Cannabinol β‑O-Glucoside Inhibitors of the Keap1-Nrf2 Interaction: A Novel Therapeutic Strategy for Parkinson’s Disease

Abstract

Oxidative stress and xenobiotic exposure disrupt redox homeostasis and contribute to neurodegenerative disease progression. The Keap1-Nrf2-ARE pathway is a key antioxidant defense mechanism, in which Nrf2 activates cytoprotective gene expression. Because disrupting the Keap1-Nrf2 interaction can enhance antioxidant responses, we assessed modified Cannabis sativa cannabinoids bearing C- and O-β-d-glucosides as potential Keap1 inhibitors. Building on our earlier fractionation and derivatization work, these compounds were evaluated for binding to the Keap1 BTB and KDC domains and for activation of Nrf2-mediated responses using in vitro and in silico methods. In a Parkinson’s disease model, CS-11 (4a) showed the strongest neuroprotective effect, with an EC₅₀ of 100 nM. Treatment significantly increased cytoprotective gene expression, indicating enhanced Nrf2 transcriptional activity. CS-11 (4a) also restored tyrosine hydroxylase levels and protected dopaminergic neurons from cell death. Computational analyses supported its Keap1-targeting activity, identifying CS-11 (4a) as a promising Parkinson’s disease therapeutic candidate.

Oxidative stress plays a crucial role in the pathophysiology of neurodegenerative diseases (NDDs). Reactive oxygen species (ROS) can cause significant molecular-level damage to proteins. The Kelch-like ECH-associated protein 1 (Keap1)nuclear factor erythroid 2-related factor 2 (Nrf2)antioxidant response element (ARE) system (Keap1-Nrf2-ARE) is an intrinsic defense mechanism activated under oxidative stress. Keap1 comprises three major domains: the BTB domain (Broad complex, Tramtrack, Bric-à-brac), the intervening region (IVR), and the C-terminal Kelch/DGR domain (KDC). Nrf2, a vital cytonuclear shuttling transcriptional activator, regulates the transcription of antioxidant genes containing AREs, which help reduce oxidative stress. Keap1 is a negative regulator of Nrf2 that directs Nrf2 toward ubiquitin-mediated proteasomal degradation under normal conditions. However, during oxidative stress, redox-sensitive cysteine residues in Keap1 undergo a conformational change, disrupting the Keap1-Nrf2 interaction. This allows Nrf2 to dissociate from the complex and translocate to the nucleus, promoting the expression of antioxidant enzymes such as HO-1, NQO1, and GCLC. In NDDs, endogenous defense mechanisms, including the Keap1-Nrf2-ARE pathway, are often insufficient to combat oxidative stress. Therefore, targeting the Keap1-Nrf2 interaction to enhance Nrf2 activity presents a promising therapeutic approach for mitigating oxidative stress in these diseases. Drug discovery aimed at disrupting the Keap1-Nrf2 interaction has garnered attention in the context of central nervous system (CNS) diseases, including Parkinson’s disease (PD). − Notable examples of potent natural molecules include sulforaphanes, as well as the antioxidants curcumin and resveratrol, along with various polyphenolic compounds and terpenoids. , Along side natural products, numerous synthetic and semisynthetic compounds have been evaluated for their ability to modulate Keap1-Nrf2 interactions. Among the most extensively studied are fumaric acid esters, triterpenoids, and specialized protein–protein interaction inhibitors such as PRL295 and NG262, as well as diverse small molecules designed from core scaffolds like coumarin, hydroxyquinoline, and indoline aimed at enhancing selectivity. , Other recent examples of Keap1-Nrf2 inhibitors include inflachromene that demonstrated in vivo PD efficacy via direct Keap1-Nrf2 disruption, N-(1H-indol-5-yl)­aryl-acrylamide derivatives that function as dual NRF2 inducers and MAO-B inhibitors and Keap1-based PROTACs designed for the degradation of Keap1. Findings from our laboratory suggest that dimethyl fumarate (DMF) may directly inhibit the Keap1-Nrf2 interaction, not only via reactive cysteine 151 but also through interactions with the β-propeller domain of Keap1 (Keap1-DC), which directly associates with Nrf2. Nonetheless, despite these advancements, only a limited number of lead compounds have progressed to clinical trials. The majority of these identified molecules function as indirect activators of Nrf2, primarily through electrophilic interactions with cysteine residues. Noncovalent inhibitors address the challenges on selectivity, toxicity, and predictability that have hindered the advancement of covalent Nrf2 activators, thereby establishing them as the more advantageous approach for clinical translation.

The cannabis-derived phytochemicals are used as antipsychotic, anti-inflammatory, and neuroprotective curing agents in various disease pathologies. The Cannabis sativa (marijuana) plant is reported in more than 110 cannabinoid classes of natural products and shows high potential therapeutic drug molecules for Alzheimer’s disease (AD), stroke, and PD. Cannabinoids are well-known for their potential in treating NDDs, traumatic brain injury, and ischemic stroke due to their antioxidative, anti-inflammatory, neuroprotective, and neurogenic properties. − Previously, we reported a method for fractionation of cannabis extracts that contain THC, CBN, THCV, and cannabivarin (CBDV) (tetrahydrocannabidiverin). Using 2,3,4,6-tetra-O-acetyl-d-glucosyl trichloroacetimidate as a donor, these cannabinoids were glucosylated, and after purification by HPLC, cannabinoid glucosides (1a–4a and 1b–4b) were identified (refer Schemes S1 and S2). , The structural confirmation of compounds 1a–4a and 1b–4b was effectively established through analysis of the coupling constants of the anomeric protons ¹H NMR spectra. Specifically, the coupling constants for compounds 1a, δH 4.93 (d, J = 7.1 Hz, 1H, δC 100.40); 2a, δH 4.96 (d, J = 7.4 Hz, 1H, δC 100.35); 3a, δH 4.92 (d, J = 7.2 Hz, 1H, δC 100.43); 4a, δH 5.06 (d, J = 7.7 Hz, 1H, δC 100.45); 1b, δH 4.66 (d, J = 9.7 Hz, 1H, δC 79.16); 2b, δH 4.54 (d, J = 9.7 Hz, 1H, δC 79.16); 3b, δH 4.53 (d, J = 9.7 Hz, 1H, δC 79.16); and 4b, δH 4.63 (d, J = 9.7 Hz, 1H, δC 79.05), respectively. These measurements clearly confirm the β-configuration of the glycosides. Additionally, we successfully identified the aglycon units in both series of compounds, enhancing our understanding of their structural characteristics (refer to Figures S2–S19). After screening all the compounds, we find that 4a is the most potent, so we synthesize CBN (4) and its O-glycoside. The experimental details were explained in the Supporting Information (pp S5–S7).

Upon screening all these compounds for binding efficiency and testing them on the human neuroblastoma cell line SH-SY5Y, several compounds demonstrated promising activity, as detailed in Table (presented later in this paper). Our results suggest that these sugar modifications, particularly 4a, enhance the binding efficiency of the compounds compared to the core molecule CBN (4), as evidenced by both biochemical assays and SH-SY5Y-based PD cell model assays.

2. Cannabinoids and Their O-Glucosidic Derivatives .

^aThe compound details and the EC₅₀ values in SH-SY5Y cells are shown.

In-Vitro Binding Studies of Cannabis-Derived Compounds against the BTB and KDC Domains of Keap1 by MST Assay

The purity of the BTB and KDC domains of Keap1 was assessed by SDS-PAGE before performing binding assays (Figure S20). Recent research revealed that cannabis-derived compounds interact with Keap1/Bach1 proteins and inhibit their function significantly. Intriguingly, cannabis-derived compounds help the Nrf2 expression and its downstream homeostasis pathway. We employed the MST assay to assess the binding efficiency of CBN and its derivatives to the BTB and KDC domains of Keap1. Surprisingly, CBN and its derivatives showed substantial bindings in the low μM to nM range with BTB and KDC (Figures and ).

1.

The quantitative binding analysis of BTB with CBN and its derivatives. The concentration-dependent binding response of various ligands with BTB shows good binding affinity: (A) BTB-CBN (11.5 μM), (B) BTB-CS11 (1.2 μM), (C) BTB-CS12 (6.5 μM), (D) BTB-CS13 (360 nM), and (E) BTB-CS14 (185 nM).

2.

Quantitative binding studies of the KDC domain with CBN and its derivatives. Concentration-dependent response behavior of the binding interaction of CBN and its derivatives with the KDC domain shows good binding: (A) CBN, 869 nM; (B) CS11, 71 nM; (C) CS12, 455 nM, (D) CS13, 155 nM, and (E) CS14, 98 nM.

For the BTB domain, the binding constants were determined to be 11.5 μM for CBN, 1.2 μM for CS-11, 6.5 μM for CS-12, 360 nM for CS-13, and 185 nM for CS-14, respectively (Figure ). The assay suggested that the CBN binds to the BTB domain less effectively than derivatives (CS-11, CS-12, CS-13, and CS-14).

For KDC, the binding constants were determined to be 869 nM for CBN, 71 nM for CS-11, 455 nM for CS-12, 155 nM for CS-13, and 98 nM for CS-14 with the KDC domain (Figure ). The MST assay indicates that adding a sugar moiety via an O-link to the CBN significantly improves its binding efficiency to the BTB and KDC domains. Moreover, the study revealed that CBN and its derivatives bind substantially to the KDC domain than to the BTB domain of Keap1.

In-Vitro Binding Studies of Cannabis-Derived Compounds against the BTB and KDC Domains of Keap1 by SPR Assay

Surface plasmon resonance (SPR) assay was also conducted to examine the interaction of the cannabinol and its derivatives with the Keap1 domains. , Kinetic analyses were conducted using a BIACore 1K+ (Cytiva, Sweden), in which both proteins were immobilized on a CM5 sensor chip. The compounds were allowed to flow over the immobilized-protein surface, and the analyte’s binding response to the protein was recorded. Moreover, the assay was performed using Keap1-DC mutants to determine whether these compounds bind to the Nrf2-binding site. The sensograms were reference subtracted, and the optimal fitting of the curves was done in accordance with the 1:1 Langmuir binding eq (see Figures , , and Figures S21–S24). For compounds exhibiting rapid on- and off-cycles, the K _D value was derived from the steady-state curves (see insets for Figures and ). The equilibrium dissociation constants and the kinetic parameters obtained for all the compounds are enumerated in Table and Table S3, respectively.

3.

Surface plasmon resonance studies for the binding of cannabinol and its glycoside derivatives with the BTB domain of Keap1. The sensograms are reference-subtracted and blank-subtracted. The colored lines show the representative multicycle kinetic sensorgrams of (A) CBN, (B) CS11, (C) CS12, (D) CS14, and (E) CS13 (right panel: Affinity curve obtained from normalized and averaged independent triplicate experiments, and affinity calculated using equilibrium 1:1 binding model) compounds interacting with the BTB protein. In panels (A), (B), (C), and (D), the thinner black lines following the curves represent the 1:1 interaction fitting. All experiments were performed in triplicate.

4.

Surface plasmon resonance studies for the binding of cannabinol and its glycoside derivatives with the KDC domain of Keap1. The sensograms are reference-subtracted and blank-subtracted. The colored lines show the representative multicycle kinetic sensorgrams of (A) CBN, (B) CS11, (C) CS12, (D) CS13, and (E) CS14 compounds interacting with the KDC protein. The affinity curves obtained from normalized and averaged independent triplicate experiments for respective compounds are shown in the respective right panels, calculated using the equilibrium 1:1 binding model. Error bars represent the standard deviation.

1. Equilibrium Dissociation Constants, K _D Value of Cannabinol and Its Glycoside Derivatives for BTB and KDC Domains Obtained using SPR Assay .

	CBN	CS11	CS12	CS13	CS14
BTB	154 ± 23 nM	171 ± 26 nM	38.6 ± 4.4 μM	125 ± 62 nM	497 ± 61 nM
KDC	222 ± 27 nM	74.1 ± 6.3 nM	197 ± 39 nM	188.5 ± 4.9 nM	304.6 ± 7.0 nM

^aThe value represents the mean of three independent experiments, with the standard deviation reported.

It is important to note that MST and SPR rely on distinct physical principlesSPR detects binding at a surface, while MST assesses interactions in free solutionwhich may result in variations in their respective K _D values. , By comparing the MST findings for BTB compounds, CBN and CS11 demonstrated low nanomolar binding affinities in SPR, whereas CS12 was identified as a weak binder. In the case of KDC, the K_D values obtained by SPR correlate with the MST results and are of the same order of magnitude, suggesting a nanomolar binding range for all compounds interacting with KDC. CS11, with K _D values of 171 ± 26 nM and 74.1 ± 6.3 nM for BTB and KDC domains, proves to be a high-affinity binder to the Keap1 protein. Overall, the SPR assay data are robust and correlate well with our earlier MST findings.

The KDC mutants at the Nrf2 site further validate that these compounds bind specifically at the Nrf2 binding site. Mutations Y334A, Y525A, Y572A, and S602A substantially weakened ligand binding, shifting affinities from the nanomolar to the micromolar range or eliminating detectable binding altogether (see Table S4 and Figures S21–S24). Together, these results demonstrate that all of the above-mentioned compounds interact with the Nrf2 binding site in a site-specific manner.

Neuroprotective Effects of CBN and Its Derivatives Using Cell Viability Assay

Recently, substances derived from Cannabis sativa, notably cannabidiol (CBD), delta-9-tetrahydrocannabinol (Δ9-THC), and CBN, have garnered attention in medical research for their neuroprotective potential in models of NDDs. Based on data from binding studies with the KDC and BTB domains, we surmised that CBN and its derivatives may have neuroprotective potential via the Nrf2-Keap1 pathway in NDDs such as PD. Cell viability experiments were carried out using SH-SY5Y cells to confirm the findings (Table ).

SH-SY5Y cells were co-treated with the known PD toxin rotenone (1 μM) and CBN (0.1, 0.5, 1.0, 2.0, 4.0, and 8.0 μM) or CS-11 (10, 100, 500, and 1000 nM), followed by assessment of cell viability (Figure A and B). The CBN showed the highest cell survival at 2 μM, whereas CS-11 showed maximum protection at 100 nM. Since the cell viability assay showed that CS-11 was relatively more potent and neuroprotective, we used this derivative for subsequent studies.

5.

(A and B) Effect of CBN and CS11 on the cell survival in SH-SY5Y cells. The SH-SY5Y cells were seeded for 12 h and, thereafter, treated with (A) CBN and (B) CS11 compounds in various concentrations (0.1, 0.5, 1, 2, 4, and 8 μM) and (10, 100, 500, and 1000 nM), respectively, for 24 h. (C) Effect of CS11 on the mRNA level of antioxidant enzymes. The mRNA levels of antioxidant enzymes (HO-1, GCLC, and NQO1) were quantified in the presence or absence of CS11 (100 nM). Each experiment was performed in triplicate, and the data shown are the averages of three independent experiments (n = 3) with the error bars representing the standard deviation. (*) p ≤ 0.05, (**) p < 0.01, and (***) p < 0.001; where the p-value has been calculated using a paired two-tailed t-test compared to the rotenone-treated wells.

Effect of CS11 on SH-SY5Y Cells at mRNA Expression Level (qRT-PCR) and Nuclear Translocation of Nrf2 (Western Blot)

The transcriptional activating protein Nrf2 enters the nucleus in response to oxidative stress. It increases the expression of many cytoprotective genes, including glutamate cysteine ligase (GCLC), heme oxygenase-1 (HO-1), and NAD­(P)H quinone oxidoreductase 1 (NQO-1). , Here, we evaluated the effect of the CS-11 compound on antioxidant enzymes at the transcriptional level, primarily regulated by Nrf2 activation. We measured mRNA expression levels of HO-1, GCLC, and NQO1 using qRT-PCR. The SH-SY5Y cells were cotreated with rotenone and CS11 at 1 μM and 100 nM, respectively, for 24 h. The mRNA expression levels of GCLC and NQO1 increased significantly compared with those in rotenone-treated cells (Figure C). This observation indicated that CS-11 significantly increases the mRNA expression levels of antioxidant enzymes, suggesting that CS-11 competitively binds to Keap1, freeing Nrf2 to activate ARE-driven antioxidant enzymes. In addition, qRT-PCR results were consistent with our Nrf2 nuclear translocation studies. Additionally, to build upon the increased mRNA expression of antioxidant enzymes observed after CS-11 administration, we investigated the corresponding protein levels and the nuclear localization of Nrf2. Western blot analysis showed a marked increase in protein levels of the enzymes GCLC and NQO1, accompanied by greater nuclear accumulation of Nrf2 in cells treated with CS-11 (Figure S25). Taken together, the marked increase in antioxidant enzyme expression and the nuclear translocation of Nrf2 together demonstrate that CS-11 activates the cell’s intrinsic defense mechanisms.

Effect of CS-11 on PD-Specific Marker Tyrosine Hydroxylase (TH), Apoptosis, and Neuronal Stability

Next, we performed a tyrosine hydroxylase (TH) assay to evaluate the neuroprotective effect of CS-11 on rotenone-induced SH-SY5Y cells. Immunoblotting data showed that TH levels were significantly decreased in rotenone-treated cells compared with control cells. The CS-11 treatment significantly increased TH levels in rotenone+CS-11-treated cells. Dopaminergic neuronal death was assessed using the terminal neuronal apoptosis marker cleaved caspase 3 in all sets. Data showed that in rotenone-treated cells, neuronal death was significant and was significantly inhibited by CS11 treatment. The immunoblot data depict that cleaved caspase 3 was significantly increased in rotenone-treated cells compared to control, whereas CS11 treatment significantly rescued the cells from rotenone-induced apoptosis (Figure A–C). The immunofluorescence data on TH expression were corroborated by the above findings (Figure F, G). Rotenone acts as a tubulin-binding agent, disrupting microtubule assembly, and can be assessed by the level of β3-tubulin, a neuron-specific protein involved in neuronal maturation and axon guidance. CS11 treatment significantly ameliorated the rotenone-induced decrease in β3-tubulin levels (Figure D, E).

6.

(A−E) Immunoblots and graphs depicting the tyrosine hydroxylase (TH), cleaved caspase 3, and β3 tubulin levels after treatment with rotenone (R) (1 μM) and CS11 (100 nM) for 24 h in SH-SY5Y cells (n = 3). (F) Representative immunofluorescence images (40×) of tyrosine hydroxylase (green) in SH-SY5Y cells after treatment with rotenone and CS11. DAPI (blue) indicates nuclear staining (scale bar = 50 μm). (G) Quantification of TH integrated density in SH-SY5Y cells is shown in panel (F). (*) p < 0.05, (***) p < 0.001 control (C) vs Rotenone (1 μM), (#) p < 0.05, (##) p < 0.001 Rotenone vs Rotenone+CS11.

Molecular Docking Studies for Cannabis-Derived Compounds against BTB and KDC

The synthesized compounds were subjected to computational studies using the KDC and BTB domains to examine their binding modes and intermolecular interactions. Molecule docking was performed using AutoDock Vina v1.1.2. For the KDC domain, the direct interaction site of Nrf2 was targeted as the established interaction site for the compounds. For the BTB domain, the binding site encompassing Cys151 and surrounding residues was considered, as this site is proposed to inhibit Cul3 binding to Keap1 by disrupting Cul3-BTB interactions. A grid box with coordinates 50 Å × 54 Å × 52 Å in the X-, Y-, and Z-directions and center_x = −90.928, center_y = 101.319, and center_z = −0.97 Å was used for the blind docking of ligands on the KDC protein. The site-specific docking on BTB was performed using a grid box of dimensions 20 Å × 20 Å × 20 Å in the X-, Y-, and Z-directions and center_x = −24.912, center_y = 35.547, and center_z = −33.248 Å. Figure and Figure S26 illustrate the best-docked binding modes for all derivatives with their respective Keap1 domains, and Table lists the binding free-energy values. The parent molecule, CBN, exhibited lower binding energy than its derivatives. CBN displayed only aromatic interactions with the domains: in the KDC domain, it engaged in a π edge-to-face interaction with Tyr525, whereas in the BTB domain, it interacted with His129 and His154 through π–π stacking interactions. In interactions with the KDC domain, all derivatives were observed to form aromatic interactions with Tyr334. Additionally, the presence of sugar moieties in these derivatives facilitated multiple polar interactions, thereby increasing their affinity for the KDC-Nrf2 binding site. Specifically, CS-11, CS-12, and CS-14 formed multiple hydrogen bonds with Arg415 and Ser555, while CS-12 and CS-14 also formed hydrogen bonds with Ser508. Compound CS-13 could interact with Arg414, Ser431, Gly364, and Ser602 through hydrogen bonding with the side chains of these amino acids.

7.

Docking interaction analysis for the top selected pose of each compound. Intermolecular interaction of (A) CBN and (B) CS11 with respective domains. The left panel indicates interaction with the KDC domain (protein shown as a gray cartoon), and the right panel indicates interactions with the BTB domain (protein shown as a light pink cartoon). Binding site residues are shown as yellow sticks.

3. Description of Binding Free Energy Values and the List of Interacting Residues for Each Protein–Ligand Complex Obtained by Docking Studies.

Compound	Protein	Binding free energy (kcal/mol)	Aromatic interaction	Hydrogen bond
CBN	KDC	–7.2	Tyr525	–
CS-11	KDC	–8.3	Tyr334	Arg415, Ser555
CS-12	KDC	–7.7	Tyr334	Arg415, Ser508, Ser555
CS-13	KDC	–7.5	Tyr334	Gly364, Asn414, Ser602
CS-14	KDC	–8.5	Tyr334, Tyr525	Arg415, Ser508, Ser555
CBN	BTB	–5.6	His129, His154	–
CS-11	BTB	–5.1	His129, His154	Tyr85
CS-12	BTB	–6.3	His154	Tyr85, Gln86, His129
CS-13	BTB	–5.6	His129, His154	–
CS-14	BTB	–5.2	His129, His154	Tyr85, Gln86

In the BTB domain, the major anchor to the binding site was due to aromatic interactions with His129 and His154 residues, where compounds are sandwiched between the two amino acids. CS-11 was also found interacting with Tyr85. CS-12 forms hydrogen bonds with the side chains of Tyr85, Gln86, and His129, and CS-14 with Tyr85 and Gln86. Detailed list of interactions are tabulated in Table .

Molecular Dynamics Simulation Studies for Cannabis-Derived Compounds against BTB and KDC

To examine the dynamics of intermolecular interactions and stability at the desired binding sites, the complexes of each domain with the designed compounds were simulated for 100 ns. The results of the complete trajectory analysis are compiled in Figure S27. The protein backbone RMSD gives the stability and structural changes of the protein during the simulation in a complex with a compound. The first frame’s protein structure was considered the reference structure for RMSD calculation. Both domains in complex with the compounds showed stable RMSD (Figure S27A), indicating stable complex formation. Protein RMSF calculation was carried out to check the effect on the flexibility of each residue upon compound binding. As observed, there was no major fluctuation, and the complexes had RMSF values similar to those of the apo structures (Figure S27B). For comparison, the average RMSD and RMSF values for the proteins in apo form and in complex with the compounds are listed in Table S6. As the proteins in the presence of the ligand exhibit a pattern similar to that in the apo form, and the averaged parameters show no significant difference compared to apo proteins (Table S6), this indicates fewer distortions and well-structured regions even upon compound binding.

Regarding ligand flexibility at the binding site, ligand RMSD with respect to the protein was plotted for the entire run. For the calculation, the protein in the first frame was used as the reference, and the protein in all 2000 frames was superposed onto it; this yielded ligand fluctuations at the protein’s binding site. Among the compounds bound to the BTB domain, CS-11 and CS-13 deviate least from their initial positions, while CBN remained stabilized until 70 ns, after which some fluctuations were observed. CS-12 and CS-14 showed small changes in orientation at the site, resulting in a high RMSD (Figure S27C, left panel). For complexes with the KDC domain, all derivatives stabilized within the initial 5 ns and showed high stability at the binding site (Figure S27C, right panel).

Interaction Analysis with BTB Domain

CBN retains the hydrophobic π-cation interaction with Lys131 throughout the run and forms a hydrophobic interaction with Cys151, which is maintained for 80 ns. The compound was mainly anchored by interactions with these residues. CS-11 showed high stability at the binding site, forming multiple hydrophobic and hydrogen bonds (Figure A and B). It contributes aromatic interactions with Tyr85, His129, and His154 and hydrophobic interactions with Val132 and Lys150. Interestingly, in addition to the hydrogen bonds with Arg135, Lys150, and His154, CS-11 also possesses hydrogen bonds with Cys151. CS-12 forms a stable hydrogen bond with Ser146. Continuous hydrophobic interactions were also observed with Val132, Cys151, and Val155. CS-12 also contributed to π-cation interactions with Lys131. CS-13 forms hydrophobic interactions with Tyr85 and Val132 and engages in continuous π-π interactions with His154. Hydrogen bond interactions were observed with residues Tyr85, Gln86, Asp87, and His154. Interaction analysis of CS-14 revealed hydrophobic interactions with Tyr85, Val132, and Cys151, aromatic interactions with His154, and hydrogen bonding with Asp87, His129, and His154 (Figure S28A and S28B).

8.

Detailed interaction analysis for CS11 in complex with BTB and KDC domains. Panels (A) and (C) demonstrate hydrogen bonds formed by CS11 in BTB and KDC, respectively; panels (B) and (D) demonstrate hydrophobic interactions made by CS11 in BTB and KDC, respectively.

For the BTB domain, trajectory cluster analysis was performed to compare the binding of the proposed compounds with that of the known triterpenoid drug 2-cyano-3,12-dioxooleana-1,9(11)-dien-28-oic acid (CDDO). The CDDO and its analog drug molecules bind covalently to Cys151 of the BTB domain and are known to reduce the affinity of Keap1-CUL3 binding. The proposed compounds, CS11, CS12, and CS14, show a similar binding pattern to CDDO (Figure C, D, F). Interestingly, CS13 showed interactions at the 3-box motif from the BTB-IVR region. The 3-box within the BTB-IVR domain forms a short helical motif that is found to be critical for high-affinity CUL3 interaction (Figure A). The binding of CS13 to this region might also be responsible for disrupting Keap1-Cul3 interactions.

9.

Cluster analysis of 100 ns Simulation trajectory for BTB-ligand complexes. (A) Cul3_NTD-KEAP1_{BTB‑3‑Box} structure (PDB-5NLB). Cul3_NTD, Keap1_BTB, and Keap1_3‑Box interacting residues are shown as dark green, blue, and brown sticks, respectively. The clusters with the largest numbers of members for (B) CBN, (C) CS11, (D) CS12, (E) CS13, and (F) CS14, superimposed on the Cul3_NTD-KEAP1_{BTB‑3‑Box} structure.

Interaction Analysis with the KDC Domain

The intermolecular interaction pattern of CBN with the KDC domain indicates the presence of continuous aromatic interactions with Tyr334, Tyr572, and Phe577. Although CBN forms a hydrogen bond with Ser602, the interaction is lost after 10 ns. It is mainly supported at the binding site by aromatic interactions. The residues Tyr334 and Tyr525 exhibited π–π interactions, and Tyr572 and Phe577 made hydrophobic interactions with CS-11 throughout the run. CS-11 also established hydrogen bonds with Tyr525, Gln530, and Tyr572 (Figure C, D). Similarly, CS-12 also made hydrophobic interactions with Tyr334, Tyr572, and Phe577. Moreover, CS-12 was found to form a constant hydrogen bond with Gln530 and Ser555. CS-13 made strong hydrogen bonds with Ser363, Asn414, and Ser602 while retaining hydrophobic bonds with Tyr334, Phe478, Tyr525, and Tyr572. Over the entire trajectory, CS-14 maintained significant hydrogen bonding with Ile461, Ser508, and Ser555. Stable hydrophobic interactions with Tyr334, Tyr525, Ala556, and Tyr572 were also noticed (Figure S28C, D).

Overall, the most active derivative, CS-11, exhibits highly stable binding to both domains. It maintains the interactions with critical residues of both domains throughout the simulation. The computational studies support the dual-mode action of derivatives, i.e., directly affecting Nrf2 interaction by blocking the KDC-Nrf2 site and indirectly hindering the BTB-Cul3 association.

In conclusion, our research demonstrated that the CBN derivative CS-11 inhibits the Keap1-Nrf2 PPI at nanomolar concentrations and substantially enhances neuroprotection in SH-SY5Y cells by activating the Nrf2-ARE pathway. Biochemical analysis reveals that CS-11 shows a notably higher binding affinity for both the Keap1-DC and BTB domains compared to the CBN (core moiety). Treatment with CS-11 led to pronounced protective effects in SH-SY5Y cells.

Quantitative RT-PCR results confirmed an upregulation of mRNA levels for Nrf2-driven antioxidant enzymes. Additionally, immunoblotting assays confirmed Nrf2’s translocation into the nucleus and a corresponding increase in antioxidant enzyme protein levels when cells were treated with CS-11. The results further demonstrate that CS-11 exhibits anti-Parkinson’s disease properties, shown by its ability to significantly prevent rotenone-induced loss of TH and reduce neuronal cell death. Furthermore, computational modeling supports these outcomes by illustrating CS-11’s dual-binding action on Keap1. This study identifies CS-11 as a potent small-molecule Keap1 inhibitor, offering new scaffolds for drug development. The compound showed marked cytoprotective efficacy in the SH-SY5Y PD model, highlighting its promise as a lead inhibitor for novel PD therapies.

Glossary

Abbreviations

AD

Alzheimer’s Disease

ARE

Antioxidant Response Element

BTB

Broad-complex, Tramtrack, Bric-à-brac domain

CBN

Cannabinol

CBD

Cannabidiol

CBDV

Cannabidivarin

CH₂Cl₂

Dichloromethane

CNS

Central Nervous System

CUL3

Cullin-3

DAPI

4′,6-Diamidino-2-phenylindole

DMF

Dimethyl Fumarate

EC₅₀

Half-maximal Effective Concentration

GCLC

Glutamate–Cysteine Ligase Catalytic Subunit

HPLC

High-Performance Liquid Chromatography

HR-ESI-MS

High-Resolution Electrospray Ionization Mass Spectrometry

HO-1

Heme Oxygenase-1

IVR

Intervening Region

KDC

Kelch/DGR C-terminal Domain

Keap1

Kelch-like ECH-associated Protein 1

K _D

Equilibrium Dissociation Constant

MAO-B

Monoamine Oxidase B

MD

Molecular Dynamics

MeOD

Deuterated Methanol

MST

Microscale Thermophoresis

NDD

Neurodegenerative Disease

NMR

Nuclear Magnetic Resonance

NQO1

NAD­(P)H Quinone Oxidoreductase 1

Nrf2

Nuclear Factor Erythroid 2-related Factor 2

PD

Parkinson’s Disease

PDB

Protein Data Bank

PROTAC

Proteolysis-Targeting Chimera

PPI

Protein–Protein Interaction

qRT-PCR

Quantitative Reverse Transcription Polymerase Chain Reaction

RMSD

Root Mean Square Deviation

RMSF

Root Mean Square Fluctuation

ROS

Reactive Oxygen Species

RT

Room Temperature

SH-SY5Y

Human Neuroblastoma Cell Line

SPR

Surface Plasmon Resonance

TH

Tyrosine Hydroxylase

THC

Δ⁹-Tetrahydrocannabinol

THCV

Tetrahydrocannabivarin

TMSOTf

Trimethylsilyl Trifluoromethanesulfonate

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.6c00156.

• Scheme S1: Reagents and Conditions: (a) TMSOTf, dry CH₂Cl₂, 0 °C to rt, 10 h; (b) NaOMe, Methanol: CH₂Cl₂ (8:2), rt, 24 h. Scheme S2: Total Synthesis of Cannabinol-O and β-d-Glucoside. Figure S1: ¹H NMR (400 MHz, CDCl₃) of CBN (4). Figure S2: ¹H NMR (400 MHz, MeOD) of CS-11 (4a). Figure S3: ¹³C NMR (100 MHz, MeOD) of CS-11 (4a). Figure S4: bDEPT-135 NMR (100 MHz, MeOD) of CS-11 (4a). Figure S5: (+)-HR-ESI-MS of CS-11 (4a). Figure S6: HPLC Chromatogram of CS-11 (4a). Figure S7: Stability Studies of Compound CS-11 (4a). Figure S8: ¹H NMR (400 MHz, MeOD) of CS-12 (2a). Figure S9: ¹³C NMR (100 MHz, MeOD) of CS-12 (2a). Figure S10: DEPT-135 NMR (100 MHz, MeOD) of CS-12 (2a). Figure S11: (+)-HR-ESI-MS of CS-12 (2a). Figure S12: ¹H NMR (400 MHz, MeOD) of CS-13 (3a). Figure S13: ¹³C NMR (100 MHz, MeOD) of CS-13 (3a). Figure S14: mDEPT-135 NMR (100 MHz, MeOD) of CS-13 (3a). Figure S15: (+)-HR-ESI-MS of CS-13 (3a). Figure S16: ¹H NMR (400 MHz, MeOD) of CS-14 (1a). Figure S17: ¹³C NMR (100 MHz, MeOD) of CS-14 (1a). Figure S18: DEPT-135 NMR (100 MHz, MeOD) of CS-14 (1a). Figure S19: (+)-HR-ESI-MS of CS-14 (1a). Figure S20: SDS–PAGE analysis of the final size exclusion chromatography purification step for both proteins. Figure S21: Overlay plot of sensograms from a steady-state affinity with five sample injections of mKDC mutant (Y334A) with ligands (CBN, CS11, CS12, CS13, and CS14), respectively (A–J). The sensograms are reference-subtracted and blank-subtracted. Sample injections include five samples from a range of different concentrations. The affinity curves obtained from normalized, averaged independent triplicate experiments for the respective compounds are shown and calculated using the 1:1 equilibrium binding model. Figure S22: Overlay plot of sensograms from a steady-state affinity with five sample injections of KDC mutant (Y525A) with ligands (CBN, CS11, CS12, CS13, and CS14), respectively (A–J). The sensograms are reference-subtracted and blank-subtracted. Sample injections include five samples from a range of different concentrations. The affinity curves obtained from normalized, averaged independent triplicate experiments for the respective compounds are shown and calculated using the 1:1 equilibrium binding model. Figure S23: Overlay plot of sensograms from a steady-state affinity with five sample injections of KDC mutant (Y572A) with ligands (CBN, CS11, CS12, CS13, and CS14), respectively (A–J). The sensograms are reference-subtracted and blank-subtracted. Sample injections include five samples from a range of different concentrations. The affinity curves obtained from normalized, averaged independent triplicate experiments for the respective compounds are shown and calculated using the 1:1 equilibrium binding model. Figure S24: Overlay plot of sensograms from a steady-state affinity with five sample injections of mKDC mutant (S602A) with ligands (CBN, CS11, CS12, CS13, and CS14), respectively (A–J). The sensograms are reference-subtracted and blank-subtracted. Sample injections include five samples from a range of different concentrations. The affinity curves obtained from normalized, averaged independent triplicate experiments for the respective compounds are shown and calculated using the 1:1 equilibrium binding model. Figure S25: Nrf2 nuclear translocation and overexpression of downstream genes. Figure S26: Docking interaction analysis for the top selected pose of each compound. Figure S27: Evolution of RMSD and RMSF during 100 ns simulation of respective protein–ligand complexes. Figure S28: Detailed interaction analysis for CS14 in complex with BTB and KDC domains. Table S1: The sequence of primers used to generate the mutants. Table S2: The expression conditions for different mKeap1-DC mutants used in the study. Table S3: Kinetic parameters of cannabinol and its glycoside derivatives for the BTB domain obtained were fit to a 1:1 interaction model. Table S4: The binding affinity of mKDC mutants with the ligands. Table S5: List of antibodies used in Western blotting experiments. Table S6: The calculated parameters for the apo proteins and all the complexes obtained after 100 ns MD simulations (PDF)

∇.

Authors P. Deshmukh and S. Aouti have contributed equally to this work.

Safety Statement: No unexpected or unusually high safety hazards were encountered.

The authors declare no competing financial interest.