Close this search box.

Investigation of the potential effects of estrogen receptor modulators on immune checkpoint molecules – Scientific Reports

Immune checkpoints play key roles in regulating the immune system response. The design and development of new inhibitors or repurposing clinically used drugs are important areas of research. Several high-impact studies have shown promising results, suggesting that flavonoids, also known as phytoestrogens, could be potent inhibitors of the PD-1/PD-L1 signaling axis. Therefore, we investigated the potential use of estrogens and estrogen receptor modulators as inhibiting ligands of PD-1, PD-L1, and CTLA-4 using molecular docking methods. The calculated binding energy values indicate that quinestrol, quercetin and bazedoxifene could potentially exhibit therapeutic effects through the inhibition of PD-1 and CTLA-4.

Estrogens are among the most important hormones that control not only reproduction in females, but also play a significant role in the overall regulation of the female organism throughout the fertile period. This was highlighted during the COVID-19 pandemic, where the ratio of infected women to men was similar, but the mortality rate among males was higher in countries with limited access to medical care or lower levels of health care resources1. In addition to their apparent anti-viral effects, estrogen factors have been shown to support the function of cardiovascular systems2 and respiratory systems3. These molecules also help prevent endothelial damage2 and minimize the risk of cytokine storm4 by reducing the binding of IL-6 to its receptor5. Estrogens, including the phytoestrogens found in the diet, have been found to influence wound healing, cancer microenvironment, and viral infections such as COVID-191,6,7. These processes intersect at an important crossing point: the microenvironment changes represented by IL-6-dependent inflammation8. These findings have sparked interest in exploring the role of estrogen receptor modulators in the immune system, particularly in relation to immune checkpoints, and their potential clinical applications.

The immune system serves as an effective protective mechanism against various pathogens, including tumor cells. In the context of anti-tumor immune response, the activation of T cells plays a key role, which requires fulfilment of two conditions9: first, an antigen-presenting cell (APC) must present antigens to T cells through the interaction between the peptide-presenting major histocompatibility complex (MHC) molecule and the T-cell receptor; second, co-stimulatory molecules must be activated. Without proper co-stimulation, T cells enter a state of clonal anergy in which they become unresponsive. Tumors often evade immune surveillance by downregulating both MHC and co-stimulatory molecules while upregulating co-inhibitory molecules. Two representative immune checkpoint proteins are programmed cell death 1 (PD-1) and T-lymphocyte-associated antigen 4 (CTLA-4)10. PD-1 shares 21–31% sequence identity with CTLA-411, but unlike PD-1, CTLA-4 contains an extracellular cysteine residue that allows it to form covalently bound homodimers.

PD-1, also known as CD279, is a surface transmembrane glycoprotein and a member of the CD28 family12. It is not expressed on resting naïve T cells but is found on the surface of TCR-stimulated T cells13. PD-1 has two known physiological ligands: programmed death-ligand 1 (PD-L1; B7-H1; CD274) and programmed death-ligand 2 (PD-L2; B7-DC; CD273)14. PD-L1 can be expressed by T and B cells, dendritic cells (DCs), macrophages, and cancer cells, while PD-L2 is present on cancer cells, macrophages, dendritic cells, and B cells. PD-1 represses the immune response by suppressing the activity of T cells and protects the body against chronic inflammation. Nevertheless, in the tumor environment, PD-1 expression induces an immunosuppressive phenotype. The interaction of PD-1 with PD-L1 activates the Src homology region 2 domain-containing phosphatase-1 (SHP-1) and, to a higher extent, SHP-215. SHP-1 and SHP-2 suppress T-cell receptor (TCR) function, leading to inhibited cell proliferation and cytokine production, such as that of interferon-γ (IFN-γ) and interleukin 2 (IL-2)13. However, in regulatory T (TREG) cells, the PD-1/PD-L1 signaling axis stimulates cell proliferation and Foxpro3 signaling14. Furthermore, IFN-γ produced by activated NK and T cells can induce PD-L1 expression in cancer cells16.

In contrast to PD-1, CTLA-4 exerts its immunosuppressive effects on T cells during the early phase of immune response. The activation of T cells involves the interaction between antigen/MHC and the T-cell receptor (TCR), or between CD80/86 on antigen-presenting cells and CD28 on T cells17. However, the presence of CTLA-4 suppresses the activation of T cells. CTLA-4 interacts with CD80/CD86 on the surface of antigen-presenting cells, including dendritic cells and macrophages10. Additionally, CTLA-4 expressed by TREGs can stimulate trans-endocytosis of CD80/CD86 in dendritic cells, thereby suppressing their activation function17.

Numerous inhibitors targeting the PD-1/PD-L1 and CTLA-4 signaling pathways have been discovered18,19,20. However, the development of novel inhibitors or repurposing existing compounds remains of great importance. Flavonoids, also known as phytoestrogens, have been reported to exhibit potent inhibitory activity against the PD-1/PD-L1 signaling axis21,22,23. Similarly, other ligands of estrogen receptors, such as selective estrogen receptor modulators (SERMs), have the potential to act as inhibitors of immune checkpoint proteins. In the process of identifying protein target ligands, molecular docking is a valuable computational tool employed for studying the interaction of a set of SERM molecules with immune checkpoints.

Considering the chemical similarity between quercetin and certain SERMs (e.g., Tanimoto similarity indices between quercetin and luteolin or genistein are 0.7927 and 0.5536, respectively), additional ligands of estrogen receptors like SERMs could serve as inhibitors of immune checkpoint proteins24. To address the need for introducing novel molecules and repurposing existing compounds to target immune checkpoint proteins, we conducted a computational analysis utilizing molecular docking. Specifically, we examined the possible interactions between a set of 14 SERM molecules and three immune checkpoint proteins, namely CTLA-4, PD-1, and PD-L1, along with one of the physiological ligands of CTLA-4, the CD80 protein. Quercetin was used as a reference compound in this in silico study.

In Fig. 1, the chemical structures of the four estrogen molecules used in the docking study are displayed, depicting Tanimoto similarity scores ranging from 0.2782 to 0.6103. The docking results indicated that these molecules exhibited similar binding locations on the receptor surface and comparable binding energy values. Consequently, we focused on the complexes with estradiol as a representative estrogen and further examined them in detail. Among the docking results, poses with higher binding affinities than those with quercetin were selected, and bazedoxifene and quinestrol were chosen as representative prototypes (Fig. 2). Notably, estrogens demonstrated better docking scores compared to bazedoxifene and quinestrol in the case of the CD80 protein, and thus the results obtained with estrogens are also presented. Additional information on the binding modes of all other docking poses can be found in the supplementary information section.

Figure 1
figure 1

Chemical structures of estrogens.

Figure 2
figure 2

Chemical structures of quercetin, bazedoxifene, and quinestrol.

Docking studies of checkpoint proteins with estrogen receptor modulators (ERMs)

In this section, we present the results of docking ERMs to checkpoint proteins. Structurally, all these proteins are based on the immunoglobulin fold domain (IgV) architecture. The human checkpoint proteins CTLA-4, PD-1, and PD-L1, along with one of the ligands of CTLA-4, CD80, were used as docking targets for the calculations. Values of Tanimoto similarity between the studied estrogens and SERMs and known ligands of checkpoint proteins were used for assessing the relevance of the binding mode. Molecular docking calculations were performed using the AutoDock Vina software25 and 3-D crystallographic structures were obtained from the Protein Data Bank26. Only docking poses with a free energy of binding lower than -5.5 kcal/mol were considered, which corresponds to an approximate “interaction constant” value of 0.1 mM or lower. The figures of relevant of docking poses are showed in Fig. S1–Fig. S17.

Molecular docking of cytotoxic T-lymphocyte protein 4 (CTLA-4)

Considering the evidence of the effects of compounds from Rhus verniciflua Stokes on the CTLA-4/CD80 axis, we performed docking studies with both CTLA-4 and CD8021. The CTLA-4 receptor (Alpha Fold entry P164010-F1) shares a similar topological organization with the PD-1 protein. Its N-terminal side contains a single extra-cellular IgV fold domain, followed by a transmembrane helix connecting to the intracellular segment. The function of the protein is to bind its physiological ligands, CD80 and CD86. For structural analysis, we selected the 3D structures of CTLA-4 complexes with two proteins, PDB ID 1I8L and 1I85, respectively. The 3D structures of hCTLA-4 N-terminal domain complexes with monoclonal antibodies (mABs, PDB ID: 5TRU, 6RP8, 7SU0, 7SU1, 5XJ3, 5GGV, 6XY2, 6RQM, 7DV4) reveal that the blocking antibodies cover the CD80 and CD86 binding surface. The CTLA-4 residues directly involved in the interactions with CD80 and CD86 are listed in Table S1.

To generate search boxes, we utilized the structural model of the residues listed in the Supplementary Information (Tables S1 and S2). Docking was performed using the 17 SERMs listed in Table 1. The most significant docking scores are highlighted in red in the Table, which summarizes the docking scores and approximate Ki values for each SERM with CTLA-4 and hCD80.

Table 1 Results of the docking calculations for hCTLA-4 and hCD80.

For CTLA-4 (depicted in green), there is one docking location where most SERM molecules are predicted to bind. There is a second, less populated location where only two molecules are docked. In contrast, for CD80 (depicted in bronze), the docking calculations predict only one docking site (Fig. 3).

Figure 3
figure 3

Docking locations of SERMs on the surface of CTLA-4 (green) and CD80 (bronze). The small molecules shown here are those with significant values of docking scores (Table 1).

The molecular docking analyses of quinestrol, bazedoxifene, quercetin, estradiol, raloxifene, and XL-147 (Fig. S1S5) with hCTLA-4 reveal their potential as inhibitors of CTLA-4 signaling through specific binding interactions to that protein. These findings provide insights into the binding modes and molecular recognition of these compounds with hCTLA-4. Overall, these docking studies shed light on the potential role of quinestrol, bazedoxifene, and quercetin as inhibitors of CTLA-4 signaling, opening avenues for further research in the development of novel therapeutics targeting immune checkpoint molecules.

In addition to the docking analysis, a steric clash analysis was conducted specifically for CTLA-4 to assess the potential interference of SERM binding with the physiological ligands CD80 and CD86. The analysis revealed extensive steric clashes between the docked SERMs and the binding sites of CD80 and CD86 on CTLA-4 (Fig. 4). Note that this steric clash analysis was specifically focused on CTLA-4 and was not performed for PD-1 or PD-L1. Given that the SERMs bind in the region of CTLA-4, PD-1, or PD-L1 that corresponds to the binding site of the physiological ligands, it is reasonable to expect similar steric clashes with the bound physiological ligands for PD-1 (such as PD-L1) as well.

Figure 4
figure 4

Steric clashes of CD80 and CD86 with bound bazedoxifene on CTLA-4.

Figure 4 illustrates the potential steric clashes between CD80 and CD86 ligands and bazedoxifene bound to CTLA-4. To generate the figures, a superposition operation was performed with CTLA-4, aligning the coordinates of CTLA-4 complexed with CD80 or CD86 to the location of CTLA-4 in the 3-D structure used for the docking (the B-subunit of PDB 3OSK). The atomic coordinates of CTLA-4 were then removed for clarity, and the steric contacts between the docked bazedoxifene and CD80 (left) or CD86 (right) are depicted as thin red lines. The extensive steric clashes observed between the docked bazedoxifene and CD80/CD86 suggest potential interference with the binding of these physiological ligands to CTLA-4. These findings highlight the potential inhibitory effects of bazedoxifene on the binding of CD80 and CD86, which are crucial for the immune response mediated by CTLA-4. Taken together, these findings highlight the potential interference of SERMs, such as bazedoxifene, with the binding of CD80 and CD86 to CTLA-4, which are crucial for the regulation of immune responses mediated by CTLA-4. The steric clashes observed emphasize the importance of further investigation to understand the implications of SERM binding for the functional interactions between immune checkpoint proteins and their physiological ligands.

Molecular docking of T-lymphocyte activation antigen CD80

CD80, one of the physiological ligands of CTLA-4, plays a crucial role in T-cell activation. It is a co-stimulatory molecule that delivers a second signal to T cells upon interaction with CD2827. Structurally, CD80 (AlphaFold entry P33681-F1) is a type 1 transmembrane protein expressed on the surface of antigen-presenting cells. Its extracellular N-terminal region consists of two sequential Ig fold domains, followed by a transmembrane helix and an intra-cytoplasmic carboxy-terminal segment. The primary function of CD80 is to enhance and sustain T-cell activation by binding to CD28. However, this activation process is inhibited when CD80 binds to CTLA-4, which outcompetes CD28. Consequently, the immune response is terminated. Given the inhibitory effect of compounds from Rhus verniciflua Stokes on the CTLA-4/CD80 axis21, conducting docking studies of ERMs with CD80 can provide valuable insights into the potential effects of these estrogens on the interaction between CTLA-4 and CD80.

By exploring the docking of ERMs to CD80, we aimed to further elucidate the impact of these estrogenic compounds on the CTLA-4/CD80 interaction and its downstream signaling. From the results presented in Table 2 and considering the chemical similarity between the four estrogens (with Tanimoto chemical similarity scores up to 0.61), it is highly likely that if these molecules possess the capability to bind to proteins within the CTLA-4/CD80 axis, their primary binding site would be CD80. The structural and functional characteristics of CD80 make it a potential target for these estrogens, as depicted in Figure S7. These findings support the notion that these four estrogens, due to their chemical similarity, would most likely interact with CD80, a key protein within the CTLA-4/CD80 axis. Moreover, the docking scores obtained for quercetin (Table 2, Figure S8) and quinestrol (Table 2, Figure S8) suggest their potential as inhibitors. The docking scores reflect the strength of the interaction between these compounds and the target protein, with lower scores indicating more favorable binding affinities. In the case of quercetin and quinestrol, the docking scores suggest a strong potential for inhibitory activity against the target protein associated with the CTLA-4/CD80 axis.

Table 2 Results of the docking calculations for the PD-L1 symmetric homodimer (N-terminal Ig fold domain).

Molecular docking of programmed cell death 1 ligand 1 (PD-L1)

PD-L1 is a protein that is anchored in the plasma membrane through a single transmembrane helix (AlphaFold entry Q9NZQ7-F1). It serves as a ligand for the Programmed Cell Death Protein 1 (PD-1) receptor28.

Symmetric homo-dimer (N-terminal IgV fold domain)

Multiple 3D structures of the N-terminal domain of PD-L1 have been determined in a complex with non-peptide small molecules (PDB IDs 5J89, 5J8O, 5N2F, 5N2D, 5NIU, 6NM7, 6NOJ, 6NOS, 6R3K, 6RPG, 6NM8, 6VQN, 7DY7, 7BEA, 7NLD). These crystal structures reveal the PD-L1 domain as a symmetric homodimer, different from the skewed homodimer crystal structure of PD-L1 (a construct that contains the two N-terminal Ig fold domains in each monomer, PDB id 4Z18). The dimerization of PD-L1 is believed to be induced by the binding of small molecules29. Importantly, this homo-dimeric form of PD-L1, when bound to small molecules, undergoes internalization, resulting in its removal from the cell surface.

The symmetric homodimer configuration of PD-L1 exhibits a central channel between the two domains, which accommodates the binding of small molecules (Figure S18). The region encompassing this channel was utilized to define the search box for docking experiments. Prior to that, blind docking was also performed with the entire homo-dimer: all ERMs were docked either within the central channel or at the “base” of this channel, away from the N-terminus of the polypeptide chain (not shown).

Table 2 presents the results of the molecular docking calculations performed with the symmetric PD-L1 homodimer. The docking scores, representing the binding affinity, are provided in kcal/mol, while the approximate Ki values in µM give an indication of the “strength of binding.” These results reveal the interaction between each ERM and the homodimer. Notably, certain ligands such as bazedoxifene, estradiol, and quinestrol exhibited strong binding affinity with lower docking scores and Ki values, indicating a potential favorable interaction with this conformation of PD-L1. Detailed results concerning these three ligands are described in the following figures.

The Tanimoto similarity indices between the 17 SERMs investigated in this study and a representative ligand, derivative of Schiff base (called compound A, R81) observed bound to the homo-dimeric form of PD-L129, indicate a relatively low chemical similarity, suggesting that these ERMs may not readily induce formation of a symmetric dimer. This finding raises caution regarding the potential restructuring of PD-L1 organization on the cell surface by the investigated SERMs. Previous studies have reported dissociation constants of 10.19 μM and 4.53 μM for PD-1 and PD-L1, respectively, in the presence of quercetin22. Similarly, kaempferol 7-O-rhamnoside showed dissociation constants of 31.1 μM and 19.7 μM. Although the 3D complex structures of these compounds with PD-1 and PD-L1 have not been determined, a reasonable hypothesis can be formulated that both quercetin and kaempferol bind to residues involved in the PD-1:PD-L1 interface, disrupting their interaction23. This suggests a potential for modulating the PD-1/PD-L1 axis by these compounds.

Based on the docking calculations for the symmetric homodimer PD-L1, the results suggest hypothetical binding of the ERMs, including quercetin, bazedoxifene, quinestrol, and others, to the central channel of the homodimer. However, it is important to note the low chemical similarity between the ERMs and the representative ligand together, with the absence of experimental complex structures, additional studies are needed to confirm the actual binding affinities and functional implications of these interactions. Overall, the docking results presented in Figures S10S12 provide insights into the potential binding modes and interactions between the investigated ERMs and the hPD-L1 symmetric homodimer. However, caution should be taken when interpreting these findings, particularly in relation to the induction of the symmetric dimer formation by the ERMs. Future experimental investigations are warranted to establish the precise binding affinities, evaluate the impact on PD-L1 organization, and determine the functional consequences of these interactions.

PD-L1 N-terminal IgV fold domain

For the docking calculations, we utilized multiple 3-D structures of PD-L1 to gain a comprehensive understanding of the binding interactions. Firstly, we employed the structure of the PD-L1 domain present in the complex with PD-1 (PDB id 4ZQK, resolution of 2.5 Å) as a starting point. Additionally, high-resolution models of PD-L1 alone (PDB id 4Z18, resolution of 1.8 Å) and in complex with a macrocyclic inhibitor (PDB id 5O45, resolution of 0.99 Å) were also incorporated into the docking calculations. The utilization of higher resolution structures is expected to improve the accuracy of the docking calculations by providing more precise positional and geometric information, potentially revealing additional docking poses that may have been missed in calculations based on the lower resolution structures. The residues involved in the interaction with the human PD-1 IgV domain are listed in Table S3.

Table 3 presents the results of the docking calculations performed with the PD-L1 N-terminal domain and PD-1 N-terminal domain. These calculations provide insight into the potential binding locations of the ERMs on the surfaces of PD-L1 and PD-1.

Table 3 Results of the docking calculations for the PD-L1 N-terminal domain and PD-1 N-terminal domain.

Figure 5 illustrates the docking locations of the ERMs on the surfaces of PD-L1 (depicted in green) and PD-1 (depicted in bronze). The small molecules shown in the figure correspond to those with significant docking scores, as listed in Table 3.

Figure 5
figure 5

Docking locations of ERMs on the surface of PD-L1 (green) and PD-1 (bronze). The small molecules shown here are those with significant values of docking scores (Table 3).

In contrast to the binding predictions for CD80, the docking calculations for PD-L1 and PD-1 indicated potential binding sites on their surfaces where the majority of the ERMs are likely to bind (Fig. 5). Furthermore, for both proteins, the calculations suggest a second, less populated binding location where only one ERM is predicted to bind. The docking results presented in Figures S13, S14, and S15 provide insights into the potential binding modes and interactions between bazedoxifene, quercetin, and quinestrol, respectively, with the PD-L1 N-terminal IgV fold domain. These computational docking simulations reveal critical interactions between the ligands and the receptor, as depicted in the schematic and 2D diagrams.

The findings suggest potential binding modes and provide a visual representation of the interactions, highlighting the contributions of specific amino acid residues and their respective atoms or groups. Consolidating these results with the overall docking findings in this section, they collectively offer valuable insights into the binding preferences and interactions of the investigated ligands with the PD-L1 N-terminal IgV fold domain. As mentioned before, further experimental investigations are warranted to validate these findings and establish the precise binding affinities and functional consequences of these interactions. The docking results presented in this entire section provide valuable insights into the binding preferences and interactions of the investigated ligands with the PD-L1 N-terminal IgV fold domain. Bazedoxifene exhibited a docking score of − 6.1 kcal/mol, indicating a moderate binding affinity to PD-L1. The interactions involved key amino acid residues, including GLY 119, TYR 118, PHE 19, VAL 44, ALA 18, THR 20/22, GLU 45, LEU 94, and ASN 96. These findings suggest that bazedoxifene holds promise as a selective modulator of PD-1 activity. Similarly, quercetin demonstrated a docking score of − 6.3 kcal/mol, indicating a relatively strong binding affinity for PD-L1. Notable interactions involved VAL 42, LYS 46, ALA 52, GLY 119, PHE 42, PRO 43, GLU 45, ASP 49, and TYR 118. These findings suggest that quercetin may be capable of disrupting the interaction between PD-1 and PD-L1, potentially affecting the PD-1/PD-L1 axis. Remarkably, quinestrol exhibited potent binding affinities not only with PD-L1 but also with PD-1, CTLA-4, and CD80. The docking scores for quinestrol (− 7.2 kcal/mol for PD-L1, − 6.0 kcal/mol for PD-1, − 6.8 kcal/mol for CTLA-4, and − 6.1 kcal/mol for CD80) indicate strong and stable interactions with these proteins. This suggests that quinestrol has the potential to modulate the activity of PD-L1, PD-1, CTLA-4, and CD80 through its high binding affinity. Quinestrol may therefore represent a versatile therapeutic candidate or serve as a lead compound for further drug development targeting these proteins. For this reason, the case of quinestrol is discussed after the results of docking with the PD-1 N-terminal domain.

The findings presented in this section provide a foundation for future investigations into the role of the PD-L1 N-terminal IgV fold domain and its interaction with ligands, paving the way for potential therapeutic strategies targeting the PD-1/PD-L1 axis. In the following section, we will explore the binding characteristics and interactions of the investigated ligands with the PD-1 extracellular domain, shedding light on their potential as modulators of PD-L1 function.

Molecular docking of Programmed Cell Death protein 1 (PD-1)

The receptor protein PD-1 (AlphaFold entry Q15516-F1) is a key player in the PD-1/PD-L1 immune checkpoint pathway30. It shares a similar topological organization with its two ligands, PD-L1 and PD-L2. PD-1 consists of a single extracellular IgV fold domain located at the N-terminal side of its amino acid sequence, followed by a transmembrane helix that connects to the intracellular segment. Like PD-L1, PD-1 is anchored in the plasma membrane of T cells and pro-B cells. Through molecular docking simulations, we investigated the binding characteristics and interactions of the investigated ligands with the N-terminal IgV domain of hPD-1. The residues of the hPD-1 N-terminal IgV domain that form the interaction surface with the ligands hPD-L1 and hPD-L2 are summarized in Table S4. These findings provide insights into the potential modulation of PD-1 activity by the investigated ligands. In this section, we will present the docking results and discuss the binding preferences and interactions of the ligands with PD-1. These findings contribute to our understanding of the PD-1/PD-L1 axis. The docking simulation results revealed the binding interactions between PD-1 and bazedoxifene, as shown in Figure S16.

Bazedoxifene demonstrated a favorable docking score of − 6.6 kcal/mol, indicating its potential binding affinity for the PD-1 protein. The docking scores for quinestrol with the four proteins under investigation (− 7.2 kcal/mol for PD-L1, − 6.0 kcal/mol for PD-1, − 6.8 kcal/mol for CTLA-4, and − 6.1 kcal/mol for CD80) indicate strong and stable interactions with these proteins. This suggests that quinestrol has the potential to modulate the activity of PD-L1, PD-1, CTLA-4, and CD80 through its high binding affinity. Quinestrol may therefore represent a versatile therapeutic candidate or serve as a lead compound for further drug development targeting these proteins. The molecular docking analysis of Programmed Cell Death Protein 1 (PD-1) in the context of the PD-1/PD-L1 axis provides valuable insights into the binding preferences and interactions of potential modulators, namely bazedoxifene (Figure S16) and quinestrol (Figure S17). The results suggest that both bazedoxifene and quinestrol exhibit potential binding affinities for PD-1, indicating their potential as modulators of PD-1 activity.

Molecular dynamics simulations of selected SERMs and estradiol against the PD-L1/PD-1 axis

The docking simulations provided valuable suggestions that the biological activity of estradiol and some SERMS (e.g., quercetin, bazedoxifene, and quinestrol) could be associated with inhibition of immune checkpoint inhibitors, especially for the PD-1/PD-L1 axis. Nevertheless, the accuracy of molecular docking may not always be sufficient. To obtain more accurate estimates of the free energies of binding, we used a more advanced method, molecular dynamics simulations. Using this method, we investigated the interactions of quercetin, bazedoxifene, estradiol, and quinestrol with both PD-L1 and PD-1. The molecular dynamics simulations were performed using CHARMM, together with the CHARMM-GUI web server for input generation31,32. The results obtained using this procedure are shown in Tables 4 and 5.

Table 4 Free energy of binding (Δ-G, kcal/mol) to PD-L1 (5O45).
Table 5 Free energy of binding (Δ-G, kcal/mol) to PD-1 (6UMV).

The values of the free energies of binding obtained after both the initial equilibration step and the molecular dynamics simulations are significantly lower than those obtained from Vina docking, except for the binding of estradiol to PD-1. It should be noted, however, that molecular dynamics simulations address the dynamics of the ensembles, and the coordinates after MD show ligands slightly displaced from the lowest energy state. Even though the value of the binding energy calculated for the estradiol interaction with PD-1 was lower than in the case of molecular docking, its value was still significant. While the docking calculations provide valuable predictions, they are based on computational models with inherent limitations. Validation through additional experimental studies is crucial to confirm the binding affinities and functional implications. The integration of 3-D crystallographic structures, manual preparation using UCSF Chimera, docking with AutoDock Vina, and visualization with UCSF Chimera and BIOVIA Discovery Studio Visualizer enhances the reliability and comprehensibility of the results. These techniques contribute to a more comprehensive understanding of the molecular interactions, driving further research in the field. These findings suggest potential modulatory effects of quercetin, bazedoxifene, estradiol, and quinestrol on the activities of immune checkpoint inhibitors, at least on the PD-1/PD-L1 axis.