR962 extend lifespan of wild-type C. elegans. Lifespan survivals are shown for wild-type C. elegans (strain Bristol-N2, DRM lineage) exposed to (A) 10 µM PNR962 or (B) 10 or 100 µM PNR886, continuously from the L4/adult molt, replenished on fresh plates every other day. In each panel, drug treatments are compared to vehicle alone (controls) containing DMSO at the same final concentration as utilized in the drug plates. Worms (two plates of 25 worms per group, N ≈ 50) were maintained at 20 °C with ample E. coli (strain OP50) as food. Significance of survival differences, drug-treated vs. controls, were assessed by Gehan–Wilcoxon log-rank tests.

Figure 5

Age-dependent loss of motility is reduced by PNR886 and PNR962. TDZD analogs were added to NGM–agar plates at 10 µM, and 50 L4 larvae (C. elegans strain CL4176) were placed on these plates and transferred to fresh plates with drugs every 3 days. Cohorts were analyzed as for lifespan survivals, except immobile and dead worms, which were combined to calculate the paralyzed fractions. Gehan–Breslow–Wilcoxon tests were used to assess significance of differences from controls to allow disproportional hazards.

Figure 6

TDZD analogues PNR886 and PNR962 bind the hydrophobic pocket of GSK3β in the inactive conformation. PNR962, PNR886, and TDZD-8 bind to the allosteric hydrophobic pocket of GSK3β. (A) Glide docking pose of TDZD-8 binding to the GSK3β allosteric pocket in the modeled inactive DFG-out conformation. (B) Glide docking of PNR962 binding to the GSK3β allosteric hydrophobic pocket in the modeled inactive DFG-out conformation. (C) Glide docking of PNR886 binding to the GSK3β allosteric hydrophobic pocket in the modeled inactive DFG-out conformation. (D) MM-GBSA-based ΔG_binding (Gibbs free energy of binding) calculated for TDZD analogues (TDZD-8, PNR962, and PNR886), each with GSK3β in the inactive (DFG-out) conformations. Molecular structures are depicted with Schrödinger Maestro 11.4 (Schrödinger Inc., New York City, NY, USA).

Figure 7

Molecular-Dynamic simulations support stable binding of TDZD analogs TDZD-8, PNR886, and PNR962 to the hydrophobic pocket in inactive GSK3β. (A–I), Snapshots from 0.5-µs simulations of full-length GSK3β in the inactive conformation, bound to TDZD-8, PNR886, or PNR962; structural images were extracted at 0 ns, 100 ns, and 200 ns. (J–L), Root Mean Square Deviations (RMSD) time-courses of GSK3β::ligand complexes are plotted for simulations of GSK3β bound to TDZD-8, PNR886, or PNR962, each superimposed over the same no-ligand control simulation (black tracings). Replicate simulations produced similar results, reaching plateaus (indicating stable conformations) in <50 ns. Simulations were conducted in Schrödinger Maestro 11.4 (New York City, NY, USA).

Figure 8

Replacing TDZD-8 with PNR886 or PNR962 is predicted to improve binding to the hydrophobic pocket in the inactive conformation of GSK3β. The binding energy of TDZD-8 to GSK3β is first computed (represented as ΔG), and then TDZD-8 is exchanged with PNR886 or PNR962 using the FEP+ module of the Schrödinger Suite, permitting calculation of the resultant shifts in free energy (denoted as ΔΔG for binding to GSK3β, in kcal/mol).