Mutation data for this study were collected from cBioPortal (Cerami et al., 2012) developed at Memorial Sloan Kettering Cancer Center (MSKCC) which hosts data from The Cancer Genome Atlas (TCGA) (Weinstein et al., 2013), National Institute of Health (NIH). TCGA catalogs the genetic mutations responsible for different cancers using genome sequencing. TCGA was supervised by the National Cancer Institute’s Center for Cancer Genomics and the National Human Genome Research Institute. Out of all the listed mutations in ABCB1, ABCC1, and ABCG2 for endometrial cancer, only mutations supported by at least two mutation functional impact prediction methods (Mutation Assessor, SIFT and PolyPhen-2) for their deleterious or damaging impact were selected (Ng and Henikoff, 2001; Adzhubei et al., 2010; Reva et al., 2011) (Supplementary Material).

The three-dimensional structures for the ABCB1, ABCC1, and ABCG2 transporters were taken from the Research Collaboratory for Structural Bioinformatics (RCSB) Protein Data Bank. (Gao et al., 2013; Kowal et al., 2021; de Bruijn et al., 2023). PDB files for docking were taken from the RCSB with ID 7NFD(31) for ABCG2, 7A69 (Nosol et al., 2020) for ABCB1, and 6UY0 (Wang et al., 2020) for ABCC1. The mitoxantrone structure was taken from ChEMBL (Gaulton et al., 2012; Zdrazil et al., 2024). The list of residues that are known to interact with mitoxantrone or other drug molecules for ABCG2, ABCB1, and ABCC1 was used during molecular docking. Molecular docking was performed between ABC transporters and mitoxantrone individually using LeDock (Liu and Xu, 2019), and the complex structures were visualized for interactions in the 2D mode using LigPlus (Laskowski and Swindells, 2011). Three-dimensional visualization was performed using PyMOL (PyMOL, 2015; Yuan et al., 2017).

All the mutations that were considered for the analysis based on their possible functional impact in ABCB1, ABCC1, and ABCG2 were further classified based on their location in the three-dimensional structure as follows: in the drug binding/transport channel, at the entry or exit, the membrane-facing regions, and the nucleotide- or ATP-binding regions. We have also taken inputs from the annotations provided at the cBioPortal for this purpose (Gao et al., 2013).

After review of the literature, a list of residues that have been shown to interact with the drugs or substrate molecules was curated for all the three ABC transporters. We found a total of 22 residues in ABCB1, 27 residues in ABCC1, and 28 residues in ABCG2, as shown in the first column of Table 1, Table 2, and Table 3, respectively. These residues were then compared with the TCGA mutation data for any potential matches, wherein we observed six matches in ABCB1 (W232L, A987T, X235_splice, G872V, A230T, and S993A), six matches in ABCC1 (M1092V, G1319R, A438V, V1325M, C1208Y, and V776M), and three matches in ABCG2 (L405M, E446N, and L595I), as shown in the second column of Tables 1–3, respectively. These matches were significant for determining the location of our grid box for molecular docking.

The PDB structure taken for ABCG2, 7NFD had already mitoxantrone bound to it, making it a reference for our results. The comparison with complex structures generated by LeDock (Liu and Xu, 2019) has shown similar amino acid residues taking part in the molecular interaction. This result allowed us to use the same docking methodology to build all the complex structures.

For structural level interaction studies between ABC transporters and substrate/drug molecules, molecular docking was performed for the selected PDB structures of ABCB1, ABCC1, ABCG2, and mitoxantrone using LeDock (Liu and Xu, 2019). After molecular docking, the complexes were then analyzed for visualizing molecular interactions between mitoxantrone and amino acid residues using a LigPlot (Laskowski and Swindells, 2011)-based two-dimensional interaction diagram (Figure 4; Figure 5; Figure 6).

In Figure 4, the structure of the complex between ABCB1 and mitoxantrone can be seen. Novel residues found to be interacting with mitoxantrone were M192, Q195, F343, Q347, M878, and T945. All the interacting residues are also listed in the third column of Table 1, and the interactions can be seen in a two-dimensional representation on the right side in Figure 4. Similarly, in Figure 5, the structure of the complex between ABCC1 and mitoxantrone can be seen. Novel residues found to be interacting with mitoxantrone were F389, Q432, M435, T439, S1084, and M1085. All the interacting residues are also listed in the third column of Table 2, and the interactions can be seen in the two-dimensional representation on the right side in Figure 5. Figure 6 shows the structure of the complex formed between ABCG2 and mitoxantrone. The two-dimensional visualization of interactions can be seen on the right side in Figure 6. Out of all the interactions observed, many amino acid residues were similar to the residues already reported to interact with the ABC transporters. However, few novel residues S420, Q424, A427, G428, S443, A606, T607, Y613, and K616 were also shown to interact with mitoxantrone, and these residues along with all the interacting residues that were observed are also listed in the third column of Table 3.

The description of mutations in ABCB1, ABCC1, and ABCG2 that are also present in other different cancers (such as cancers of the brain, lung, colon, prostate, and kidney) in addition to endometrial cancer is provided in Table 4.

We found that three residues D77, V974, and R958 found on the exit site of the channel were shown to be mutated into hydrophobic residues. All these mutations will lead to the increase of drug efflux because it is known that ABCB1 majorly transports hydrophobic and amphipathic compounds, and all these mutations will enhance ABCB1’s capability to transport such compounds that can possibly lead to chemoresistance (Figure 7; Table 5).

Mutations A987T, G872V, A230T, S237Y, and D188G are located close to the binding site of the drug molecule inside the channel, and mutations A819T, A246V, and R789Q indicate that it will lead to an increase in the amphiphilic nature of the channel that can lead to binding of diverse drug compounds and seems important in explaining the polyspecificity in ABCB1 important for chemoresistance. The R789Q mutation was also shown to be important in the drug response as per ClinVar (Landrum et al., 2020). Mutations S1177F, V801A, and G892C are present close to the entry and intramembrane region, with the closest being S1177F. S1177F and V801A can lead to an increase in hydrophobicity and entry volume of the channel, which is important in entry of the drug from the intracellular environment, making it more efficient, which in turn helps in chemoresistance (Figure 7; Table 5).

Mutations E566K and D562G are very closely located on the entry from the intracellular side. Common drugs like paclitaxel and cyclosporine A (known to have an effect in ABCB1) contains hydroxyl and keto group and aspartic acid residues, respectively, which are negatively charged; therefore, due to mutations from glutamate to lysine, they can effectively be impacted via charged mutated and hydrogen bonding interactions that can lead to better drug efflux. The other mutation D562G can be important in increasing the volume of the intracellular entry side, important for accessing the drug-binding channel (Figure 7; Table 5) (Sissung et al., 2006; Chang et al., 2009; García et al., 2013; Zhang et al., 2013; Gao et al., 2014; Damont et al., 2016; Vallo et al., 2017; van Eerden et al., 2021).

According to the hydrophobic match/mismatch theory, transmembrane proteins avoid unfavorable exposure of the hydrophobic surface present on the transmembrane region of the protein to a hydrophilic environment present on the lipid bilayer surfaces. Biological membranes tend to adapt to such mismatch by expanding and shrinking. Lipid bilayers get adjusted elastically to the hydrophobic length of the transmembrane helices where a decreased hydrophobic transmembrane length would lead to shrinkage of the lipid bilayer along the transmembrane region, reducing the curvature of the overall membrane, whereas increased hydrophobic transmembrane length would lead to elongation of the bilipid layer along the transmembrane region, increasing the curvature of the overall membrane. We discussed the possible effects of this phenomenon with respect to mutations in the ABCB1, ABCC1, and ABCG2 transporters (Weiss et al., 2003; Janmey and Kinnunen, 2006; Soubias et al., 2008; de Jesus and Allen, 2013; Engberg et al., 2022).

With mutations I848S, L843I, and N842H, here threonine to serine will change from non-polar to polar residues, leucine to isoleucine will change from large hydrophobic residues to smaller hydrophobic residues, and asparagine to histidine will change from polar to positively charged amino acids. All these three mutations can lead to a favorable interaction with the non-polar membrane component and the polar component of the cholesterol molecule. For example, asparagine to histidine mutation will introduce an imidazole side chain which can interact favorably with the polar part cholesterol and non-polar part of the phospholipid membrane. Overall, these mutations will help in increasing the fluidity of the cell membrane.

Other mutations like S766Y, G763V, and S298Y would lead to an increase in the non-polar interaction between these residues and the non-polar part of the membrane. However, the hydroxyl group present in tyrosine can also interact favorably with cholesterol molecules. These mutations can lead to a probable increase in the rigidity of the membrane. With mutations R276M and G269E, arginine to methionine will change from charged to non-polar residues, which will lead to favorable interactions for non-polar membrane components. With mutation G269E, glycine to glutamic acid will lead to favorable interactions with cholesterol molecules present on the membrane, important for maintaining the fluidity and stability of the membrane. With mutation Q799H, glutamine to histidine will favor interactions with cholesterol molecules, leading to an increase in fluidity and stability of the membrane.

Q158H, R489H/C, and N494H mutations from polar to positively charged residues like Q158H and N494H can lead to favorable interactions with cholesterol and other polar molecules present on the membrane, important for maintaining the membrane fluidity and stability. On the other hand, R489H and R41H mutations will lead to a favorable interaction with cholesterol and polar molecules, and also an imidazole ring can interact favorably with the non-polar components of cholesterol and phospholipid membrane, signifying a dual role in histidine. Mutation F697C which is facing outwards toward the membrane may favorably interact with cholesterol and polar molecules required for maintaining the fluidity of the cell membrane and mutation R929I with hydrophobic nature would have favorable interactions with the non-polar component of the membrane and cholesterol.

The mutations L843I, S766Y, G763V, S298Y, R489H/C, F697C, and R929I present in the transmembrane region of ABCB1 would result in a change in hydrophobic interactions, which can modulate the hydrophobic mismatch. This would lead to a change in membrane curvature and/or fluidity affecting the ABCB1 conformational dynamics important for the drug transport and efflux.

In the ATP-binding site 1, we observed mutations V437A and R467W, both of which will lead to an enhanced ATP binding or release through the Walker motif A and B. Since V437A will lead to decrease in the side chain volume, it will allow more conformational freedom for the adjacent residues T435 of the Walker motif A. The R467W mutation will lead to an increase in the volume, and due to its hydrophobic nature, it would lead to reorientation of the R441 residue located opposite to that, which can restrict the conformational freedom of the Walker A motif important in ATP binding and hydrolysis (Loo et al., 2002; Orelle et al., 2003; Rai et al., 2006; Szöllősi et al., 2020).

V472M and G604E mutation will lead to an increase in the side chain volume, leading to the restriction of movement of the Walker A motif, thus favoring the ATP bound conformation which may have importance in chemoresistance. G449R mutation to the positively charged residue can form favorable interactions with adjacent negatively charged glutamic acid residues (E448), which will in turn lead to stabilization of an adjacent long loop which is directly connected to the Walker A motif. Mutations R543H and R547H which are located adjacent to each other upon mutation to histidine will lead to overall greater conformational freedom, allowing an optimal interaction between Walker A motif and Q loop for better ATP binding and hydrolysis. The R543H mutation was also shown to be important in the drug response as per ClinVar (Landrum et al., 2020). Mutations K550N and T582A will lead to a decrease in the overall volume of the side chain of amino acid residues, allowing more conformational freedom for the interaction between Walker A motif and ATP for optimal binding and hydrolysis.

Mutation R1188C located on the ATP-binding domain 2 is present in the vicinity of Q loop and Walker B motif; therefore, the mutation from R to C will allow more conformational freedom for better interaction and thus better binding with ATP molecules. The G604E mutation, which is located close to the cystine C431 residue of the Walker A motif, may lead to formation of polar interactions with C431 and would restrict the movement of the Walker A motif that may stabilize the ATP bound form of the ATP-binding domain. Mutations in NBD regions can be targeted to inhibit or reduce the ATP hydrolysis which is important for drug transport and hence chemoresistance.

D355Y mutation present on the exit site of this transporter may facilitate efficient efflux of diverse drug compounds since the side chain of tyrosine containing the hydroxyl group can act as a hydrogen bond donor or acceptor. Furthermore, the aromatic side chain of tyrosine can facilitate a hydrophobic interaction with diverse drug compounds. V584A mutation which is present on the opposite site of the D335 mutation will lead to increase in exit cavity volume, leading to efficient drug efflux (Figure 7; Table 6).

Mutation Q1239H located close to the binding site will allow both the charge-mediated and non-polar interaction with diverse drug compounds. Since this mutation is located just above the Y1242 residue of the ABCC1 binding pocket, it can also stabilize the binding pocket through pi–pi stacking between the histidine and tyrosine residues. Thus overall, this mutation may enhance the efflux of diverse drug compounds. Mutation M1093V takes place in the H-pocket which comprises W1245, Y1242, M1093, T550, and W553. Upon mutation to valine, there will be a reduction in the site chain volume; therefore, the H-pocket can accommodate binding of both smaller and larger drug molecules mediated by a hydrophobic interaction (Figure 7; Table 6).

Mutations R1142C present on the channel lining will lead to an increase in the volume of the channel and will also facilitate the transport of diverse amphipathic drug compounds. E1144K mutation, which is located in the vicinity of R1146, K1139, and R1140, will lead to repulsion due to like charges and, thus, may lead to an increase in channel volume and flexibility.

The R1148H mutation is again located on the channel lining. Histidine’s imidazole ring would lead to better non-polar interactions in addition to charge-mediated interactions with diverse amphipathic drug molecules. It would also lead to the reduction in the side chain volume, resulting in a probable increase in channel volume, favoring efficient drug efflux. In residue M495, it is found to be mutated in two different cancer samples to threonine and valine. In both cases, it will lead to a decrease in the volume of the side chain, leading to easy efflux of diverse amphipathic drug molecules. R501Q mutation will lead to a decrease in the side chain volume without losing the capability of forming a polar interaction with drug molecules for efficient efflux. G421R mutation located adjacent to E422 might lead to favorable charge-mediated interaction, leading to the stabilizing of the helix which forms the efflux channel (Figure 7; Table 6).

The A398T mutation is surrounded by residues V253, V256, and L257. So this mutation will change the hydrophobic nature of the side chain to polar and, therefore, may form unfavorable interactions with the surrounding residues while pushing itself toward the drug efflux channel, thus facilitating the interaction with amphipathic drug molecules. E507K mutation is located in close proximity to K503 and is involved in the formation of polar contact. Thus, this mutation leading to a lysine residue in this position pushes this residue toward the entry of the channel, which may be important in regulating the drug axis and efflux.

In the NBD2 region of ABCC1, mutation V1326M will lead to increase in the hydrophobic side chain volume, leading to better packing with the Walker A motif and surrounding hydrophobic residues.

Residues V776 and L778 are sandwiched between LSGGQ and Walker motif B in the NBD1 region. Mutations V776M/M and L778M, where the residues are directly interacting with the LSGGQ motif, upon mutation to methionine, will lead to a better interaction with the LSGGQ motif and the other surrounding residues. G844S, A690S, and A690T are located close to Walker motif A in NBD1. All these mutations will lead to a hydrophobic to polar side chain that will lead to a favorable polar interaction with the charged and polar residue of Walker motif A.

The G971R mutation will lead to formation of polar interactions with cholesterol and other polar molecules present in the phospholipid membrane required for maintaining membrane fluidity. G226W mutation located in the vicinity of two other tryptophan residues W223 and W242 will lead to the formation of strong hydrophobic clusters, thus stabilizing the L₀ region (205–268) which is known to be essential for folding of ABCC1 (Weinstein et al., 2013). P1121S and G1123D both are located facing the membrane. These mutations will lead to a better interaction with cholesterol and the polar component of the phospholipid membrane, thus providing membrane fluidity required for efficient drug efflux.

We observed a higher number of key mutations in the NBD1 region as compared to the NBD2 region of ABCC1. It is reported that in ABCC1, only the NBD2 region is catalytically active (Sajid et al., 2023). Therefore, a higher number of mutations in NBD1 might be important in retaining the activity of NBD2, which is important for drug transport, while increasing the binding affinity of the NBD1 region through mutations.

R482K, D217Y, and G80R mutations are located close to the entry site from the intracellular side of ABCG2. R482K will lead to an increase in channel volume without affecting the formation of a charge-mediated interaction with diverse drug compounds, while mutation of D217Y to tyrosine can facilitate the efflux of diverse amphipathic drug molecules through side-chain flipping mediated by favorable interactions. G80R mutation located to arginine provides a stable interaction with the charged substrate and drug molecule (Figure 7; Table 7).

The E138K mutation would probably make contact with the hydrophilic head of the membrane which is often negatively charged; therefore, this mutation will facilitate favorable charge-mediated interactions. L140V mutation will allow more flexible hydrophobic interactions with the nearby residues such as I161, thereby helping modulate the flexibility of this region. K157T mutation will lead to formation of a better interaction with nearby residues such as T150, N154, and M152, therefore playing a crucial role for enhancing the stability of this whole region. Mutation I161S will lead to the formation of favorable polar interactions with E138 and K157 (Figure 7; Table 7).

Mutation L648F facing toward the membrane region and located in the vicinity of F650 will lead to stabilization of the helix through hydrophobic interactions and can also interact favorably with the non-polar region of the phospholipid membrane. Mutation K653E facing toward the membrane region may interact favorably with the phospholipid membrane and also the polar molecule present in the phospholipid membrane and therefore may be crucial for membrane stability.

In ABCG2, mutations L140V, K157T, and L648F changing to hydrophobic residues may lead to hydrophobic matching by increasing and elongating the bilipid layer across the transmembrane region, contributing in its flexibility and hence important in drug transport and efflux.

The Y247C, A270T, E285K, and A304D mutations are located on the entry region on the intracellular side. Mutations Y247C and A304D, wherein A304D mutation also becomes larger due to an increase in the side chain volume, will allow flexible polar interactions with diverse drug substrates. The A270T mutation will replace a smaller residue with a larger residue therefore facilitating interaction with amphipathic drug molecules. E285K mutation will lead favorable interactions with nearby negatively charged residues such as D296, D292, and D301.