Integrative Network Pharmacology and ADMET Modeling Reveal the Multitarget Therapeutic Potential of Geraniol

Abstract

Background: Geraniol is an acyclic monoterpene widely distributed in the essential oils of aromatic species such as Cymbopogon citratus, Pelargonium graveolens, and Rosa damascena, It is known for its antioxidant, anti-inflammatory, neuroprotective, and antitumor activities. Methods: This study aimed to investigate, through network pharmacology and computational ADMET modeling, the molecular mechanisms and pharmacological potential of geraniol, integrating drug-likeness parameters, toxicity prediction, and multitarget interactions. Results: A total of 25 core targets were identified, mainly involved in inflammation, oxidative stress, apoptosis, and transcriptional regulation. Geraniol exhibited a favorable drug-likeness profile, high predicted intestinal absorption, and low systemic toxicity, supporting its pharmaceutical applicability. Mechanistically, it modulates the Nrf2/HO-1 ↔ NF-κB axis, reducing reactive oxygen species, pro-inflammatory cytokines (TNF-α, IL-1β, IL-6), and apoptotic markers (caspases, Bax), while enhancing antioxidant enzymes (SOD, CAT, GPx) and antiapoptotic proteins (Bcl-2). Conclusions: These findings confirm its multitarget and pleiotropic nature, highlighting its potential as a therapeutic candidate for inflammatory, metabolic, and neurodegenerative disorders. Furthermore, this study provides a robust mechanistic rationale for future in vitro and in vivo validation, as well as for the design of nanostructured formulations to improve geraniol’s bioavailability and therapeutic safety.

1. Introduction

Geraniol (3,7-dimethyl-2,6-octadien-1-ol) is a natural acyclic monoterpene widely distributed in essential oils of several plant species, including geranium, rose, citronella, and palmarosa [1]. This compound is found mainly in the flowers, leaves, bark, and stems of aromatic plants and is responsible for characteristic aromatic notes used in perfumery, cosmetics, and phytotherapy [2]. Plants with the highest concentrations of geraniol belong to families such as Poaceae (lemongrass Cymbopogon citratus, palmarosa Cymbopogon martinii), Rosaceae (Rosa cinnamomea and other rose species), Geraniaceae (Pelargonium graveolens, popularly known as geranium), and Lamiaceae (lavender, mint). In lemongrass, for example, geraniol can make up to 80% of volatile oils [3,4,5].

These plants are native to and cultivated in tropical and temperate regions and are cultivated extensively in Asia (India, China, Indonesia), Africa (Egypt, Madagascar), Central America, and also in Brazil, which has great potential for exploiting native aromatic species cultivated in biomes such as the Cerrado and Atlantic Forest [6]. In Brazil, species such as lemongrass (C. citratus) and citronella (Cymbopogon winterianus) are widely cultivated and stand out for their high yield of geraniol-rich essential oil, representing strategic sources for both the cosmetic and pharmaceutical industries [3,7].

Geraniol plays a prominent role among monoterpenes, functioning as a fundamental bioactive constituent of several aromatic plants used in traditional medicine. The diversity of sources reinforces the importance of understanding its pharmacological actions, since the presence of geraniol in different plant species is directly associated with its therapeutic activities [8]. As a bioactive component, geraniol demonstrates broad pharmacological activities, including antioxidant, anti-inflammatory, antimicrobial, hepatoprotective, and antitumor properties [7,8,9].

According to the literature, geraniol participates in a signaling pathway involving the redox–inflammation–apoptosis coupling [10,11]. By reducing ROS and activating Nrf2/HO-1, the compound simultaneously deactivates NF-κB and lowers cytokines such as TNF-α, IL-1β, and IL-6; this same axis decreases MPO (Myeloperoxidase)/NO, lipid peroxidation (MDA—Malondialdehyde), and caspases, while increasing SOD (Superoxide Dismutase), CAT (Catalase), GPx (Glutathione Peroxidase), and Bcl-2 (with Bax reduction), which explains why the antioxidant, anti-inflammatory, and antiapoptotic effects appear together [12,13,14].

The most consistent evidence for the central role of geraniol in the Nrf2/HO-1—NF-κB axis comes from hepatic and renal models of oxidative damage. In hepatic ischemia–reperfusion, geraniol activates Nrf2 and induces HO-1, reducing histological damage and apoptosis, which is associated with a decrease in MDA (lipid peroxidation) and an increase in SOD, CAT, and GPx activities. Simultaneously, there is attenuation of TNF-α and COX-2 and a reduction in cleaved caspase-3, evidencing the antioxidant–anti-inflammatory–antiapoptotic connection [15,16]. Similar findings were observed in hepatotoxicity by CCl₄, in which geraniol reversed the elevation of inflammatory cytokines TNF-α, IL-1β, IL-6, MDA and MPO/NO, restored SOD/CAT/GPx, decreased caspases and normalized Bcl-2/Bax, confirming that the blockade of oxidative stress has repercussions on the inhibition of NF-κB and the containment of apoptosis [7].

The Nrf2–NF-κB crosstalk is well established in the literature and is associated with antioxidant, anti-inflammatory, and antiapoptotic effects [17,18,19]. Nrf2 activation induces phase II genes (HO-1, NQO1, GCLC/GCLM), reducing ROS and, consequently, the pro-inflammatory signal that stabilizes NF-κB. Conversely, HO-1 and its products can directly interfere with inflammatory transduction, while NF-κB knockdown decreases caspases and Bax, preserving Bcl-2 [18,20].

In the central nervous system, the same coupling supports neuroprotection. In models of stress/neural injury, geraniol reduces ROS, decreases IL-6/IL-8/IL-1β, corrects antioxidant imbalance, and attenuates neuroinflammation. This translates into normalization of oxidative and neurochemical markers, improvement in the electroencephalogram and anxiogenic behaviors, and in epilepsy, there is a reduction in oxidative stress/neuroinflammation and activation of the GABAergic pathway. These findings demonstrate that geraniol’s antioxidant effect is the initial event that disarms inflammation and limits cell death, resulting in functional protection [21,22]. Furthermore, recent evidence demonstrates that geraniol exhibits orofacial antinociceptive activity mediated by modulation of TRPV1 channels, reinforcing its role in the ionic mechanisms of pain [23]. These effects reflect a mechanism of suppression of the NF-κB/NLRP3 axis, through which geraniol attenuates neuroglial inflammation and reduces nitric oxide (NO) production. The compound improves synaptic plasticity and memory in experimental models of Alzheimer’s disease, while reducing oxidative and neuroinflammatory markers in the hippocampus, highlighting the translational relevance of its primary antioxidant effect [24].

Geraniol modulates the nitrergic pathway, reducing excessive NO production and attenuating oxidative stress in a pentylenetetrazol-induced seizure model [25]. The study demonstrated increased seizure latency and recovery of behavioral parameters, supporting the compound’s multifactorial role in redox homeostasis and inhibitory neurotransmission. In a D-galactose-induced brain aging model, Rajendran et al. 2024 demonstrated that geraniol restores systemic and cerebral antioxidant parameters, reduces the expression of inflammatory cytokines, and improves cognitive performance, confirming its neuroprotective action in conditions associated with neuronal senescence and chronic oxidative stress [26].

The growing demand for more effective therapeutic approaches with fewer adverse effects has directed pharmacological research toward natural compounds with multitarget potential, such as monoterpenes. The traditional paradigm of one drug, one target has gradually been replaced by the understanding that many complex diseases require multitarget interventions to achieve optimal therapeutic efficacy [27]. Network pharmacology has emerged as a powerful tool for understanding the mechanisms of action of complex bioactive compounds, enabling the systematic identification of molecular targets and modulated signaling pathways [28,29]. This integrative approach combines information from multiple biological databases with advanced computational analyses to map the complex interactions between drugs, proteins, and metabolic pathways [30].

Network pharmacology has elucidated complex molecular mechanisms of several natural compounds, revealing connectivity patterns that explain their pleiotropic effects [31]. For geraniol, preliminary analyses suggested interactions with multiple molecular targets, including metabolic enzymes, nuclear receptors, and cell signaling proteins [11,32]. Among the most recurrent candidates are enzymes such as HMG-CoA reductase (HMGCR), COX-2/PTGS2, iNOS/NOS2, and NOX-2/CYBB oxidase, in addition to the Nrf2 (NFE2L2)–HO-1 (HMOX1) axis and the NF-κB factor, whose redox–inflammatory balance geraniol modulates in multiple tissues [8,12]. As receptors and signaling proteins, CHRM3 (muscarinic M3) and the kinases PRKCA/PRKCD stand out, in addition to JAK1/JAK2 and components of the PI3K/AKT pathway, targets that emerge from network analyses and functional studies with the compound [33].

Drug-likeness assessment is a fundamental aspect of drug development, enabling early prediction of pharmacokinetic and safety properties [32]. The Lipinski, Ghose, and Veber rules, among other pharmacological guidelines, establish physicochemical parameters associated with adequate oral bioavailability and tissue penetration [34,35]. However, these classical rules have limitations, as many natural bioactive compounds violate one or more rules and still exhibit good bioavailability. Therefore, recent efforts propose quantitative methods and hybrid assessment models. For example, DBPP-Predictor is an algorithm that integrates physicochemical property profiles and ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) to calculate the drug-likeness metric with good discriminative capacity [36]. Therefore, the present study aims to perform a comprehensive computational analysis of geraniol, integrating drug-likeness assessment, ADME (absorption, distribution, metabolism and excretion) properties, molecular target identification and protein–protein interaction network analysis, providing a solid scientific basis for understanding its multitarget pharmacological mechanisms.

2. Results

2.1. Analysis of Structure and Physicochemical Properties and Drug-Likeness

The molecular structure of geraniol was characterized using ChemSketch software (ACD/Labs, Toronto, ON, Canada), generating two-dimensional (2D) and three-dimensional (3D) representations of the molecule, including visualization of the electron cloud to analyze the molecular electron density distribution (Figure 1).

Structural analysis reveals that geraniol has the molecular formula C₁₀H₁₈O, characterizing it as an acyclic monoterpene with a primary alcohol functional group [1]. Figure 1A shows the two-dimensional (2D) representation of geraniol, highlighting its linear monoterpene structure composed of ten carbon atoms, a terminal hydroxyl group, and two conjugated double bonds in the isoprene chain. This configuration gives geraniol amphiphilic characteristics and high chemical reactivity, mainly due to the presence of high electron density sites in the C=C double bonds and the electron donation capacity of the hydroxyl group (–OH). This organization gives geraniol an amphiphilic nature; that is, it simultaneously presents a hydrophobic portion (nonpolar chain) and a polar region (–OH group). This duality is essential for its ability to interact with biological membranes, facilitating both penetration into the lipid bilayer and the establishment of hydrogen bonds at the polar interface [37]. Furthermore, this same configuration is responsible for the compound’s high polarizability and ability to participate in hydrophobic and electrostatic interactions in complex biological systems [38].

The three-dimensional ball-and-stick model (Figure 1B) allows visualization of the molecule’s spatial conformation, revealing the actual molecular geometry and interatomic distances. This representation is fundamental for understanding possible intermolecular interactions and molecular recognition by specific biological targets. The 3D molecular surface representation (Figure 1C) highlights the spatial distribution of the hydrophobic and polar regions of the molecule. The hydrocarbon chain is shown predominantly in cyan, corresponding to the less polar portion of geraniol, whereas the oxygen-containing hydroxyl region is highlighted in pink/red, representing the polar region capable of participating in hydrogen bonding and electrostatic interactions. This distribution is consistent with the amphiphilic nature of geraniol and helps explain its ability to interact with both lipid environments and polar biological interfaces [39].

Studies on antioxidant activity also point to the role of conjugated double bonds as important components in this function. In a simple colorimetric method for evaluating antioxidants, it was demonstrated that blocking conjugated double bonds in monoterpenes reduces their antioxidant capacity [40]. This result suggests that the presence of double bonds contributes to geraniol’s ability to donate electrons or stabilize free radicals. The physicochemical properties and drug-likeness descriptors of geraniol were obtained using the ADMETlab 3.0 platform and are summarized in

Table 1. Physical–chemical properties and drug-likeness of geraniol.

Parameter	Value	Reference Range *	Compliance
Molecular properties
Molecular Weight (MW)	154.14	100–600	Compliant
LogP	3.428	≤5.0	Compliant
LogD (pH 7.4)	2.853	−2.0 to 5.0	Compliant
TPSA (Ų)	20.23	≤140	Compliant
Donors of H	1	≤5	Compliant
Acceptors H	1	≤10	Compliant
Rotatable connections	4	≤10	Compliant
Quality scores
QED †	0.617	0–1	Moderate
Fsp3 ‡	0.60	≥0.42	Compliant
Pharmacological rules
Lipinski	0 Violations	0	Compliant
GSK	0 Violations	0	Compliant
Pfizer	1 Violation	0	Non-compliant

* Established ranges for oral drugs; † quantitative estimate of drug-likeness; ‡ carbon fraction sp³.

.

The molecular weight of 154.14 Da places geraniol in a favorable range according to drug-likeness criteria, in line with the literature that points to smaller compounds as more prone to absorption [1]. The LogP values ≈ 3.4 and moderate LogD (physiological pH) suggest sufficient lipophilicity for cellular permeation without making it excessively hydrophobic, which is compatible with reports of good intestinal transport and bioavailability in animal models [41].

The moderate LogP lipophilicity value is consistent with intermediate cell permeability profiles, although experimental quantitative data for permeation in intestinal epithelia or bioavailability in humans are not evident. However, the modulating effect on the permeability of endothelial cells of the blood–brain barrier (BBB) in an OGD/R model (reduction in permeability via Nrf2/HO-1 activation) suggests that geraniol interacts significantly with endothelial barrier mechanisms [42]. Neuroprotective model studies indicate that geraniol crosses the blood–brain barrier and exerts beneficial effects on the central nervous system, implying significant permeability and modulation of oxidative stress and inflammation [43]. It is worth noting that some studies have shown that high doses of geraniol can directly alter membrane integrity, as evidenced by membrane depolarization and reduced resistance in Caco-2 cells [44].

Drug-likeness analysis revealed complete compliance with Lipinski and GSK rules, with only one violation of Pfizer’s rule related to the combination of high LogP and low topological polar surface area (TPSA) [45]. The compound did not present structural warnings for PAINS, ALARM NMR, or other problematic structures. Figure 2 shows the graphical representation of the results.

The radar chart in Figure 2 (ADMET) summarizes that most of the geraniol descriptors fall within the optimal window for oral drugs (blue area), with occasional moderate deviations consistent with lipophilic monoterpenes (yellow line). This visualization format is standard in ADMETlab 2.0, which defines lower/upper limits for key properties (molecular mass, LogP/LogD, TPSA, HBD/HBA, rotatable bonds, etc.) and compares the compound to the “drug-likeness zone” (green/blue) [46]. In the case of geraniol, a molecular weight of ~154 Da and low TPSA (~20 Ų) support high permeability, while LogP ~3.4 maintains the permeability/solubility balance that favors oral absorption, exactly what was observed in experimental pharmacokinetics, with an absolute oral bioavailability of 92% for an emulsified formulation and 16% when adsorbed onto fiber [41]. These experimental data validate the radar reading that the compound is neither excessively polar nor too hydrophobic, remaining in the drug-likeness corridor.

The intermediate lipophilicity highlighted in the radar is consistent with CNS effects reported for geraniol when formulated to cross the blood–brain barrier (e.g., polymeric micelles or intranasal administration), which broaden cerebral distribution and neuroprotective effect in animal models [47,48]. These results suggest that, even with a small and slightly polar molecule, the pharmaceutical form is crucial for optimizing delivery to the brain. In parallel, the radar behavior for risk markers (e.g., alerts related to lipophilicity in “Pfizer/GSK” type filters) is compatible with the need to pay attention to metabolic interactions (CYPs/P-gp) and rational formulation, points already recognized in the compound and investigated in in vitro prediction/assays [41,49]. Thus, integrated radar reading not only confirms ADMET compatibility for oral/topical use, but also offers rationale for advanced (nano/micellar) formulations when the therapeutic target requires targeted tissue delivery (CNS or anti-angiogenesis).

2.2. Pharmacokinetic Profile ADME

2.2.1. Absorption and Permeability

Geraniol demonstrated favorable absorption properties, based on the results of analyses performed on the ADMETlab platform, with predicted human intestinal absorption of 60.9% and adequate permeability through Caco-2 cells (−4.426 log cm/s) (

Table 2. Pharmacokinetic properties of geraniol.

Parameter	Value	Interpretation
Absorption
HIA (%)	60.9	Good intestinal absorption
Caco-2 permeability (log cm/s)	–4.426	Adequate permeability
P-gp substrate (%)	0.2	Non-substrate
P-gp inhibitor (%)	92.7	Probable inhibitor
Distribution
Plasma protein binding (%)	69.6	Moderate binding
VDss (log L/kg)	–0.185	Limited distribution
BBB penetration (%)	3.7	Low brain penetration
Metabolism
CYP2C9 substrate (%)	75.0	Probable substrate
CYP2C19 substrate (%)	97.5	Probable substrate
CYP3A4 inhibitor (%)	30.2	Possible inhibitor
Microsomal stability (%)	92.0	Stable
Excretion
Clearance (mL/min/kg)	10.99	Moderate clearance
Half-life (h)	1.497	Speedy deletion

). The analysis indicated a low probability (0.2%) of being a substrate of P-glycoprotein, a value that, according to Xiong et al. (2021b) in ADMETlab, classifies the compound as “excellent” (category 0–0.3), suggesting the absence of active efflux and consequent better cellular retention [46]. However, the compound showed a high probability (92.7%) of inhibiting P-glycoprotein, a value that, according to the same platform, indicates a strong potential for clinically relevant drug interactions through the blockade of the efflux of other P-gp substrate drugs.

Figure 3 presents the predicted probabilities of inhibition of cytochrome P450 isoenzymes (CYPs) by geraniol, obtained using the ADMETlab 3.0 platform. Higher probabilities of inhibition were observed for CYP2C8 (0.92), CYP2C9 (0.84), and CYP2B6 (0.80), while the CYP1A2 (0.70) and CYP2C19 (0.06) isoforms showed moderate and low values, respectively. CYP2D6 (0.01) and CYP3A4 (0.004) showed virtually no inhibition. These results indicate that geraniol may interact with secondary metabolic pathways of xenobiotic biotransformation, but with low potential for interference with the most clinically relevant enzymes, such as CYP3A4 and CYP2D6, responsible for approximately 70% of human drug metabolism [50].

2.2.2. Distribution and Metabolism

Estimates obtained by ADMETlab (Table 2) indicate that the plasma protein binding (bound fraction) of geraniol is approximately 69.6%, which implies that approximately 30.4% would remain as a potentially active free fraction. This proportion suggests that a significant portion of the molecule may be available to cross membranes and reach biological targets, since only the free fraction is considered pharmacologically active. In general, it is known that plasma protein binding regulates the distribution, clearance, and half-life of compounds, limiting their diffusion into tissues [51].

Regarding the volume of distribution at steady state (VDss), the estimated value of -0.185 log L/kg (≈0.65 L/kg) suggests a predominantly plasma distribution or only modest tissue penetration. Low VDss values indicate that the drug tends to remain in the vascular compartment, which may limit its effectiveness in peripheral tissues. This behavior is consistent with general ADME principles, in which compounds with high protein binding generally exhibit smaller volumes of distribution [52,53]. Finally, the low BBB estimate presented by ADMETlab (3.7%) suggests limited capacity for action in the central nervous system, which may restrict direct effects on the nervous system. However, it should be noted that predictions in tools such as ADMETlab, Protox3.0, and other computational platforms may diverge, as they are based on machine learning models, structural fragments, and training data. In cases of disagreement, such as the divergence observed between ADMETlab and Protox3.0, it is necessary to seek experimental validation (e.g., in vivo brain penetration studies) to confirm or refute these estimates.

2.3. Toxicological Assessment

In silico toxicological analysis was performed using two platforms (ADMETlab 3.0 and ProTox-3.0) for cross-validation of safety predictions. The multidimensional toxicological profile of geraniol, visualized through the radar chart (Figure 4), demonstrates a predominantly low risk for most of the endpoints evaluated.

The analysis reveals that geraniol presents consistently low probabilities for most of the toxicological endpoints evaluated, positioning itself near the center of the radar in categories such as hepatotoxicity, neurotoxicity, cardiotoxicity, mutagenicity, and carcinogenicity. The only endpoints that stand out with greater activity are penetration into the BBB, metabolism via CYP2C9, and potential ecotoxicity. It has already been highlighted that monoterpenes like geraniol are generally safe for humans with respect to hepatotoxicity, cardiotoxicity, mutagenicity, carcinogenicity, and even endocrine disruption.

The low probability of hepatotoxicity estimated by ProTox is aligned with in vivo evidence of geraniol’s hepatoprotective effect. In ischemia–reperfusion and chemical assault models, the compound activates the Nrf2/HO-1 axis, reduces oxidative stress and inflammation (decrease in TNF-α/iNOS/COX-2), with improvement in histological and biochemical markers. These experimental findings provide biological support for the benign predictive profile observed in the radar [15].

The BBB-related peak is consistent with the low polarity of geraniol and with studies demonstrating delivery to the CNS when appropriate pharmaceutical forms are used, such as polymeric micelles and intranasal administration, which increase brain penetration and enhance neuroprotective effects in ischemia models [48]. Recent studies have critically reviewed the main QSPR and machine-learning strategies used for BBB permeability prediction [54] and have sought to validate and refine these models by combining in silico approaches and three-dimensional human models [55].

Regarding the signal for CYP2C9/CYP2C, it reinforces the need for experimental validation with in vitro inhibition/induction assays using index substrates, since the CYP1A2, 2C9, 2C19, 2D6, and 3A4 isoenzymes account for most drug metabolism in humans and are central to drug-induced dysplasia (DIDs). The low involvement of CYP3A4 suggested by the radar is favorable from a clinical risk perspective, given the dominant role of this isoform in drug biotransformation [56,57]. Cluster analysis of toxicological endpoints (Figure 5) suggests 47 parameters evaluated in two functionally distinct groups, providing a complementary perspective on the safety profile of geraniol.

These parameters represent the most relevant points of attention in the safety assessment of geraniol, suggesting the need for experimental validation, especially for interaction with CYP2C9 and brain permeability. The colors on the map correspond to the predicted probabilities of activity, according to the scale presented in the platform legend.

The Inactive Cluster (top) groups 44 endpoints with high-confidence predictions of inactivity (probability ≥0.7), including all nuclear receptors (AhR, AR, ER, PPAR-γ), stress response pathways (nrf2/ARE, HSE, p53), neurotransmitter receptors (GABA, NMDA, AMPA), and most cytochrome P450 enzymes (CYP1A2, CYP2C19, CYP2D6, CYP3A4). This clustering confirms the low potential for systemic toxicity of geraniol. This pattern is consistent with the absence of strong mechanistic signals of genotoxicity and cardiotoxicity specifically attributed to geraniol in modern ingredient safety studies and with reports of protective effects in cellular/tissue models when tested within usual experimental ranges [58].

The Active Cluster (bottom) contains only three endpoints: metabolism via CYP2C9 (moderate probability), penetration of the blood–brain barrier (high probability), and ecotoxicity (moderate probability). This segregation indicates that the main safety concerns are concentrated on specific pharmacokinetic aspects and environmental impact.

CYP2C9 is a clinically relevant liver enzyme responsible for metabolizing approximately 15% of drugs and is subject to polymorphisms with a significant impact on drug interactions and interindividual variability [59,60]. Therefore, any inhibition/competition for monoterpenes can have relevant pharmacological consequences. However, specific direct evidence for geraniol and its interaction with CYP2C9 is still limited. Studies suggest that geraniol and other terpenoids inhibit CYP2B6 in human microsomes [61]. These inhibitory interactions may be correlated with CYP2C9 inhibition [62]. Therefore, the prediction for CYP2C9 should be treated as an operational hypothesis to be verified with HLMs/recombinant cells and 2C9 index substrates [63,64,65].

The high probability of penetration into the BBB derived from the ProTox-3.0 model is consistent with the moderate lipophilicity of geraniol, but the recent literature reinforces that predicting BBB requires multiple lines of evidence (in silico + in vitro PAMPA-BBB/endothelial cultures + in vivo), as isolated lipophilicity does not guarantee effective access to the CNS [66,67]. Toxicological analysis using ADMETLab demonstrated a relatively favorable safety profile (

Table 3. Predicted toxicological profile of geraniol.

Toxicology Endpoint	Probability (%)	Risk Rating
Systemic toxicity
DILI	20.7	Low
Hepatotoxicity	68.1	Moderate
Nephrotoxicity	29.3	Low
Mutagenicity (Ames)	22.8	Low
Genotoxicity	1.4	Very low
Cardiotoxicity
hERG Lock	4.9	Low
hERG (10 μM)	30.4	Moderate
Topical toxicity
Skin sensitization	98.3	High
Eye irritation	99.7	High

). The risk of drug-induced hepatotoxicity (DILI) was low (20.7%), as was mutagenicity by the Ames test (22.8%). The compound presented a low cardiotoxic risk, with a minimal probability of hERG channel blockade (4.9%).

Significant warnings were identified for topical applications, with a high probability of skin sensitization (98.3%) and eye irritation (99.7%), suggesting the need for precautions for external use formulations. The findings predicted in Table 3 indicate that geraniol toxicity is predominantly associated with topical exposure, with a high likelihood of skin sensitization and eye irritation. This profile is consistent with experimental evidence demonstrating positive response in LLNA assays and classifying geraniol as a dose-dependent skin sensitizer [68].

A fundamental mechanistic aspect to interpret the high probability of sensitization observed in Table 3 is the oxidative instability of the geraniol. Studies have shown that auto-oxidation generates highly reactive hydroperoxides that have greater allergenic potential than the parent compound [69]. The translational relevance of this risk is confirmed by clinical patch test studies that demonstrated a higher frequency of positive reactions to oxidized geraniol in patients with fragrance-associated contact dermatitis [70].

Table 4. Comparison of critical toxicological endpoints between ADMETlab and ProTox-3.0.

Endpoint	ADMETlab 3.0	ProTox-3.0	Agreement	Interpretation
Systemic Toxicity
Hepatotoxicity (DILI)	20.7% (low)	Inactive (79%)	High	Low liver risk
Nephrotoxicity	29.3% (low)	Inactive (74%)	High	Low renal risk
Cardiotoxicity (hERG)	4.9% (very low)	Inactive (84%)	High	Low cardiac risk
Neurotoxicity	-	Inactive (78%)	-	Low neurological risk
Genotoxicity
Mutagenicity (Ames)	22.8% (low)	Inactive (97%)	High	Non-mutagenic
Genotoxicity	1.4% (very low)	-	-	Low genetic risk
Carcinogenicity	-	Inactive (76%)	-	Non-carcinogenic
Special Properties
BBB Penetration	3.7% (low)	Active (91%)	Discordant	Requires validation
Immunotoxicity	-	Inactive (99%)	-	Non-immunotoxic
Acute Toxicity
Oral LD50	-	2100 mg/kg	-	Class 5 (Low Toxicity)

compares the critical toxicological endpoints obtained from ADMETlab and ProTox. The acute oral toxicity predicted by ProTox-3.0 (LD50 = 2100 mg/kg) classifies geraniol in Class 5 toxicity according to the Globally Harmonized System (GHS), indicating low acute toxicity.

The comparative analysis between ADMETlab 3.0 and ProTox-3.0 revealed substantial agreement in the safety predictions of geraniol, strengthening the reliability of the conclusions. Both platforms converge in the classification of low risk for hepatotoxicity, nephrotoxicity, cardiotoxicity, and mutagenicity, establishing a robust safety profile [2]. The differences observed between the platforms reflect distinct methodological approaches: ADMETlab uses machine learning algorithms based on experimental data, while ProTox employs QSAR models based on molecular fragments. Experimental validation remains essential, especially for discordant endpoints.

2.4. Identification and Analysis of Molecular Targets

Three complementary databases were used to identify and compare potential molecular targets of geraniol. The Comparative Toxicogenomics Database (CTD) identified 425 possible interacting genes, SwissTargetPrediction identified 100 predicted targets, and GeneCards identified 199 associated genes. After removal of duplicated entries and normalization of gene symbols, the three lists were analyzed using the JVenn tool to visualize overlapping targets among the databases (Figure 6). A pairwise consensus intersection strategy was applied, in which only genes identified in at least two of the three databases were retained for subsequent analyses. This approach resulted in 25 consensus molecular targets. The selection strategy was adopted to reduce false-positive associations derived from individual databases and to prioritize targets supported by complementary evidence sources, including chemical–gene interaction data, ligand-based prediction, and gene-centric biomedical annotation. Figure 6 presents these consensus targets, which were subsequently organized according to their predominant biological functions, including metabolic enzymes, nuclear receptors, protein kinases, transcriptional regulators, and inflammatory mediators.

The functional classification of the 25 consolidated targets was performed using ontological analysis based on Gene Ontology (GO) and Enzyme Commission (EC) classification, resulting in the identification of eight distinct functional classes with specific quantitative distribution (Figure 7). Molecular taxonomy demonstrates that the predicted molecular targets for geraniol are mainly distributed among kinases (36%), general-purpose enzymes (20%), and oxidoreductases (12%), followed by smaller proportions of cytochrome P450 (8%), G protein-coupled receptors (GPCRs, 8%), nuclear receptors (4%), proteases (4%), and ion channels (4%). This structural and functional diversity reflects the pleiotropic and multimodal nature of geraniol interaction, typical of low molecular weight bioactive monoterpenes capable of modulating multiple cellular and enzymatic signaling pathways [1,71].

Before functional classification and network construction, the 25 consensus targets were standardized to their corresponding Homo sapiens orthologs. Although some target annotations retrieved from the source databases were originally linked to non-human organisms, all downstream analyses were performed using Homo sapiens as the reference organism. Thus, the targets presented in

Table 5. Systematic molecular characterization of the 25 consensus molecular targets of geraniol standardized as Homo sapiens orthologs for network analysis.

ID	Symbol	Systematic Naming	Organism	Ontological Class	Main Molecular Function
1	PGR	Nuclear progesterone receptor	H. sapiens	Hormone receptor	Ligand-dependent transactivation
2	HMGCR	3-hydroxy-3-methylglutaryl-CoA reductase	H. sapiens	Oxidoreductase	Catalysis of mevalonate biosynthesis
3	HMOX1	Heme oxygenase (decycling) 1	H. sapiens	Oxidoreductase	Catalysis of heme group degradation
4	NR3C1	Nuclear glucocorticoid receptor	H. sapiens	Hormonal receptor	Steroid-dependent transcriptional regulation
5	SLC6A3	Na+/Cl− dependent dopamine transporter	H. sapiens	Transporter	Synaptic dopamine reuptake
6	MAPK3	Mitogen-activated protein kinase 3 (ERK1)	H. sapiens	Serine/threonine kinase	Phosphorylation of cytoplasmic substrates
7	CHEK1	Checkpoint kinase 1	H. sapiens	Serine/threonine kinase	Activation of G1/S and G2/M checkpoints
8	GSK3B	Glycogen synthase kinase 3 beta	H. sapiens	Serine/threonine kinase	Inhibitory phosphorylation of multiple substrates
9	CDK1	Cyclin-dependent kinase 1	H. sapiens	Serine/threonine kinase	G2/M cell cycle progression
10	MPO	Myeloperoxidase	H. sapiens	Peroxidase	Formation of reactive oxygen species
11	TRPV1	TRPV1 thermosensitive cation channel	H. sapiens	Ion channel	Transduction of nociceptive stimuli
12	POLA1	DNA polymerase alpha 1, catalytic subunit	H. sapiens	DNA polymerase	DNA synthesis during replication
13	MAP2K1	MAP kinase kinase 1 (MEK1)	H. sapiens	Tyrosine/threonine kinase	Activation of MAPK3/1 via dual phosphorylation
14	PTGS2	Prostaglandin G/H synthase 2 (COX-2)	H. sapiens	Cyclooxygenase	Prostaglandin biosynthesis
15	IL1B	Interleukin 1 beta	H. sapiens	Pro-inflammatory cytokine	Activation of acute inflammatory response
16	DDIT3	CHOP transcription factor (C/EBP homologous protein)	H. sapiens	Transcription factor	Regulation of stress-induced apoptosis
17	GNA15	Guanine nucleotide-binding protein G(q) subunit alpha-15	H. sapiens	Signal transducer	Activation of phospholipase C beta
18	EHF	ETS homologous factor	H. sapiens	Transcription factor	Tissue-specific regulation of gene expression
19	IL1A	Interleukin 1 alpha	H. sapiens	Pro-inflammatory cytokine	Initiation of the inflammatory cascade
20	IRF6	Interferon regulatory factor 6	H. sapiens	Transcription factor	Regulation of epithelial differentiation
21	KLF5	Kruppel-like factor 5	H. sapiens	Transcription factor	Regulation of cell proliferation
22	KLF6	Kruppel-like factor 6	H. sapiens	Transcription factor	Tumor suppression and cell cycle regulation
23	TFAP2A	Transcription factor AP-2 alpha	H. sapiens	Transcription factor	Control of cell differentiation
24	TRIM29	Tripartite motif-containing protein 29 (E3 ubiquitin ligase)	H. sapiens	Ubiquitin ligase	Post-translational regulation via ubiquitination
25	EGFR	Epidermal growth factor receptor	H. sapiens	Receptor tyrosine kinase	Mitogenic signal transduction

Note: All targets are presented as Homo sapiens orthologs because GeneMANIA, STRING, MCL clustering, and functional enrichment analyses were performed using Homo sapiens as the reference organism. Target entries retrieved from non-human annotations in the source databases were standardized to their corresponding human orthologs using official human gene symbols before downstream analyses.

correspond to the human ortholog gene symbols used in GeneMANIA, STRING, MCL clustering, and functional enrichment analyses. This standardization ensured that the network topology, interaction metrics, and functional clusters were generated from a species-consistent human target set.

The predominance of protein kinases is consistent with experimental evidence describing geraniol as a modulator of intracellular phosphorylation pathways, particularly the PI3K/AKT and MAPK/ERK cascades, involved in the regulation of the cell cycle, survival, and inflammatory response. In prostate cancer models, geraniol reduces the expression of oncogenic genes and inhibits the activation of ERK1/2 and AKT, culminating in apoptosis and tumor suppression [72]. Similarly, Cho et al. (2016) showed that the compound interferes with the signaling of multiple kinases, including VEGFR-2, p38 MAPK, and JNK, indicating a broad-spectrum modulation profile on pathways involved in cell proliferation and differentiation [73].

Table 5 consolidates the 25 most trusted molecular targets of geraniol, revealing five main functional clusters. Kinases constitute the dominant group (nine targets, 36%), including critical cell cycle regulators (CHEK1, CDK1), the MAPK pathway (MAPK3, MAP2K1), and GSK3B—a pleiotropic kinase associated with neurodegeneration, diabetes, and cancer. Oxidoreductases (three targets, 12%) include HMGCR (a target of statins), HMOX1 (a cytoprotective enzyme), and PTGS2/COX-2 (a target of nonsteroidal anti-inflammatory drugs). Transcription factors (five targets, 20%) encompass proliferation regulators (KLF5), tumor suppressors (KLF6), and differentiation modulators (TFAP2A, IRF6). Hormone receptors include PGR and NR3C1, suggesting endocrine modulation. Pro-inflammatory cytokines (IL1A, IL1B) and other specialized targets complete the profile.

2.5. Analysis of Protein–Protein Interaction Networks

Molecular network analysis of the 25 consolidated targets was conducted using the GeneMANIA platform to elucidate the functional association pattern among the predicted molecular targets of geraniol (Figure 8). To improve visual clarity and biological interpretability, the network representation was reorganized and restricted to the 25 consolidated target genes identified by the integrative database analysis, without displaying the additional secondary genes automatically suggested by GeneMANIA. All GeneMANIA evidence categories were retained in the visualization, including co-expression, physical interactions, pathway associations, colocalization, shared protein domains, and genetic interactions. The revised network uses gene labels and color-coded edges to distinguish the different interaction types, allowing clearer visualization of the relationships among targets involved in inflammation, oxidative stress, cell signaling, cell cycle regulation, and transcriptional control. Gene co-expression networks can be used to associate genes with biological processes and prioritize candidate genes related to disease mechanisms, indicating that many of these targets may be jointly regulated under specific physiological or pathological conditions [74].

According to the legend, the interactions are classified into six main categories, each represented by a distinct color linking one gene to another. Co-expression (38.71%, purple) represents the highest percentage of interactions in the network, indicating that geraniol may exhibit a tendency, in relation to the expression of these genes under certain physiological or pathological conditions [74]. This suggests that geraniol may modulate regulatory pathways that influence the coordinated expression of these genes.

From a methodological standpoint, co-expression networks reflect the correlation between transcript levels across multiple experimental conditions and are widely used to infer functional modules and shared regulatory pathways. In this context, van Dam et al. (2018) demonstrated that co-expressed genes tend to participate in common biological processes, sharing transcriptional regulation, belonging to the same metabolic pathways, or integrating protein complexes [74].

In the pharmacological context of geraniol, the predominance of co-expression is biologically plausible and consistent with experimental findings showing coordinated modulation of inflammatory and signaling pathways. Studies have demonstrated that geraniol reduces the expression of pro-inflammatory mediators (IL-1β, TNF-α, and COX-2) and interferes with regulatory cascades such as NF-κB and MAPK, indicating a direct impact on inflammatory and proliferative transcriptomic programs [75,76].

Physical interactions (19.14%, red) are the second most prevalent category, representing direct protein–protein interactions. This is particularly relevant for understanding how the products of these genes interact functionally in protein complexes and signaling cascades. Interactions via pathways (14.58%, turquoise blue) indicate genes that participate in the same metabolic or signaling pathways, reinforcing the concept that geraniol can affect multiple components of specific pathways simultaneously.

Colocalization interactions (12.51%, blue) suggest genes whose protein products are expressed in the same cellular compartments or tissues, suggesting related functions in specific contexts. Shared protein domain interactions (10.20%, light yellow) show structural similarities between the encoded proteins, indicating possible functional or evolutionary similarities. Genetic interactions (4.85%, green) are those that occur least frequently in the selected genes, referring to functional relationships where a change in one gene affects the phenotype associated with another gene.

Complementary analysis was conducted using the STRING platform (Search Tool for the Retrieval of Interacting Genes/Proteins) to validate and further explore the molecular interactions identified by GeneMANIA. STRING analysis of the 25 consolidated targets revealed a highly interconnected network with significant enrichment of protein interactions (p = 7.31 × 10⁻¹²), using medium-level confidence parameters (0.4) and multiple sources of evidence including experimental data, curated databases, co-expression, and text mining (Figure 9).

Quantitative analysis of the STRING network confirmed the high molecular connectivity of geraniol targets, with 25 nodes connected by 74 edges, resulting in an average degree of 5.92 connections per protein. The local clustering coefficient of 0.518 indicates a strong tendency for the formation of functional modules, where connected proteins tend to form cohesive clusters. The p-value of 7.31 × 10⁻¹² for interaction enrichment confirms that this network exhibits significantly more connections than would be expected for a random set of proteins of similar size, validating the biological relevance of the identified associations.

The application of the MCL (Markov Cluster Algorithm) identified four distinct functional clusters that elucidate the molecular mechanisms of geraniol’s pharmacological properties (Figure 10). According to Shih et al. (2012), MCL has emerged as an effective algorithm for clustering biological networks, particularly for clustering protein–protein interaction (PPI) networks, and is widely used for identifying functional modules [77]. The MCL decomposition into four functional modules is methodologically appropriate for PPI, as demonstrated in classic evaluations and reviews that establish MCL as an effective algorithm for detecting modules/complexes in biological networks [78,79].

Cluster 1 (red, 11 proteins)—inflammatory/immune: It includes IL1A, IL1B, PTGS2 (COX-2), and HMOX1, proteins directly related to inflammation, oxidative stress, and the innate immune response. It represents the largest functional module, focused on regulating cell networks and proliferation, including critical kinases and transcriptional regulators. The experimental literature converges with this arrangement, as geraniol suppresses inflammatory mediators (IL-1β, TNF-α, COX-2) and deactivates NF-κB, reducing MMP-1 and pro-inflammatory enzymes in skin and cartilage models [80]. In liver and kidney, it activates the Nrf2/HO-1 axis (HMOX1), reduces lipid peroxidation, and normalizes damage markers, connecting antioxidant and anti-inflammatory effects [15,81]. These findings support the functional cohesion of the IL1A/IL1B/PTGS2/HMOX1 module observed in the network.

Cluster 2 (yellow, nine proteins)—cell signaling: It involves EGFR, MAP2K1, MAPK3, and NR3C1, central components of proliferative pathways and growth receptor-mediated signal transduction. It is associated with bladder cancer and upregulation of miRNA maturation, suggesting epigenetic mechanisms in antitumor action. Evidence shows that geraniol blocks VEGF/VEGFR-2 signaling and suppresses subsequent AKT/ERK pathways, reducing PCNA and increasing caspase-3, with inhibition of angiogenesis [82]. Since EGFR/MEK (MAP2K1)–ERK (MAPK3) governs proliferation and survival, the sensitivity of these pathways to geraniol is consistent with the identified signaling module; furthermore, the literature on tumor models with coordinated EGFR/VEGF inhibition reinforces the decrease in AKT/ERK as a central antiproliferation mechanism [83].

Cluster 3 (green, three proteins)—cell cycle and checkpoint: It brings together CDK1, CHEK1, and GSK3B, key elements in the control of mitotic progression and response to DNA damage. It includes the activation of the cyclin B1-CDK1 complex and regulation of the quiescent barrier, representing specific control of the cell cycle. In prostate carcinoma, geraniol induces cell cycle arrest and apoptosis, modulating cell cycle regulators [84], in line with the prioritization of CDK1/CHEK1 in the network. The observed AKT/ERK suppression affects GSK3β, a node that integrates mitogenic signals and controls cyclin D1, a canonical mechanism associated with G1/S arrest and proliferative reduction (AKT GSK3β cyclin D1 ratio). Although this axis is extensively validated in analogous monoterpenes, such a specific connection with geraniol is mechanistically plausible in light of its AKT/ERK inhibition [85].

Cluster 4 (blue; four proteins)—transcriptional regulation: it is composed of KLF5, KLF6, TFAP2A, and IRF6, nuclear regulators associated with epithelial differentiation and inflammatory response. It groups specialized transcription factors involved in cell differentiation. The enrichment for transcriptional regulation is consistent with the fact that modulation of MAPK/AKT and NF-κB culminates in gene reprogramming. Although direct geraniol-KLFs/TFAP2A/IRF6 linkages remain scarce, these factors are downstream mediators of the ERK, AKT, and NF-κB pathways modulated by the compound in epithelium and inflammation, a biologically consistent hypothesis to be validated in gene expression and reporter assays.

The comparison between GeneMANIA and STRING analyses reveals convergence in key topological characteristics: high connectivity, formation of functional modules, and significant enrichment of interactions. The predominance of co-expression identified by GeneMANIA (38.71%) is corroborated by the identification of cohesive clusters in STRING, validating the hypothesis of coordinated transcriptional regulation. The identification of similar functional modules in both platforms (cell cycle control, inflammatory response, transcriptional regulation) confirms the robustness of the molecular architecture of geraniol targets. According to van Dam et al. (2018), gene co-expression networks can be used to associate genes of unknown function with biological processes and prioritize candidate genes for diseases, being fundamental for validating hypotheses of coordinated regulation [74].

2.6. Functional Enrichment Analysis

The functional enrichment analysis of the 25 consolidated geraniol targets was conducted using the g:Profiler platform. The analysis was configured for the human organism (Homo sapiens) with a significance threshold of 0.05 and using the g:SCS correction method (g:Limit SCS). Data sources included Gene Ontology ontologies for molecular function, cellular component, and biological process, as well as selected biological pathway databases. The most significantly enriched terms (Figure 11 and

Table 6. Functional enrichment results showing the top 26 most significant entries.

N°	Source	Term ID	Name of the Term	p_adj (Query_1)
1	GO:MF	GO:0043167	Ionic binding	2.665 × 10−4
2	GO:MF	GO:0001228	DNA-binding transcription activator activity	4.312 × 10−4
3	GO:MF	GO:0004674	Serine/threonine protein kinase activity	1.564 × 10−2
4	GO:MF	GO:0106310	Serine protein kinase activity	2.239 × 10−2
5	GO:MF	GO:0003690	Double-stranded DNA binding	3.024 × 10−2
6	GO:MF	GO:0034056	Estrogen response element binding	2.613 × 10−2
7	GO:BP	GO:0009891	Positive regulation of biosynthetic process	1.444 × 10−7
8	GO:BP	GO:0007267	Cell–cell signaling	5.541 × 10−7
9	GO:BP	GO:0001660	Fever generation	6.694 × 10−7
10	GO:BP	GO:0150076	Neuroinflammatory response	4.135 × 10−5
11	GO:BP	GO:0009612	Response to mechanical stimulus	2.323 × 10−4
12	GO:BP	GO:0048313	Golgi inheritance	1.132 × 10−3
13	GO:BP	GO:0032310	Prostaglandin secretion	1.955 × 10−3
14	GO:BP	GO:0038128	ERBB2 signaling pathway	3.562 × 10−3
15	GO:BP	GO:0010575	Positive regulation of growth factor production	4.135 × 10−3
16	GO:BP	GO:0014805	Smooth muscle adaptation	1.169 × 10−2
17	GO:BP	GO:0071216	Cellular response to biotic stimulus	2.017 × 10−2
18	GO:BP	GO:0033092	Positive regulation of immature T cell proliferation	2.918 × 10−2
19	GO:BP	GO:0044187	Nucleic acid biosynthetic process	3.531 × 10−2
20	GO:BP	GO:0001394	Positive regulation of protein biosynthetic process	4.083 × 10−2
21	GO:BP	GO:0060440	Trachea formation	4.083 × 10−2
22	GO:BP	GO:0030335	Positive regulation of cell migration	4.806 × 10−2
23	GO:CC	GO:0000785	Chromatin	2.560 × 10−3
24	GO:CC	GO:0005901	Caveola	2.164 × 10−2
25	GO:CC	GO:0005654	Nucleoplasm	2.748 × 10−2
26	GO:CC	GO:0005737	Cytoplasm	2.502 × 10−2

Legend: GO:MF = Gene Ontology: Molecular Function; GO:BP = Gene Ontology: Biological Process; GO:CC = Gene Ontology: Cellular Component; p_adj = p-value adjusted for multiple comparisons.

) included “response to oxidative stress” (GO:0006979), “inflammatory response” (GO:0006954), “MAPK cascade” (GO:0000165), and “regulation of transcription, DNA-templated” (GO:0006355), with adjusted p-values less than 0.05 after multiple correction (g:SCS). g:Profiler is a reliable and up-to-date functional enrichment analysis tool that supports various types of evidence, identifier types, and organisms, and is widely recognized in the scientific community with over 924 citations [86].

The results revealed multiple significantly enriched functional terms (Figure 11 and Table 6). Among the molecular functions (GO:MF), ion binding (GO:0043167) showed the lowest adjusted p-value (2.665 × 10⁻⁴), followed by DNA-binding transcription activating activity (GO:0001228, p_adj = 4.312 × 10⁻⁴) and serine/threonine kinase protein activity (GO:0004674, p_adj = 1.564 × 10⁻²). According to Pellarin et al. (2025), cyclin-dependent kinases (CDKs) are closely connected to the regulation of cell cycle progression, with kinases capable of directing cell cycle transitions [87].

In the KEGG pathway set, the “IL-17 signaling pathway”, “TNF signaling pathway”, “PI3K-AKT signaling pathway”, and “MAPK signaling pathway” stood out, all recognized as being involved in inflammation, apoptosis, and cell remodeling processes. The recurrence of these terms reinforces the multitarget and integrative nature of geraniol, capable of simultaneously modulating inflammatory, oxidative, and proliferative axes. The Manhattan plot (Figure 11) shows the distribution of enriched terms by functional category, with different colors representing the analyzed databases. The points with the highest elevation on the y-axis correspond to the terms with the lowest adjusted p-value, indicating greater statistical significance of the enrichment. The functional enrichment analysis demonstrates that the 25 identified targets are functionally organized in related biological processes, including transcriptional regulation, cell signaling, inflammatory response, and cell cycle control. The statistical significance of the enriched terms confirms the biological relevance of the set of targets identified for geraniol.

Topological analysis of the PPI network demonstrated that proteins such as MAPK3, HMOX1, and PTGS2 exhibited high degrees of connectedness and high values of betweenness, characterizing them as hub genes. These results indicate that such targets function as strategic convergence points for the pharmacological effects of geraniol. The combination of functional modularity, GO/KEGG enrichment, and network centrality supports the hypothesis that geraniol acts in a coordinated manner on several interconnected molecular axes, which explains its ability to regulate multiple biological phenotypes, including inflammation, antioxidant response, and cell cycle control.

3. Discussion

Geraniol is an acyclic monoterpene widely distributed in nature, present in high concentrations in the essential oils of aromatic plants such as Cymbopogon martinii (palmarosa), Pelargonium graveolens (geranium), Rosa damascena (rose), Cymbopogon citratus (lemongrass), and Cymbopogon winterianus (citronella), among other tropical and subtropical species [7,8,11]. These plants are cultivated in warm and humid climates, especially in Asia (India, China, Indonesia), Africa, and South America, with particular emphasis on Brazil, where species of the genus Cymbopogon show high yields of essential oil rich in geraniol [88]. This wide botanical and phytochemical abundance makes geraniol easily obtainable from low-cost natural sources, reinforcing its strategic value as a model molecule for pharmacological and biotechnological investigations [38,71].

The diversity of plant matrices containing geraniol allows for the exploration of chemical and biological variations that contribute to its multitarget profile, justifying the growing scientific and industrial interest [89]. This ubiquity and phytochemical richness underpin the integration of in silico approaches, including ADME/predictive toxicity analyses and network pharmacology (PPI, GO, and KEGG), as rational tools to prioritize experimental hypotheses and understand the molecular basis of its biological activities. The safety of geraniol has been extensively evaluated in different in vitro and in vivo models, and in toxicological assessments of fragrance ingredients. A recent report from RIFM (Research Institute for Fragrance Materials) concluded that geraniol is not genotoxic, and based on repeated-dose toxicity and reproductive toxicity data, a Margin of Exposure (MOE) greater than 100 was calculated [68].

This study provides a comprehensive computational analysis of geraniol through network pharmacology approaches, revealing a multitarget pharmacological profile consistent with its diverse biological activities reported in the literature. The results demonstrate that geraniol possesses favorable physicochemical and pharmacokinetic properties, acting through multiple molecular targets organized in integrated functional networks. The molecular characterization of geraniol performed in ChemSketch (ACD/Labs, Toronto, ON, Canada) enabled a detailed analysis of its chemical architecture and structural properties. The two-dimensional (2D) representation highlights its acyclic monoterpene character (C₁₀H₁₈O), with two conjugated double bonds and a terminal hydroxyl group (–OH). This structural combination determines its behavior in biological systems, allowing reversible association with membranes and proteins, who demonstrated how linear monoterpenes, such as geraniol, exhibit conformational mobility and modulable polarity, facilitating coupling to hydrophobic enzymatic sites and bioinspired intermolecular interactions [90].

The amphiphilic properties and low toxicity of geraniol and its esters explain their high penetration capacity in lipid membranes, preserving cellular structural integrity and indicating an ideal balance between lipophilicity and solubility, crucial for passive diffusion and pharmacological bioavailability [91]. Thus, the structural analyses obtained in ChemSketch corroborate the experimental and theoretical findings in the literature, consolidating the importance of the linear and polarized configuration of geraniol for its physicochemical and pharmacokinetic properties. The excellent conformity of geraniol with the drug-likeness rules, particularly Lipinski’s Rule of Five, corroborates previous studies that demonstrated its potential as a drug candidate [45]. The molecular weight of 154.14 Da and the moderate lipophilic properties (LogP = 3.428) position the compound within the ideal range for adequate oral bioavailability, consistent with experimental observations of efficient intestinal absorption [41].

The identification of 25 consolidated molecular targets through integrative analysis of multiple databases provides a solid foundation for understanding the multitarget mechanisms of geraniol. This methodological approach, using intersections between different data sources, increases the reliability of predictions and reduces false positives. The functional diversity of the identified targets, including metabolic enzymes (HMGCR, HMOX1), nuclear receptors (PGR, NR3C1), protein kinases (MAPK3, CHEK1, GSK3B), and inflammatory mediators (IL1A, IL1B, PTGS2), explains the broad spectrum of pharmacological activities of geraniol reported in the experimental literature. Network analysis revealed a highly significant modular organization (p = 7.31 × 10⁻¹²), with four distinct functional clusters representing the main mechanisms of action of geraniol. Among these, the inflammatory/immune and cell signaling clusters were the most relevant to the biological interpretation of the present study and are discussed below:

-

Inflammatory/immune cluster: The presence of IL1A, IL1B, PTGS2 (COX-2), and HMOX1 in this cluster corroborates the well-documented anti-inflammatory effects of geraniol [80]. The functional association with “fever generation” suggests modulation of pyrogenic pathways, consistent with experimentally observed antipyretic properties.

-

Cell signaling cluster: The EGFR-MAP2K1-NR3C1 connection represents a plausible mechanism for the antitumor effects of geraniol, through modulation of the MAPK cascade and growth receptor signaling [1].

The predicted metabolic profile, with predominant metabolism via CYP2C9 and CYP2C19, is consistent with experimental pharmacokinetic studies that have demonstrated the extensive hepatic metabolism of geraniol [41]. The possible inhibition of P-glycoprotein (92.7%) and CYP3A4 (30.2%) suggests potential for drug interactions, an aspect that requires careful consideration in pharmaceutical formulations. The network results highlight IL1A/IL1B and PTGS2 (COX-2) in the inflammatory cluster, with HMOX1 as a redox–cytoprotective node, an arrangement consistent with the literature. In inflammatory models, geraniol reduces IL-1β and TNF-α and attenuates tissue damage, confirming the immunomodulatory effect on the NF-κB/COX-2 axis (in arthritis and skin, respectively) [75]. In parallel, in ischemia/reperfusion liver injury, geraniol activates Nrf2/HO-1 (HMOX1), reduces oxidative stress and inflammation, integrating the antioxidant arm with the anti-inflammatory arm of the compound [15].

The literature suggests that the Nrf2–HO-1 ↔ NF-κB/COX-2 coupling is a common mechanism by which monoterpenes (including geraniol) decrease cytokines and MMPs in inflamed tissues, supporting the coherence between their cluster (IL1A/IL1B/PTGS2/HMOX1) and the described physiology [11]. In the signaling cluster, EGFR–MAP2K1–MAPK3 (ERK1/2) and the nuclear receptor NR3C1 (GR) appear as central axes. Geraniol modulates subsequent growth receptor cascades, including inhibiting AKT/ERK when it blocks VEGF/VEGFR-2. This decrease in PCNA and increase in caspase-3 suggests that ERK/AKT pathways are sensitive targets to the presence of the monoterpene [82]. Since the EGFR → MEK (MAP2K1) → ERK (MAPK3) pathway governs proliferation/survival, and NR3C1 crosses with MAPKs/PI3K in various tissues, its network nodes are biologically plausible to explain the antimitogenic signaling observed with geraniol [92,93].

The cell cycle cluster comprises CDK1, CHEK1, and GSK3B. Experimental studies suggest that geraniol induces cell cycle arrest and apoptosis in prostate cancer in vitro/in vivo, modulating cell cycle regulators such as cyclin D1/CDKs, consistent with the prioritization of these nodes in its network [84]. By suppressing ERK/AKT, geraniol affects GSK3β, a kinase that integrates mitogenic signals and checkpoints, offering a mechanistic pathway for G1/S arrest and proliferative reduction (AKT–GSK3β–cyclin D1 relationship; analogous evidence in monoterpenes) [89,94]. Thus, the presence of CDK1/CHEK1/GSK3B in the analysis is consistent with the antiproliferative phenotype reported in multiple models. The identification of KLF5/KLF6 and TFAP2A/IRF6 suggests transcriptional regulatory points linked to epithelial differentiation and the inflammatory response. Although direct evidence of interaction between geraniol and these factors is still limited, there is a solid biological basis for considering them downstream mediators of the ERK, AKT, and NF-κB pathways modulated by the compound [95,96].

By converging the clusters found, it is assumed that geraniol attenuates oxidative stress by activating Nrf2/HO-1 and reducing lipid peroxidation, as well as reducing pro-inflammatory cytokines (IL-1β, TNF-α) and COX-2, uncoupling NF-κB, in addition to suppressing ERK/AKT, impacting PCNA and the cell cycle, and resulting in immunomodulation with less tissue activation [15,80,97]. From a translational perspective, standardized herbal medicines and oils rich in geraniol can be rationally targeted to conditions with a redox–inflammatory–proliferative nexus, such as arthritis, inflammatory dermatitis, pain syndromes, metabolic disorders, and supportive oncology, provided they are standardized (geraniol content, chemotype) and evaluated for safety/ADME (P-gp/CYP interactions). Experimental evidence of anti-angiogenesis (VEGF/VEGFR-2→AKT/ERK), anti-inflammation (IL-1β/TNF-α/COX-2), and neuroprotection (Nrf2/HO-1) supports the categorization of geraniol as a natural drug with multitarget mechanisms, aligned with pharmacological network trends for complex diseases [15,80,82].

Furthermore, recent evidence in a streptozotocin-induced Alzheimer’s disease model indicates that geraniol, in association with limonene, may exert neuroprotective effects, reinforcing its relevance in neuroinflammatory and neurodegenerative contexts [98]. The computational toxicology profile demonstrates relatively favorable systemic safety, with a low risk of hepatotoxicity (DILI: 20.7%) and mutagenicity (Ames: 22.8%). These findings are consistent with experimental toxicity studies that reported an acceptable safety profile for geraniol at therapeutic doses [1]. The warnings for skin sensitization (98.3%) and eye irritation (99.7%) are relevant for topical applications, corroborating clinical observations of hypersensitivity reactions in some formulations containing geraniol [70]. These limitations can be mitigated through appropriate formulation strategies and optimized concentrations.

Functional enrichment analysis using g:Profiler confirmed the biological relevance of the identified targets, revealing significant enrichment for processes related to the inflammatory response, cell signaling, and transcriptional regulation. The most significant terms include cell–cell signaling (p = 5.541 × 10⁻⁷) and neuroinflammatory response (p = 4.135 × 10⁻⁵), consistent with established pharmacological activities. The identification of “positive regulation of T cell proliferation” (p = 2.918 × 10⁻⁵) suggests immunomodulatory potential [99].

The results of the network pharmacology analysis provide a robust scientific basis for the development of geraniol-based pharmaceutical formulations. The multitarget nature of its action suggests potential for complex therapeutic indications, including inflammatory diseases, metabolic disorders, and neoplasms. The excellent drug-likeness and favorable ADME profile indicate viability for the development of oral formulations, while limitations for topical use require specific formulation strategies [41]. The potential for drug interactions, particularly through inhibition of P-glycoprotein and CYP3A4, should be considered in future clinical studies.

The present study is based on an integrative computational framework combining ADMET prediction, network pharmacology, and bioinformatics analyses using publicly available databases. Although these approaches are valuable for generating mechanistic hypotheses and prioritizing molecular targets, some limitations should be acknowledged. First, the co-expression data retrieved from GeneMANIA are derived from heterogeneous datasets encompassing multiple tissues, developmental stages, physiological conditions, and disease contexts. Therefore, the observed co-expression relationships do not necessarily reflect direct or context-specific biological interactions and should be interpreted as indicators of potential functional association rather than causal evidence.

Furthermore, the molecular targets included in the network were obtained through database integration and computational prediction, which may be influenced by differences in data curation, the literature coverage, and prediction algorithms. The consensus-based strategy adopted in this study was designed to reduce false-positive associations but may also exclude biologically relevant targets identified by only a single source.

Finally, the functional modules and enriched pathways identified through GeneMANIA, STRING, and g:Profiler analyses should be considered hypothesis-generating and require experimental validation through molecular, cellular, and pharmacological studies to confirm the predicted interactions and biological effects of geraniol.

4. Materials and Methods

4.1. Structural Analysis and Physicochemical Properties

The molecular structure of geraniol was characterized using ACD/ChemSketch (Freeware), version 2025.2.5 (Advanced Chemistry Development, Inc., ACD/Labs, Toronto, ON, Canada), generating two-dimensional (2D) and three-dimensional (3D) representations of the molecule, including electron cloud visualization for analysis of the molecular electron density distribution. Physicochemical properties were calculated using the same program, including the octanol/water partition coefficient (LogP), elemental composition, molar refractivity, molar volume, parachor, refractive index, surface tension, density, and polarizability.

4.2. Drug-Likeness

The theoretical oral bioavailability of geraniol was evaluated by applying Lipinski’s Rule of Five, using the ArgusLab software, version 4.0.1 (Planaria Software LLC, Seattle, WA, USA) [100]. The analyzed parameters included molecular mass (MM ≤ 500 Da), water/oil partition coefficient (cLogP ≤ 5), number of hydrogen bond donors (nDLH ≤ 5) and number of hydrogen bond acceptors (nALH ≤ 10). The pharmacological properties ADME (absorption, distribution, metabolism and excretion) were analyzed using the SwissADME online platform (Molecular Modeling Group, SIB Swiss Institute of Bioinformatics, Lausanne, Switzerland; accessed on 4 July 2025; https://www.swissadme.ch), developed by the Molecular Modeling Group of the University of Lausanne and the Swiss Institute of Bioinformatics [101,102]. The analysis included gastrointestinal (GI) absorption, blood–brain barrier (BBB) permeation, skin permeation coefficient (LogKp), and topological polar surface area (TPSA). Additional analyses of ADMET properties were performed using the ADMETlab 3.0 online platform (CBDD Group, Central South University, Changsha, Hunan, China; https://admetlab3.scbdd.com; accessed on 4 July 2025), including assessment of toxicity, metabolic stability, and interactions with cytochrome P450 enzymes [103,104]. For cross-validation of toxicological predictions, the ProTox-3.0 online platform (Charité–Universitätsmedizin Berlin, Berlin, Germany; https://tox.charite.de) accessed on 4 July 2025 was used, developed by Charité–Universitätsmedizin Berlin [105].

4.3. Identification of Molecular Targets

The identification of geraniol molecular targets was performed using three complementary databases: the Comparative Toxicogenomics Database version 2025.1 (CTD; North Carolina State University, Raleigh, NC, USA; http://ctdbase.org/; accessed on 4 July 2025), which provided 425 possible interacting genes; SwissTargetPrediction online platform, accessed on 4 July 2025 (Molecular Modeling Group, University of Lausanne and SIB Swiss Institute of Bioinformatics, Lausanne, Switzerland; https://www.swisstargetprediction.ch), which provided 100 predicted targets; and GeneCards version 5.3 (Weizmann Institute of Science, Rehovot, Israel; https://www.genecards.org), accessed on 4 July 2025, which provided 199 associated genes. Searches were performed using the descriptor “geraniol”. The use of these databases was based on their complementary evidence frameworks: CTD provides curated chemical–gene interaction information mainly derived from toxicogenomic and literature-based evidence; SwissTargetPrediction predicts potential targets using ligand-based chemical similarity; and GeneCards integrates gene-centric biomedical information from multiple biological and clinical sources.

After data retrieval, all gene lists were manually curated to remove duplicated entries, normalize gene symbols, and harmonize target names across databases. This screening strategy is widely used in network pharmacology studies and was adopted according to the methodology described by Noor et al. (2022) [31]. The normalized gene lists were then analyzed using the jVenn online tool (INRAE/GenoToul Bioinfo, Toulouse, France; http://jvenn.toulouse.inra.fr/app/example.html) accessed on 4 July 2025 to visualize overlaps among databases and identify consensus targets [106]. A pairwise consensus intersection criterion was adopted: genes were included in the final target set when they were identified in at least two of the three databases. Thus, genes did not need to be present simultaneously in all three databases, but they needed to be supported by more than one independent source.

Genes identified in only one database were excluded from the main network analysis to reduce false-positive associations and improve reproducibility. This exclusion criterion was selected to prioritize targets supported by complementary prediction or annotation frameworks, while maintaining a conservative and reproducible target selection strategy. Using this approach, 25 consensus molecular targets were selected for subsequent functional classification, protein–protein interaction analysis, and enrichment analysis. This intersection strategy is widely used in network pharmacology to reduce false positives from different prediction approaches (chemical, genomic, and toxicological) [106]. Although this strategy increases confidence in the selected target set, it may exclude biologically relevant genes reported by only one database; therefore, the final target list should be interpreted as a prioritized consensus set rather than an exhaustive catalog of all possible geraniol targets.

Before downstream network analyses, all selected targets were standardized to Homo sapiens orthologs. This step was necessary because some target annotations retrieved from the source databases were originally associated with non-human organisms, whereas GeneMANIA, STRING, MCL clustering, and functional enrichment analyses were performed using Homo sapiens as the reference organism. Therefore, non-human target annotations were mapped to their corresponding human orthologs and harmonized using official human gene symbols. Only targets with unambiguous human ortholog correspondence were retained for downstream analyses. This procedure ensured species consistency across the protein–protein interaction and enrichment analyses and reduced potential bias related to cross-species annotation differences.

4.4. Analysis of Interaction Networks

The 25 selected target genes were subjected to network analysis using the GeneMANIA online platform (University of Toronto, Toronto, ON, Canada; http://www.genemania.org), accessed on 4 July 2025, after selecting the organism Homo sapiens, adopting a minimum confidence level of 0.40 (moderate confidence) and including known physical and functional interactions. The generated network categorized the interactions as: co-expression, physical interactions, pathways, colocalization, shared protein domains, and genetic interactions. The protein–protein interaction network was analyzed using STRING online database, version 12.0 (STRING Consortium; https://string-db.org), accessed on 4 July 2025, with confidence parameters set at a medium level (0.4) [107]. Interactions were determined based on experimental data, curated databases, co-expression, and text mining.

4.5. Network Topological Analysis and Clustering

Topological analysis of the network was conducted using quantitative metrics implemented in the STRING platform, including the number of nodes and edges, average node degree, local clustering coefficient, and statistical significance of interaction enrichment [78]. Functional clusters were identified using the MCL (Markov Cluster Algorithm) implemented in STRING, with an inflation parameter set to 3. This algorithm detects natural clusters based on stochastic flow in the interaction network, identifying protein communities with high internal connectivity. The statistical significance of protein interaction enrichment was assessed using the PPI (protein–protein interaction) enrichment test, comparing the observed number of interactions with the expected number for a random set of proteins of similar size. Network characterization metrics included connectivity density, clustering coefficient, node centrality, and network modularity. All analyses were performed using default parameters of the respective software, except where specifically mentioned. Results were considered statistically significant when p < 0.05.

4.6. Functional Enrichment Analysis Methodology

Functional enrichment analysis of molecular targets was performed using the g:Profiler online platform, version e114_eg62_p19_27110d83, accessed on 10 August 2025 (University of Tartu, Tartu, Estonia; https://biit.cs.ut.ee/gprofiler/gost) which provides statistical enrichment analyses for Gene Ontology (GO), metabolic pathways (KEGG—Kyoto Encyclopedia of Genes and Genomes, Reactome) and other functional annotations. Significance parameters followed the g:SCS (Set Counts and Sizes) multiple correction method, and only terms with adjusted p (FDR) < 0.05 were considered statistically significant [108].

5. Conclusions

This integrative investigation, combining in silico predictions, network pharmacology analysis, and ADMET modeling, shows that geraniol is a bioactive compound with a broad pharmacological spectrum, whose versatility results from its coordinated action on multiple interconnected biological pathways. The identification of 25 central molecular targets, organized into functional modules related to the inflammatory response, oxidative stress, cell cycle control, and transcriptional signaling, demonstrates that the molecule operates as a systemic modulator capable of restoring cellular homeostasis in the face of harmful stimuli.

The drug-likeness parameters and predicted pharmacokinetic properties reinforce its viability for pharmaceutical development, particularly in oral and nanostructured formulations aimed at targeted delivery. In parallel, the computational toxicological profile suggests an acceptable safety margin, compatible with therapeutic and cosmetic applications when associated with rational formulation strategies. From a mechanistic point of view, geraniol stands out for modulating the Nrf2/HO-1 ↔ NF-κB axis, reducing ROS, pro-inflammatory cytokines and caspases, while increasing antioxidant enzymes (SOD, CAT, GPx) and antiapoptotic proteins (Bcl-2). This molecular plasticity explains the multitarget character of the compound, supporting its experimentally reported antioxidant, anti-inflammatory, neuroprotective and antitumor effects.

The results obtained also offer a solid mechanistic rationale for the integration of in vitro and in vivo approaches that validate the computational predictions described here. The association of omics data (transcriptomics, proteomics and metabolomics) and systems biology techniques represents the next step to consolidate geraniol as a promising pharmacological candidate, suitable for clinical and industrial applications. In summary, this work broadens the understanding of the molecular basis of geraniol, reaffirming it as a model molecule in the field of network pharmacology and natural product biotechnology, and outlines a strategic scientific roadmap for its translational advancement.