Radicicol as a Dual-Site Inhibitor of Isoprenylcysteine Carboxyl Methyltransferase: A Computational Study

Abstract

Background:

Isoprenylcysteine carboxyl methyltransferase (ICMT) is an enzyme crucial for the post-translational processing of Ras oncoproteins. Pharmacological inhibition of ICMT can mislocalize Ras and disrupt oncogenic signaling, exhibiting anticancer effects. Radicicol (RAD) is a 14-membered resorcylic acid lactone biosynthesized by fungi and possesses diverse bioactivities.

Objective:

The current work explored RAD’s interaction with ICMT through molecular docking and molecular dynamics simulations, comparing Radicicol’s binding mode to that of the native ICMT ligand.

Method:

Docking simulations were performed using Glide XP mode, and binding energies were refined by MM-GBSA calculations. Molecular dynamics simulation of 100 ns was conducted to assess the binding stability of the Radicicol–ICMT complex using Desmond, with analysis of RMSD and protein–ligand interactions.

Results:

Molecular docking and molecular dynamics simulations revealed that RAD stably bound within the ICMT active site by bridging both the S-adenosylmethionine cofactor pocket and the hydrophobic prenyl substrate tunnel. Its key interactions included a persistent hydrogen bond with Val116 and an induced-fit engagement of Arg125, supporting a snug and stable RAD–ICMT complex.

Conclusion:

These computational insights suggested that RAD inhibited ICMT by dual-site binding, simultaneously occupying the cofactor and substrate pockets. Such a dual engagement could mislocalize prenylated proteins (like Ras) and represent a novel mechanism of action for RAD.

INTRODUCTION

Naturally derived metabolites have drawn a great deal of attention because of their promise as essential constituents in the developing of new pharmaceuticals.[1] Fungi are widely distributed in nature and have been identified as one of the major sources of natural products due to their abundance of secondary metabolites biosynthetic clusters of genes.[2,3] Fungal metabolites demonstrate wealthy pharmacological characteristics due to their diverse structural features.[4,5,6,7,8,9]

Isoprenylcysteine carboxyl methyltransferase (ICMT) is an integral membrane enzyme that catalyzes the final step of the post-translational modification of CAAX-motif proteins such as Ras.[10] In this reaction, ICMT uses the cofactor S-adenosyl-L-methionine (SAM) to methylate the carboxylate of prenylated cysteine, resulting in a methyl ester.[10] Structurally, ICMT is unusual among methyltransferases, consisting of multiple transmembrane helices with a conserved C-terminal catalytic domain that encloses a polar SAM-binding pocket and an adjacent hydrophobic tunnel for the lipid substrate.[10] This unique architecture enables recognition of both a hydrophilic cofactor and a lipophilic prenyl group within the same active site.[10] Because proper Ras processing and membrane localization require ICMT activity, ICMT has emerged as an anticancer target of interest. Indeed, pharmacological inhibition of ICMT can mislocalize Ras and disrupt oncogenic signaling, exhibiting anti-cancer effects in various models.[11] For example, the indole compound cysmethynil was identified via high-throughput screening as a small-molecule ICMT inhibitor that competes with the prenylated substrate, causing Ras mislocalization and growth inhibition in an ICMT-dependent manner.[11] However, cysmethynil and related early inhibitors suffer from suboptimal potency and drug-like properties,[11] motivating the search for new scaffolds that can more effectively target ICMT’s active site.

Radicicol is structurally distinct from other Hsp90 inhibitors like geldanamycin, yet it fits into a nucleotide pocket and interferes with Hsp90 function.[10] Given its ability to bind tightly to a protein pocket and its amphipathic character (a chlorinated aromatic moiety and multiple polar groups), we hypothesized that radicicol might also bind to ICMT’s active site. If RAD or analog can occupy ICMT’s SAM or substrate-binding regions, it could act as a novel ICMT inhibitor. This study explored the binding of RAD to ICMT through molecular docking and molecular dynamics (MD) simulations, comparing RAD’s binding mode to that of the native ICMT ligand. The native ligand in the crystal structure of Methanosarcina acetivorans ICMT (PDB 4A2N) is S-adenosyl-L-homocysteine (SAH, the cofactor product) along with a bound lipid (palmitate) in the prenyl substrate tunnel.[10] For our comparisons, we focused on SAH as the representative “native” ligand as it occupies the active site and provides key polar interactions for catalysis. By analyzing docking poses and MD trajectories, we aimed to identify the detailed residue interactions that stabilize RAD versus the native ligand in ICMT. These insights will help assess RAD’s potential as a lead compound or inspire the design of new ICMT inhibitors leveraging RAD’s scaffold.

MATERIALS AND METHODS

The X-ray crystal structure of M. acetivorans ICMT (PDB ID: 4A2N) was used as the protein model[10] as no human ICMT structure is yet available. Importantly, the archaeal ICMT has a conserved active site and ~28% sequence identity in the catalytic domain to human ICMT, ensuring that the binding insights are translationally relevant.[10,12] All heteroatoms were removed except for the native ligand (SAH) during initial analysis. Protein preparation involved adding missing hydrogens, assigning proper protonation states at pH ~7, and minimizing side-chain conformations, using the Schrödinger Maestro Protein Preparation Wizard (Schrödinger, LLC, New York, NY, USA). For docking experiments, two ligand structures were considered: (1) Radicicol, retrieved and prepared with proper bond orders and protonation (assumed neutral in physiological conditions, given its lack of ionizable groups), and (2) the native cocrystallized ligand from 4A2N. The native ligand was taken as SAH without the palmitate since SAH occupies the active-site pocket; this “prepared native ligand” was geometry-minimized for docking. Partial charges for ligands were assigned using the OPLS3e force field in Maestro.

Molecular docking

Docking simulations were performed with Glide (Schrödinger) in extra-precision (XP) mode. The binding site was defined as a receptor grid encompassing the crystallographic ligand (SAH) position, including the adjacent hydrophobic tunnel entrance. No constraints were applied so as to allow RAD to explore the pocket freely. RAD and the native ligand were each docked into the ICMT binding site; poses were scored by GlideScore. The top-scoring pose for each was analyzed. Binding energies were further refined by an implicit solvent MM-GBSA (molecular mechanics Generalized Born Surface Area) calculation (Prime, Schrödinger) on the docked complexes to estimate binding free energy (ΔG_bind).

Molecular dynamics simulation

To assess binding stability, the RAD–ICMT complex from docking was subjected to all-atom MD simulation. The system was prepared with the Desmond MD System (Schrödinger) using the OPLS3e force field. The protein–ligand complex was embedded in a POPC lipid bilayer (to mimic the native membrane environment of ICMT) and solvated with explicit TIP3P water molecules. Periodic boundary conditions and 0.15M NaCl were applied. After a short minimization and equilibration, an unbiased 100 ns MD production run was conducted at 300 K and 1 atm (NPT ensemble) with a 2-fs time step. Coordinates were saved every 100 ps. Simulation analysis was performed using the Simulation Interaction Diagram tool: protein and ligand root-mean-square deviation (RMSD) were computed to monitor stability, and protein–ligand contact frequencies were calculated. Hydrogen bonds were defined by a distance cutoff of 3.5 Å and donor–acceptor angle >120° and considered persistent if present in >30% of frames. Hydrophobic contacts were counted when any ligand atom was within 4.0Å of a hydrophobic residue’s side chain. Halogen bond occurrence was monitored specifically for RAD’s chlorophenyl group (with geometry criteria of ~3.5 Å distance and 150°–180° C–Cl·acceptor angle).

Analysis and visualization

Docking poses and interactions were visualized in Maestro. Two-dimensional (2D) ligand interaction diagrams were generated for both RAD and the native ligand to identify key residues within 3Å of the ligand (hydrophobic contacts, hydrogen bonds, etc.). These 2D schematics were used to guide the initial interaction analysis. Molecular graphics (3D) were rendered in PyMOL and Maestro for figure preparation.

RESULTS

Docking scores and binding energies

RAD was predicted to bind ICMT with high affinity, yielding a Glide XP docking score of –10.39, which was slightly more favorable than that of the native ligand (–9.46). The docking pose of RAD was also associated with a favorable MM-GBSA ΔG_bind (–60.13 kcal/mol). These computational binding energies suggested that RAD can fit into the ICMT active site and interact strongly, potentially rivaling the native ligand in binding strength. The docked orientation of the native ligand (SAH) closely superimposed with its crystallographic conformation, indicating the docking procedure reliably reproduces the known binding mode. The heavy-atom RMSD between the redocked SAH and the crystal pose was low (within ~1.0 Å). This validation gave confidence that the docking protocol is suitable for predicting RAD’s pose.

Radicicol binding mode and interactions

Upon docking into the ICMT active site, RAD adopted a conformation that spanned the cofactor pocket and extended partially into the prenyl-binding tunnel. The 2D interaction diagram and 3D pose revealed a network of hydrogen bonds and hydrophobic contacts that stabilized RAD in the binding site. RAD formed multiple hydrogen bonds within the ICMT active site. Notably, the resorcinol ether portion of RAD engaged in a hydrogen bond with His126 (HIE126), a residue in the catalytic domain. In the docked pose, the ND-H of His126 donated a hydrogen bond to one of RAD’s carbonyl oxygens (O═C). Additionally, RAD’s cyclic dihydroxyphenyl moiety acted as a hydrogen bond donor to the backbone carbonyl of Val116 on ICMT. The docked model also suggested a possible hydrogen bond involving His113 (HIE113): The imidazole of His113 was positioned near one of RAD’s oxygen atoms, consistent with a hydrogen bond. In summary, the key polar contacts predicted for radicicol included H-bonds with His126, His113, and Val116. His126 and His113 both reside in the active-site cavity that normally coordinates the adenine and ribose of SAM/SAH; in the RAD complex, they appeared to form H-bonds analogous to those that stabilized the cofactor (see below for comparison to SAH). Val116 was part of a loop (residues 114–117) that lined the binding pocket; RAD’s interaction with Val116’s backbone mimicked the way the adenine ring of SAM/SAH was anchored by backbone interactions.

In addition to hydrogen bonding, RAD is stabilized by hydrophobic and aromatic interactions within ICMT’s active site. The macrocyclic and aromatic portions of radicicol are surrounded by a pocket of nonpolar residues. The chlorophenyl side chain of RAD penetrated into the prenyl-binding tunnel region, where it was embraced by hydrophobic residues such as Tyr179, Tyr182, Leu171, and Leu115 (all shown in green in the 2D schematic). The dichlorophenyl ring made van der Waals contacts with Tyr179 in particular, a tyrosine located at the entrance of the lipid tunnel. Meanwhile, RAD’s naphthalene-like core (resorcylic lactone ring system) is flanked by aromatic side chains Tyr129 and Tyr121 on one side and Met128 and Pro127 on the other. These residues formed a predominantly hydrophobic surface that complements RAD’s polyaromatic scaffold, likely contributing to π–π stacking or general van der Waals stabilization. For example, Tyr129’s aromatic ring was nearly parallel to RAD’s benzene ring system in the docked pose, suggesting a π–π stacking interaction. Leu99 and Leu115 also made hydrophobic contacts with radicicol’s core. Overall, residues Tyr121, Pro127, Met128, Tyr129, Leu99, Leu115, Leu171, Tyr179, and Tyr182 formed a hydrophobic binding pocket enveloping radicicol.

An interesting specific interaction observed was a halogen bond involving RAD’s chlorinated phenyl group. The chloride substituent was positioned toward the imidazole of His126. The geometry in the docked pose was consistent with a halogen bonding interaction: The chlorine atom on RAD pointed toward the lone pair of His126’s ND1, at a distance of ~3.3 Å and an approximately linear C–Cl·N angle. This suggested that His126 engaged RAD’s chloroaromatic group via a halogen bond, further anchoring the ligand. His126 was thus a bifunctional interaction partner for RAD- donating a hydrogen bond and accepting a halogen bond, underscoring its importance in recognizing this ligand.

In summary, docking indicated that RAD bound in ICMT’s active site by mimicking key interactions of the native ligand while also exploiting unique contacts. It hydrogen-bound with His126, His113, and Val116, occupied the hydrophobic tunnel with its chlorophenyl ring (packing against Tyr179/Leu171), and even formed a halogen bond with His126. These interactions collectively confered a tight fit for radicicol in the SAM/substrate pocket of ICMT.

Comparison of radicicol and native ligand interactions

Both RAD and the native ligand (SAH) engaged the ICMT active site residues, but there were notable differences aligned with their distinct structures. RAD, lacking the charged carboxylate and cationic amine of SAH, did not form the Lys114 salt bridge. In our docked RAD pose, Lys114 was not involved in binding (consistent with RAD being neutral and not reaching that region of the pocket). Instead, RAD compensated by reaching deeper into the prenyl tunnel (contacting Tyr179, which SAH does not significantly contact) and by forming a halogen bond and an extra H-bond with His126. SAH, on the other hand, relied on Lys114 and Tyr121 to stabilize its polar terminus, interactions that RAD cannot make. Conversely, both ligands shared the Val116 backbone hydrogen bond motif – radicicol’s ring oxygen and SAH’s adenine each accept an H-bond from Val116’s NH – highlighting that any ligand occupying the SAM pocket will likely need to satisfy this key interaction. Both ligands also interacted with His113, though in different ways: SAH’s ribose OH was H-bonded by His113, whereas RAD, which has an –OH in a different position, appeared to donate a H-bond to either His113 or Val116 (our docking suggests RAD’s –OH to Val116’s carbonyl, whereas MD analysis below will show involvement of Arg125 instead). In terms of hydrophobic contacts, RAD and SAH shared many partners (Tyr129, Met128, Pro127, Leu115, Tyr182, etc.) as they occupied overlapping space in the pocket. However, RAD’s chlorophenyl group extended toward Tyr179/Leu171, tapping into the hydrophobic tunnel where the lipid tail of a substrate would go – an area SAH by itself did not reach. This suggested RAD may simultaneously block the substrate lipid-binding tunnel in addition to the SAM-binding site, a potentially advantageous feature for inhibition. In summary, RAD imitated the native ligand’s binding to the polar pocket (Val116, His113, etc.) while also extending into the hydrophobic tunnel, engaging residues like Tyr179 that the native cofactor/product did not. These combined interactions likely underline RAD’s strong docking score.

Molecular dynamics simulation of radicicol–ICMT complex

To evaluate the stability of RAD’s binding and to observe if any new interactions emerged due to protein flexibility, a 100 ns MD simulation was carried out on the RAD-bound ICMT complex. Figure 1a shows the RMSD of the protein (blue) and radicicol (magenta) over time. The protein backbone remained stable (fluctuating around ~1.5 Å RMSD after an initial equilibration period), indicating no large-scale unfolding. RAD’s positional RMSD (magenta, right-hand y-axis) averaged around ~2.0 Å with moderate fluctuations, suggesting that RAD stayed in the binding pocket without dissociation, though it explored slight adjustments in pose. There was no indication of RAD unbinding; the ligand RMSD remained below ~3 Å throughout the simulation, which is consistent with a stably bound ligand.

Figure 1.

Molecular dynamics (MD) simulation analysis of the Radicicol–ICMT complex over 100 ns: (a) RMSD profiles showing stable binding of Radicicol within the ICMT active site. (b) Interaction timeline demonstrating persistent residue contacts, primarily with Val116, Arg125, and Ile124. (c) Histogram representing interaction fractions, confirming stable hydrogen bonds with Val116 (99% occupancy) and Arg125 (~80%), and frequent hydrophobic contacts with Ile124 (~90%). (d) Representative MD snapshot illustrating stable binding pose and critical interactions involving Val116 and Arg125

Critically, the MD simulation confirmed the persistence of several key interactions identified in docking and revealed some dynamic behavior of the binding site [Figure 1b-d]. A timeline of contacts [Figure 1b] and interaction fraction analysis [Figure 1c] were generated to quantify how frequently each residue interacted with radicicol. Val116 and Arg125 emerged as the top interaction partners during MD. In particular, Val116 maintained a hydrogen bond with RAD for nearly the entire 100 ns (interaction fraction ~99%, Figure 1c green bar for Val116). This indicated that the H-bond between RAD’s hydroxyl and the Val116 backbone carbonyl (as seen in docking) is very robust in the dynamic environment. Interestingly, Arg125, which was not a primary contact in the initial docked pose, formed a stable interaction with radicicol during the simulation (interaction fraction ~80%, Figure 1c). The contact analysis suggests that Arg125 (a residue located on a flexible loop in the active site region) reoriented to form a hydrogen bond (or salt bridge) with RAD in a majority of simulation frames. A closer examination of the trajectory indicated that Arg125’s side chain engaged RAD’s second ring hydroxyl (the one that was hydrogen-bonded to His126 in docking). In the MD-refined pose [Figure 1d], Arg125 (blue sphere label) is seen hydrogen-bonding to radicicol’s resorcinol O– (with an occupancy ~ 82% of the time, magenta arrow in Figure 1d). This Arg125–RAD interaction replaces the His126–RAD hydrogen bond that was seen initially, suggesting a shift: As the protein relaxes, Arg125 (which carries a positively charged guanidinium) can provide a favorable H-bond/electrostatic interaction to radicicol’s phenolate oxygen (RAD’s ring OH likely deprotonated in the binding pocket environment), whereas His126 might pivot to a slightly different role. Indeed, the halogen bond with His126 was somewhat transient in the simulation – the halogen bond occupancy was about 10–15% (orange fraction on His126 in Figure 1c), indicating the chloride…His126 interaction broke and reformed occasionally. This is not surprising as subtle shifts in geometry can disrupt a halogen bond; nonetheless, His126 remained in close contact (via hydrophobic or π interactions) ~50% of the time (Figure 1b shows frequent contact events for His126).

The MD results also highlighted Ile124 as consistent hydrophobic contact, with an interaction fraction ~90% (purple bar in Figure 1c). Ile124 (part of the same loop as Arg125) appeared to pack against RAD’s macrocyclic ring throughout the simulation, emphasizing the role of that loop (residues 121–129) in cradling the ligand. Other hydrophobic contacts observed in docking persisted as well: Tyr129 and Met128 showed notable hydrophobic contact fractions (~40% and ~20%, respectively), indicating RAD remained nestled against these residues. Tyr182 and Leu99/Leu115 had lower contact fractions (<20%), consistent with RAD staying closer to the core pocket and not fully extending to Tyr182 or deeply to Leu99 once equilibrated. Glu167, which is at the pocket periphery, occasionally interacted (likely via water-mediated contacts) but not strongly (fraction <10%).

Another important observation from the simulation is that RAD remained tightly bound without inducing large disturbances in the protein. The protein’s average RMSD (~1.5 Å) and the relatively stable contact pattern suggest that ICMT’s active site accommodated RAD without significant structural rearrangement (aside from the local loop movement of Arg125). This implied that the binding site was inherently flexible enough to accept RAD’s bulky scaffold, or that RAD’s shape is well-aligned with the existing pocket. The fact that Arg125 can move to hydrogen-bond RAD hints at the plasticity of the active site, the enzyme can adjust to maximize interactions with a non-native ligand. Arg125 was not required for SAH binding (since SAH already has Lys114/Tyr121 for its carboxylate), but for RAD, Arg125’s ability to form a stabilizing H-bond was a fortuitous gain.

Overall, the MD simulation corroborated the docking-predicted interactions, with Val116 and His126 remaining key anchors and Arg125 stepping in as a new stabilizer. The hydrophobic contacts remain largely as initially identified, and RAD did not dissociate or drift, reinforcing the notion that RAD was a snug fit for the ICMT active site. Figure 1d shows a representative snapshot at the end of the simulation, summarizing the prominent interactions: Val116 (green sphere) still hydrogen-bonded (~89% occupancy) to RAD’s lactone oxygen, Arg125 (blue sphere) hydrogen-bonded (~82% occupancy) to RAD’s phenolic oxygen, and the chlorophenyl group of RAD positioned in the hydrophobic tunnel (with the chlorine oriented toward where His126 would be, though at this frame the halogen bond was not engaged).

DISCUSSION

In this study, we investigated the binding of RAD, a known Hsp90 inhibitor to ICMT, in comparison to ICMT’s native ligand (SAH). Docking and MD simulations show RAD stably binds ICMT’s active site, replicating many of the key interactions required for ligand binding. RAD achieved a docking score comparable to (even slightly better than) the native ligand, suggesting it is energetically favorable for RAD to bind in the SAM/SAH pocket. This was a striking finding given how different RAD’s structure is from SAH; RAD lacks the charged groups of SAH, yet through a combination of hydrogen bonding and hydrophobic insertion, it anchored itself in the dual-character cavity of ICMT.

A central theme that emerged is that RAD mimicked the cofactor (SAM/SAH) in how it binds to ICMT’s polar pocket. Specifically, RAD’s interactions with His113 and Val116 mirrored those of SAH’s adenine and ribose with the enzyme. Val116 is part of a conserved glycine-rich loop (in many methyltransferases) that recognizes the adenine moiety of SAM.[10] In ICMT (4A2N), Val116 (preceded by Leu115 and followed by Lys117 in sequence) plays a similar role, and our results showed that any ligand occupying the adenine site is likely to form H-bonds with Val116’s backbone. SAH did so via its adenine N1 and N6, and RAD did so via an oxygen (as an acceptor) and possibly via its hydroxyl (as a donor to Val116’s carbonyl). This highlighted Val116 as a conserved anchor point for ligand binding in ICMT. Moreover, the presence of His113 (part of a His-Glu motif often seen in Class I methyltransferases) provided a polar residue to interact with the 2′/3′-OH of the ribose in SAH and in the RAD complex, His113 also engaged one of RAD’s polar groups (either directly or via water). This suggested that His113 can adapt to hydrogen bonds with different donors/acceptors in the active site, contributing to the flexibility of ligand recognition.

On the other hand, RAD also exploited the hydrophobic prenyl-binding tunnel of ICMT more than the native cofactor did. The native ligand SAH only covered the cofactor half of the active site; the prenyl-binding tunnel was normally occupied by the lipid tail of the substrate (a farnesyl or geranylgeranyl group attached to a cysteine). In the crystal structure 4A2N, a palmitate molecule was bound in that tunnel,[10] but in our docking of SAH alone (without palmitate), the tunnel remained empty. RAD, in contrast, has a dichlorobenzene moiety that extended into this hydrophobic tunnel and interacted with residues like Tyr179, Leu171, and Tyr182, effectively plugging the tunnel. This dual engagement, polar contacts in the cofactor pocket and hydrophobic contacts in the tunnel, may make RAD a more comprehensive inhibitor, potentially preventing both SAM binding and substrate binding simultaneously. In other words, RAD’s binding mode suggested it could act as a bi-substrate-mimetic inhibitor, spanning both the SAM site and part of the substrate site. This is a desirable property in inhibitor design as it often leads to higher affinity and specificity.

One unanticipated result from the MD simulation was the strong, persistent bonding of Arg125 with RAD. In hindsight, this makes sense: RAD has a 2,5-dihydroxyphenyl moiety that, once deprotonated (radicicol can ionize at one of the dihydroxy positions under certain conditions), could carry a negative charge or strong polarity, attracting Arg125. Even if not fully deprotonated, the phenolic O⁻ of RAD would have a partial negative character that an Arg side chain could stabilize. This interaction is analogous to how Lys114 stabilizes the negative carboxylate of the substrate in RAD’s case, Arg125 played the role of stabilizing a negatively polarized group. From a drug design perspective, this is insightful because it suggested that adding or positioning a negative charge on an ICMT inhibitor could enhance interactions with Arg/Lys residues in the active site, possibly improving potency. Conversely, it also implied that RAD’s binding might depend on microenvironment pH (as the protonation state of its phenolic OH could modulate Arg125 binding).

Comparing our findings to known ICMT inhibitors, some parallels and differences were observed. The prototype inhibitor cysmethynil is much smaller and primarily occupies the substrate pocket (it is an indole that likely binds near where the prenyl cysteine would go).[11] Cysmethynil does not mimic SAM; instead, it competes with the prenylated cysteine.[11] RAD, on the other hand, appeared to straddle both the SAM site and part of the substrate site, which might grant it a different mechanism possibly a noncompetitive or mixed type inhibition if it prevented the enzyme from binding the required component properly. If RAD (or an analog) were to inhibit ICMT, it might not be outcompeted simply by high SAM levels or high substrate levels alone since it touched both sites. This could make it a more robust inhibitor under varying cellular conditions. That said, RAD is a relatively large natural product (~360 Da) and somewhat lipophilic; its binding orientation suggested a snug but tight fit, meaning there might be limited room to chemically modify radicicol without losing fit. However, the RAD scaffold could possibly be simplified or tuned to better exploit the ICMT pocket, for instance, one might consider analogs that retain the resorcinol lactone core (for Val116 and His interactions) but have different substituents that target Lys114 or Arg125 more optimally, or that fill the tunnel deeper.

It is also worth noting that ICMT’s active site must accommodate both a methyl donor (SAM) and a long lipid substrate; thus, it is quite spacious and bifurcated. RAD’s successful binding demonstrates that non-substrate-like molecules can bridge these two subpockets. This feature holds promise for drug discovery as it broadens the range of viable chemotype. Many early ICMT inhibitors were substrate analogs (indole derivatives, tetrapeptides, etc.), primarily targeting the lipid site.[11] Our findings suggest an alternative strategy: targeting the SAM binding pocket, which is more conserved across species and enzymes, simultaneously blocking the substrate-binding site. RAD inadvertently accomplishes this dual inhibition. Mechanistically, if Radicicol binds ICMT within a cell, it may function similarly to SAH, the reaction product known to act as a feedback inhibitor for many methyltransferases when accumulated. By mimicking SAH binding, Radicicol could serve as a potent product-analog inhibitor, with the additional advantage of occluding the lipid substrate channel. While our simulations employ an archaeal ICMT structure, the key SAM-binding residues and overall fold are highly conserved in mammalian ICMT, supporting the applicability of our results to the human enzyme. Selectivity Considerations: Because RAD binds in the SAM pocket, it could theoretically inhibit other SAM-dependent methyltransferases (recalling that SAH is a broad inhibitor of many methyltransferases[13]). However, ICMT is unique in possessing an adjacent hydrophobic prenyl-binding channel; RAD’s bulky macrocyclic scaffold simultaneously engages this hydrophobic subpocket. Other methyltransferases generally lack a comparable lipid-binding cavity, potentially limiting RAD’s affinity for those enzymes. Thus, RAD’s dual-site binding mechanism may impart a measure of selectivity for ICMT, though off-target effects on other methyltransferases cannot be ruled out without further testing. Analogs in Multi-Target Inhibitor Design: Our dual-pocket binding hypothesis for RAD aligns with recent strategies in anticancer drug development. For example, quinazolinone–amino acid hybrids have been reported as single molecules that simultaneously inhibit EGFR kinase and tubulin polymerization, effectively targeting two distinct sites to improve therapeutic outcomes. Such dual-site inhibitors illustrate the potential benefits of engaging multiple enzyme pockets, supporting the notion that RAD’s concurrent occupancy of the SAM pocket and prenyl substrate tunnel could yield enhanced inhibitory potency.

CONCLUSION

Radicicol (RAD) is a naturally occurring fungal metabolite with a wide range of biological activities.

Furthermore, the reported computational investigation here suggested RAD as a potential ICMT inhibitor, providing a new pathway for its anticancer effectiveness through dual-site binding in the active region of the enzyme. More computational docking and molecular dynamics simulation investigations are required to discover novel targets of RAD and to improve its molecular contacts. In vitro and in vivo studies are needed to prove the molecular docking and dynamics simulation results. RAD may also be used as a scaffold for multitarget therapeutic medicines and has potential for drug repurposing strategies due to its multiple pharmacological properties. In future studies, we will examine RAD’s activity against other methyltransferases to empirically determine its selectivity, ensuring readers understand this as a considered aspect of the work.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.