Molecular mechanisms underlying HRK interaction with BCL-XL and BCL-2 reveal specificity determinants for BH3 mimetics

Summary

BH3 mimetics targeting the BCL-2 family hold broad promise for cancer therapy. High similarity between the anti-apoptotic proteins BCL-XL and BCL-2 challenges the engineering of selective inhibitors. The BH3-only protein HRK is a natural selective inhibitor of BCL-XL and to a less extent of BCL-2. The detailed interaction mechanism remains elusive. Our structural and mutational analyses show that the discrepant conformational changes and non-conserved residues in the α2-α3 region are crucial for the preferential binding between BCL-XL and HRK. BCL-XL tolerates hydrophilic Thr33 or hydrophobic substitutions at the h1 position of HRK, whereas BCL-2 favors hydrophobic interactions, resulting in a weaker affinity for HRK. In addition, we design HRK-derived stapled peptides with improved helicity and activity against BCL-XL and BCL-2, and further elucidate the structural mechanism. Our findings reveal the binding specificity of HRK interactions with BCL-XL and BCL-2, and provide advanced insights into the development of BH3 mimetics.

Graphical abstract

Highlights

• •Structural analyses reveal the interaction property of BCL-XL and BCL-2 to HRK
• •The α2-α3 region of BCL-XL facilitates its more preferential binding to HRK
• •Structural and computational approaches help to design HRK-derived peptides
• •HRK-derived stapled peptides show improved inhibitory activity to BCL-XL and BCL-2

Pharmacology; Natural sciences; Biological sciences; Biochemistry

Introduction

The removal of cells by apoptosis is a highly regulated process essential for the maintenance of tissue homeostasis, development, and defense mechanisms in multicellular organisms.¹ The mitochondrial apoptosis (also known as the intrinsic apoptosis pathway) is tightly controlled by a complicated network of interactions between anti-apoptotic and pro-apoptotic proteins of the B-cell lymphoma-2 (BCL-2) family.² The dysregulation of the BCL-2 family helps malignant cells evade apoptosis and contribute to resistance to multiple treatments.³ The anti-apoptotic members BCL-2 and BCL-XL are overexpressed in a wide range of cancers, and are believed to be highly promising drug targets.⁴ Selective inhibitors targeting BCL-2 or BCL-XL and dual BCL-2/BCL-XL inhibitors have shown excellent preclinical anticancer efficacy in various hematological malignancies and solid tumors.^5,6 Some of them have entered clinical trials, and only the BCL-2 selective inhibitor venetoclax has been approved by the U.S. Food and Drug Administration (FDA).⁷

BCL-2 and BCL-XL are highly conserved anti-apoptotic proteins with 45% sequence identity (Figure S1). They share a similar BCL-2 core structure consisting of eight α-helices and a C-terminal transmembrane domain localized to the mitochondria. They also exert highly similar anti-apoptotic activities by sequestering the pro-apoptotic BH3-only proteins and inhibiting the activation of the apoptotic effector proteins BAX/BAK.⁸ Residues from the BH1, BH2, and BH3 homology domains form a hydrophobic groove on their protein surface.⁹ The hydrophobic groove is responsible for protein interactions with the BH3 domain of pro-apoptotic BCL-2 family proteins and is therefore referred to as the BH3-binding groove.¹⁰ Remarkably, inhibitors of BCL-2 and BCL-XL, such as venetoclax, are engineered to occupy this groove with high affinity, thereby inhibiting their interaction with pro-apoptotic proteins and leading to the activation of pro-apoptotic proteins.¹¹ This class of inhibitors is therefore known as BH3-mimetic drugs.

ABT-737 and Navitoclax (ABT-263), designed to target BCL-2 and BCL-XL with high affinity, represent the first generation of small molecule BH3 mimetics.^12,13 Subsequently, Venetoclax (ABT-199) was developed as a selective inhibitor of BCL-2 and has been approved by the FDA for the treatment of chronic lymphocytic leukemia and acute myeloid leukemia.¹⁴ BCL-XL-selective inhibitor, such as A1331842,¹⁵ has also been developed. Despite great progress, concerns about selectivity, resistance, and toxicity still persist and limit their widespread clinical use of BH3 mimetics.¹⁶ Meanwhile, stabilized α-helices of BH3 domains (SAHBs), which are derived from natural BH3 domains (e.g., BID, BIM), exert potential anticancer functions by blocking anti-apoptotic proteins or directly activating BAK/BAX.^{17,18,19,20,21} They are often used as structural templates or lead tool molecules for drug screening, as well as for BH3 profiling to predict apoptosis sensitivity. Other peptide-based and even small protein-based inhibitors are also under continuous development because they have high potency and selectivity in interfering with protein-protein interactions that are difficult to target.^22,23,24

The pro-apoptotic BH3-only proteins, which share only a single BH3 domain with other BCL-2 family members, are considered to play an essential role in apoptotic initiation.²⁵ They sense and transform diverse cellular stress signaling by activation through transcription or post-translation. These BH3-only proteins have different binding specificity for other multi-domain proteins of the BCL-2 family, and function either as direct activators of BAK/BAX or/and as sensitizers of anti-apoptotic proteins.²⁶ HRK (harakiri, DP5) is a BH3-only protein that was firstly defined as a depressor interacting with BCL-2 and BCL-XL, but not with other members.^27,28 HRK is expressed in central and peripheral nervous systems, and has a critical role in neuronal cell death.^29,30,31 Studies have also shown its role in some cancers.^32,33,34 In prostate cancer cells, HRK is a critical effector of 2-methoxyestradiol-induced apoptosis, which displaced BAK from the complex with BCL-XL.³⁵ HRK expression induce glioblastoma cell apoptosis through regulating BCL-2 and/or BCL-XL activity.³⁶ In addition, HRK and BCL-XL interaction is involved in therapy-induced senescence of melanoma cells.³⁷ Senescence induction leads to HRK downregulation, prompting BCL-XL to inhibit BAK and prevent apoptosis.

It should be emphasized that BCL-2 and BCL-XL are highly similar in sequence and structure, particularly the highly conserved BH3 binding groove. Although some selective BCL-2 or BCL-XL inhibitors have been developed, the structural mechanisms underlying their selectivity have yet to be elucidated in depth. HRK is widely recognized as a BCL-XL-selective inhibitor, and while it displays BCL-2 inhibitory activity in cells, in vitro affinity suggests that HRK does not have a high affinity for BCL-2.^{27,28,38,39,40,41} A detailed understanding of the distinct and overlapping interaction specificity of HRK binding to BCL-XL and BCL-2 is important for the design of advanced classes of selective inhibitors and advancing the clinical development of BH3 mimetic drugs. In this study, binding affinities and structural analysis of the interactions of HRK-derived BH3 peptide with BCL-XL and BCL-2 revealed that conformational dynamics in the α2-α3 helical domains and differential binding modes mediated by non-conserved residues constitute the structural basis for binding specificity. Further structure-based mutational analysis demonstrated that BCL-XL tolerates hydrophilic substitutions at Thr33 in the HRK, whereas BCL-2 exhibits a strict preference for hydrophobic interactions at this site, explaining its weaker affinity for HRK. Guided by the structure of the HRK complex and the computational FoldX procedure, we performed sequence optimization and engineered hydrocarbon-stapled HRK peptides, achieving dual-specific suppression with inhibitory activity in the nanomolar range. These findings not only delineate the atomic-level determinants of binding specificity within the BCL-2 family but also provide insights for the development of next-generation BCL-2 or/and BCL-XL inhibitors.

Results

Structural mechanism of Harakiri binding to B-cell lymphoma-XL

HRK is an intrinsically unstructured protein for which the BH3 region is crucial for interaction with BCL-XL and BCL-2, as well as for its apoptotic activity.^27,42 A typical BH3 domain is featured with an invariant aspartic acid and four hydrophobic residues (h1-h4) that can occupy the corresponding P1-P4 pockets on the hydrophobic BH3-binding groove of multi-domain proteins of the BCL-2 family. Interestingly, HRK-BH3 possesses a hydrophilic Thr33 at the h1 position (Figure S1). In the fluorescence polarization (FP) assays, a 25-residue HRK-BH3 peptide showed potent affinity to the recombinant BCL-XL proteins with an EC₅₀ value of 3.1 nM (Figure 1A). The fusion 6×His tag at the C-terminal of proteins had little influence on the inhibitory activity (Figure S2). To elucidate the detailed binding specificity, we determined the structure of HRK in complex with BCL-XL at a resolution of 2.32 Å (Table S1).

Figure 1.

Structural basis of HRK binding to BCL-XL

(A) FP assay was performed to determine the EC₅₀ values of FITC-labelled HRK-BH3 peptide (FITC-HRK) for BCL-XL. Data are represented as the mean ± SD of n = 3 independent experiments.

(B) The overall structure of HRK-BH3 with BCL-XL. BCL-XL is shown as a green cartoon and a white surface. HRK-BH3 is shown as a yellow cartoon with the electron density map (2mFo-DFc) at 1σ.

(C) Superposition of BCL-XL/HRK-BH3 structure with apo BCL-XL structure (PDB: 1MAZ, pink).

(D) BCL-XL residues involved in HRK binding. The P1-P4 pockets that interact with h1-h4 residues of HRK are colored by violet, salmon, pale green, and slate, respectively. The residues located in the conserved α4-α5 BH1 loop are colored by cyan.

(E–G) Detailed interactions between HRK-BH3 and BCL-XL. Key residues are shown as sticks with BCL-XL residues marked in black font and HRK residues in red font. The black lines and red lines, respectively represent hydrogen bond and salt bridge interactions. Water molecule is shown as red spheres. Also see Figures S1 and S2 and Table S1.

HRK-BH3 peptide is inserted into the hydrophobic groove of BCL-XL as expected (Figures 1B and 1C). It covers a BCL-XL surface of approximately 2100 Ų (Figure 1D). The Thr33 residue and three leucine residues (Leu37, Leu40, Leu44) at positions h1-h4 were oriented into the four hydrophobic pockets of BCL-XL. The invariant Asp42 and other polar residues (Gln31, Arg36, Lys38, Glu43, and His45) are arranged to make hydrophilic interactions with both sides of the BCL-XL groove. In detail, Asp42 establishes conserved salt bridge contacts with the highly conserved residue Arg139 on the α4-α5 BH1 loop of BCL-XL, as well as a hydrogen bond with Asn136 (Figure 1E). Gln31 and Lys38 provide additional polar contacts with BCL-XL residues Glu129 and Asp133. In addition, residue Tyr195 at the edge of BCL-XL P4 pocket interacts with His45 and Thr48 at the rear of the HRK helix via water-mediated and direct hydrogen bonds (Figure 1F). Specifically, the α3 helix of BCL-XL moves and becomes less helical upon HRK-BH3 binding (Figure 1C). The hydrophobic Phe105 is flipped outwards and forms a cation-π interaction with the positively charged HRK residue Arg36, while Tyr101 is pointed toward the interior of BCL-XL and interacts with HRK Arg36 and the h1 residue Thr33 through water-mediated hydrogen bonds (Figure 1G). A salt bridge interaction is also observed between BCL-XL Arg100 and HRK Glu43. Overall, these extensive hydrophobic and hydrophilic interactions stabilize the BCL-XL/HRK interactions.

The differences in the binding patterns of Harakiri to B-cell lymphoma-XL and B-cell lymphoma-2

Previous studies have employed isothermal titration calorimetry (ITC), surface plasmon resonance (SPR), and FP assay to detect the interaction of HRK peptides with anti-apoptotic proteins.^38,39,40,41 Although the results showed differences in the K_d or EC₅₀ values of these HRK-derived BH3 peptides for binding BCL-XL, probably due to different peptide lengths and different detection methods, HRK peptides had a higher affinity for BCL-XL than other anti-apoptotic proteins studied (including BCL-2, MCL-1, A1, and BCL-w). Specifically, the affinity of these HRK peptides for BCL-XL is more than an order of magnitude higher than that of the highly homologous BCL-2, although intracellular studies have demonstrated that HRK has inhibitory activity against BCL-2.^27,28 In our fluorescence polarization (FP) assays (Figure 2A), the EC₅₀ value of the 25-residue HRK-BH3 peptide for recombinant BCL-2 protein was 30 nM, which was 10-fold weaker than the affinity of BCL-XL. To analyze the differences in the binding patterns of HRK to BCL-XL and BCL-2, we modeled the BCL-2/HRK-BH3 structure using AlphaFold (BCL-2/HRK-BH3 AF structure) (Figure 2B). Most of the hydrophobic and hydrophilic residues within the groove are conserved between BCL-2 and BCL-XL, while the pocket surfaces are somewhat different (Figure S3). It seems that HRK could form similar polar contacts with the conserved BCL-2 residues located in the α4-α5 BH1 loop and the tyrosine (Try202) in the C-terminus of the α8 BH2 helix (Figures 2C and 2D). In particular, the BCL-2/HRK-BH3 model suggests that the α3 helix of BCL-2 remains well folded, and the residues Tyr108 and Phe112 located in the α2-α3 corner do not undergo orientation flip, which is different from BCL-XL (Figure 2B). Correspondingly, the conserved Tyr108 can contribute to hydrogen bond interaction with HRK Glu43, whereas the non-conserved Asp111 (Ala104 in BCL-XL) has the potential to form an electrostatic interaction with HRK Arg36 (Figure 1E). Overall, these differences in conformational changes, as well as discrepant residues and interactions contribute to the different binding affinity of HRK-BH3 toward BCL-2 and BCL-XL. Noteworthy, the predicted models may not reflect subtle conformational differences relevant to BCL-2/HRK-BH3 interactions. The key interactions it predicts need to be validated experimentally.

Figure 2.

Structural basis of HRK binding to BCL-2

(A) FP assay was performed to determine the EC₅₀ values of FITC-HRK for BCL-2. Data are represented as the mean ± SD of n = 3 independent experiments.

(B) Superposition of BCL-XL/HRK-BH3 structure with BCL-2/HRK-BH3 AF model. The BCL-XL/HRK-BH3 structure is represented by green cartoons. In the BCL-2/HRK-BH3 AF model, BCL-2 and HRK-BH3 are colored gray and yellow, respectively.

(C–E) The detailed interaction in the BCL-2/HRK-BH3 AF model. Key residues are shown as sticks with BCL-2 residues marked in black font and HRK residues in red font. The black lines and red lines, respectively, represent hydrogen bond and salt bridge interactions. Also see Figure S3.

Effect of B-cell lymphoma-2 and B-cell lymphoma-XL mutations on the interaction with Harakiri-BH3

To better understand the hot spots of BCL-2 and BCL-XL for HRK binding, we performed computational alanine-scanning using BAlaS⁴³ using the BCL-XL/HRK-BH3 structure and the BCL-2/HRK-BH3 model. The results suggested that BCL-XL residues Phe97, Tyr101, Phe105, Glu129, Arg139, and Tyr195 were major contributors to the interaction energy (ΔΔGs>5 kJ/mol). For HRK, residues Arg36, Leu37, Lys38, Leu40, Asp42, Glu43 and Leu44 were major hot spots (Figure 3A). Similar results were obtained for BCL-2/HRK-BH3 AF structure, except for Phe112 and Glu136 (corresponding to Phe105 and Glu129 in BCL-XL). Since the interactions of HRK with the α2 and α3 helices of BCL-XL and BCL-2 are more differentiated (Figures 1 and 2), we focused on the residues Tyr101, Phe105, Arg100, and the non-conserved residue Ala104 (corresponding to Tyr108, Phe105, Arg107, and Asp111 in BCL-2). These residues of BCL-2 or BCL-XL were mutated, respectively, and the mutants were purified. Their affinity to HRK-BH3 was then analyzed by FP assay.

Figure 3.

Effect of BCL-2 and BCL-XL mutations on the binding affinity to HRK-BH3

(A) The interaction energy (ΔΔGs) predicts hot-spot residues of BCL-XL or BCL-2. Equivalent residues of BCL-XL and BCL-2 are labeled as BCL-XL/BCL-2 below the horizontal axis. The BCL-XL residues with ΔΔGs value less than 0.5 are not shown, and the BCL-XL residues with ΔΔGs value more than 5 are highlighted in red square.

(B) Binding affinity of BCL-XL mutants to FITC-HRK was determined by FP assay. The EC₅₀ values were obtained by nonlinear regression analysis. Data are represented as the mean ± SD for 3–4 independent experiments.

(C) Surface representation of these mutated residues in the BCL-XL/HRK-BH3 structure. BCL-XL is shown as a white surface with the studied BCL-XL residues colored and labeled as indicated. HRK-BH3 is shown as a yellow cartoon. The black lines and red lines, respectively, represent hydrogen bond and salt bridge interactions. Water molecule is shown as red spheres.

(D) Binding affinity of BCL-2 mutants to FITC-HRK was determined by FP assay. Data are represented as the mean ± SD for 3–4 independent experiments.

(E) Comparison of the effect of BCL-XL or BCL-2 mutations on their binding affinity to HRK-BH3. The fold change in HRK-BH3 binding affinity of BCL-XL or BCL-2 mutants versus the wild type is shown in histograms.

The mutants BCL-XL R139A and BCL-2 R146A were also tested as controls because the invariant arginine in the α4-α5 BH1 loop is known to be an essential site for ligand binding.⁹ Changes in affinity are normalized against their respective wild-type (WT) counterparts (BCL-XL WT: EC₅₀ = 3.1 nM; BCL-2 WT: EC₅₀ = 30 nM). Compared to BCL-XL WT, the mutant F105A caused an approximately 10-fold decrease in the affinity for HRK-BH3, and Y101A and E129A caused a 3-fold decrease, indicating their important role in interaction with HRK (Figures 3B and 3C). Though Arg100 forms an electrostatic interaction with HRK in the complex structure, BCL-XL R100A had no effect on the affinity. In addition, the substitution of BCL-XL Ala104 with aspartic acid reduced its affinity for HRK-BH3 by more than 3-fold. However, among these corresponding residues of BCL-2, only the F112A and E129A mutants decreased its affinity for HRK-BH3 by about 2–3-folds compared to BCL-2 WT, whereas R107A and D111A had little influence (Figure 3D). The affinity of BCL-2 Y108A for HRK-BH3 was even slightly increased. In comparison, mutations in these conserved and non-conserved residues in the α2 and α3 helices had a greater effect on BCL-XL/HRK interaction than on BCL-2/HRK interaction (Figure 3E).

Structure-based optimization of Harakiri-derived BH3 peptides

BH3-only proteins and their α-helical BH3 domains show binding specificity to multi-domain BCL-2 family proteins and exhibit natural death-inducing functions.⁴⁴ The derived BH3 peptides are applied as peptide inhibitors that target anti-apoptotic proteins with high affinities and specificities, and also as useful tools in BH3 profiling assays for predicting dependency on specific anti-apoptotic proteins and sensitivity to the BH3 mimetics. HRK is a natural inhibitor of BCL-XL and BCL-2, and its efficacy against BCL-XL is approximately 10-fold higher than BCL-2 (Figures 1 and 2). It could be a potential candidate for the design of BH3 peptide inhibitors. To further improve the activity and selectivity of HRK-derived peptides toward BCL-XL or/and BCL-2, we rationally optimized the peptide sequences on the basis of the determined BCL-XL/HRK-BH3 structure and artificial intelligence-assisted mutation analysis.

The computational tool FoldX was used to predict HRK-BH3 mutations for improved stability and binding affinity to BCL-XL. Potential mutation of each HRK-BH3 residue to the other nineteen amino acids was generated and computed by FoldX. The change in free energy of the mutant protein and the wild-type protein (ΔΔG) provides an estimate of the impact of the mutation on the protein’s thermodynamic stability and interaction energy. The top ten stabilizing mutations (negative ΔΔG) were located at four sites (Thr33, Arg36, His45, and Thr48) with ΔΔG< −1.8 kJ/mol (Figure 4A). Interestingly, the computational alanine-scanning analysis indicated that Thr33 at position h1 contributed less to the interaction energy (Figure S4), probably due to its hydrophilicity affecting the hydrophobic interactions in the P1 pocket. The other three residues contribute moderately (Arg36) or weakly (His45 and Thr48) to the interaction energy. Sequence optimization on these interfacial residues has the possibility of providing an increase in affinity. At the same time, the BCL-2/HRK-BH3 AF structure was also used to predict mutation energy, and the results suggested that these ten mutations also had the potential to stabilize the BCL-2/HRK-BH3 interactions in varying degrees.

Figure 4.

Optimization and binding affinity of HRK-derived peptides to BCL-2 and BCL-XL

(A) The computational tool FoldX predicts HRK-BH3 mutations for improved stability and binding affinity to BCL-XL. The difference between the free energy of the designed HRK mutant (ΔG^mutant) and that of HRK WT (ΔG^WT) (ΔΔG = ΔG^mutant − ΔG^WT) was shown. The IC₅₀ or K_i values of these peptides against BCL-2 and BCL-XL were determined by competitive FP assay.

(B–G) The fitting curves of competitive FP assay. In the competitive FP assay, non-fluorescent BH3 peptides were used as a competitor for FITC-HRK to interact with BCL-XL (B–D) or BCL-2 (E–G). The IC₅₀ values were determined by nonlinear regression analysis, and then Ki was calculated as described in STAR Methods. Data are represented as the mean ± SD of n = 3 independent experiments. Also see Figure S4.

Subsequently, ten HRK peptides with the top ten indicated stabilizing mutations were synthesized. Competition FP assays were performed to measure their inhibitory activity toward BCL-XL and BCL-2 (Figure 4). A wild-type HRK-BH3 peptide (HRK-WT) had an IC₅₀ of 38 nM and 96 nM for BCL-XL and BCL-2, corresponding to competitive inhibition constant (K_i) values of 1.8 nM and 36 nM, respectively. Among these peptides, only HRKM1 and HRKM7, carrying the substitution of Thr33 by Ile or Leu, showed a slight increase in potency to compete for BCL-XL. However, it appears that BCL-2 is more dependent on the hydrophobicity within the P1 pocket. The replacement of HRK Thr33 at h1 position with hydrophobic residues T33I, T33M or T33L significantly increased its potency to compete for BCL-2 compared to HRK-WT. The peptides with H45F or T48I mutations also had a slightly increased potency toward BCL-2. It is noticeable that the replacement of HRK Arg36 with hydrophobic residues R36M, R36W, R36L or R36F reduced the potency against BCL-XL (>10-fold) and BCL-2 (∼2-fold). This indicated that HRK Arg36 is actually an important residue for the interactions with BCL-XL and BCL-2. Remarkably, the peptide HRKM7 (T33L) showed potent inhibitory activity against both BCL-XL and BCL-2, which may be a potential candidate for a dual inhibitor of BCL-XL and BCL-2, whereas the original HRK-BH3 is more selective in inhibiting BCL-XL.

Activities of stapled Harakiri-derived peptides toward B-cell lymphoma-XL and B-cell lymphoma-2

Natural short peptides are usually unstructured in solution, which makes them susceptible to proteolysis. Covalent linkage of i, i+4 or i, i+7 by cross-linking is a common method to obtain SAHBs and therefore confer peptides stabilized α-helical structure and better biological activity. On the basis of the peptide HRKM7, we tried the hydrocarbon-stapled method by incorporating a pair of non-natural amino acids containing olefin tethers at the indicated i, i+4 positions of the non-interacting residues, followed by ruthenium-catalyzed olefin metathesis (Figure 1, Figure 2A and S5). Secondary structure of these peptides was then analyzed by circular dichroism (CD). The helicity of the constrained peptides HRKM7-S1 and HRKM7-S2 in solution was 50.6% and 56.9%, respectively, which was significantly higher than that of the unrestricted HRKM7 (26.2% helicity) (Figure 5B). This suggested that the hydrocarbon stapling is helpful in stabilizing the α-helical conformation of these peptides. Their inhibitory activities against BCL-XL and BCL-2 were measured by competition FP assays (Figure 1, Figure 2C and 5D). All the stapled peptides showed high affinity for BCL-XL and BCL-2, with K_i close to the detection limit of the FP assay (K_i<1 nM). The IC₅₀ values of HRKM7-S1 and HRKM7-S2 against BCL-2 and BCL-XL were around ten nanomolar, which were comparable to those of the unconstrained peptides. HRKM7-S1 showed improved affinity for BCL-2 with an IC₅₀ of 4.9 nM.

Figure 5.

Design and inhibitory activity of the HRK-derived stapled peptides

(A) The HRK-derived SAHBs were designed. A pair of cross-linking non-natural amino acids (X) was substituted at the indicated i, i+4 position and stapled by ruthenium-catalyzed olefin metathesis.

(B) CD spectra of HRKM7 and stapled peptides HRKM7-S1, HRKM7-S2.

(C and D) The inhibitory activity of stapled peptides HRKM7-S1 and HRKM7-S2 to BCL-XL (C) or BCL-2 (D) is determined by competitive FP assay. HRKM7-S1 and HRKM7-S2 were used as a competitor for FITC-HRK to interact with BCL-XL or BCL-2. Data are represented as the mean ± SD of n = 3 independent experiments. Also see Figures S5 and S8.

A detailed binding mechanism of the stapled peptides was analyzed by determining the structure of the BCL-XL/HRKM7-S1 complex at 2.97 Å. Superposition with the BCL-XL/HKR-BH3 structure give an RMSD of about 0.35 Å for Cα atoms (Figure 6A). The notable differences are located in the α3 helix region of BCL-XL, which may be due to the accommodation of the different residue T33L of HKRM7-S1, as well as the constrained helical conformation of the stapled peptide. Correspondingly, some hydrophobic BCL-XL residues, such as Leu108 and Leu112, undergo minor positional shifts and make hydrophobic interactions with the residue Leu33 of HKRM7-S1 (Figure 1, Figure 2B, 6C, and S6). In addition to the conserved polar interactions within the α4-α5 BH1 loop, BCL-XL Glu97 forms a salt bridge with Arg47 of HRKM7-S1, and Tyr101 forms a hydrogen bond with the main chain of HRKM7-S1 to help strengthen the interaction (Figure 6D). Meanwhile, BCL-2/HRKM7-S1 structure was modeled by AlphaFold to help explain the binding mechanism. BCL-2 possesses a compact hydrophobic P1 pocket, and the residue Leu33 of HRKM7-S1 would interact hydrophobically with the BCL-2 residues Met115, Leu119, and Val133 (Figure 1, Figure 2E and S7), thus boosting the binding affinity. Although the stapled peptide is expected to reduce the entropic penalty upon binding, and thus possess the potential to increase binding affinity and reduce degradation by proteases, the constrained helical conformation may also disrupt the local favorable interactions.^45,46 Binding of BCL-XL or BCL-2 to peptides is dynamic. The more stable helical conformation of stapled peptides may also perturb the native binding conformation, resulting in BCL-XL binding to the HRKM7-S1 and HRKM7-S2 with slightly lower affinity than unrestricted HRKM7.

Figure 6.

Structural analysis of the stapled HRKM7 peptide binding to BCL-XL and BCL-2

(A) Superposed structures of BCL-XL/HRKM7-S1 (blue/magenta) with BCL-XL/HRK-BH3 (green/yellow).

(B) Detailed interaction of the P1 pocket in the BCL-XL/HRK-BH3 structure.

(C) Detailed interaction of the P1 pocket in the BCL-XL/HRKM7-S1 structure.

(D) Detailed hydrophilic interaction in the BCL-XL/HRKM7-S1 structure.

(E) Structural model of BCL-2/HRKM7 (gray/magenta) predicts the detailed interaction of the P1 pocket. Key residues are shown as sticks. BCL-XL or BCL-2 residues are marked in black font and HRK residues in red font. The distinct Thr33 and T33L of HRK peptides are shown on the surface in the B, C, and E. The black lines and red lines, respectively, represent hydrogen bonds and salt bridges. Also see Figures S6–S8 and Table S1.

Subsequently, the cellular activity of HRK-derived peptides was evaluated in the NCI-H146 cell line, a BCL-XL and BCL-2 co-dependent small cell lung cancer cell line (Figure S8). Unfortunately, 10 μM of hydrocarbon-stapled HRKM7-S1 had little cytotoxicity to NCI-H146 cells. However, unconstrained HRKM7 with a C-terminal membrane-penetrating peptide R8 (HRKM7-R8) dose-dependently inhibited the growth of H146 cells, with cell death reaching 70–80% at 10 μM. This suggests that the HRKM7 peptide carrying the T33L substitution possesses the potential to inhibit the growth of NCI-H146 cells, whereas the stapled HRKM7-S1 peptide has limited activity, possibly due to cell permeability, and still needs to be optimized.

Discussion

Cancer cells always achieve survival and drug resistance by overexpressing anti-apoptotic proteins. Targeted therapies that block the protein-protein interactions between the BCL-2 family via BH3 mimetics are under pharmacological investigation.^3,8 BCL-2 and BCL-XL share highly conserved structure and a similar binding profile for pro-apoptotic proteins, which challenges the development of highly selective BH3 mimetic inhibitors targeting BCL-2 or BCL-XL. HRK is a less-characterized BH3-only protein. It promotes cell death primarily through neutralizing the anti-apoptotic functions of BCL-XL and also BCL-2 to a less extent.^27,28,38 The elucidation of the binding specificity and interaction mechanism of HRK to BCL-2 and BCL-XL is significant for the development and clinical application of highly potent and selective BH3 mimetics.

Our structural and mutational analysis indicate that the leucine at h2-h4 positions, Asp42, and the other three charged residues (Arg36, Lys38, and Glu43) are the major hot spots for BCL-XL and BCL-2 binding. Interestingly, the high-affinity interaction between BCL-XL and HRK is highly tolerant to a hydrophilic Thr33 or hydrophobic substitutions at h1 position, whereas for BCL-2, HRK Thr33 may be an important reason for the lower selectivity. Substitution of Thr33 with hydrophobic residues T33I, T33M or T33L greatly improves the affinity for BCL-2. A noticeable difference between BCL-XL and BCL-2 that may be related to this phenotype is located in the α2-α3 region. Upon HRK binding, the α3 helix of BCL-XL is less helical, and residues Tyr101 and Phe105 undergo a conformational transition. The transition introduces variations in the P1-P3 pockets of BCL-XL. In the BCL-XL/HRK-BH3 structure, Tyr101 constitutes the edge of P2 pocket and the bottom of P3 pocket, and forms water-mediated interaction with the hydrophilic HRK Thr33 and Arg36. Alternately, Phe-105 forms the edge of the P3 pocket, and establishes the cation-π interaction with HRK Arg36. Mutations of these residues in BCL-XL or HRK resulted in significant changes in affinity, indicating an important role of these contacts for the high-affinity interaction between BCL-XL and HRK. As for BCL-2, the α3 helix probably remains helical, and mutations of the homologous Tyr108 and Phe112 have less influence on binding HRK. Correspondingly, the P1 pocket of BCL-2 is more compact and hydrophobic, which may be one of the reasons why BCL-2 favors a small hydrophobic leucine at the h1 position.

The binding of BH3 peptides always induces dramatic conformational changes in the α2-α4 regions of BCL-XL and BCL-2, along with unfolding or partially unfolding of the α3 helix (Figure S8).^9,44 In comparison with BCL-2, it seems that BCL-XL α3 helix is more flexible to accommodate different ligands, with conformational change in Tyr101 and Phe105 being more common.^{9,47,48,49,50} In the structures of BCL-XL with BIM-BH3, BID-BH3 or BMF-BH3, the site-shifted Tyr101 and Phe105 are also involved in interactions with the polar residues preceding the h2 position.^47,49 Upon PUMA-BH3 binding, BCL-XL has the most dramatic conformational change.⁵⁰ Both α2 and α3 become highly disordered, and His113 in the α3-α4 loop of BCL-XL forms a critical hydrophobic contact with a unique tryptophan residue of PUMA. For BCL-2, the α2-α4 regions also move to accommodate BH3 ligands, and the α3 helix sometimes becomes less helical, whereas the conformational shift in Tyr108 and Phe112 is only observed in the previous structure of BCL-2/BAX (PDB: 2XA0).⁴⁰ In addition, the BCL-XL-selective inhibitor A-1331842 also induces a dramatic conformational change in BCL-XL α3 helix and forms important hydrophobic interactions with Tyr101 and position-flipped Phe105.¹⁵ The conformation is not observed for Navitoclax, a BCL-2 and BCL-XL dual inhibitor.^13,51 These conformational changes and variations in pockets appear to have an important role in the binding selectivity.

Moreover, BCL-XL and BCL-2 possess a markedly discrepant residue in this α2-α3 region, with an Ala104 in BCL-XL and a charged Asp111 in BCL-2. Although structural predictions indicate that the BCL-2 residue Asp111 is able to form a salt bridge interaction with HRK Arg36, whereas BCL-2 mutation D111A had little impact on the affinity. This suggests that the electrostatic interaction of BCL-2 D111 with HRK R36 captured by structural predictions may be dynamic and fragile, and actually contribute less to the complex formation. However, for the high-affinity BCL-XL/HRK-BH3 and BCL-2/BMF-BH3 interactions, the BCL-XL mutation A104D resulted in an approximately 3-fold reduction in affinity for HRK-BH3, while the BCL-2 mutation D111A caused in a dramatic reduction in affinity for BMF-BH3.⁴⁹ In addition, a recent study has reported this discrepant residue (BCL-2 D111/BCL-XL A104) as a switch that may regulate the binding specificity of BCL-XL and BCL-2 for cyclic peptides and the BCL-2-selective inhibitor S55746 through different hydrogen bond interaction and binding interface.²² Overall, these differences in the contacting residues, the topology, and electrostatic character of the binding groove are likely to lead to differences in the affinity of BCL-2 and BCL-XL for binding HRK and some other BH3 ligands.

SAHBs mimic the structure and function of natural α-helical peptides, and are particularly useful in disrupting or modulating protein interactions that are difficult to target with traditional small molecules. Many computational and experimental strategies are engineered to optimize the binding characteristics of peptides or proteins, such as binding affinity and specificity, stability, and cell permeability. For example, hydrocarbon stapling of BID- and BIM-derived peptides has been shown to activate the apoptotic pathway to kill hematologic cancers and overcome apoptotic resistance to chemotherapy and radiation treatments.^17,21,46 Edwards et al. yield a biologically and pharmacologically enhanced BIM SAHB that by replacing a salt bridge with an i, i + 4 hydrocarbon staple.⁵² MS1, which is a BIM-derived peptide with side-chain substitution and targets MCL-1 with high affinity and specificity, is used to detect cellular MCL-1 dependence in BH3 profiling assays.^53,54 Zhang et al. also reported HRK-derived peptides, which were stapled using dibromomaleimide constraining and screened using a combination of homology modeling and virtual alanine scanning.⁵⁵ In this study, by analyzing the binding affinity, structure, and computational data of the interactions of HRK with BCL-XL and BCL-2, we found that the hydrophilic Thr33 or hydrophobic Leu33 in HRK-derived peptides are of great significance for the specificity. Subsequently, we designed two stapled peptides that exhibited highly potent activity for BCL-XL and BCL-2. This structure-based hotspot optimization approach can be easily applied to protein interaction investigations. Although HRKM7 did not show significant anti-cancer activity at the cellular level, which might be related to its cell membrane permeability, it may serve as a lead molecule for subsequent optimization or as a structural model in the search for advanced lead molecules.

When BCL-2-selective inhibitor Venetoclax has been used to treat a number of hematological tumors, the clinical application of BCL-XL inhibitors has been hampered by their toxic side effect, which trigger platelet apoptosis leading to severe thrombocytopenia.³ However, BCL-XL overexpression is associated with Venetoclax resistance, and BCL-XL is highly expressed in some solid tumors, whereas Venetoclax has limited efficacy in solid tumors.⁵ Therefore, researchers have not given up on the development of BCL-XL inhibitors. BCL-XL-specific inhibitors, BCL-2/BCL-XL dual inhibitors, and BCL-XL/MCL-1 dual inhibitors have been reported consecutively.^15,24,56,57 Selective drug delivery, such as VHL-recruiting PROTACs and antibody-drug conjugates, may offer new opportunities.^56,58 Furthermore, these HRK-derived peptides may also show a high affinity for BCL-w, another anti-apoptotic member that exhibits high structural conservation with BCL-XL/BCL-2 in the BH3-binding groove (Figure S1B). It showed a 2- to 10-fold weaker affinity for the BH3 peptides of HRK in previous studies.^38,40 In addition, the improved hydrophobicity of the h1 residue of HRKM7 may be compatible with the preference of the P1 pocket of BCL-w to accommodate hydrophobic residues with bulky side chains.⁵⁹ Such multi-targeting properties may offer synergistic advantages in certain therapeutic contexts.

The interaction modes of BH3 peptides with the BCL-2 partners share high similarity, but also exhibit binding specificity. These similarities and differences are important for understanding the functional mechanisms and designing targeted strategies. In this study, our structure determination and homology modeling of HRK-BH3 with BCL-XL and BCL-2 elucidate the detailed molecular basis of the binding specificity. Additionally, structure-based and AI-assisted mutational analysis facilitate the understanding of the selective determinants of BCL-XL and BCL-2 for BH3 ligands. On the other hand, a previous study has reported that HRK can directly activate BAK.⁶⁰ The binding site is located at an alternative groove formed by BAK α4, α6, and α7 helices, and involves the three leucines at h2-h4 positions of HRK-BH3. Taken together, the information provides advanced insights into the development of selective BH3 mimetic drugs targeting the BCL-2 family.

Limitations of the study

The limitations of this study include the need for a precise structural detail of the BCL-2/HRK interaction and a more mechanistic intracellular study of the HRK-derived peptides. Structural predictions and experimental validation by mutational assays have helped to understand the hotspots of BCL-2/HRK interaction, but the predicted BCL-2/HRK-BH3 model may not account for conformational flexibility or dynamic changes that occur upon binding. The structural determination of the BCL-2/HRK complex (which we have attempted, but have not yet obtained) would help to provide more precise structural details. Additionally, the HRK-derived peptides obtained through structure-based optimization showed potent activity against both BCL-2 and BCL-XL in biochemical assays, but their biological activity, such as cell specificity and cell permeability, needs to be further characterized and optimized.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Hudie Wei (hudie_wei@csu.edu.cn).

Materials availability

Plasmids generated in this study will be available on request through the completion of a material transfer agreement.

Data and code availability

• •The coordinates and structure factors are deposited in the Protein DataBank under the accession codes “PDB: 9LI8” (BCL-XL/HRK-BH3 structure) and “PDB: 9LGU” (BCL-XL/HRKM7-S1 structure), and are publicly available as of the date of publication.
• •This article does not report the original code.
• •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

We thank the staff at BL17B1/BL18U1/BL19U1 beamlines at SSRF of the National Facility for Protein Science in Shanghai (NFPS), Shanghai Advanced Research Institute, Chinese Academy of Sciences, for providing technical support in X-ray diffraction data collection and analysis. The present study was financially supported by the National Natural Science Foundation of China (grants 82470176, 82273496 and 31900880), the Natural Science Foundation of Hunan Province (grants 2023JJ20092 and 2023JJ30863) and Central South University Innovation-Driven Research Program (2023CXQD076).

Author contributions

H.W. conceived the project. J.W. performed the experiments. M.G. and H.W. performed data collection and structure determination. S.D. helped with data analysis. J.W. and H.W. prepared the article.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and virus strains
E. coli BL21(DE3)	TransGen Biotech	Cat# CD601-02
Chemicals, peptides, and recombinant proteins
Isopropyl β-D-1-thiogalactoside (IPTG)	BioFroxx	Cat# 1122GR100
Ni Sepharose 6 Fast Flow	Cytiva	Cat# 17531802
HiTrap Q HP	Cytiva	Cat# 29051325
Superdex 75 10/300 GL	GE Healthcare	Cat# 17-5175-01
Tris	Sigma	Cat# V900866
Sodium Chloride	Sinopharm Chemical Reagent	Cat# 10019318
Kanamycin	Aladdin	Cat# K103024
Imidazole	DINGGUO	Cat# BI170
TCEP	MeilunBio	Cat# MB2601
Magnesium chloride hexahydrate	Sigma	Cat# M9272
Polyethylene glycol 3350	Sigma	Cat# P4338
Tween 20	BioFroxx	Cat# 1247
DTT	DINGGUO	Cat# CD116-5G
RPMI 1640 Medium	Gibco	Cat# C11875500BT
Fetal Bovine Serum Premium	NEWZERUM	Cat# FBS-CP500
Recombinant proteins of BCL-XL and the mutants (sequence in Table S2)	This paper	N/A
Recombinant proteins of BCL-2 and the mutants (sequence in Table S2)	This paper	N/A
HRK peptides	GenScript	N/A
Critical commercial assays
Cell Counting Kit-8	Ecotop	Cat# EK-5103-500T
ClonExpress II One Step Cloning Kit	Vazyme	Cat# C112
KOD Plus mutagenesis kit	TOYOBO	Cat# SMK-101
Deposited data
BCL-XL/HRK-BH3 structure	This paper	PDB: 9LI8
BCL-XL/HRKM7-S1 structure	This paper	PDB: 9LGU
Experimental models: Cell lines
Human: NCI-H146 cells	National Collection of Authenticated Cell Cultures	SCSP-5075; RRID:CVCL_1473
Oligonucleotides
See Table S3 for Oligonucleotides	TsingkeBiotechnology	N/A
Recombinant DNA
pET28a	Novagen	Cat# 69864-3
pET28a C-his-BCL-XL	This paper	N/A
pET28a C-his-BCL-XL R100A	This paper	N/A
pET28a C-his-BCL-XL Y101A	This paper	N/A
pET28a C-his-BCL-XL A104D	This paper	N/A
pET28a C-his-BCL-XL F105A	This paper	N/A
pET28a C-his-BCL-XL E129A	This paper	N/A
pET28a C-his-BCL-XL R139A	This paper	N/A
pET28a C-his-BCL-2	This paper	N/A
pET28a C-his-BCL-2 R107A	This paper	N/A
pET28a C-his-BCL-2 Y108A	This paper	N/A
pET28a C-his-BCL-2 D111A	This paper	N/A
pET28a C-his-BCL-2 F112A	This paper	N/A
pET28a C-his-BCL-2 E136A	This paper	N/A
pET28a C-his-BCL-2 E146A	This paper	N/A
Software and algorithms
HKL2000	Otwinowski et al.61	https://www.hkl-xray.com/hkl-2000
Phenix	Liebschner et al.62	https://phenix-online.org/download
Coot	Emsley et al.63	https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/
PyMOL	Schrodinger, LLC	https://pymol.org/
Graphpad Prism 6.01	GraphPad Software	https://www.graphpad.com/
AlphaFold 3	Abramsonet al.64	https://alphafold.com/
BUDE Alanine Scan	Wood et al.43	https://pragmaticproteindesign.bio.ed.ac.uk/balas/
FoldX 5.0	Delgado et al.65	https://foldxsuite.crg.eu/
CDPro	Sreerama et al.66; Sreerama et al.67	https://www.bmb.colostate.edu/cdpro/

Experimental model and study participant details

Cell lines and cell culture

The human small-cell lung cancer (SCLC) cell line NCI-H146 was sourced from the National Collection of Authenticated Cell Cultures (SCSP-5075). NCI-H146 is an epithelial-like cell that was isolated from the lung of a 59-year-old, White, male with lung carcinoma. Cells were cultured in RPMI 1640 medium (GIBCO) supplemented with 10% fetal bovine serum (NEWZERUM), incubated at 37°C under a humidified 5% CO₂ atmosphere. Routine quality control was performed through continuous monitoring of cellular morphology and functional characteristics. All experiments were performed in accordance with the relevant regulatory standards. Cell line authentication was performed by National Collection of Authenticated Cell Cultures using short tandem repeat (STR) profiling. All cell lines were tested mycoplasma-free.

Method details

Cell viability assays

Cell viability was assessed using the Cell Counting Kit-8 (CCK-8, Ecotop) according to the manufacturer’s guidelines. 2.0 × 10⁴ cells/well were seeded in 96-well microplates and treated 1–10 μM BH3 peptides, utilizing dimethyl sulfoxide (DMSO) as the negative vehicle control. Following 48-h incubation at 37°C under 5% CO₂, 10 μL of CCK-8 reagent was introduced to each well containing 100 μL culture medium. After 60-min incubation at 37°C under 5% CO₂, absorbance was measured at 450 nm using a microplate spectrophotometer.

Plasmids

Human BCL-XL removing the C-terminal 22 residues and human BCL-2 removing the C-terminal 32 residues were cloned into vector pET28a (Novagen, 69864-3) with a C-terminal 6×His-tag (Table S2). All plasmids were performed according to the manufacturer’s instructions for the ClonExpress II One Step Cloning Kit (Vazyme, C112). All mutants were constructed by PCR mutagenesis using the wild-type plasmid as the template following the KOD Plus mutagenesis kit (TOYOBO, SMK-101). All plasmids were confirmed by DNA sequencing (Tsingke).

Peptides

All peptides were synthesized and purchased from GenScript with a purity of over 95%, as analyzed by HPLC. The peptide sequences are as shown in Figures in the main text. All fluorescent-labeled peptides were N-terminally FITC-Ahx labeled and C-terminally amidated. Other peptides were unmodified at both ends. The stapled peptides were synthesized by incorporating a pair of non-natural amino acids S5 at the indicated i, i+4 positions and stapling them as the hydrocarbon linker (Figure S5).

Protein expression and purification

Plasmids were transfected into in Escherichia coli Rosetta BL21 (DE3) cells. The cells were cultured at 37 °C in LB culture medium and then induced by 0.5 mM isopropyl β-D-1-thiogalactoside (IPTG) at 18 °C for 12 h 6×His-tagged proteins were purified by nickel affinity chromatography (Cytiva, 17531802) and anion-exchange chromatography (HiTrap Q HP, Cytiva, 29051325). Proteins used for crystallization after further purification by size exclusion chromatography (Superdex 75 10/300 GL, GE Healthcare, 17-5175-01) with buffer containing 20 mM Tris pH 8.0, 150 mM NaCl and 0.5 mM TCEP. Peak fractions were collected and concentrated to 20 mg/mL, and stored at −80 °C until use.

Crystallization and structure determination

In the crystallization process, BCL-XL was combined with an excess of BH3 peptides at a ratio of 1:1.5 (protein to peptide). Crystallization was achieved at a temperature of 18°C using the hanging drop vapor diffusion method, where drops consisted of 1 μL protein solution mixed with 1 μL reservoir solution. Crystals of BCL-XL/HRK-BH3 complex were grown with a reservoir buffer of 0.2 M Magnesium chloride hexahydrate, 0.1 M Tris pH 8.5 and 25% w/v Polyethylene glycol 3350. Crystals of the HRKM7-S1 complexed with BCL-XL were grown with a reservoir buffer of 0.2 M Magnesium chloride hexahydrate, 0.1 M Tris pH 8.5, 25% w/v Polyethylene glycol 3,350 (w/v).

For data collection, a single crystal was equilibrated in a cryo-protectant solution comprising a reservoir solution with the addition of 20% (v/v) ethylene glycol, and then the crystal was flash-frozen in liquid nitrogen. The data were collected at beamline BL19U1 of the Shanghai Synchrotron Radiation Facility (SSRF). The processing of the data was conducted using HKL2000.⁶¹ Structures were determined by molecular replacement using Phenix.phaser,⁶² with the structure of BCL-XL/BMF (PDB: 8IQK)⁴⁹ as search template. The initial model was manually built by the program Coot,⁶³ and refinement was performed with Phenix.refine. Translational-liberation-screw (TLS) refinement was used during the final stages of refinement. Graphical representations of the structures were generated using PyMOL.

Fluorescence polarization (FP) assay

Serial dilutions of proteins were prepared in a binding buffer composed of 1×PBS, 0.005% Tween 20, and 1 mM DTT. The 1×PBS was diluted with 10×PBS (80 g NaCl, 2 g KCl, 2.4 g KH₂PO₄, 14.2 g Na₂HPO₄·7H₂O dissolved in 1 L distilled water) These protein dilutions were subsequently injected into 384-well black Corning plates (Corning #4815), along with FITC-HRK at a concentration of 15 nM. Multiwell plates were then agitated on a shaker for 1 min. After 10 min of dark incubation at room temperature, the plates were ready. Polarization was measured in mP using a microplate reader (PerkinElmer Envision). Each experiment was performed at least three times. All experimental data were analyzed for EC₅₀ by sigmoidal dose-response nonlinear regression model.

To conduct the inhibition assay, the BH3 peptide was gradiently diluted and mixed with 15 nM FITC-HRK. Then 30 nM BCL-2 or BCL-XL were added to the mixture in binding buffer. The 384 plates were subjected to a 1-min mixing process on a shaker, followed by incubation in the dark at room temperature until equilibrium was attained. Each experiment was replicated at least three times. Polarization in milli-polarization units (mP) was measured with a microplate reader (PerkinElmer Envision). The experimental data were then analyzed for IC₅₀ using the sigmoidal dose-response nonlinear regression model.

AlphaFold

The structure models of BCL-2 in complex with HRK-BH3 or HRKM7 were predicted with AlphaFold 3 through its public web server.⁶⁴ In brief, the amino acid sequences of human BCL-2 removing the C-terminal 32 residues and the indicated HRK-BH3 peptide were used as the input. The rest parameters were default. In the results, the value of ipTM is about 0.9, the value of pTM is about 0.7, and the plDDT scores of the alpha helix in the complex structure are over 70, indicating that the results have confident high-quality prediction. PyMOL was used to visualize and analyze the predicted protein structures.

Virtual alanine screen

A virtual Alanine Scan is performed using the BUDE Alanine Scan.⁴³ To submit a Scan job, the BCL-XL/HRK-BH3 structure and BCL-2/HRK-BH3 AF structure were uploaded, respectively. The output files contain the energetic contribution of a residue to the interaction energy (ΔΔGs). The more positive the ΔΔGs value is for a residue, the larger a contribution it makes to forming the interaction.

FoldX

The calculation of interaction energies that contribute to the stability of proteins and protein complexes was performed by FoldX web server.⁶⁵ The BCL-XL/HRK-BH3 structure and BCL-2/HRK-BH3 AF structure was used as the design template for FoldX 5.0 algorithm calculations of the change in free energy of unfolding. The amino acids 29 to 52 of HRK were scanned for mutations to the other 19 amino acids, respectively. The ΔΔG values, which were computed from the difference between the free energy of the designed mutant (ΔG^mutant) and that of WT (ΔG^WT) (ΔΔG = ΔG^mutant − ΔG^WT), were calculated using the PositionScan command. Amino acid sites with ΔΔG greater than zero were considered to be beneficial for the overall stability of the complex. Official documentation and tutorials are available here (https://foldxsuite.crg.eu/command/PositionScan).

Circular dichroism

BH3 peptides (50 μM) were dissolved in 50 mM potassium phosphate (pH 7.5). CD spectra in the far-UV region (190–260 nm) were performed on a Jasco J-815CD spectrometers using a 0.1 cm cell at room temperature. Each spectrum was obtained by averaging 3 scans at 0.5-nm increments and helical content was calculated from the mean residue molar ellipticity value at 222 nm, ([θ]222). The secondary structure of the protein was calculated using SELCON software. The SELCON program is included in the CDPro software package, which is available from the website: https://www.bmb.colostate.edu/cdpro/.^66,67

Quantification and statistical analysis

Quantification and statistical analysis were conducted in GraphPad Prism 6. The experimental data from the fluorescence polarization (FP) assay was fitted using the sigmoidal dose-response nonlinear regression model in GraphPad Prism 6.01 to determine the IC₅₀. The values were expressed by the mean ± standard deviation (SD). All data points were obtained in three to four replicate experiments.

Supplemental information

Document S1 contains Figures S1–S9, Tables S1–S3, and supplemental reference.

Published: August 6, 2025