Article

1681

ISSN 0006-2979, Biochemistry (Moscow), 2024, Vol. 89, No. 10, pp. 1681-1691 © Pleiades Publishing, Ltd., 2024.

ISSN 0006-2979, Biochemistry (Moscow), 2024. © Pleiades Publishing, Ltd., 2024.

Computational Assessment of Carotenoids

as Keap1-Nrf2 Protein–Protein Interaction Inhibitors:

Implications for Antioxidant Strategies

Alessandro Medoro

, Tassadaq Hussain Jafar

, Fabio Sallustio

Giovanni Scapagnini

, Luciano Saso

, and Sergio Davinelli

1,a

Department of Medicine and Health Sciences “V. Tiberio”, University of Molise, Campobasso, Italy

Department of Precision and Regenerative Medicine and Ionian Area (DIMEPRE-J), University of Bari, Bari, Italy

Department of Physiology and Pharmacology “Vittorio Erspamer”, La Sapienza University, Rome, 00185, Italy

e-mail: sergio.davinelli@unimol.it

Received May 8, 2024

Revised July 21, 2024

Accepted July 23, 2024

Abstract— The Keap1-Nrf2 pathway is an essential system that maintains redox homeostasis and modulates key

metabolic processes, including metabolism of amino acids to promote the synthesis of antioxidant enzymes.

Inhibitors of the protein-protein interaction (PPI) between Keap1 and Nrf2 have emerged as a promising strategy

for developing novel classes of antioxidant agents that selectively activate this pathway without off-target effects.

Carotenoids, a large family of lipophilic isoprenoids synthesized by all photosynthetic organisms, are well-known

for their antioxidant activities. However, the ability of carotenoids to inhibit the Keap1-Nrf2 PPI through the

involvement of specific amino acid residues has not yet been revealed. We utilized molecular docking, molecular

dynamic simulations, and pharmacokinetic prediction techniques to investigate the potential of eight oxygenat-

ed carotenoids, known as xanthophylls, to inhibit Keap1. Among the compounds investigated, fucoxanthin and

astaxanthin established multiple hydrogen-bonding and hydrophobic interactions within the Kelch domain of

Keap1, showing remarkable binding affinities. Furthermore, fucoxanthin and astaxanthin displayed the ability

toform astable complex with Keap1 and fit into the binding pocket of its Kelch domain. These analyses led to the

identification of critical amino acid residues in the binding pocket of Keap1 which are involved in the interac-

tion with carotenoid xanthophylls. Our analyses further revealed that fucoxanthin and astaxanthin demonstrate

a favorable safety profile and possess pharmacokinetic properties consistent with acceptable drug-like character-

istics. These findings lay the preliminary foundation for developing a novel class of Keap1-Nrf2 PPI inhibitors with

potential applications against oxidative stress-related diseases.

DOI: 10.1134/S0006297924100031

Keywords: Keap1, Nrf2, carotenoids, xanthophylls, antioxidants

* To whom correspondence should be addressed.

INTRODUCTION

As maintaining redox homeostasis is a continu-

ous challenge, multicellular organisms have evolved

various defense systems to mitigate the toxic effects

of numerous oxidants and electrophiles to which they

are exposed. A prominent aspect of oxidative stress

is the induction of a cellular stress response, mediat-

ed by molecular redox switches. These switches are

crucial in the activation of an extensive network of an-

tioxidant enzymes. Accordingly, this activation coun-

teracts the oxidative challenge, thereby preserving

redox homeostasis [1, 2]. The Nrf2 (nuclear factor

erythroid 2 [NF-E2]-related factor 2)-Keap1 (Kelch-like

ECH-associated protein 1) pathway is one of the ma-

jor regulators of oxidative and electrophilic stress re-

sponse, playing a pivotal role in the maintenance of

the cellular redox homeostasis. Nrf2, ubiquitously ex-

pressed across various organs, is a cap’n’collar (CNC)

basic region-leucine zipper (bZIP) transcription factor.

MEDORO et al.1682

BIOCHEMISTRY (Moscow)

It regulates theexpression levels of hundreds of genes

associated with homeostatic functions and oxidative

stress response. This regulation occurs through the

binding of Nrf2 to a DNA enhancer region known as

the antioxidant response element (ARE). Furthermore,

Nrf2 is involved in amino acid metabolism, promoting

the synthesis of antioxidant enzymes and increasing

the availability of amino acids such as cysteine, ser-

ine, glycine, and asparagine. In humans, Nrf2 consists

of 605 amino acids and includes seven distinct and

highly conserved Nrf2-ECH homology (Neh) domains

(Neh1–Neh7) [3-5].

Although the activity of Nrf2 is regulated by vari-

ous mechanisms, including at transcriptional and epi-

genetic levels, its proteasomal degradation is mainly

mediated by the repressor Keap1. Under basal condi-

tions, Nrf2 protein levels remain low due to its inter-

action with Keap1 in the cytoplasm. This interaction

involves the Neh2 domain of Nrf2, which binds to

Keap1 using the DLG and ETGE motifs. This binding is

essential for the ubiquitination and subsequent prote-

asomal degradation of Nrf2. Briefly, Keap1 recruits the

Cullin 3 (Cul3)-RING-box protein 1 (RBX1) E3 ubiquitin

ligase complex with subsequent ubiquitination of Nrf2

by the 26S proteasome [6]. Human Keap1 contains 627

amino acids and consists of five distinct domains: the

N-terminal region (NTR), the broad complex, tramtrack

and bric-a-brac (BTB) domain, the intervening region

(IVR), double glycine repeats (DGR) or Kelch domain,

and the C-terminal region (CTR). The BTB domain fa-

cilitates dimerization and interaction with Cul3, while

the Kelch domain, binds to Neh2 of Nrf2 [7]. Moreover,

Keap1 contains several highly reactive Cys residues

in these domains, particularly within the IVR domain,

that act as cellular redox status sensors. The most com-

mon electrophilic Nrf2 inducers trigger conformation-

al changes in Keap1 due to modifications of its cysteine

residues, disrupting the Keap1-Nrf2 protein–protein

interaction (PPI) and leading to the nuclear accumula-

tion of Nrf2. This, in turn, activates a cellular defense

response against oxidative stress [8,9]. However, since

the mechanism of action of these compounds involves

the covalent targeting of cysteine thiols, Nrf2 inducers

may lack selectivity for Keap1, potentially affecting

other cysteines within the cells and leading to unpre-

dictable off-target side effects [10].

Currently, alternative strategies to target Nrf2

with higher selectivity have attracted increasing at-

tention. These alternative approaches focus on devel-

oping Keap1–Nrf2 PPI inhibitors to block their interac-

tion, thereby increasing Nrf2 activity and potentially

improving the safety and efficacy of treatments for

oxidative stress-related diseases. To date, several non-

covalent small-molecule Keap1-Nrf2 inhibitors have

been described [11]. These Keap1-Nrf2 PPI inhibitors

typically bind to the Kelch domain of Keap1, target-

ing the sites where the ETGE and DLG motifs of Nrf2

would normally bind. The Keap1‐Kelch domain, a

six‐bladed β‐propeller structure, is divided into six

subpockets (P1-P6). These subpockets accommodate

different amino acid residues that interact with DLG

and ETGE motifs located in the Neh2 domain of Nrf2.

Therefore, the higher selectivity of Keap1-Nrf2 PPI

inhibitors is determined by interactions between the

DLG and ETGE motifs in the Neh2 domain of Nrf2 and

the corresponding subpockets within the Keap1-Kelch

domain [12,13]. This mechanism of action selectively

inhibits the binding between Keap1 and Nrf2, thereby

enhancing Nrf2 activity and ensuring a more targeted

modulation of the pathway.

Carotenoids, which are terpenoid-based com-

pounds, are widely distributed in algae, fungi, bacteria,

and plants. They constitute a large family of fat-solu-

ble plant pigments known for their health-promoting

effects against major chronic diseases, including dia-

betes, cancer, and dementia [14-16]. Carotenoids are

typically classified into two groups, carotenes, and

xanthophylls, based on chemical and biochemical cri-

teria. Xanthophylls, such as fucoxanthin, astaxanthin,

lutein, zeaxanthin, and canthaxanthin, which are rich

in double bonds, exhibit higher antioxidant properties

than carotenes [17, 18]. Several experimental studies

have shown that marine-derived xanthophylls, such

as astaxanthin, zeaxanthin, and lutein, can activate

the Nrf2 pathway and induce the expression of anti-

oxidant enzymes, including glutathione (GSH), heme

oxygenase-1 (HO-1), and superoxide dismutase (SOD)

[19-21]. However, there is limited evidence suggesting

that these xanthophylls may disrupt the Keap1-Nrf2

PPI by interacting with Keap1, leading to the release

and activation of Nrf2 [22]. We selected from the

CHEMnetBASE – Dictionary of Marine Natural Prod-

ucts eight microalgal xanthophylls extensively studied

for their antioxidant characteristics (astaxanthin, can-

thaxanthin, echinenone, fucoxanthin, lutein, neoxan-

thin, violaxanthin, and zeaxanthin) [23-32]. We used

computational methods, including molecular docking

and molecular dynamic simulations (MDS), to inves-

tigate whether these xanthophyll carotenoids might

serve as potential candidates for developing selective

Keap1-Nrf2 PPI inhibitors.

MATERIALS AND METHODS

3D structural retrieval and preparation of the

Keap1 protein. Keap1 protein (2FLU) was retrieved

from the RCSB Protein Data Bank (https://www.rcsb.org/

structure/2FLU). In detail, it is the crystal structure of

the Kelch domain of Keap1 bound to a 16-mer pep-

tide from Nrf2 containing a highly conserved DxETGE

motif [33]. After conducting a 3D structural retrieval,

CAROTENOIDS AS KEAP1-NRF2 PPI INHIBITORS 1683

BIOCHEMISTRY (Moscow)

the protein underwent preparation using UCSF Chimera

software (v1.12) [34]. The retrieved structure was

checked for missing residues and refined removing

water molecules, additional protein chains, and li-

gands. To ensure protein stability, energy minimi-

zation, and geometry optimization were carried out

using UCSF Chimera with 1000 steps (step size 0.02 Å)

and the conjugate gradient method. The protonation

of wild-type histidine was done following the AMBER

ff98 method [35, 36]. Energy minimization was also

performed to maintain the stability of Keap1 using the

Chiron energy minimization online tool [37].

Xanthophyll carotenoids selection and struc-

ture preparation. A total of eight marine-derived

xanthophyll carotenoids were selected from the

CHEMnetBASE– Dictionary of Marine Natural Products

(https://dmnp.chemnetbase.com/chemical/Chemical

Search.xhtml?dswid=-7750) for the study: astaxanthin,

canthaxanthin, echinenone, fucoxanthin, lutein, neox-

anthin, violaxanthin, and zeaxanthin. The two-dimen-

sional (2D) chemical structures of the compounds were

searched in the PubChem database [38]. In addition,

2D chemical structures were converted into a .pdb for-

mat by Chem3D Pro, and the structural optimization

was performed by UCSF Chimera software (v1.12).

Molecular docking studies. AutoDock Vina soft-

ware was used for molecular docking [39]. The grid

size was set at 75×75×75 Å in the X, Y, and Z axes, with a

grid spacing of 0.650 Å covering Keap1 active domains.

Polar hydrogen atoms were added to the Keap1 pro-

tein structure, and each docking experiment consist-

ed of 100 runs. Default parameters for van der Waals

forces, electrostatic forces, AMBER force field, and in-

termolecular forces were used for the ligand-protein

complex. The genetic algorithm was employed as the

primary search protocol. All docked complexes were

analyzed, and the best-docked complex based on in-

termolecular interactions was visualized and analyzed

using Discovery Studio.

Molecular dynamics simulations (MDS). The

NAMD2 simulation software (version 2.14) was used

to conduct Molecular Dynamics Simulations (MDS) of

docked complexes for 120 ns in a water environment

[40]. The simulations utilized the Amber ff14SB force

field for proteins and a General Amber Force Field

(GAFF) for ligands [41, 42]. The antechamber package

generated topology and coordinate files for the li-

gands, while Xleap prepared the simulation system

[43]. The system was solvated in a cubic box with

TIP3P water molecules, and the protein was positioned

within 1 nm of the box edge to adhere to the mini-

mum image convention. Neutralization was achieved

by adding 10 Na

ions to the system. The complexes of

astaxanthin and fucoxanthin contained a total of 41363

and 41385 atoms, respectively, and were protonated

at a pH of 7.4. Energy minimization was performed

with the conjugate gradient method to eliminate ste-

ric clashes. The equilibration process involved settling

water molecules, gradual heating, and equilibration

in an NPT ensemble. The production run maintained

a constant temperature of 310.15 K and pressure of

1 atm with weak coupling using a Langevin thermostat

and barostat. The time step used is 0.2 fs. Bond lengths

involving hydrogen atoms were constrained using the

SHAKE algorithm, and Particle Mesh Ewald (PME) was

employed for electrostatic interactions beyond a cutoff

distance.

Prediction of drug-likeness, pharmacokinetics,

and toxicity. To estimate the physicochemical proper-

ties, absorption, distribution, metabolism, excretion,

toxicity (ADMET) profile, pharmacodynamic and drug-

likeness parameters of all the eight xanthophyll ca-

rotenoids, the mCule server, SwissADME, AdmetSAR,

Molsoft, and the ProTox-II server were used as previ-

ously described [44].

RESULTS AND DISCUSSION

Molecular docking and interaction analyses.

The molecular docking studies showed the best-fitted

conformational binding pose of the tested xanthophyll

carotenoids (astaxanthin, canthaxanthin, echinenone,

fucoxanthin, lutein, neoxanthin, violaxanthin, and zea-

xanthin) within Keap1. The identified binding pocket

is located in the Kelch domain of Keap1, in the shared

binding region with Nrf2. This domain is a β-propeller

structure composed of six blades, each consisting of a

four-stranded antiparallel β-sheet connected by loops.

This structure serves as the binding pocket for the

Nrf2 ETGE and DLG motifs and is divided into six sub-

pockets (P1-P6) based on interactions observed in co-

crystal structures of Keap1 with these motifs. Subpock-

ets P1 and P2 are polar, containing residues such as

Ser363, Arg380, Asn382, Asn414, Arg415, Ile461, Gly462,

Phe478, Arg483, and Ser508. In contrast, P4 and P5 are

hydrophobic, with residues Tyr525, Gln530 in P4, and

Tyr334, Tyr572, Phe577 in P5. P3, located at the center

of the channel formed by small polar residues, consists

of Gly509, Ser555, Ala556, Gly571, Ser602, and Gly603.

Finally, P6 includes residues Asp389, Ser431, His432,

Gly433, Cys434, Ile435, and His436, which interact with

the DLGex peptide but not with the ETGE motif [11].

Moreover, Nrf2-ETGE-motif-containing peptides, when

co-crystallized with the Keap1-Kelch domain, adopt a

β-hairpin structure within the binding site. This site

includes residues Asp77, Glu78, Glu79, Thr80, Gly81,

and Glu82. This β-hairpin conformation is stabilized

by a network of intramolecular hydrogen bonds in-

volving the backbone atoms of residues Gln75, Asp77,

Asp79, Thr80, Glu82, Leu89, and the sidechain atoms

of residues Asp77 and Thr80 [33, 45].

MEDORO et al.1684

BIOCHEMISTRY (Moscow)

Table 1. Binding affinities calculation of astaxanthin,

canthaxanthin, echinenone, fucoxanthin, lutein, neox-

anthin, violaxanthin, and zeaxanthin

Compounds Binding Affinities (Kcal/mol)

Astaxanthin –9.0

Canthaxanthin –8.1

Echinenone –8.8

Fucoxanthin –9.5

Lutein –8.6

Neoxanthin –8.9

Violaxanthin –8.5

Zeaxanthin –7.1

The molecular docking assessments of the eight

xanthophyll carotenoids revealed variations in the re-

corded binding energies, with the docked complex ex-

hibiting the lowest binding energy being chosen from

all the docking complexes created for each compound,

as detailed in Table 1. The generated docked complex-

es were evaluated based on docking energy scores and

the number of hydrogen bonds and hydrophobic inter-

actions. Fucoxanthin and astaxanthin (Fig. 1) showed

the lowest binding affinity scores (–9.5 to –9.0 kcal/mol,

respectively), while those of canthaxanthin, echinen-

one, lutein, neoxanthin, violaxanthin, and zeaxanthin,

were lower than –7.0 kcal/mol (ranging from –7.1 to

–8.9 kcal/mol), although good values that predicted a

stable binding.

All intermolecular interactions, including hydro-

gen bonds, and hydrophobic and electrostatic inter-

actions, contribute to the stability of the docked com-

plex and lower binding affinity score. In particular,

hydrogen bonds and hydrophobic interactions play a

significant role in ligand-receptor stability. Notably, fu-

coxanthin and astaxanthin, the two compounds with

the lowest binding scores, were firmly bound, predom-

inantly in a planar manner, in the Kelch domains with

potential hydrogen bonding interactions, as shown

in Figs. 2-3 and Table 2. In detail, in the docking com-

plex between fucoxanthin-Keap1, Arg326, Arg415, and

Arg416 were the three residues involved in hydrogen

bonding. Cys368, Val370, Val418, Val420, and Val467

are involved in hydrophobic interactions, contributing

to the fucoxanthin stability in the binding pocket of

Keap1 (Fig. 3, Table 2). Several potential residues iden-

tified in this docking analysis are already reported as

crucial for Keap1–Nrf2 interaction and the PPI inhib-

itors, including Arg415, Gly509, Ala556, and Gly603,

suggesting, due to the peculiar structure of this mole-

cule, interaction with polar and hydrophobic subpock-

ets [11]. These identified key residues for the fucoxan-

thin–Keap1 complex also showed good overlap with

8 previously reported potential residues, including

Arg415, Val465, Val512, and Ala556 [22]. In this study,

Wu et al. reported that fucoxanthin, extracted from

seaweeds, exhibits a neuroprotective effect by the po-

tential inhibition of the Keap1–Nrf2 complex. Biolayer

interferometry (BLI) assays showed that fucoxanthin

binds to Keap1 with a potential dissociation constant

(Kd) of 51.6 µM, confirming its potential as Keap1-Nrf2

PPI and inducer of Nrf2 nuclear translocation and acti-

vation of the antioxidant response element (ARE). Sim-

ilar to fucoxanthin, it was observed that astaxanthin

also showed a strong binding interaction with residues

in the Kelch domains of Keap1, including key residues

such as Arg415, Gly509, Ala556, and Gly603 (Table 2),

with one hydrogen bond observed at Arg415 and three

hydrophobic interactions with Val467, Val512, and

Cys513 (Fig. 3). For the neoxanthin–Keap1 complex,

Fig. 1. 2D chemical structures of (a) fucoxanthin, and (b) astaxanthin.

CAROTENOIDS AS KEAP1-NRF2 PPI INHIBITORS 1685

BIOCHEMISTRY (Moscow)

Table 2. Molecular interactions of astaxanthin and fucoxanthin with Keap1

Docked complexes Interactive residues of docking studies

Astaxanthin–Keap1

Gly364, Leu365, Ala366, Gly367, Cy368, Arg415, Ile416, Gly417, Val418, Gly419, Val420,

Asp422, Gly423, Gly462, Val463, Gly464, Val465, Ala466, Val467, Gly509, Ala510, Gly511,

Val512, Cys513, Leu557, Ile559, Gly603, Val604

Fucoxanthin–Keap1

Gly364, Leu365, Ala366, Gly367, Cys368, Val369, Val370, Gly371, Gly372, Arg415, Ile416,

Gly417, Val418, Gly419, Val420, Ile421, Asp422, Gly423, Val463, Gly464, Val467, Gly509,

Ala510, Gly511, Ala556, Leu557, Gly558, Ile559, Gly603, Val604, Gly605

Note. Bold residues indicate residues that are involved in potential hydrogen bonding.

a unique hydrogen bond with the residue Ile416 was

identified. No hydrogen bonding interactions were

reported in the docking complexes of canthaxanthin–

Keap1, echinenone–Keap1, lutein–Keap1, violaxanthin–

Keap1, and zeaxanthin–Keap1, although in the dock-

ing complex of echinenone–Keap1 some hydrophobic

interactions were detected (Fig. S1 in the Online Re-

source 1).

Moreover, studied carotenoids may bind many

Keap1 residues shared with Nrf2, including Arg415,

Gly509, Ala556, and Gly603, which are all residues lo-

cated in P1-P3 subpockets. Overall, these data may of-

fer further interpretation of the antioxidant activity

of xanthophyll carotenoids. Carotenoids are known to

be efficient antioxidants that directly scavenge singlet

molecular oxygen and peroxyl radicals, contributing

to the antioxidant defense system [46]. However, there

is multiple experimental evidence that xanthophyll

carotenoids, particularly fucoxanthin, and astaxan-

thin, may inhibit Nrf2 degradation and increase Nrf2-

dependent activity [20, 47-49]. Overall, our findings

provide evidence for a new potential mechanism by

which carotenoids may exert their antioxidant activi-

ty through the Nrf2 pathway. Among the carotenoids

analyzed in this study, astaxanthin and fucoxanthin

demonstrated potentially higher binding stability

with Keap1, as indicated by their binding energies

and the greater number of identified hydrogen bonds

and hydrophobic interactions.

MDS studies. To gain further insights into the

study of the interaction of the two most potential

xanthophyll carotenoids, fucoxanthin and astaxan-

thin, with Keap1, MDS studies were performed. MDS

is a method for analyzing the physical movements of

atoms and molecules, providing information on con-

formational changes in docking complexes. As ex-

pected, some differences in docking interactions of

protein- ligand complexes between AutoDock Vina and

MDS analyses were detected, considering that MDS

offers a dynamic view of the protein-ligand complex,

Fig. 2. 3D molecular representation of fucoxanthin (green) and astaxanthin (cyan) docking complexes with Keap1.

MEDORO et al.1686

BIOCHEMISTRY (Moscow)

Fig. 3. 2D molecular interaction analyses of (a)fucoxanthin, and (b)astaxanthin with Keap1.

aswell as revealed comprehensive and more realistic

molecular interactions during the simulation system

(Fig. S2 in the Online Resource 1). Overall, fucoxan-

thin and astaxanthin showed significant potential in

the binding of the Kelch domains of Keap1, generat-

ing a very stable and tight-docked complex, as shown

by the data below.

Root Mean Square Deviation (RMSD) is a measure

used to quantify the structural analysis of the docking

complex. In Fig. 4 RMSD values indicated the average

deviation for both ligands (astaxanthin and fucoxan-

thin) and protein. Both fucoxanthin and astaxanthin

simulation complexes with Keap1 showed that the

structure remained stable throughout the simulation

time with some fluctuation within the range of ~1 Å,

which is an expected behavior. Therefore, the RMSD

values (1.54 ± 0.08 Å in complex with astaxanthin and

1.67 ± 0.14 Å with fucoxanthin) indicate that the Keap1

underwent small local conformational changes with

apparently no change in its folding and stability

(Fig. 4a). Astaxanthin and fucoxanthin showed high

RMSD values (2.44 ± 0.47 Å and 3.39 ± 0.35 Å, respec-

tively), suggesting conformational changes. Fucoxan-

thin exhibited higher variation in ligand conformation,

as indicated by its higher RMSD value, and suggests at

least two different clusters for ligand conformations.

The ligand’s RMSD plot immediately reached a pla-

teau, indicating stability in conformation. Astaxanthin

at least exhibited two ensembles of ligands during sim-

ulation, which transformed into each other back and

forth (Fig.4b).

In addition, Root Mean Square Fluctuation (RMSF)

was used to estimate the binding orientations of the

ligands and structural alternations of the residues

that linked to targeted protein at a specific tempera-

ture and pressure. Fluctuations in the protein residues

CAROTENOIDS AS KEAP1-NRF2 PPI INHIBITORS 1687

BIOCHEMISTRY (Moscow)

Fig. 4. a, b) RMSD detailed calculation analyses of astaxanthin (black line), and fucoxanthin (red line) complexes with Keap1

during 120ns MDS.

ofKeap1 and both changes in the ligand atom position

of astaxanthin and fucoxanthin were calculated from

the 120 ns trajectory. These RMSF studies endorsed

that both complexes in our simulations exhibited

≤2.0 Å protein RMSF values indicating firmly bound li-

gands with great protein stability and rigidity in the

structures (Fig.S3A in the Online Resource1).

Both complexes showed a very stable radius of

gyration with a fluctuation of 1 Å, which indicates the

stability and the compactness of the 3D protein struc-

ture. The slight fluctuation within the 1 Å radius-of-

gyration value during the MDS time indicated a slight

opening and closing of the N- and C-terminal domains

(Fig. S3B in the Online Resource1). Notably, these were

very small fluctuations in the normal range of radi-

us-of-gyration, which indicates that the protein–ligand

system remains compact during molecular dynamics

simulation and that no unfolding event is observed.

In detail, the radius of gyration of astaxanthin and fu-

coxanthin indicated no change in the protein compact-

ness, remaining highly compact with consistent values

and smaller standard deviations (18.09 ± 0.07 Å and

18.06 ± 0.06 Å, respectively).

To deepen the molecular interactions between

Keap1 active site residues and astaxanthin and fucox-

anthin, the free energy of the docked complexes was

calculated using the MM/GBSA method. Molecular in-

teractions between Keap1 binding sites residues and

astaxanthin, and fucoxanthin after every 1 ns (120

frames in total) portrayed highly favorable electrostatic

energies, and polar solvation-free energies, and showed

that van der Waals forces were highly favorable inter-

actions contributing to the stability of the molecular

docking systems (Fig. 5). The overall least binding en-

ergies obtained for astaxanthin (–46.84 ± 7.39 kcal/mol)

and fucoxanthin (–52.66 ± 4.77 kcal/mol) docking com-

Fig. 5. MM/GBSA protein–ligand binding energy calculated for astaxanthin (black line), and fucoxanthin (red line) after every

1ns (120 frames in total).

MEDORO et al.1688

BIOCHEMISTRY (Moscow)

Table 3. Drug-likeness analyses of astaxanthin, canthaxanthin, echinenone, fucoxanthin, lutein, neoxanthin,

violaxanthin, and zeaxanthin

Compounds RO5 Veber Rule Drug-likeness Bioavailability

Astaxanthin no yes –0.13 0.17

Canthaxanthin no yes 0.08 0.17

Echinenone no yes 0.19 0.17

Fucoxanthin no no –0.35 0.17

Lutein no yes –0.33 0.17

Neoxanthin no yes –0.78 0.17

Violaxanthin no yes –1.49 0.17

Zeaxanthin no yes –0.18 0.17

plexes represented binding flexibility. Although these

findings showed that both ligands remained in bound

form throughout the simulation time with very good

binding energy, fucoxanthin may bind with slightly

more favorable energy to Keap1 with respect to astax-

anthin. Moreover, ligand-protein hydrogen bonds play

a significant role in determining the binding effica-

cy of a ligand on the binding pockets of the protein.

Thetotal number of hydrogen bonds formed between

ligand and protein is shown in Fig.S4 in the Online Re-

source 1. Astaxanthin and fucoxanthin, during 120 ns

simulation time, maintained almost two hydrogen

bonds formed at the interface of protein at most points

of the simulation time.

Drug-likeness, pharmacokinetics, and toxicity

analyses. All compounds investigated in this study

were subjected to drug-likeness, pharmacodynamic,

and toxicity profiling to assess their potential safety,

with overall very good results for all the xanthophylls

tested. Table S1 in the Online Resource 1 shows the

cheminformatics analysis of the compounds. The mo-

lecular weight (g/mol) of all compounds was also com-

parable with the standard value (<5000 g/mol). The

SwissADME analyses presented in Table 3 show that

all the analyzed carotenoids have more than one viola-

tion in RO5, the five rule-based filters allow prediction

of whether a molecule is defined as an orally active

drug. However, as shown in Table 3 and Table S1 in

the Online Resource 1, these carotenoids, except fu-

coxanthin, met the requirements of drug-likeness of

Veber filters. Moreover, also drug-likeness scores of

all the compounds calculated using MolSoft and syn-

thetic bioavailability scores calculated by SwissADME

were in contrast to the RO5 results. Fucoxanthin and

astaxanthin displayed good drug-likeness scores (–0.35

and –0.13) and bioavailability scores within the range

of >10%. All the drug-likeness scores of the other

carotenoids were comprised between 0.08 and –1.49

(Table3 and TableS1 in the Online Resource1).

All compounds were predicted to be absorbed by

the human intestine and displayed promising results

for the blood brain barrier (BBB) permeability. More-

over, non-toxic and non-carcinogenic behaviors were

observed for all the xanthophyll carotenoids. Accord-

ing to the predicted results, overall, these compounds

showed lower toxicity (hepatotoxicity, immunotoxic-

ity, mutagenicity, and cytotoxicity), and carcinogenic-

ity. The overall predicted results justified their good

ADMET behavior (Table S2 in the Online Resource 1).

According to the predicted ADMET analysis, some of

the compounds had a very high lethal dose50(LD

such as astaxanthin (4600 mg/kg), canthaxanthin

(10,000 mg/kg), and echinenone (10,000 mg/kg), which

were classified as non-toxic for acute oral toxicity.

On the other side, because of their lower LD

, fu-

coxanthin (130 mg/kg), lutein (10 mg/kg), neoxanthin

(50 mg/kg), violaxanthin (55 mg/kg), and zeaxanthin

(not available) were classified as harmful if swallowed

(50 < LD

 ≤ 300) or fatal if swallowed (5 < LD

 ≤ 50).

In the toxicity analyses, astaxanthin confirmed its

non-toxic potential, showing no hepatotoxicity, car-

cinogenicity, mutagenicity, or cytotoxicity. On the oth-

er hand, fucoxanthin showed carcinogenicity and im-

munotoxicity potential with more probability.

In addition, all compounds were screened with

ProTox-II to predict whether they could interfere with

various biological pathways. The results revealed that

fucoxanthin and astaxanthin were inactive against all

the targeted biological pathways. In detail, astaxan-

thin, echinenone, fucoxanthin, neoxanthin, and vio-

laxanthin showed no potential activity against nucle-

ar receptor pathways and stress response pathways,

including aryl hydrocarbon receptor (Ahr), andro-

gen receptor (AR), androgen receptor ligand binding

CAROTENOIDS AS KEAP1-NRF2 PPI INHIBITORS 1689

BIOCHEMISTRY (Moscow)

domain (AR-LBD), estrogen receptor alpha (ER), Estro-

gen receptor ligand binding domain (ER-LBD), peroxi-

some proliferator-activated receptor gamma (PPAR-γ),

and p53. Canthaxanthin could interfere with ER and

ER-LBD, and lutein and zeaxanthin with p53 (TableS3

in the Online Resource1).

CONCLUSIONS

Keap1, a highly redox-sensitive member of the

BTB-Kelch substrate adaptor protein family, is known

to mediate the ubiquitination of Nrf2, a master regu-

lator of the antioxidant response. This study employs a

computational approach to explore the potential of ox-

ygenated carotenoids, known as xanthophylls, as can-

didates for inhibiting the Keap1-Nrf2 PPI. Recently, in-

hibitors of the Keap1-Nrf2 PPI have gained increasing

attention as a means to selectively target Nrf2 without

off-target effects, leading to the development of novel

classes of potential preventive and therapeutic agents

for a variety of diseases characterized by oxidative

stress. Although the reliability of in silico predictions

greatly depends on the accuracy of molecular dock-

ing simulations and scoring functions, computational

techniques remain crucial for accelerating the screen-

ing of Keap1-Nrf2 PPI inhibitors and diversifying their

chemical scaffolds. In this study, three-dimensional

structural information, molecular docking screening,

and interaction analysis revealed the structural deter-

minants underpinning the binding process between

carotenoid xanthophylls, especially fucoxanthin and

astaxanthin, and Keap1. These analyses led to the iden-

tification of critical amino acid residues in the binding

pocket of Keap1 which are involved in the interaction

with carotenoid xanthophylls. However, fucoxanthin

and astaxanthin exhibited a stronger binding interac-

tion with residues in the Kelch domains of Keap1 than

other carotenoid xanthophylls. We then performed

MDS studies to elucidate the dynamic behavior and

stability of the interactions between fucoxanthin, as-

taxanthin, and Keap1. Through a 120 ns simulation

trajectory, the results offered insights into the tempo-

ral evolution of protein–ligand complexes, deciphering

the forces underlying their associations. Notably, fu-

coxanthin and astaxanthin formed a highly stable and

tightly docked complex throughout the simulation pe-

riod, establishing multiple hydrogen bonds with Keap1

and characterized by very good binding energy. More-

over, favorable pharmacokinetic and safety profiles

have demonstrated the drug-like properties of these

compounds, enhancing their potential as Keap1-Nrf2

PPI inhibitors. Overall, this study provides theoretical

support that could stimulate further research into an-

tioxidant strategies involving carotenoid xanthophylls,

particularly focusing on their potential to inhibit the

Keap1-Nrf2 PPI in experimental models of oxidative

stress.

Supplementary information. The online version

contains supplementary material available at https://

doi.org/10.1134/S0006297924100031.

Contributions. Conceptualization, A.M., S.D., and

G.S.; methodology, A.M. and S.D.; formal analysis, A.M.

and T.H.J.; writing – original draft preparation, A.M.,

T.H.J., and F.S.; writing– review and editing, G.S., L.S.,

and S.D.

Funding. This work was supported by ongoing in-

stitutional funding. No additional grants to carry out

or direct this particular research were obtained.

Ethics declarations. This work does not contain

any studies involving human and animal subjects.

Theauthors of this work declare that they have nocon-

flicts of interest.

REFERENCES

1. Go, Y.M., and Jones, D.P. (2013) The redox proteome,

J. Biol. Chem., 288, 26512, https://doi.org/10.1074/

jbc.r113.464131.

2. Brigelius-Flohé, R., and Flohé, L. (2011) Basic prin-

ciples and emerging concepts in the redox control

of transcription factors, Antioxid. Redox Signal., 15,

2335-2381, https://doi.org/10.1089/ars.2010.3534.

3. Hybertson, B.M., Gao,B., Bose, S.K., and McCord, J.M.

(2011) Oxidative stress in health and disease: the ther-

apeutic potential of Nrf2 activation, Mol. Aspects Med.,

32, 234-246, https://doi.org/10.1016/j.mam.2011.10.006.

4. Medoro,A., Saso,L., Scapagnini,G., and Davinelli,S.

(2023) NRF2 signaling pathway and telomere length

in aging and age-related diseases, Mol. Cell. Biochem.,

https://doi.org/10.1007/s11010-023-04878-x.

5. Malhotra, D., Portales-Casamar, E., Singh, A., Sri-

vastava, S., Arenillas, D., Happel, C., Shyr, C., Wak-

abayashi, N., Kensler, T. W., Wasserman, W. W., and

Biswal, S. (2010) Global mapping of binding sites

for Nrf2 identifies novel targets in cell survival re-

sponse through chip-seq profiling and network analy-

sis, Nucleic Acids Res., 38, 5718-5734, https://doi.org/

10.1093/nar/gkq212.

6. Baird, L., and Yamamoto, M. (2020) The molecular

mechanisms regulating the KEAP1-NRF2 pathway,

Mol. Cell. Biol., 40, e00099-20, https://doi.org/10.1128/

mcb.00099-20.

7. Canning, P., Sorrell, F. J., and Bullock, A. N. (2015)

Structural basis of Keap1 interactions with Nrf2, Free

Radic. Biol. Med., 88, 101-107, https://doi.org/10.1016/

j.freeradbiomed.2015.05.034.

8. Davinelli,S., Medoro,A., Intrieri,M., Saso,L., Scapag-

nini,G., and Kang, J.X. (2022) Targeting NRF2-KEAP1

axis by Omega-3 fatty acids and their derivatives:

MEDORO et al.1690

BIOCHEMISTRY (Moscow)

emerging opportunities against aging and diseases,

Free Radic. Biol. Med., 193, 736-750, https://doi.org/

10.1016/j.freeradbiomed.2022.11.017.

9. Dayalan Naidu,S., and Dinkova-Kostova, A.T. (2020)

KEAP1, a cysteine-based sensor and a drug target for

the prevention and treatment of chronic disease, Open

Biol., 10, 200105, https://doi.org/10.1098/rsob.200105.

10. Lee,S., and Hu,L. (2020) Nrf2 activation through the

inhibition of Keap1-Nrf2 protein-protein interaction,

Med. Chem. Res., 29, 846-867, https://doi.org/10.1007/

s00044-020-02539-y.

11. Crisman, E., Duarte, P., Dauden, E., Cuadrado, A.,

Rodríguez-Franco, M. I., López, M. G., and León, R.

(2023) KEAP1-NRF2 protein-protein interaction inhib-

itors: design, pharmacological properties and ther-

apeutic potential, Med. Res. Rev., 43, 237-287, https://

doi.org/10.1002/med.21925.

12. Pallesen, J.S., Tran, K.T., and Bach,A. (2018) Non-co-

valent small-molecule Kelch-like ECH-associated

protein 1-nuclear factor erythroid 2-related Factor2

(Keap1-Nrf2) Inhibitors and their potential for target-

ing central nervous system diseases, J. Med. Chem.,

61, 8088-8103, https://doi.org/10.1021/acs.jmedchem.

8b00358.

13. Leung, C.H., Zhang, J.T., Yang, G.J., Liu,H., Han, Q.B.,

and Ma, D.L. (2019) Emerging screening approaches

in the development of Nrf2-Keap1 protein-protein in-

teraction inhibitors, Int.J. Mol. Sci., 20, 4445, https://

doi.org/10.3390/ijms20184445.

14. Davinelli, S., Ali, S., Solfrizzi, V., Scapagnini, G., and

Corbi, G. (2021) Carotenoids and cognitive out-

comes: a meta-analysis of randomized intervention

trials, Antioxidants, 10, 223, https://doi.org/10.3390/

antiox10020223.

15. Saini, R. K., Keum, Y. S., Daglia, M., and Rengasamy,

K. R. (2020) Dietary carotenoids in cancer chemo-

prevention and chemotherapy: a review of emerg-

ing evidence, Pharmacol. Res., 157, 104830, https://

doi.org/10.1016/j.phrs.2020.104830.

16. Medoro,A., Intrieri,M., Passarella,D., Willcox, D.C.,

Davinelli, S., and Scapagnini, G. (2024) Astaxanthin

as a metabolic regulator of glucose and lipid homeo-

stasis, J. Funct. Foods, 112, 105937, https://doi.org/

10.1016/j.jff.2023.105937.

17. Thomas, S.E., and Johnson, E.J. (2018) Xanthophylls,

Adv. Nutr., 9, 160, https://doi.org/10.1093/advances/

nmx005.

18. Pereira, A. G., Otero, P., Echave, J., Carreira-Casa-

is, A., Chamorro, F., Collazo, N., Jaboui, A., Lou-

renço-Lopes, C., Simal-Gandara, J., and Prieto, M. A.

(2021) Xanthophylls from the sea: algae as source

of bioactive carotenoids, Mar. Drugs, 19, 188, https://

doi.org/10.3390/md19040188.

19. Davinelli, S., Saso, L., D’angeli, F., Calabrese, V., In-

trieri, M., and Scapagnini, G. (2022) Astaxanthin

as a modulator of Nrf2, NF-κB, and their crosstalk:

molecular mechanisms and possible clinical appli-

cations, Molecules, 27, 502, https://doi.org/10.3390/

molecules27020502.

20. Zou,X., Gao,J., Zheng,Y., Wang,X., Chen,C., Cao,K.,

Xu,J., Li,Y., Lu,W., Liu,J., and Feng,Z. (2014) Zeaxan-

thin induces Nrf2-mediated phase II enzymes in pro-

tection of cell death, Cell Death Dis., 55, e1218, https://

doi.org/10.1038/cddis.2014.190.

21. Chang, J., Zhang, Y., Li, Y., Lu, K., Shen, Y., Guo, Y.,

Qi,Q., Wang,M., and Zhang,S. (2018) NrF2/ARE and

NF-κB pathway regulation may be the mechanism

for lutein inhibition of human breast cancer cell,

Futur. Oncol., 14, 719-726, https://doi.org/10.2217/

fon-2017-0584.

22. Wu, W., Han, H., Liu, J., Tang, M., Wu, X., Cao, X.,

Zhao,T., Lu,Y., Niu, T., Chen, J., and Chen, H. (2021)

Fucoxanthin prevents 6-OHDA-induced neurotox-

icity by targeting Keap1, Oxid. Med. Cell. Longev.,

2021, 6688708, https://doi.org/10.1155/2021/6688708.

23. Ahmed, F., Fanning, K., Netzel, M., and Schenk,

P. M. (2015) Induced carotenoid accumulation in

Dunaliella salina and Tetraselmis suecica by plant

hormones and UV-C radiation, Appl. Microbiol. Bio-

technol., 99, 9407-9416, https://doi.org/10.1007/s00253-

015-6792-x.

24. Pasquet,V., Morisset,P., Ihammouine,S., Chepied,A.,

Aumailley, L., Berard, J. B., Serive, B., Kaas, R., Lan-

neluc, I., Thiery, V., Lafferriere, M., Piot, J. M.,

Patrice,T., Cadoret, J.P., and Picot,L. (2011) Antiprolif-

erative activity of violaxanthin isolated from bioguid-

ed fractionation of Dunaliella tertiolecta extracts, Mar.

Drugs, 9, 819-831, https://doi.org/10.3390/md9050819.

25. Abe, K., Hattori, H., and Hirano, M. (2007) Accumu-

lation and antioxidant activity of secondary carot-

enoids in the aerial microalga Coelastrella striolata

var. multistriata, Food Chem., 100, 656-661, https://

doi.org/10.1016/j.foodchem.2005.10.026.

26. Xia,S., Wang,K., Wan,L., Li,A., Hu,Q., and Zhang,C.

(2013) Production, characterization, and antioxi-

dant activity of fucoxanthin from the marine diatom

Odontella aurita, Mar. Drugs, 11, 2667-2681, https://

doi.org/10.3390/md11072667.

27. Přibyl,P., Pilný,J., Cepák, V., and Kaštánek,P. (2016)

The role of light and nitrogen in growth and carot-

enoid accumulation in Scenedesmus sp., Algal Res.,

16, 69-75, https://doi.org/10.1016/j.algal.2016.02.028.

28. Neumann, U., Derwenskus, F., Flister, V. F., Schmid-

Staiger,U., Hirth,T., and Bischoff, S.C. (2019) Fucox-

anthin, A carotenoid derived from phaeodactylum

tricornutum exerts antiproliferative and antioxidant

activities in vitro, Antioxidants, 8, 183, https://doi.

org/10.3390/antiox8060183.

29. Medoro, A., Davinelli, S., Milella, L., Willcox, B. J.,

Allsopp, R. C., Scapagnini, G., and Willcox, D. C.

(2023) Dietary astaxanthin: a promising antioxi-

dant and anti-inflammatory agent for brain aging

CAROTENOIDS AS KEAP1-NRF2 PPI INHIBITORS 1691

BIOCHEMISTRY (Moscow)

and adult neurogenesis, Mar. Drugs, 21, 643, https://

doi.org/10.3390/md21120643.

30. Terao,J. (2023) Revisiting carotenoids as dietary an-

tioxidants for human health and disease prevention,

Food Funct., 14, 7799-7824, https://doi.org/10.1039/

d3fo02330c.

31. Srivastava, R. (2021) Physicochemical, antioxidant

properties of carotenoids and its optoelectronic and

interaction studies with chlorophyll pigments, Sci. Rep.,

11, 18365, https://doi.org/10.1038/s41598-021-97747-w.

32. Miller, N.J., Sampson,J., Candeias, L.P., Bramley, P.M.,

and Rice-Evans, C.A. (1996) Antioxidant activities of

carotenes and xanthophylls, FEBS Lett., 384, 240-242,

https://doi.org/10.1016/0014-5793(96)00323-7.

33. Lo, S. C., Li, X., Henzl, M. T., Beamer, L. J., and Han-

nink, M. (2006) Structure of the Keap1:Nrf2 inter-

face provides mechanistic insight into Nrf2 signal-

ing, EMBO J., 25, 3605-3617, https://doi.org/10.1038/

sj.emboj.7601243.

34. Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch,

G.S., Greenblatt, D.M., Meng, E.C., and Ferrin, T. E.

(2004) UCSF Chimera – a visualization system for ex-

ploratory research and analysis, J.Comput. Chem., 25,

1605-1612, https://doi.org/10.1002/jcc.20084.

35. Huang, Z., Lin, Z., and Song, C. (2007) Protonation

processes and electronic spectra of histidine and re-

lated ions, J. Phys. Chem. A, 111, 4340-4352, https://

doi.org/10.1021/jp067280a.

36. Uranga, J., Mikulskis, P., Genheden, S., and Ryde, U.

(2012) Can the protonation state of histidine resi-

dues be determined from molecular dynamics sim-

ulations? Comput. Theor. Chem., 1000, 75-84, https://

doi.org/10.1016/j.comptc.2012.09.025.

37. Ramachandran,S., Kota,P., Ding,F., and Dokholyan,

N.V. (2011) Automated minimization of steric clash-

es in protein structures, Proteins, 79, 261-270, https://

doi.org/10.1002/prot.22879.

38. Kim, S., Thiessen, P. A., Bolton, E. E., Chen, J., Fu,G.,

Gindulyte,A., Han,L., He,J., He,S., Shoemaker, B.A.,

Wang, J., Yu, B., Zhang, J., and Bryant, S. H. (2016)

PubChem substance and compound databases, Nucle-

ic Acids Res., 44, D1202-D1213, https://doi.org/10.1093/

nar/gkv951.

39. Trott, O., and Olson, A. J. (2010) AutoDock Vina: im-

proving the speed and accuracy of docking with a

new scoring function, efficient optimization and

multithreading, J. Comput. Chem., 31, 455, https://

doi.org/10.1002/jcc.21334.

40. Phillips, J. C., Braun,R., Wang,W., Gumbart,J., Tajk-

horshid, E., Villa, E., Chipot, C., Skeel, R.D., Kalé,L.,

and Schulten,K. (2005) Scalable molecular dynamics

with NAMD, J. Comput. Chem., 26, 1781-1802, https://

doi.org/10.1002/jcc.20289.

41. Wang, J., Wolf, R. M., Caldwell, J. W., Kollman, P. A.,

and Case, D. A. (2004) Development and testing of

a general amber force field, J. Comput. Chem., 25,

1157-1174, https://doi.org/10.1002/jcc.20035.

42. Maier, J.A., Martinez,C., Kasavajhala,K., Wickstrom,L.,

Hauser, K. E., and Simmerling, C. (2015) ff14SB: im-

proving the accuracy of protein side chain and back-

bone parameters from ff99SB, J.Chem. Theory Comput.,

11, 3696-3713, https://doi.org/10.1021/acs.jctc.5b00255.

43. Yu,Y., Xu,S., He,R., and Liang,G. (2023) Application

of molecular simulation methods in food science: sta-

tus and prospects, J.Agric. Food Chem., 71, 2684-2703,

https://doi.org/10.1021/acs.jafc.2c06789.

44. Medoro, A., Jafar, T. H., Ali, S., Trung, T. T., Sorren-

ti, V., Intrieri, M., Scapagnini, G., and Davinelli, S.

(2023) In silico evaluation of geroprotective phyto-

chemicals as potential sirtuin 1 interactors, Biomed.

Pharmacother., 161, 114425, https://doi.org/10.1016/

j.biopha.2023.114425.

45. Padmanabhan,B., Tong, K.I., Ohta,T., Nakamura,Y.,

Scharlock,M., Ohtsuji,M., Kang, M.I., Kobayashi,A.,

Yokoyama, S., and Yamamoto, M. (2006) Structural

basis for defects of Keap1 activity provoked by its

point mutations in lung cancer, Mol. Cell, 21, 689-700,

https://doi.org/10.1016/j.molcel.2006.01.013.

46. Stahl, W., and Sies, H. (2023) Antioxidant activity of

carotenoids, Mol. Aspects Med., 24, 345-351, https://

doi.org/10.1016/s0098-2997(03)00030-x.

47. Roselli, S., Pundavela, J., Demont, Y., Faulkner, S.,

Keene, S., Attia, J., Jiang, C. C., Zhang, X. D., Walker,

M.M., and Hondermarck,H. (2015) Sortilin is associat-

ed with breast cancer aggressiveness and contributes

to tumor cell adhesion and invasion, Oncotarget, 6,

10473-10486, https://doi.org/10.18632/oncotarget.3401.

48. Frede, K., Ebert, F., Kipp, A. P., Schwerdtle, T., and

Baldermann,S. (2017) Lutein activates the transcrip-

tion factor Nrf2 in human retinal pigment epithe-

lial cells, J. Agric. Food Chem., 65, 5944-5952, https://

doi.org/10.1021/acs.jafc.7b01929.

49. Li, Z., Dong, X., Liu, H., Chen, X., Shi, H., Fan, Y.,

Hou, D., and Zhang, X. (2013) Astaxanthin protects

ARPE-19 cells from oxidative stress via upregulation

of Nrf2-regulated phaseII enzymes through activation

of PI3K/Akt, Mol. Vis., 19, 1656.

Publisher’s Note. Pleiades Publishing remains

neutral with regard to jurisdictional claims in published

maps and institutional affiliations. AI tools may have

been used in the translation or editing of this article.