β-Carboline Compounds, Including Harmine, Inhibit DYRK1A and Tau Phosphorylation at Multiple Alzheimer’s Disease-Related Sites

DYRK1A is a dual-specificity kinase encoded on chromosome 21 within the Down syndrome critical region and has been implicated in neurodevelopmental abnormalities and cognitive deficits when overexpressed. Previous cellular and transgenic studies have linked DYRK1A activity to both amyloid and tau pathologies observed in Down syndrome and sporadic Alzheimer's disease (AD), and DYRK1A protein has been detected in neurofibrillary tangles. Earlier work by the authors and others indicated that DYRK1A phosphorylates tau at several sites associated with AD-related tau dysfunction, raising the kinase as a potential therapeutic target to modify tau pathology. Frost and colleagues set out to test whether harmine, a b-carboline alkaloid reported to inhibit DYRK1A, and a panel of related b-carboline derivatives could inhibit DYRK1A-dependent phosphorylation of tau at multiple AD-relevant sites. The study combines RNAi, cell-based assays in a tau-overexpressing neuroglioma line, and in vitro kinase assays with recombinant proteins to (1) map DYRK1A-dependent tau phosphorylation sites, (2) measure the effect of harmine and related compounds on phosphorylated and total tau, and (3) relate compound structure to inhibitory potency against DYRK1A-mediated phosphorylation.

Cellular work used an H4 neuroglioma cell line stably overexpressing 4R0N tau. Cells were maintained in Dulbecco's Modified Eagle Medium with 10% fetal bovine serum, antibiotics, geneticin and L-glutamine, split regularly and used at 70–75% confluence for experiments. For DYRK1A knockdown, siRNA was complexed with lipid reagent and applied in a six‑well format at a final effective molarity reported as 22.85 nM per well; cells were cultured for 96 hours before lysate preparation. Western blotting was performed on cell lysates after 96 hours of treatment. Lysates were prepared with phosphatase inhibitors, protein quantified by BCA assay, and 30 µg (text reports 30 mg, but likely 30 µg) loaded per lane for SDS–PAGE and transfer to nitrocellulose. Blocking conditions differed for phospho- versus total-protein detection. Primary antibodies included pan-tau, 12E8 (pS262/pS356), pT231, pS396, and anti-DYRK1A; tubulin was used to verify loading. Signal development employed HRP-conjugated secondaries and chemiluminescence with care taken to use exposures within the instrument dynamic range. Compound treatments and viability assays: all compounds were solubilised in DMSO and diluted to final working concentrations (reported as 0.01 mM, 0.1 mM, 1 mM and 10 mM in the text) with a final DMSO concentration of 0.1% for all conditions. For viability, alamarBlue metabolic assays were read 12 hours after reagent addition in 96‑well plates. For tau phosphorylation assays, cells were treated in six‑well plates with freshly prepared drug-containing medium replaced every 24 hours for four days, and lysates were collected after 96 hours. In vitro kinase assays used recombinant human DYRK1A (0.08 µg) and recombinant 4R2N human tau (0.15 µg) incubated in kinase buffer with 1 mM ATP at 30°C for 30 minutes in a 20 µl volume. DYRK1A was pretreated with compounds for 10 minutes before adding tau and ATP when testing inhibitors. Reactions were stopped with sample buffer and heating, resolved on 7% Tris‑Acetate gels and probed for pS396 and total tau for quantitation. For some assays both bands of a pS396 doublet were quantified.

RNAi-mediated reduction of DYRK1A in H4 cells decreased DYRK1A protein to 38% of control and produced concomitant reductions in tau phosphorylation: pT231 was 48% of non‑silencing control and pS396 was 35% of control. This extended the authors' prior observation at the 12E8 epitope (pS262/pS356) and supports DYRK1A involvement at multiple AD‑related tau sites. Harmine produced a dose‑dependent reduction in phosphorylated tau species in cell assays. The extracted text reports that 12 mM harmine yielded 50% cell viability in the toxicity assay and that selected test doses included 80 nM, 800 nM and 8 mM. Significant (P<0.05) reductions in 12E8 and pT231 tau occurred at 0.8 mM and 8 mM; however, harmine at those concentrations also lowered total tau. After adjusting for reductions in total tau, 12E8 remained at 58% of control and pT231 at 44% of control, both significant. Effects on pS396 were largely accounted for by the total‑tau decrease. To test whether harmine effects might be mediated via monoamine oxidase A (MAO‑A) inhibition, the MAO‑A antagonist moclobemide (reported IC50 3.9 mM) was tested up to 500 mM and did not change total or phospho‑tau, arguing against MAO‑A mediation of harmine's effects. A panel of b‑carboline derivatives (harmol, harmane, harmaline, norharmane, 9‑ethylharmine, and two proprietary compounds MPP‑021 and MPP‑313) was evaluated. Toxicity rank order reported in the text was: 9‑ethylharmine, harmine, harmol, harmane, harmaline, MPP‑021, norharmane, MPP‑313. Only 9‑ethylharmine, harmine and harmol produced significant reductions (P<0.05) in total and phosphorylated tau at doses ≥1 mM; MPP‑021 and MPP‑313 reduced tau at 50 mM. The authors report a general correlation between compound toxicity and the sensitivity with which total and phospho‑tau were reduced. In vitro phosphorylation assays confirmed that DYRK1A directly phosphorylates tau on serine 396 and produced a pS396 doublet. Harmine inhibited DYRK1A‑dependent pS396 with an IC50 reported as 0.7 mM. Across the derivative panel, IC50s for inhibition of DYRK1A‑mediated pS396 were: harmol 90 nM, 9‑ethylharmine 400 nM, harmine 700 nM, with harmane, norharmane and harmaline showing more than an order of magnitude lower affinity than harmine. MPP‑313 inhibited the lower pS396 band with an IC50 of 5 mM but produced a striking biphasic effect on the higher molecular weight pS396 band: levels declined to about 50% at 5 mM but then increased to 255% and 640% of control at 50 mM and 100 mM, respectively. The authors note one exception in rank order between in vitro potency and cell assays—harmol was most potent in vitro (90 nM) but ranked third in cell‑based assays, suggesting possible cellular metabolism or other factors. Additional observations: baseline phosphorylation of recombinant tau at T231 and the 12E8 epitope precluded testing direct DYRK1A phosphorylation at those sites in vitro. The pattern across compounds showed that decreases in pS396 often paralleled reductions in total tau, whereas reductions in 12E8 and pT231 generally remained significant after correction for total tau changes.

Frost and colleagues interpret their findings as supporting a model in which DYRK1A is a promiscuous tau kinase that can influence phosphorylation at multiple sites relevant to AD pathology, including T231 and S396 in addition to previously reported sites such as T212, S202, and S404. The authors argue that pharmacological inhibition of DYRK1A could modify tau function and the progression of tau pathology in AD and related tauopathies. Structure–activity relationships across the b‑carboline series indicated that aromaticity, substitutions at carbon 7, the methyl group on carbon 1, and alkylation at N‑9 affect potency and toxicity; for example, harmol (a 7‑OH derivative) showed highest in vitro potency, and N‑9 ethylation increased affinity relative to harmine. However, the authors note a consistent correlation between reductions in tau levels and cellular toxicity across compounds and were unable to identify derivatives that dissociated tau effects from toxicity. They suggest this may reflect a causative link in which excessive tau depletion produces toxicity, consistent with prior siRNA‑based tau reductions. Several uncertainties and limitations are acknowledged. High baseline phosphorylation of recombinant tau prevented direct in vitro assessment of DYRK1A activity at some epitopes (T231 and the 12E8 epitope). The discrepancy between in vitro potency (e.g. harmol) and cellular efficacy could reflect differential cellular metabolism of functional groups such as the 7‑hydroxyl. The biphasic activation pattern observed with MPP‑313 at high concentrations is unexplained and implies potential compound‑specific effects on DYRK1A activity at suprapharmacological doses. The authors state that further experiments are required to clarify the relationship among compound exposure, pS396, total tau levels and cellular toxicity. Overall, the investigators position b‑carboline alkaloids as a chemical class that inhibits DYRK1A and reduces multiple phosphorylated tau species, but they caution that toxicity associated with extensive tau reduction and compound‑specific behaviours will need to be addressed in future optimisation and mechanistic studies.

The authors conclude that b‑carboline alkaloids, including harmine, inhibit DYRK1A kinase activity and reduce levels of multiple phosphorylated tau species implicated in AD pathology. They propose that refining chemical substituents that determine affinity for DYRK1A could yield higher‑affinity inhibitors of tau phosphorylation, but that toxicity linked to substantial tau depletion and compound‑specific effects will need resolution in further development.