Discovery of a KLHL41 Ligand for Muscle Specific Protein Degradation Authors: Junhyeong Yim, Jaeseok Lee, Solbi Kim, Jieun Choi, Soyoung Yoon, Hana Cho, Sunbin Jung, GaYeon Yoo, Sanghee Lee, Hankum Park, Juyong Lee, Jongmin Park Published: 2026 ABSTRACT AbstractDespite the recent advancement of proteolysis-targeting chimera (PROTAC) development, they remain predominantly dependent on two E3 ligases, CRBN and VHL, which are ubiquitously expressed in all types of cells. Recently, efforts to discover tissue-specific E3 ligase ligands get attention as a promising strategy to enable tissue-specific protein degradation and avoid off-target tissue effects. Advancing this line of research, we discover a ligand of KLHL41, a muscle-specific E3 ligase, through virtual screening. Building on the KLHL41 ligand, we develop KBD-1, a muscle-specific BRD4-targeting PROTAC with micromolar activity. To enhance degradation efficiency, we employ a two-body kinetic strategy, resulting in the covalent PROTAC cKBD-1, which achieves sub-nanomolar activity. cKBD-1 demonstrates muscle-specific BRD4 degradation through KLHL41 recruitment both in vitro and in vivo. Moreover, the KLHL41 ligand enables AR-targeting PROTAC development, demonstrating its broad applicability. These findings highlight the potential of KLHL41 as a platform for tissue-specific protein degradation and its applicability in therapeutic development.Subject terms: Small molecules, Proteins, Ubiquitins Introduction In recent years, targeted protein degradation (TPD) has emerged as a therapeutic approach to overcome the limitations of traditional drug discovery strategy by selectively degrading disease-related proteins through ubiquitin-proteasome system (UPS)1–3. Proteolysis-targeting chimeras (PROTACs) are one of the main modalities of TPD strategy consisting of a target protein ligand, an E3 ligase ligand, and a linker conjugating the two ligands. This design generates the proximity between the target protein and the E3 ligase for the ubiquitination and subsequent proteasomal degradation of the target protein. While the PROTAC approach has demonstrated promise, current development is largely reliant on two E3 ligases: Cereblon (CRBN) and Von Hippel-Lindau (VHL)4,5. Although CRBN and VHL based PROTACs are actively developed and currently in clinical trials6, either CRBN ligand or VHL ligand is not always the best choice for the efficient TPD. By choosing a ligand of a different E3 ligase, target protein selectivity7 and target protein degradation efficiency8–10 can be enhanced. Moreover, drug resistance of CRBN or VHL based PROTACs is observed, which can be overcome by using other E3 ligase ligands11. In this context, expansion of E3 ligase toolbox is one of the main interest in TPD considering the existence of over 600 E3 ligases in human4,12,13. Tissue-dependent expression of E3 ligases is another issue in PROTAC design. CRBN and VHL are ubiquitously expressed across most human tissues, which makes them useful E3 ligase ligands for protein degradation in wide-range of diseases14. At the same time, CRBN or VHL based PROTACs can induce protein degradation in off-target tissue, potentially contributing to side effects. Among over 600 E3 ligases, their expression varies significantly across tissues and diseases, offering opportunities for tissue-, cell-, and disease-specific TPD1,13,15. In this context, the discovery of tissue-specific E3 ligase ligands could enable more efficient and precise therapeutic strategies, while minimizing off-target tissue effects16–18. The Kelch protein superfamily, characterized by the presence of Kelch-repeat-containing domains that interact with Cullin-3 (CUL3) E3 ligase19, is implicated in a wide range of biological functions20. Several members of this family including KLHL41, are specifically involved in regulating muscle cell proliferation, differentiation, and maintenance, and are therefore predominantly expressed in muscle tissue21. The tissue-specific expression of KLHL41 presents a promising avenue for the development of muscle-specific PROTACs, offering a targeted approach for treating muscle-related disorders. In this study, we discover EN10, a ligand for the muscle-specific E3 ligase, KLHL41, through virtual screening and verify EN10 as a KLHL41 ligand using cellular thermal shift assay (CETSA). By conjugating EN10 and JQ1, we develop KBD-1, a BRD4 targeting PROTAC, which achieves muscle cell specific protein degradation. Furthermore, we improve the degradation efficiency of KBD-1 by converting it to a covalent PROTAC, cKBD-1 with sub-nanomolar half-maximal degradation concentration (DC50) through a two-body kinetic PROTAC strategy. We also demonstrate the muscle specific targeted protein degradation of cKBD-1 in vivo, which suggests its potential as a tissue specific protein degrader. We believe that this approach could offer a precise treatment modality for muscle-related disorders, such as muscular dystrophy and sarcoma, with both improved potency and reduced off-target tissue effects compared to current CRBN- and VHL-dependent PROTACs. Results Virtual screening for the discovery of KLHL41 ligand candidates To identify potential ligands targeting KLHL41, virtual screening was performed using the Enamine hit locator library. The initial structure of KLHL41 was modeled using AlphaFold222. Due to the absence of its crystal structure. The β-propeller domain of KLHL41 was predicted as the most probable binding site for small-molecule interaction using the COACH-D binding site prediction server23. Subsequently, a three-stage docking simulation was performed targeting the β-propeller domain for virtual screening24. Specifically, a total of 230,000 small molecules was docked to KLHL41 using AutoDock-GPU as a first-stage25,26. This was followed by refinement with two additional docking programs: Glide SP and Glide XP27–29, which offer increasingly precise docking capabilities (Fig. 1a). From the AutoDock-GPU screening (Fig. 1b), the top 10,000 compounds were selected and evaluated using Glide SP (Fig. 1c), and the top 1000 from Glide SP were further assessed using Glide XP for higher-precision docking analysis (Fig. 1d). Finally, the top 60 compounds from those processes underwent absolute binding free energy (ABFE) simulations30,31 for quantitatively accurate estimation of their binding affinities. Based on a combination of docking scores and ABFE calculation results, 34 promising hit candidates were selected for further validation (Fig. 1e). Virtual screening for the discovery of KLHL41 ligands. a Virtual screening process: After the KLHL41 protein structure was predicted using AlphaFold2, KLHL41 ligand candidates were selected using sequential docking- and molecular dynamics-based virtual screening. b 230,000-membered virtual library was screened using AutoDock. c The top 10,000 molecules from virtual screening using AutoDock were subjected to further docking simulation using Glide SP. d The top 1000 molecules screened using Glide SP were subsequently analyzed using Glide XP. e Finally, the top 60 molecules screened from the Glide XP underwent molecular dynamics analysis, leading to the selection of 34 KLHL41 ligand candidates. CETSA identifies EN10 as a KLHL41 ligand To validate the initial hits from our virtual screening, we employed the CETSA, a widely used biophysical method for assessing ligand-target protein engagement in cells without any ligand modification by measuring changes in protein thermal stability32,33. Among the 34 hit candidates, EN10 demonstrated a clear thermal stability shift of KLHL41 (Fig. 2a, b). KLHL41 remained stable up to 50 °C and denatured at 55 °C. However, in the presence of EN10, KLHL41 remained stable even at 55 °C, indicating that EN10 binds to KLHL41 (Fig. 2c, d, and Supplementary Fig. 1). Additionally, a dose-dependent thermal stability shift of KLHL41 by EN10 was observed in CETSA at 55 °C, without affecting thermal stability of GAPDH (Supplementary Fig. 2a, b). These results further confirmed that the interaction between KLHL41 and EN10 in live RH30 rhabdomyosarcoma cells. Discovery of EN10 as a KLHL41 ligand. a Cellular Thermal Shift Assay (CETSA) results for 34 KLHL41 ligand candidates. RH30 cells were treated with 20 µM of each hit candidate for 1.5 h, followed by heat denaturation at 55 °C. The cells were lysed and KLHL41 in the resulting cell lysates was analyzed by western blot. b Chemical structure of EN10. The left (Blue) and right (Green) moieties of the compound are highlighted in different colors to facilitate visual distinction. c EN10 induces the thermal stability shift of KLHL41 in RH30 cells. RH30 cells were treated with 20 µM of EN10 for 1.5 h, followed by heat denaturation at various temperatures. The cells were lysed and KLHL41 in the resulting cell lysates was analyzed by western blot. d Quantification of KLHL41 levels at 55 °C in c (n = 3, biological replicates). e Chemical structures of the EN10 derivatives for linker position optimization. f CETSA results of EN10 and EN10 derivatives in RH30 cells, following the same procedure as in c. g Quantification of KLHL41 levels at 55 °C in f (n = 3, biological replicates). h Chemical structure of the EN10 photoaffinity based probe, EN10-DA, where EN10 is conjugated with diazirine and alkyne. i Photoaffinity based pulldown assay demonstrated the interaction between EN10-DA and KLHL41. j Quantification of KLHL41 levels after pulldown in i (n = 3, biological replicates). k Competitive KLHL41 labeling using EN10-DA and EN10 further confirms KLHL41 as a target protein of EN10. l Quantification of KLHL41 levels after competitive labeling in k (n = 3, biological replicates). Data in (c, f, i, k) are representative of three independent experiments. Data in (d, g, j, l) are presented as the mean ± standard deviation, with error bars indicating the standard deviation. Statistical significance was determined using a two-tailed unpaired Student’s t-test. ns, not significant (P ≥ 0.05); *P < 0.05; **P < 0.01. After identification of EN10 as a KLHL41 ligand, we synthesized EN10 derivatives for a linker conjugation (Fig. 2e). To determine the optimal linker position, the EN10 derivatives were evaluated for their KLHL41 binding with CETSA. We found that EN10-L1 and EN10-L3 maintained their interaction with KLHL41, indicating that both linker positions are suitable for the conjugation of target protein ligands (Fig. 2f, g, and Supplementary Fig. 3). Compared to EN10-L3, EN10-L4 showed a loss of interaction with KLHL41, highlighting the crucial role of a single carbon unit between the amide linker and the phenyl-tetrahydrofuran group. To further validate KLHL41 as a target protein of EN10 in live cells, we synthesized the EN10-DA probe by conjugating a diazirine photo-crosslinker and an acetylene functional handle for click chemistry to EN10 (Fig. 2h). After treatment of EN10-DA to RH30 cells and following UV irradiation, the cells were lysed and the photo-crosslinked proteins were conjugated to biotin-azide with click reaction. After following incubation of the resulting cell lysates with streptavidin bead, EN10 binding proteins were pulled down and immunoblotted with a KLHL41-specific antibody. This revealed a dose-dependent interaction between EN10-DA and KLHL41 (Fig. 2i, j, Supplementary Fig. 4a). Furthermore, competitive photo-affinity labeling with EN10-DA and EN10 confirmed KLHL41 as a target protein of EN10 in live cells (Fig. 2k, l, and Supplementary Fig. 4b). Given that KLHL40 is a close homolog of KLHL41 and both are highly expressed in muscle34, we sought to examine whether EN10 exhibits selective binding toward KLHL41. To verify the specificity of EN10 for KLHL41, we performed an additional pulldown assay using EN10-DA with KLHL40. The results demonstrated that EN10-DA selectively engaged KLHL41 but not KLHL40 (Supplementary Fig. 5a, b), confirming the ligand’s specificity. In addition, proteomic profiling further confirmed that EN10-DA selectively labeled KLHL41 (Supplementary Fig. 6 and Supplementary Fig. 7). Development of a KLHL41-based PROTAC for BRD4 degradation via CRL-mediated ubiquitination After identification of EN10 as a KLHL41 ligand, we designed a KLHL41-based PROTAC targeting BRD4, a transcriptional coactivator from the BET protein family. BRD4 was chosen due to the extensive studies and successes of BRD4 targeting degraders1, as well as its critical role in promoting blood cancers35 and solid tumors36,37. Accordingly, we designed and synthesized a series of KLHL41 based BRD4 targeting degraders (KBDs), conjugating EN10 to the BET bromodomain inhibitor (+)-JQ1 (JQ1) using various linkers (Fig. 3a). KLHL41-based PROTACs induce muscle-specific BRD4 degradation. a Chemical structures of KLHL41-based PROTACs, referred to as KLHL41-BRD4 Degraders (KBDs). Carbon atoms in the linker are color-coded in yellow for clear visualization. b BRD4 degradation by KBDs is structure-dependent. KBDs were treated to RH30 cell with the indicated concentrations for 12 h. BRD4 degradation by KBDs were monitored by western blot. c Quantification of BRD4 band intensities at 40 μM in b (n = 3, biological replicates). d KBD-1-induced BRD4 degradation is inhibited by co-treatment of either MG132 or MLN4924 in RH30 cells, indicating that BRD4 degradation by KBD-1 is mediated by CRL-mediated ubiquitination. e Quantification of BRD4 band intensities in (d) (n = 3, biological replicates). f KBD-1 induced BRD4 degradation only in myosarcoma cell lines (RD and RH30 cells) after 24 h treatment. BRD4 was not degraded by KBD-1 in non-muscle related cancer cell lines (OV90, A549, SW480, and HeLa cells). g Quantification of BRD4 band intensities at 20 μM in f (n = 3, biological replicates). h KBD-1 (40 μM) exhibited cytotoxicity only in myosarcoma cell lines after 24 h treatment (n = 5, technical replicates). i–l KLHL41-dependent BRD4 degradation by KBD-1. i KLHL41 knockdown disturbed KBD-1 mediated BRD4 degradation in RH30 cells. j Quantification of BRD4 band intensities in i (n = 3, biological replicates). k KLHL41 overexpression induced BRD4 degradation by KBD-1 in HeLa cells. l Quantification of BRD4 band intensities in k (n = 3, biological replicates). Data in (b, d, f, i, k) are representative of three independent experiments. Data in (c, e, g, h, j, l) are presented as the mean ± standard deviation, with error bars indicating the standard deviation. Statistical significance was determined using a two-tailed unpaired Student’s t-test. ns, not significant (P ≥ 0.05); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Interestingly, when JQ1 was conjugated to the phenyl-tetrahydrofuran group of EN10, BRD4 degradation showed linker length dependency (KBD-1, 2, 3; Fig. 3b, c); shorter linker results in enhanced degradation activity. KBD-1, which has the shortest linker, achieved complete BRD4 degradation at 20 μM after 12-h treatment of RH30 cells. KBD-2 with a C-3 linker resulted in complete BRD4 degradation at 30 μM and KBD-3 with a C-5 linker showed complete BRD4 degradation at 40 μM. In contrast, when JQ1 was conjugated to the phenyl-pyrazole group of EN10 (KBD-4 and KBD-5), their BRD4 degradation activity was lost, even at 40 μM, 12-h treatment conditions. When the treatment time was increased to 48 h, KBD-4 induced slight BRD4 degradation at 30 μM, whereas KBD-5 induced complete BRD4 degradation at 40 μM (Supplementary Fig. 8a, b). This observation aligns with the trends observed in KBD-1 to KBD-3 in terms of linker length. Our results suggest that the linker length is critical for targeted protein degradation by KLHL41-based PROTACs, as has been widely recognized for CRBN- or VHL-based PROTACs38. Moreover, shorter linkers exhibit higher potency in this series of KBDs. To further validate direct binding of EN10 to KLHL41, we performed biolayer interferometry (BLI) using KBD-1 and recombinant KLHL41 protein. This analysis yielded a dissociation constant (KD) of 37.36 µM, consistent with the micromolar-range activity observed in our cellular assays (Supplementary Fig. 9). These results confirm that EN10-based PROTACs directly interact with KLHL41 and provide quantitative support for the EN10–KLHL41 interaction. KLHL41 functions in conjunction with the CUL3 adapter protein and forms part of the Cullin-RING ligase (CRL) complex, which plays a crucial role in the ubiquitination and subsequent degradation of target proteins39. To investigate CRL-mediated targeted protein degradation of our most potent KLHL41 based BRD4 degrader, KBD-1, we co-treated KBD-1 to RH30 cells with either the NEDD8-activating enzyme inhibitor MLN4924 or the proteasome inhibitor MG132. We observed that BRD4 degradation activity of KBD-1 was mitigated by co-treatment with either MLN4924 or MG132 (Fig. 3d, e), indicating that KBD-1 induces BRD4 degradation via CRL-mediated ubiquitination and proteasomal degradation. KLHL41-based PROTACs induce muscle cell-specific BRD4 degradation We next investigated whether the activity of KBDs is KLHL41-dependent in cancer cell lines. Considering KLHL41 is specifically expressed in muscle-related cells21, we selected six cancer cell lines for this study: two muscle related myosarcoma cell lines (RD and RH30) and four non-muscle related cancer cell lines (OV90, A549, SW480, and HeLa). Western blot analysis confirmed that KLHL41 was specifically expressed in the myosarcoma cell lines (Supplementary Fig. 10). Next, we treated KBD-1 and KBD-2 to all six cell lines and found that they induced BRD4 degradation in the KLHL41-expressing cell lines (RD and RH30), but not in the non-muscle related cancer cell lines (Fig. 3f, g, and Supplementary Fig. 11a). Additionally, both KBD-1 and KBD-2 showed myosarcoma cell-specific cytotoxicity (Fig. 3h, and Supplementary Fig. 11b), with 20 μM IC50 values, while they showed no cytotoxicity in non-muscle related cancer cell lines, even with treatment up to 40 μM (Supplementary Fig. 11c, and Supplementary Table 1). To further confirm that this myosarcoma cell-specific effect of KBD-1 and KBD-2 is KLHL41-dependent, we conducted KLHL41 knockdown studies in RH30 cells and KLHL41 overexpression studies in HeLa cells (Fig. 3i–l). In RH30 cells, both KBD-1 and KBD-2 successfully induced BRD4 degradation prior to KLHL41 knockdown; however, after KLHL41 knockdown, BRD4 degradation was disturbed (Fig. 3i, j, Supplementary Fig. 12a). In contrast, neither KBD-1 nor KBD-2 induced BRD4 degradation before KLHL41 overexpression in HeLa cells. However, after KLHL41 overexpression, both compounds successfully degraded BRD4 (Fig. 3k, l, Supplementary Fig. 12b). We also demonstrated BRD4 degradation activity of KBD-1 and KBD-2 in a C2C12 mouse myoblast cell line differentiation study (Supplementary Fig. 13a). In myoblast state of C2C12, KLHL41 is not expressed, but KLHL41 was upregulated after differentiation of C2C12 cells into myocyte as expected (Supplementary Fig. 13b). In myoblast state of C2C12 cells, neither KBD-1 nor KBD-2 induced BRD4 degradation. However, in myocyte state of C2C12 cells, both KBD-1 and KBD-2 successfully induced BRD4 degradation (Supplementary Fig. 13b). These results validate that our KLHL41-based PROTACs induce protein of interest (POI) degradation in a KLHL41-dependent manner, offering potential for muscle cell-specific TPD. To further evaluate the cellular consequences of KBD-1, we performed quantitative proteomic profiling following compound treatment. Although KBD-1 induced measurable BRD4 degradation at 40 μM, the analysis revealed broader proteomic changes, including proteins associated with cellular stress and adaptive response pathways (Supplementary Fig. 14). For example, ATF340,41, HEXIM142, MDM243,44, BCL2L145, and CYB46 were upregulated upon KBD-1 treatment. These observations suggest that the relatively high working concentrations of KBD-1 may lead to broader cellular perturbations. These results highlight the intrinsic limitations of EN10-based PROTACs for practical applications. Development of tPROTAC for improving the potency of target protein degradation Although KBD-1 and KBD-2 show their potential for muscle cell-specific TPD, their degradation activity remains in the micromolar range in RH30 cells, which limits their potential clinical applications (Fig. 3f, g). This potency is significantly less effective compared to that of clinically advanced PROTACs such as ARV-825 (DC50 < 1 nM)47, ARV-471 (DC50 = 1.8 nM)48 and ARV-771 (DC50 < 5 nM)49. Moreover, the micromolar activity of EN10 based PROTAC induces broader perturbations in cellular stress pathways. To overcome these limitations, we employed a two-body kinetic PROTAC (tPROTAC) strategy. Inspired by previous report about covalent E3 ligase ligands (DCAF150, DCAF1151, DCAF1652, RNF453, RNF11454, and FEM1B55), we hypothesized that covalent bonding of PROTAC to E3 ligase would enhance the target protein degradation efficiency56,57. Conventional PROTACs function based on three-body kinetics, where three bodies (E3 ligase, PROTAC, and target protein) generate a ternary complex. After the target protein degradation, the PROTAC molecule is recycled for the formation of a new ternary complex. In tPROTAC, a PROTAC molecule covalently binds to an E3 ligase, and the resulting PROTAC-E3 ligase complex subsequently binds to the target protein for a binary complex formation. After the target protein degradation, the PROTAC-E3 ligase complex will be recycled and recruit another target protein molecule for its degradation. In this way, tPROTAC can work based on two-body kinetics instead of three-body kinetics. Considering the complexity of cellular environment, a binary complex formation would be easier and faster than a ternary complex formation, thereby enhancing catalytic efficiency and degradation of the target protein (Fig. 4a)6,58. tPROTAC strategy for the highly efficient muscle cell-specific target protein degradation. a Concept of two-body kinetic PROTAC (tPROTAC). A covalent based E3 ligase ligand generates a E3 ligase-PROTAC complex, which subsequently binds to the POI for a binary complex formation. After POI degradation, the E3 ligase-PROTAC complex can be recycled to bind to POI again. Partially created in BioRender. Park, J. BioRender.com/5rvnse7. b Chemical structure of the covalent EN10 (cEN10) probe, cEN10-Alkyne-Probe, where EN10 is conjugated with α-chloroacetyl and alkyne functional group. α-chloroacetyl group is color-coded in yellow for clear visualization. c Pull-down assay demonstrated the interaction between cEN10-Alkyne-Probe and KLHL41. Competitive KLHL41 labeling using cEN10-Alkyne-Probe and EN10 further confirmed that both compounds share the binding site of KLHL41. d Quantification of KLHL41 levels after pulldown in c (n = 3, biological replicates). e Chemical structures of KLHL41-based tPROTACs, referred to as covalent KBDs (cKBDs). f BRD4 degradation by cKBDs is structure-dependent. cKBDs were treated to RH30 cell with the indicated concentrations for 12 h. BRD4 degradation by cKBDs were monitored by western blot. g Quantification of BRD4 band intensities at 5 μM in f (n = 3, biological replicates). h cKBD-1-induced BRD4 degradation is inhibited by co-treatment of either MG132 or MLN4924 in RH30 cells, indicating that cKBD-1 induces BRD4 degradation via CRL-mediated ubiquitination. i Quantification of BRD4 band intensities in h (n = 3, biological replicates). j cKBD-1 showed a dose-response BRD4 degradation. k Quantification of BRD4 band intensities in j. cKBD-1 showed a half-maximal degradation concentration (DC50) of 0.64 ± 0.14 nM and a maximum degradation (Dmax) of 83.4 ± 2.60 % for BRD4 (n = 3, biological replicates). Data shown in (c, f, h, j) are representative of three independent experiments. Data shown in (d, g, i, k) are presented as the mean ± standard deviation, with error bars indicating the standard deviation. Statistical significance was determined using a two-tailed unpaired Student’s t-test. Exact P values are provided in the Source Data file. ns, not significant (P ≥ 0.05); *P < 0.05. To develop a KLHL41-based tPROTAC molecule, we first analyzed the sequence of the β-propeller domain of KLHL41, the potential binding site of EN10. The β-propeller domain has multiple Cys and Lys residues, which could act as nucleophiles and may react with an electrophile conjugated to EN10 to form a covalent bond (Supplementary Fig. 15). Especially, Cys593 was expected to closely locate near the secondary amine of EN10. Based on this rationale, we designed a covalent-binding KLHL41 ligand (cEN10) by introducing an electrophile into the secondary amine of EN10. To evaluate the binding of KLHL41 to cEN10, we synthesized a cEN10-Alkyne-Probe by conjugating an α-chloroacetyl group to the secondary amine group of EN10 and an alkyne to the phenyl-tetrahydrofuran moiety of EN10 (Fig. 4b). After treatment of 10 μM of cEN10-Alkyne-Probe to RH30 cells for 30 min, the cells were lysed and the cEN10-Alkyne-Probe binding proteins were conjugated to biotin-azide with click reaction. After following incubation of the cell lysates with streptavidin bead, cEN10-Alkyne-Probe binding proteins were pulled down and analyzed by immunoblotting with a KLHL41 antibody. The results confirmed a covalent bond formation between cEN10-Alkyne-Probe and KLHL41 (Fig. 4c, d). Additionally, competitive labeling using cEN10-Alkyne-Probe and EN10 further validated the covalent binding of cEN10-Alkyne-Probe to KLHL41 (Fig. 4c, d). Proteomic profiling also supported this result, showing selective enrichment of KLHL41 by the cEN10-Alkyne-Probe (Supplementary Fig. 16 and Supplementary Fig. 17). Importantly, consistent with our previous specificity studies, cEN10-Alkyne-Probe selectively engaged KLHL41 without detectable binding to its close homolog KLHL40 (Supplementary Fig. 18a, b), further supporting the selectivity of the cEN10. Moreover, we synthesized a negative probe, cEN10-Alkyne-Negative, lacking the chloride in cEN10-Alkyne-Probe (Supplementary Fig. 19a). As expected, the negative probe did not show any covalent binding to KLHL41 in the pulldown assay (Supplementary Fig. 19b, c). To modify our KBD compounds to tPROTAC, an α-chloro acetyl group was conjugated to the secondary amine of the series of KBDs (Fig. 4e). Interestingly, cKBD-1, the most potent KBD-1-based tPROTAC, completely degraded BRD4 even at 5 μM, which could not be achieved by KBD-1 (Fig. 4f, g). It is also noteworthy that KBD-5 induced BRD4 degradation at 40 µM after 48 h of treatment (Supplementary Fig. 8b), whereas cKBD-5 achieved complete degradation at concentrations as low as 20 µM within only 6 h of treatment. In case of cKBD-2, cKBD-3, and cKBD-4, there was no improvement in BRD4 degradation, compared to KBD-2, KBD-3, and KBD-4. To verify BRD4 degradation mechanism by cKBD-1, we co-treated RH30 cells with cKBD-1 and either MLN4924 or MG132. From this result, we could conclude that cKBD-1 induces BRD4 degradation via CRL-mediated ubiquitination and proteasomal degradation, as KBD-1 does (Fig. 4h, i). After identifying cKBD-1 as the most potent tPROTAC molecule, we monitored its dose-dependent BRD4 degradation to determine its exact degradation activity. Using dose-response western blot analysis, we demonstrated that the DC50 of cKBD-1 is 0.64 ± 0.14 nM and the Dmax is 83.4 ± 2.60% (Fig. 4j, k), representing more than 10,000-fold improvement compared to the activity of KBD-1. Moreover, we could not observe a hook effect of cKBD-1, which is one of the major limitations of non-covalent based PROTACs. To summarize these findings, we validated that tPROTAC strategy is a promising approach for an effective target protein degradation. cKBD-1 induces muscle-specific BRD4 degradation in cell lines and in vivo To investigate the muscle cell-specific BRD4 degradation of cKBD-1, we treated cKBD-1 again in six cancer cell lines (RD, RH30, OV90, A549, SW480, and HeLa). cKBD-1 showed BRD4 degradation in RD and RH30 myosarcoma cell lines, achieving complete BRD4 degradation at 5 μM, but did not induce BRD4 degradation in non-myosarcoma cancer cell lines (Fig. 5a, b). Additionally, cKBD-1 exhibited myosarcoma cell-specific cytotoxicity (Fig. 5c, and Supplementary Fig. 20) due to BRD4 degradation. In non-myosarcoma cell lines, cKBD-1 showed no cytotoxicity even at 5 μM concentration (Supplementary Fig. 20, and Supplementary Table 1). cKBD-1 induced muscle-specific BRD4 degradation. a cKBD-1 induced BRD4 degradation only in myosarcoma cell lines (RD and RH30 cells). BRD4 was not degraded by cKBD-1 in non-muscle related cancer cell lines (OV90, A549, SW480, and HeLa cells). b Quantification of BRD4 band intensities at 0.1 μM in a (n = 3, biological replicates). c cKBD-1 exhibited cytotoxicity only in myosarcoma cell lines at 5 μM (n = 5, technical replicates). d–i KLHL41-dependent BRD4 degradation by cKBD-1. d KLHL41 knockdown disturbed cKBD-1 mediated BRD4 degradation in RH30 cells. e Quantification of BRD4 band intensities in d (n = 3, biological replicates). f KLHL41 overexpression induced BRD4 degradation by cKBD-1 in HeLa cells. g Quantification of BRD4 band intensities in f (n = 3, biological replicates). h cKBD-1 (5 μM) did not induce BRD4 degradation in undifferentiated C2C12 cells (myoblast, low KLHL41 expression). In differentiated C2C12 cells (myocyte, high KLHL41 expression), cKBD-1 (5 μM) induced BRD4 degradation. i Quantification of BRD4 band intensities in h (n = 3, biological replicates). j Mass spectrometry analysis of the tryptic peptide of KLHL41 containing Cys593 (amino acids 587–597): cKBD-1 binding peptide. k Docking analysis showed that Cys593 of KLHL41 is proximal to EN10. l BLI analysis revealed KD between the KLHL41/cKBD-1 complex and BRD4 to be 27.04 ± 0.93 nM (n = 5, technical replicates). Experimental data using a series of BRD4 concentrations (colored lines) are overlaid with globally fitted curves (black lines) using a 1:1 Langmuir binding model. The fitting yielded an association rate constant (kon) of 9.64 × 10³ M⁻¹ s⁻¹ and a dissociation rate constant (koff) of 2.73 × 10⁻⁴ s⁻¹. The maximum binding response (Rmax) was 0.36 nm, and the goodness of fit was indicated by the the coefficient of determination (R² = 0.998) and a residual sum of squares (RSS) value of 0.50. m Quantitative proteome profiling showed that cKBD-1 selectively degraded BRD4 without affecting KLHL41 level (n = 4 per group, technical replicates). n Schematic of in vivo experiment. Mice were treated with vehicle or cKBD-1 (5 mpk) via IV or IP injection and sacrificed after 24 h for tissue collection (n = 3 mice per group, biological replicates). Created in BioRender. Park, J. BioRender.com/wlh07p2. o Immunoblot showing muscle-specific KLHL41 expression in vivo. Representative data are shown from three independent experiments with similar results. p cKBD-1 induced BRD4 degradation in muscle tissue, but not in other organs in in vivo mouse model. Representative data are shown from three independent experiments with similar results. q Quantification of BRD4 band intensities in (p). Data shown in (a, d, f, h) are representative of three independent experiments. Data shown in (b, c, e, g, i) are presented as the mean ± standard deviation, with error bars indicating the standard deviation. Statistical significance was determined using a two-tailed unpaired Student’s t-test. ns, not significant (P ≥ 0.05); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. To confirm whether the muscle cell-specific activity of cKBD-1 is KLHL41-dependent, we performed KLHL41 knockdown studies in RH30 cells (Fig. 5d, e). After KLHL41 knockdown with siRNA, BRD4 degradation activity of cKBD-1 was not observed in RH30 cells. To further confirm the KLHL-dependency of cKBD-1 activity, we overexpressed KLHL41 in HeLa cells (Fig. 5f, g). Upon KLHL41 overexpression, we could observe BRD4 degradation activity of cKBD-1 even in a non-muscle related cancer cell line, HeLa. We further demonstrated KLHL41 dependent BRD4 degradation of cKBD-1 in C2C12 cells with and without differentiation into myocyte. KLHL41 expression was observed only in myocyte but not in myoblast state of C2C12. BRD4 degradation of cKBD-1 was only observed in KLHL41 enriched myocyte state of C2C12 cells as we expected (Fig. 5h, i). All those results confirm that cKBD-1 induces BRD4 degradation in a KLHL41-dependent manner, just like KBD-1. We also conducted mass spectrometry to decipher the binding site of cKBD-1. Recombinant KLHL41 proteins were incubated with cKBD-1 for 30 min to generate a KLHL41/cKBD-1 complex, after which excess cKBD-1 was removed using Amicon filter. The KLHL41/cKBD-1 complex was then subjected to peptide digestion, and the resulting peptides were analyzed. Interestingly, Cys593 of KLHL41 was modified by covalent binding with cKBD-1 (Fig. 5j and Supplementary Fig. 21), which is a residue located near the predicted binding site of the initial hit, EN10 in the docking analysis (Fig. 5k). To validate the functional importance of this residue, we performed mutagenesis studies by substituting Cys593 with serine (C593S). In HeLa cells with KLHL41 C593S overexpression, BRD4 degradation by cKBD-1 was remarkably reduced. However, robust degradation by cKBD-1 was observed in HeLa cells with wild-type KLHL41 overexpression, confirming that Cys593 is essential for cKBD-1-mediated degradation (Supplementary Fig. 22a, b). These results, together with mass spectrometry data, directly demonstrate the covalent engagement of cKBD-1 with KLHL41 at Cys593. To observe the physical interactions of cKBD-1, KLHL41, and BRD4, we conducted biolayer interferometry (BLI). The KLHL41/cKBD-1 complex was prepared in the same manner as for the binding site study. The complex was biotinylated and then loaded onto a streptavidin biosensor. The sensor was subsequently dipped into varying concentrations of BRD4 solution to determine the KD value for the interaction between KLHL41/cKBD-1 and BRD4, which was measured to be 27.04 ± 0.93 nM (Fig. 5l). We subsequently conducted unbiased tandem mass tag (TMT)-based quantitative proteomic profiling studies in RH30 cells to evaluate the target protein degradation selectivity of cKBD-1. Over 6000 proteins were quantified in the samples from RH30 cells treated with either DMSO or 10 nM of cKBD-1 for 12 h. cKBD-1 treatment induced a significant reduction in BRD4 protein levels (Fig. 5m) without affecting KLHL41 protein levels, which is consistent with the immunoblotting data (Fig. 5d). Notably, cKBD-1 treatment did not cause significant changes in the protein levels of other bromodomain-containing proteins, including BRD1, BRD2, BRD3, BRD7, BRD8, and BRD9. This BRD4 selectivity of cKBD-1 may be attributed to the cKBD-1 induced specific interaction between KLHL41 and BRD4. Notably, we also identified three proteins, including SEMA6B, SAMD11, and RAD51AP1, that were downregulated by the treatment of cKBD-1. Previous reports have demonstrated that downregulation of BRD4 by shRNA, siRNA, or PROTACs leads to decreased expression of SEMA6B59, SAMD1160, and RAD51AP161. Thus, the observed reductions in these proteins are due to the BRD4 degradation by cKBD-1 rather than direct off-target activity of cKBD-1. To further evaluate whether ligand engagement affects the endogenous function of KLHL41, we performed TMT-based proteomic profiling after the treatment of cKBD-1 and cEN10. The expression of known KLHL41 substrates and KLHL41-related proteins34 remained unaffected (Supplementary Fig. 23a, b), and cEN10 treatment did not induce noticeable proteome-wide changes (Supplementary Fig. 23b). These findings indicate that cEN10 does not disrupt KLHL41 function and does not cause meaningful off-target perturbation. Thus, cKBD-1-mediated BRD4 depletion arises from induced proximity rather than intrinsic inhibition of KLHL41. Collectively, these results show that cKBD-1 induces BRD4 degradation in a muscle cell-specific manner while maintaining high proteome-level selectivity for BRD4. After elucidating the mechanism of action of cKBD-1, we further evaluated its in vivo activity. Mice were treated with either vehicle or cKBD-1 (5 mpk) via intravenous (IV) or intraperitoneal (IP) injection for 24 h (Fig. 5n). To assess the tissue-specific expression of KLHL41, we harvested various organs from the mice including the brain, lung, stomach, liver, kidney, and muscle. Western blot analysis revealed that KLHL41 is exclusively expressed in muscle tissue (Fig. 5o). Consistent with KLHL41 expression level, BRD4 degradation was observed only in the muscle, with no significant degradation detected in other tissues (Fig. 5p, q). These findings are consistent with our in vitro results in muscle cell lines, further supporting the muscle-specific protein degradation by cKBD-1. Together, these results demonstrate that cKBD-1 selectively induces BRD4 degradation in muscle both in vitro and in vivo through KLHL41 recruitment. We next evaluated the pharmacokinetic properties of cKBD-1 in rats. Following intravenous administration (5 mg/kg), cKBD-1 exhibited a half-life (t1/2) of 2.5 h and a volume of distribution (Vss) of 1.44 L/kg, indicating substantial tissue distribution (Supplementary Table 2). Importantly, the plasma concentration remained well above the DC50 for BRD4 degradation (0.64 nM) even after 8 h (Supplementary Fig. 24). In addition, cKBD-1 remained stable in mouse plasma (Supplementary Fig. 25). These results suggest that cKBD-1 achieves sufficient systemic exposure to support efficient in vivo targeted protein degradation. KLHL41-dependent degradation of androgen receptor To evaluate the potential application of our KLHL41 ligand for TPD of other proteins, we designed and synthesized a series of KLHL41-AR degraders (KADs) by conjugating EN10 to an aryloxy cyclohexane androgen receptor (AR) antagonist with diverse linker structures (Fig. 6a). Interestingly, AR degradation by KADs was linker structure-dependent. Shorter linkers demonstrated higher degradation potency, which is consistent with the results observed for KBDs (Fig. 6b, c). KAD-1 with the shortest linker achieved complete degradation even at 30 μM after a 12 h treatment of RH30 cells. KAD-2 with a C-3 linker, resulted in slight degradation at 50 μM, while KAD-3 with a C-5 linker started to show AR degradation at 50 μM. Development of cKAD-1 as an androgen receptor (AR) degrader. a Chemical structures of KLHL41-based AR targeting PROTACs, referred to as KLHL41-AR degraders (KADs). b BRD4 degradation by KADs is structure-dependent. KADs were treated to RH30 cell with the indicated concentrations for 12 h. BRD4 degradation by KADs were monitored by western blot. c Quantification of AR band intensities at 50 μM in b (n = 3, biological replicates). d Chemical structures of KLHL41-based AR targeting tPROTAC, referred to as covalent KAD-1 (cKAD-1). e cKAD-1 showed a dose-response AR degradation. f Quantification of AR band intensities in e. cKAD-1 showed a half-maximal degradation concentration (DC50) of 202.2 ± 43.9 nM and a maximum degradation (Dmax) of 81.4 ± 6.14 % for AR (n = 3, biological replicates). Data shown in (b, e) are representative of three independent experiments. Data shown in (c, f) are presented as the mean ± standard deviation, with error bars indicating the standard deviation. Statistical significance was determined using a two-tailed unpaired Student’s t-test. **P < 0.01; ***P < 0.001. To improve the potency of KAD-1 with tPROTAC strategy, we also synthesized cKAD-1 by attaching an α-chloroacetyl group to the secondary amine group of KAD-1 (Fig. 6d). Using dose-response western blot analysis, we demonstrated that the DC50 of cKAD-1 is 202.2 ± 43.9 nM and the Dmax is 81.4 ± 6.14% in RH30 cells (Fig. 6e, f), representing more than 1000-fold improvement compared to that of KAD-1. Moreover, cKAD-1 has no hook effect as we observed in cKBD-1. From these results, we could demonstrate that our KLHL41 ligand has potential for TPD of various POIs with the tPROTAC strategy. Discussion Conventional PROTACs mostly rely on E3 ligases like CRBN and VHL, which is facing limitations due to their ubiquitous expression, often resulting in off-target tissue effects4,5. Our study addresses this challenge by leveraging the tissue specific expression of KLHL41, a muscle-specific E3 ligase21, which enables selective protein degradation in muscle cells. Through virtual screening and CETSA (Fig. 1, Fig. 2a, and Fig. 2c), we discovered a KLHL41 ligand, EN10, that binds to KLHL41 (Fig. 2b), enabling the development of muscle cell-specific PROTACs (Fig. 3a and Fig. 6a). The successful BRD4 degradation specifically in KLHL41-expressing muscle cells (Fig. 3f, g) presents promising implications for the treatment of muscle-related diseases such as muscular dystrophy and sarcoma while maintaining a favorable safety profile by avoiding unwanted degradation in non-muscle cells (Fig. 3f, g). This specificity could mitigate the systemic side effects, which could be an issue in conventional CRBN or VHL dependent PROTACs. We believe that this muscle cell specificity of KLHL41-based PROTACs could open avenues for precision medicine in treatment of muscle related diseases. Furthermore, the combinatorial strategy of virtual screening and CETSA validation to discover a KLHL41 ligand offers a cornerstone for discovery of other tissue- or tumor-specific E3 ligases, thereby providing a pathway for a more tailored therapeutic approach using PROTACs13,16–18. Our findings also demonstrate the utility of the tPROTAC strategy (Fig. 4a), which improves the potency of PROTACs by facilitating the formation of the E3 ligase/PROTAC complex that can be recycled after the degradation of the target protein6,58. Compared to conventional PROTACs, tPROTACs exhibit enhanced potency by overcoming kinetic constraints, particularly through their transition from ternary to binary complex formation, which shift the dynamic range of protein degradation activity to lower concentrations38,39. The dramatic increase in potency observed with our KLHL41-based tPROTACs represents over a 10,000-fold improvement in degradation potency compared to non-covalent KLHL41 ligand-based PROTACs (Fig. 3b, and Fig. 4j), underscoring the potential of this strategy to yield therapeutics with enhanced activity and specificity. Interestingly, we also observed no hook-effect with our tPROTACs, which is hard to be achieved with non-covalent based PROTACs. With the covalent-based tPROTAC, the PROTAC and E3 ligase function as a single unit once covalently bound, which dominantly exists in the cells, efficiently binds to target protein and leads to the degradation of POI in more efficient way. However, high concentration of non-covalent PROTACs facilitates the generation of either PROTAC-E3 ligase complex or PROTAC-POI complex, which commonly causes the hook effect. We believe that this strategy could be extended to other PROTAC molecules to enhance their activity. Moreover, our cKBD-1 showed the tissue selective BRD4 degradation in an in vivo mouse model, which suggests its potential for the treatment of muscle-related diseases. Both IV and IP administration at 5 mg/kg resulted in effective BRD4 degradation only in muscle tissues, which support that our KLHL41 ligand could be useful for the development of muscle-specific PROTACs. Despite these promising results, several challenges remain in the application of our KLHL41-based PROTACs (cKBD-1 and cKAD-1) for drug development. First, pharmacokinetic properties of our KLHL41-based PROTACs must be optimized for effective in vivo applications. Second, selection of an appropriate target protein and identification of its ligand are essential for the treatment of muscle-related diseases using KLHL41-based PROTACs. Although there is a growing demand for the treatment of muscle-related diseases, the number of suitable target proteins for TPD is currently limited to a few, such as DUX462, Dysferlin63, and MSI264. Therefore, the discovery of other muscle disease related proteins for TPD must precede the development of muscle specific PROTACs. Third, expression levels of KLHL41 in patients have not yet been thoroughly investigated. Considering its high abundance in muscle tissue, KLHL41 is expected to be highly expressed in most patients. Nevertheless, detailed studies examining inter-patient and inter-tissue variability in KLHL41 expression would be highly valuable for guiding the potential application of KLHL41-based PROTACs in personalized medicine65,66. Fourth, considering the potential resistance that may arise from long-term PROTAC treatment similar to what has been observed in traditional therapies11, discovery of multiple muscle-specific E3 ligase ligands and subsequent development of multiple muscle-specific PROTACs are crucial for the successful treatment of muscle-related diseases. In conclusion, our study highlights the transformative potential of KLHL41-based PROTACs as a therapeutic strategy for muscle-related disorders. By utilizing the muscle-specific E3 ligase KLHL41 and implementing the tPROTAC strategy, we achieved a picomolar degradation activity of cKBD-1 in a muscle cell-specific manner. With our proof-of-concept study targeting BRD4 and AR in muscle cells, our next goal is to develop muscle-specific degraders of DUX4, Dysferlin and MSI2 for the treatment of muscular dystrophy. Leveraging KLHL41 for muscle specific protein degradation would establish a solid foundation for the treatment of muscle-related disease. We believe that our approach not only suggests a tool for the discovery of E3 ligase ligands but also provides the groundwork for future exploration of tissue-selective PROTACs in the treatment of various diseases. Methods Animal experiments All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Kangwon National University (Approval No. KW-250313-3 and KIACUC-22-096) and Korea Institute of Science and Technology (Approval No. KIST-IACUC-2024-052-5). In vivo experiments were carried out in accordance with the guidelines in the US National Institutes of Health Guide for the Care and Use of Laboratory Animals. 8-week-old C57BL/6 male mice (20–25 g), 7-week-old ICR male mice (25–30 g) and 8-week-old SD male rats (280–320 g) from Samtako Bio Korea were supplied by DooYeol Biotech. The animals were maintained in temperature- and humidity-controlled conditions (22°C and 52%) with 12 h light and 12 h dark cycles. Life science study design No statistical methods were used to predetermine sample size. Experiments were typically performed with three independent biological replicates, which is consistent with standard practice in the field and was sufficient to ensure reproducibility. Samples or animals were randomly assigned to experimental groups. No additional covariate-based allocation was necessary. Cell culture HeLa (human cervical adenocarcinoma, CCL-2), A549 (human lung carcinoma, CCL-185), and OV90 (human ovarian adenocarcinoma, CRL-3585) cells were obtained from the American Type Culture Collection (ATCC) and cultured in Roswell Park Memorial Institute (RPMI) medium (Welgene; LM011-01) supplemented with 1% (v/v) antibiotic-antimycotic solution (Welgene; LS203-01) and 10% (v/v) heat-inactivated fetal bovine serum (FBS, Gibco; #16000-044). RH30 (human muscle rhabdomyosarcoma, CRL-2061) cells were obtained from ATCC and cultured in RPMI medium with 25 mM HEPES (N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid) (Welgene; LM011-03) supplemented with 1% (v/v) antibiotic-antimycotic solution and 10% (v/v) heat-inactivated FBS. C2C12 (mouse myoblast cells, CRL-1772) and SW480 (human colon adenocarcinoma, CCL-228) cells were also obtained from ATCC and cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 1% (v/v) antibiotic-antimycotic solution and 10% (v/v) heat-inactivated FBS. RD (human muscle rhabdomyosarcoma, #10136) cells were obtained from the Korea Cell Line Bank (KCLB) and cultured in DMEM supplemented with 1% (v/v) antibiotic-antimycotic solution and 10% (v/v) heat-inactivated FBS. All cell lines were maintained in 100-mm cell culture dishes (SPL; #20100) in a humidified 5% CO2 incubator at 37 °C. Cells were passaged every 2 days using trypsin (Gibco; #12605-010). STR sequencing (ATCC, 135-XV) and mycoplasma testing (Lonza, LT07-418) were carried out routinely. For C2C12 differentiation, the growth medium was replaced with differentiation medium. C2C12 cells were differentiated in DMEM supplemented with 1% (v/v) antibiotic-antimycotic solution and 2% (v/v) heat-inactivated horse serum (Gibco; #16050-130). The differentiation medium was changed every 2 days for a total of 6 days. Cell viability assay RD, RH30, OV90, A549, SW480, and HeLa cells were seeded at a density of 10,000 cells per well in a clear, flat-bottom 96-well plate (Corning; #3599) and sealed with breathable sealing tape (AeraSeal; SM-KIT-B). After 24 h of incubation, the cells were treated with KBD-1, KBD-2, cKBD-1, and cKBD-5 for 24 h. The culture medium was then replaced with 100 μl of fresh medium, and 10 μl of Cell Counting Kit-8 solution (Dojindo; CK04-13) was added to each well. After a 2 h incubation, the absorbance at 455 nm was measured using a microplate reader (Molecular Devices; SpectraMax i3x). Immunoblotting All cells were harvested and lysed in modified radioimmunoprecipitation assay (RIPA) buffer [150 mM NaCl, 1% IGEPAL CA-630 (Sigma; I8896), 0.5% deoxycholate (TCI; C0316), 5 mM NaF (Acros; AC201290250), 2 mM Na3VO4 (Thermo; S454-50), 1× protease inhibitor cocktail (PIC, Thermo; #87786), and 50 mM Tris–HCl (Sigma; T1503), pH 7.8] for 20 min on ice. The lysates were clarified by centrifugation at 20,000 g for 20 min at 4 °C. The protein concentration of the soluble fraction was measured using the BCA protein assay kit (Thermo; #23225). Samples were then mixed with 5× SDS loading buffer (Biosesang; SF2002-110-00) and heated at 95 °C for 5 min. Equal amounts of protein were analyzed by SDS-PAGE and transferred to PVDF membranes (Merck; IPVH00010). The membranes were blocked for 1 h at room temperature with 2% bovine serum albumin (BSA, MP Biomedicals; #0216006980) in Tris-buffered saline containing 0.1% Tween-20 (TBS-T, Sigma; P9416). To detect the target proteins, membranes were incubated overnight at 4 °C with primary antibodies at the following dilutions: 1:1000 for KLHL41 (Novusbio; NBP1-80787), KLHL40 (MyBioSource; MBS9215030), BRD4 (CST; #13440S), Flag (Sigma; #F1804) and AR (CST; #5153S); and 1:2000 for GAPDH (CST; #2118S). After washing with TBS-T, the membranes were incubated with HRP-conjugated secondary antibody (anti-rabbit, CST; #7074) at a 1:5000 dilution for 1 h at room temperature. After further washing with TBS-T, membranes were treated with the EZ-Western Lumi Pico Alpha Kit (Dogen; DG-WPAL250), and chemiluminescent signals were detected using a ChemiDoc XRS+ system (Bio-Rad; # 1708265). Western Blot images were analyzed by Imagelab 4.0 (Bio-Rad). Protein band intensities were quantified by ImageJ 1.53a. Cellular thermal shift assay (CETSA) RH30 cells were treated with either vehicle or 20 μM compounds for 3 h. After treatment, the RH30 cells were heated for 3 min at various temperature ranges, followed by incubation at 25 °C for 3 min. The heated cells were washed with 1× phosphate buffered saline (PBS, Welgene; LB001-02) and re-suspended in lysis buffer [PBS containing 0.4% IGEPAL CA-630 (Sigma; I8896) and 1× PIC (Thermo; #87786)]. Cells were lysed by three freeze-thaw cycles in liquid nitrogen. The cell lysates were clarified by centrifugation at 20,000 g at 4 °C for 20 min. The soluble protein fraction was mixed with 5× SDS (Biosesang; SF2002-110-00), heated at 95 °C for 5 min, and subjected to immunoblotting. siRNA transfection For KLHL41 knockdown experiments, short interfering RNA (siRNA) duplexes targeting KLHL41 (Bioneer; #10324-1) were used. siRNA oligonucleotides were transfected into RH30 cells using Lipofectamine RNAiMAX (Invitrogen; #13778-150) and Opti-MEM (Gibco; #31985-070), following the manufacturer’s instructions. Two days after transfection, cells were treated with either vehicle or KLHL41-BRD4 Degraders (KBDs) for 12 h to assess BRD4 degradation post-KLHL41 knockdown. siRNA Oligonucleotide sequences for KLHL41 are as follows siKLHL41 sense 5′- CUG AUC GCU UUG UAC AGA U TT -3′ antisense 5′- AUC UGU ACA AAG CGA UCA G TT -3′ KLHL41 overexpression The hKLHL41 plasmid (Origene; RC200295) and hKLHL41 C593S plasmid (Cloned by Cosmogenetech via overlap PCR method) were transiently transfected into HeLa cells using Lipofectamine LTX Plus (Invitrogen; #15338-100) and Opti-MEM (Gibco; #31985-070) according to the manufacturer’s protocol. One day after transfection, cells were treated with either vehicle or KLHL41-BRD4 Degraders (KBDs) for 12 h to observe BRD4 degradation. Click chemistry for pulldown assay For the diazirine probe, RH30 cells were treated with 15 μM of EN10-DA in the presence or absence of 100 μM of EN10 for 30 min, followed by photo-crosslinking using 365-nm UV irradiation on ice for 30 min. For the covalent probe, RH30 cells were treated with 10 μM of cEN10-Alkyne-Probe in the presence or absence of 100 μM of EN10 for 30 min. Afterward, the cells were washed with 1× PBS (Welgene; LB001-02) and lysed in RIPA buffer containing 1× protease inhibitor cocktail (PIC, Thermo; #87786). Next, 100 μg of protein lysate was labeled with 100 μM of Biotin-PEG4-picolyl azide (Broadpharm; BP-22954) along with 100 μM of tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (TBTA, Sigma; #678937), 1 mM of CuSO4 (Sigma; C8027), 1 mM of tris(2-carboxyethyl)phosphine (TCEP, Sigma; C4706), and 5% t-BuOH (TCI; B0706) for 1 h. The biotin-conjugated proteins were precipitated using cold acetone at –20 °C for 20 min, followed by centrifugation at 20,000 g, 4 °C for 7 min. After washing with acetone three times, the pellets were re-suspended in 50 μl of 1.2% sodium dodecyl sulfate (SDS, TCI; S0588) in PBS and diluted with 250 μl of PBS to a final concentration of 0.2% SDS in PBS. For bead blocking, 30 μl of streptavidin-agarose beads (Sigma; S1638) were incubated in 2% BSA (MP Biomedicals; #0216006980) in PBS for 2 h. The beads were then added to the protein lysate samples and incubated for 4 h at room temperature. Afterward, the beads were washed three times with PBS. Finally, the beads were mixed with 20 μl of 1× SDS sample buffer (Biosesang; SF2002-110-00), heated at 95 °C for 5 min, and subjected to immunoblotting. Preparation of recombinant KLHL41 and BRD4 The hKLHL41 plasmid (Origene; RC200295) and hBRD4 plasmid (Sino Biological; HG17954-CY) were cloned into the pET20b-hTPI vector (Addgene; #50723) with a C-terminal His10-tag by Cosmogenetech [Korea]. Each plasmid was transformed into BL21 (DE3) competent E. coli cells (Enzynomics; CP110). The BL21 cells were grown in 250 mL of LB broth (Biobasic; SD7002.250) at 37 °C with shaking until the OD600 reached 0.6. The cultures were then induced with a final concentration of 0.4 mM and 1 mM IPTG (Sigma; I5502), respectively, and incubated at 37 °C for 4 h. Cells were harvested by centrifugation at 5000 g, 4 °C for 10 min and re-suspended in protein buffer [150 mM NaCl, 1 mM DTT (Millipore; #1.11474.0025), and 10 mM Tris-HCl, pH 8.0] containing 1× PIC (Roche; #11873580001). Cells were lysed using a sonosmasher (Ulsso Hitech; ULH-700S) and clarified by centrifugation at 13,000 g, 4 °C for 30 min. The clarified lysates were incubated with Ni-NTA agarose beads (Qiagen; #30210) at 4 °C for 1 h, and the proteins were purified by elution with various concentrations of imidazole (Sigma; I5513) in protein buffer. Imidazole was removed by buffer exchange using a 10 K Amicon filter (Millipore; UFC501096). The proteins were further purified using an ÄKTA pure chromatography system (Cytiva) equipped with a Superdex 200 column (Cytiva; GE17-5175-01). The column was equilibrated with 1× PBS (Welgene; LB001-02), and the eluted protein fractions, identified by a UV absorption peak at 280 nm, were analyzed by SDS-PAGE electrophoresis. After electrophoresis, the proteins were concentrated using a 10 K Amicon filter by centrifugation at 13,000 g, 4 °C for 15 min, and the concentrations were measured using a BCA protein assay kit (Thermo; #23225). The purity of the proteins was determined by SDS-PAGE with the final purity over 90% in Coomassie staining (Biorad; #1610406). The purified proteins were stored at -80 °C for further experiments. Bio-Layer Interferometry (BLI) BLI experiments were performed using an Octet R8 protein analysis system (Sartorius) at 25 °C with a shaking speed of 1000 rpm. For KBD-1 and KLHL41 complex analysis, His-tagged KLHL41 proteins were diluted to 100 μg/ml in 1× PBS (Welgene; LB001-02) and loaded onto NTA biosensors (Sartorius; #18-5101) for 120 s, followed by a wash in PBS for 120 s. Subsequently, his-tagged KLHL41 proteins were incubated in varying concentrations of KBD-1 diluted in assay buffer (PBS containing 5% (v/v) DMSO) for 300 s to determine the association rate constant (kon), followed by a wash for 300 s in 5% (v/v) DMSO in PBS to determine the dissociation rate constant (koff). For ternary complex analysis, 10 μM of recombinant KLHL41 proteins were incubated with 500 μM (50×) of cKBD-1 in 5% (v/v) DMSO in PBS for 30 min to facilitate covalent binding between KLHL41 and cKBD-1. Subsequently, the KLHL41/cKBD-1 complex was biotinylated with 30 μM biotin for 30 min. Biotinylated proteins were then loaded onto SA biosensors (Sartorius; #18-5019) for 600 s, washed in 5% (v/v) DMSO in PBS for 300 s, quenched with 50 μg/mL biocytin for 120 s, and washed again in 5% (v/v) DMSO in PBS for 300 s. The preformed KLHL41/cKBD-1 complex was subsequently incubated in varying concentrations of BRD4 diluted in 5% (v/v) DMSO in PBS for 300 s to determine kon, followed by a wash with 5% (v/v) DMSO in PBS for 300 s to determine koff. All binding curves were processed by reference subtraction, and kinetic parameters were calculated using Octet Analysis Studio (version 13.0.0.29). Virtual KLHL41 ligand screening Data Preparation Due to the absence of the crystal structure of KLHL41 (UniProt ID: O60662), its model structure with AlphaFold ID AF-O60662-F1 was obtained from the AlphaFold database (https://alphafold.ebi.ac.uk/)67. The model was accessed on September 23, 2021 and was generated using the canonical sequence downloaded from UniProt (ID: O60662). Notably, the majority of the predicted structure is of very high confidence, with a predicted local distance difference test (pLDDT) score exceeding 90 suggesting that the model structure is accurate enough for virtual screening68. The potential ligand binding sites of KLHL41 were predicted using the COACH-D server23. Potential KLHL41 ligands were screened from the Hit Locator Library (HLL-230) retrieved on July 20, 2020, comprising a total of 230,240 molecules. The library is structured to include a minimum of three compounds per cluster, representing the entire Enamine screening collection. The molecules in the library adhere to the following properties: molecular weight: < 450 Da, clogP: < 5, topological polar surface area < 120 Å2, number of hydrogen bond acceptors ≤ 7, number of hydrogen bond donors ≤ 3, number of rotatable bonds ≤ 8, and number of heavy atoms ≤ 30. For AutoDock-GPU calculations, the 3D structures and partial charges of molecules were obtained from SMILES format to PDBQT format using the OpenBabel software69. Ligand Docking Simulation To prepare the binding pocket for protein-ligand interactions, AutoGrid4, a package within AutoDock4, was employed. The center of the grid was set to the center of mass of the ligand IQK as the origin. A 3D grid box was generated with dimensions of 31 grid points in the x, y, and z directions spanning 30 Å to define the binding pocket. A three-tiered hierarchical protein-ligand docking simulation was conducted to predict the interactions and binding energies between KLHL41 and potential ligands. First, docking calculations were performed using AutoDock-GPU with the default setting on Enamine Hit Locator Library. From the docking results, the top 10,000 molecules with the best docking scores were selected. Second, the 10,000 selected ligands were further docked using the Glide SP (Standard Precision) program. From this step, the top 1000 ligands were chosen for further refinement. Third, the top 1000 ligands underwent Glide XP (Extra Precision) docking calculations for enhanced accuracy. Glide XP incorporates additional scoring functions and thorough sampling to achieve precise binding predictions. This stepwise approach ensures both computational efficiency and accuracy in identifying promising candidate molecules. Molecular modeling Amber19 was used for molecular dynamics simulation. To refine the predicted binding energies and validate the strongest binding ligands identified through Glide XP docking, absolute binding free energy (ABFE) calculations were conducted using Yank (www.getyank.org). Yank is a GPU-accelerated Python framework designed to facilitate alchemical free energy calculations. The simulation inputs, including the ligand topology and force field parameters, were prepared using Antechamber. All ABFE simulations were conducted using the default setting of Yank. The protein and ligands were represented using the Amber ff14sb70 and GAFF271 force field. All systems were solvated in a rectangular water box consisting of TIP4P-Ew water molecules72 whose edges are 10 Å away from any atom of the protein and ligand. All systems were neutralized with sodium and chloride ions. Long-range interactions were calculated with the particle mesh Ewald method73. A time step of 2 femtoseconds was used. The system temperature was maintained at 300 K, and the pressure was set to 1 atm. For each ligand, a total of 10,000 Markov chain Monte Carlo iterations were performed, with each iteration consisting of 500 dynamics steps. For ABFE calculations, restraints, electrostatic, and van der Waals interactions changed from one to zero with decreasing λ values with an interval of 0.05 corresponding to 63 windows. Sample preparation and mass spectrometry for cKBD-1 binding site study For covalent binding between KLHL41 and cKBD-1, 20 μM of purified recombinant KLHL41 proteins were incubated with 1 mM cKBD-1 in 10% (v/v) DMSO in PBS for 30 min. After incubation, excess cKBD-1 was removed, and the buffer was changed to lysis buffer from the EasyPep Mini sample preparation kit (Thermo; A40006) using a 10 K Amicon filter (Millipore; UFC501096) by centrifugation at 13,000 g and 4 °C for 15 min. Protein reduction, alkylation, digestion, and peptide clean-up were conducted according to the manufacturer’s instructions. Briefly, 100 μg of the protein sample in 100 μL of lysis buffer was mixed with 50 μL of reduction solution and 50 μL of alkylation solution, with gentle mixing. Samples were heated at 95 °C for 10 min to reduce and alkylate the protein sample. After incubation, samples were cooled to room temperature. For digestion, 50 μL of the Trypsin/Lys-C protease mix solution was added to the reduced and alkylated protein sample solution. The samples were incubated with shaking at 37 °C for 3 h to digest. After incubation, 50 μL of digestion stop solution was added to the sample and gently mixed followed by peptide clean-up. Following the peptide clean-up, the peptides were dried using an IR vacuum concentrator (N-Biotek; NB-503CIR). Dried peptides were desalted by C18 StageTip (Supelco, #66883-U) and resuspended in 3% acetonitrile (ACN) (Sigma; #34851)/0.1% formic acid (FA) (Thermo Scientific, #85178) for mass spectrometry. Mass spectrometry was performed by an Orbitrap Exploris 480 coupled with a Vanquish Neo UHPLC system (Thermo Scientific). Peptides were loaded onto a trap column Acclaim PepMap™ (150 mm × 75 µm, particle size 3 µm, bed length 20 mm, Thermo Scientific, 164535) and separated on a C-18 reversed phase (RP) µPAC™ Neo HPLC column (180 µm bed; 110 cm length; 2.5 × 16 µm pillar diameter, Thermo Scientific, COL-NANO110NEOB) with 2–4% (2–5.5 min), 4–20.5% (5.5–55.5 min), 20.5–33% (55.5–85.5 min), 33–90% (85.5–88.5 min), 90% (88.5–103.5 min) gradient of mobile phase B (ACN/0.1% FA) at 250 nL/min flow rate. FAIMS Pro Interface was used with a cycle type of 1.25 s/CV with FAIMS CV of −40/ − 60/ − 80 V). MS1 full scans were acquired by Orbitrap (resolution 60,000, 375–1500 m/z, automatic gain control (AGC) target 1 × 106, maximum injection time 150 ms). Advanced peak determination algorithm was used, and precursor ions were isolated with a 2-Th window in the quadrupole and fragmented by normalized collision energy (NCE) 30%. Dynamic exclusion (±10 ppm) was applied for 43 s. MS2 spectra were analysed with the Orbitrap (resolution 22,500, AGC 7.5 × 104, maximum injection time 150 ms). RAW files were analysed by Proteome Discoverer3.1.0.638 (Thermo Scientific). The reference database was constructed from the human proteome downloaded from UniProt on November 8, 2023 containing both Swiss-Prot and TrEMBL entries followed by an appendage with common contaminants) and reversed for target-decoy false discovery rate (FDR) estimation. Database searching was done by Sequest HT with a 10-ppm precursor ion tolerance and 0.02 Da fragment ion tolerance. Dynamic modifications were set including oxidation at methionine (+15.995 Da), cKBD-1 at cysteine (+784.271 Da), cysteine carbamidomethylation (+57.021 Da), peptide N-terminal modifications, including acetylation (+42.011 Da), methionine-loss (-131.040 Da), and acetylation at methionine-loss (-89.030 Da). Maximum two missed cleavages were allowed. Peptide-spectrum matches (PSMs) were filter by a Percolator with 1% FDR. Then 1% FDR controls were applied to both peptide and protein level. Peptide abundances were calculated based on the precursor intensity after normalization by the total peptide amount. Sample preparation and mass spectrometry for quantitative proteome analysis RH30 cells were treated with either vehicle, 40 μM KBD-1, or 10 nM of cKBD-1 for 12 h. After washing twice with 1 × PBS (Welgene; LB001-02), the cells were harvested and lysed in lysis buffer from the EasyPep Mini sample preparation kit (Thermo; A40006). The lysates were clarified by centrifugation at 20,000 g for 20 min at 4 °C, and the protein concentration of the soluble fraction was measured using the BCA protein assay kit (Thermo; #23225). For each condition, 100 μg of protein was used for quantitative proteome analysis. Protein reduction, alkylation, digestion, and peptide clean-up were conducted using the EasyPep Mini sample preparation kit. Briefly, 100 μg of the protein sample in 100 μL of lysis buffer was mixed with 50 μL of reduction solution and 50 μL of alkylation solution, with gentle mixing. Samples were heated at 95 °C for 10 min to reduce and alkylate the protein sample. After incubation, samples were cooled to room temperature. For digestion, 50 μL of the Trypsin/Lys-C protease mix solution was added to the reduced and alkylated protein sample solution. The samples were incubated with shaking at 37 °C for 3 h to digest. After incubation, 50 μL of digestion stop solution was added to the sample and gently mixed followed by peptide clean-up. Following the peptide clean-up, the peptides were dried using an IR vacuum concentrator (N-Biotek; NB-503CIR). For quantitative labeling, dried samples were resuspended in 80 μL of 200 mM EPPS buffer, pH 8.5 (Alfa Aesar; J6147). Then, 80 μg of peptide sample was incubated with 300 μg of TMT (Thermo; A44520) in a final concentration of 30% (v/v) LC-MS grade acetonitrile (ACN) for 1 h. 5% hydroxylamine (Sigma; #438227) in LC-MS grade water was added to each labelling reaction to quench it, resulting in a final hydroxylamine concentration of 0.3%, and the samples were incubated for 15 min at room temperature. All labeled samples were combined into a new LoBind tube (Eppendorf; #22431081) and dried using an IR vacuum concentrator. Dried samples were pre-fractionated using a high pH reversed-phase peptide fractionation kit (Thermo; #84868). Samples were fractionated with varying percentages of ACN and then dried with an IR vacuum concentrator. The fractions were desalted using C18 spin tips and columns (Thermo; #89870) according to the manufacturer’s protocol. Desalted samples were dried using an IR vacuum concentrator. Dried fractions were desalted by C18 StageTip, and resuspended in 3% ACN/0.1% FA for mass spectrometry. Mass spectrometry was performed by an Orbitrap Exploris 480 coupled with a Vanquish Neo UHPLC system (Thermo Scientific). Peptides were loaded onto a trap column Acclaim PepMap™ (150 mm × 75 µm, particle size 3 µm, bed length 20 mm, Thermo Scientific, 164535) and separated on a C-18 reversed phase (RP) µPAC™ Neo HPLC column (180 µm bed; 110 cm length; 2.5 × 16 µm pillar diameter, Thermo Scientific, COL-NANO110NEOB) with 4.5–9% (4–22 min), 9-21% (22–101.5 min), 21–40% (101.5–136.5 min), 40-90% (136.5–139.5 min), 90% (139.5-159.5 min) gradient of mobile phase B (ACN/0.1% FA) at 250 nL/min flow rate. FAIMS Pro Interface was used with a cycle type of 1.25 s/CV with FAIMS CV of −40/ − 60/ − 80 V). MS1 full scans were acquired by Orbitrap (resolution 12,000, 375–1500 m/z, automatic gain control (AGC) target 3 × 106, maximum injection time 50 ms). Precursor fit filter (50% for a fit window 0.5 Th) was used, and precursor ions were isolated with a 0.5-Th window in the quadrupole and fragmented by normalized collision energy (NCE) 36%. Dynamic exclusion (±10 ppm) was applied for 43 s. MS2 spectra were analysed with the Orbitrap (resolution 45,000, AGC 7.5 × 104, maximum injection time 87 ms). RAW files were analyzed by MSFragger 4.1 with FragPipe (v22.0) user interface. The reference database was constructed from the human proteome downloaded from UniProt on November 18, 2024 containing Swiss-Prot entries, followed by an appendage with common contaminants) and reversed for target-decoy FDR estimation. Database searching was done with a ± 20-ppm precursor ion tolerance and 20-ppm fragment ion tolerance. Static modifications included cysteine carbamidomethylation (+57.021 Da) and TMTpro on lysine residues (+304.207 Da), while variable modifications included methionine oxidation (+15.995 Da), and TMTpro at peptide N-terminus (+304.207 Da). Maximum two missed cleavages were allowed. Peptide-spectrum matches were validated by a Percolator, and proteins were inferred by ProteinProphet with default settings. 1% FDR controls were applied, and proteins were filtered with a probability threshold of over 0.99. For the reporter ion quantification, IonQuant was applied with a tolerance of 20 ppm, and protein abundances were normalized by median centering method. Protein-wise linear models combined with empirical Bayesian statistics from Limma was performed followed by multiple testing p value correction with a Benjamini-Hochberg method. Sample preparation and mass spectrometry for probe pulldown analysis RH30 cells were treated with EN10-DA (15 μM) in the presence or absence of EN10 (50 μM). Cells were pretreated with EN10 for 30 min, followed by co-treatment with EN10-DA for 30 min, and then photo-crosslinked using 365-nm UV irradiation on ice for 30 min. For covalent probe experiments, RH30 cells were treated with cEN10-Alkyne-Probe (10 μM) in the presence or absence of cEN10 (25 μM). Cells were pretreated with cEN10 for 3 h, followed by co-treatment with the probe for 30 min. After treatment, cells were washed with 1× PBS (Welgene; LB001-02), and lysed in RIPA buffer containing protease inhibitor cocktail. For each condition, 200 μg of total protein was subjected to click chemistry with Biotin-PEG4-picolyl azide (Broadpharm; BP-22954) along with 100 μM of tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (TBTA, Sigma; #678937), 1 mM of CuSO4 (Sigma; C8027), 1 mM of tris(2-carboxyethyl)phosphine (TCEP, Sigma; C4706), and 5% t-BuOH (TCI; B0706) for 1 h. Excess biotin reagent was removed using 10 kDa Amicon Ultra filters (Millipore; UFC501096) by buffer exchange with PBS four times. Biotinylated proteins were then enriched using streptavidin magnetic beads (Thermo; #88816) for 3 h. The beads were washed three times with PBS, and the buffer was replaced with 100 μL of lysis buffer from the EasyPep Mini sample preparation kit (Thermo; A40006). Samples were subsequently processed using the EasyPep Mini sample preparation kit according to the manufacturer’s protocol. Peptides were labeled with TMT reagents, pooled, and desalted using C18 StageTips prior to LC–MS/MS analysis. Mass spectrometry was performed by an Orbitrap Exploris 480 coupled with a Vanquish Neo UHPLC system (Thermo Scientific). Peptides were loaded onto a trap column Acclaim PepMap™ (150 mm × 75 µm, particle size 3 µm, bed length 20 mm, Thermo Scientific, 164535) and separated on a C-18 reversed phase (RP) µPAC™ Neo HPLC column (180 µm bed; 110 cm length; 2.5 × 16 µm pillar diameter, Thermo Scientific, COL-NANO110NEOB) with 4.5–9% (4–22 min), 9-21% (22–101.5 min), 21–40% (101.5–136.5 min), 40–90% (136.5-139.5 min), 90% (139.5–159.5 min) gradient of mobile phase B (ACN/0.1% FA) at 250 nL/min flow rate. FAIMS Pro Interface was used with a cycle type of 1.25 s/CV with FAIMS CV of −40/ − 60/ − 80 V). MS1 full scans were acquired by Orbitrap (resolution 12,000, 375–1500 m/z, automatic gain control (AGC) target 3 × 106, maximum injection time 50 ms). Precursor fit filter (50% for a fit window 0.5 Th) was used, and precursor ions were isolated with a 0.5-Th window in the quadrupole and fragmented by normalized collision energy (NCE) 36%. Dynamic exclusion (±10 ppm) was applied for 43 s. MS2 spectra were analysed with the Orbitrap (resolution 45,000, AGC 7.5 × 104, maximum injection time 87 ms). RAW files were analyzed by MSFragger 4.1 with FragPipe (v22.0) user interface. The reference database was constructed from the human proteome downloaded from UniProt at November 18, 2024 containing Swiss-Prot entries followed by an appendage with common contaminants) and reversed for target-decoy FDR estimation. Database searching was done with a ± 20-ppm precursor ion tolerance and 20-ppm fragment ion tolerance. Static modifications included cysteine carbamidomethylation (+57.021 Da) and TMTpro on lysine residues (+304.207 Da), while variable modifications included methionine oxidation (+15.995 Da), and TMTpro at peptide N-terminus (+304.207 Da). Maximum two missed cleavages were allowed. Peptide-spectrum matches (PSMs) were validated by a Percolator, and proteins were inferred by ProteinProphet with default settings. 1% FDR controls were applied, and proteins were filtered with a probability threshold of over 0.99. For the reporter ion quantification, IonQuant was applied with a tolerance of 20 ppm, and protein abundances were normalized by median centering method. Only proteins supported by at least two peptide–spectrum matches (PSMs) were included in downstream analyses. Protein-wise linear models combined with empirical Bayesian statistics from Limma was performed followed by multiple testing p value correction with a Benjamini-Hochberg method. In vivo cKBD-1 treatment After a one-week adaptation period, all experiments were carried out using 9-week-old C57BL/6 male mice (20–25 g). The animals were divided into four experimental groups, with three mice in each group: group 1, treated with vehicle (IV); group 2, treated with vehicle (IP); group 3, treated with cKBD-1 (IV, 5 mpk); group 4, treated with cKBD-1 (IP, 5 mpk). cKBD-1 (5 mpk) or vehicle {distilled water containing 5% DMSO and 40% polyethylene glycol 400 (TCI; N0433)} was administered via either intravenous (IV) or intraperitoneal (IP) injection. All mice were sacrificed after 24 h after treatment. Pharmacokinetic analysis The pharmacokinetic properties of cKBD-1 were evaluated using three 8-week-old SD male rats (~300 g). cKBD-1 was formulated in DMSO:PEG400:water (5:40:55, v/v) and administered by intravenous injection at 5 mpk (n = 3, biological replicate). Blood samples were collected at 0, 2 min, 10 min, 0.5, 1, 2, 4, and 8 h after dosing, and plasma was separated for analysis. For plasma sample preparation, 50 μL of plasma was mixed with 150 μL of internal standard solution (carbamazepine, 200 pg/mL in acetonitrile), vortexed for 3 min, and centrifuged at 15,000 g for 5 min. The supernatant was collected, and 5 μL was injected for LC–MS/MS analysis. LC–MS/MS analysis was performed using a Vanquish UHPLC system (Thermo Scientific) coupled to a TSQ Quantis Plus quadrupole mass spectrometer (Thermo Scientific) with positive electrospray ionization. Chromatographic separation was achieved on a Hypersil GOLD C18 column (2.1 mm × 100 mm, 3 μm, Thermo Scientific) using a gradient of 0.1% formic acid in water and 0.1% formic acid in acetonitrile at a flow rate of 0.3 mL/min. Quantification was performed in MRM mode using the transitions m/z 821.304 → 665.05 for cKBD-1 and m/z 237 → 194 for carbamazepine. Pharmacokinetic parameters were calculated from the plasma concentration–time profiles using Analyst software. Plasma stability After a one-week adaptation period, all experiments were carried out using 8-week-old ICR male mice (25–30 g). Plasma stability of cKBD-1 was assessed using plasma from fourteen mice. The reaction was initiated by spiking a 2 mM compound stock solution into plasma, yielding a final compound concentration of 20 μM. The incubation mixture was maintained at 37 °C in a heat block. At each designated time point (0, 10, 30, 60, and 120 min), aliquots of the reaction mixture were withdrawn and immediately transferred into separate microcentrifuge tubes containing ice-cold acetonitrile:methanol (1:1, v/v) to terminate the reaction and precipitate plasma proteins. Samples were vortex-mixed and centrifuged at 4000 g for 20 min. Following centrifugation, the clear supernatant was collected and transferred to a fresh tube for LC–MS analysis. The remaining concentration of the compound at each time point was quantified by LC–MS, and the plasma half-life (t1/2) was estimated from the compound depletion profile assuming first-order reaction kinetics. Half-life values were derived from the fitted rate constant (k) according to the following equation: t1/2 = 0.693 / k. For the LC–MS experiments, Agilent InfinityLab LC/MSD XT single quadrupole with the Agilent 1260 infinity II LC system was used with electrospray, positive ion mode. The chromatographic conditions are as follows: column, 50 × 3.0 mm I.D., packed with 2.7 μm C18 (Agilent Poroshell 120); mobile phase A, 0.1% formic acid in purified deionized water; mobile phase B, 0.1% formic acid in MeCN; flow rate: 0.8 mL/min; Injection Volume: 10 μL; run time: 11.01 min. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review file Source data Source Data Acknowledgements Figures 4a, 5n are partially created with BioRender.com and finalized with powerpoint. Author contributions J.Y. and J.P. designed the study. J.Y., Juyong Lee and J.P. prepared the figures, and wrote the manuscript. J.Y. and S.K. performed the CETSA screening and molecular biology work. Jaeseok Lee synthesized the compounds. J.C. and Juyong Lee conducted virtual screening and docking analysis. S.J., H.C., and S.Y. carried out the proteomics study under supervision of H.P. G.Y. and S.L. conducted plasma stability and pharmacokinetic (PK) studies. All authors critically reviewed the manuscript. Peer review Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding This work was supported by the Creative Challenging Research Grant (RS-2023-00250887 to J.Y.), the Basic Science Research Program (2023R1A2C1007899 to J.P., NRF-2022R1C1C1005080 to J.C., RS-2022-NR071908 to H.P., RS-2025-02214577 to H.P., 2022M3E5F3081268 to J.L. and RS-2023-00256320 to J.L.), Priority Research Institute Program (RS-2023-00271205 to J.P.), Institute of Information & communications Technology Planning & Evaluation (IITP) grant (RS-2023-00220628 to J.L., Artificial intelligence for prediction of structure-based protein interaction reflecting physicochemical principles), and Bio & Medical Technology Development Program (RS-2024-00352229 to J.C. and J.L.) funded by the Ministry of Science and ICT for the research of this work. This work was also supported by Korea Basic Science Institute (KBSI) National Research Facilities & Equipment Center (NFEC) grant (2019R1A6C1010006 to J.P.), Global-Learning & Academic research institution for Master’s·PhD students, and Postdocs (G-LAMP) program (RS-2023-00301850 to J.P. and RS-2023-00301976 to H.P.) and BK21FOUR Program (5120200513755 to J.C.) from the Ministry of Education, South Korea for the research and publication of this work. J.Y. was supported by Sookmyung Women's University Research Grants (1-2603-2010). This work was also supported by Creative-Pioneering Researchers Program through Seoul National University (No. 860-20250088 to H.P.), by Yuhan Innovation Program (J.P.) funded by Yuhan Corporation, by Sookmyung Women's University Research Grant (102603-2010 to J.Y.), and by National Institute of Health (NIH) research project (2025-ER-2011-00 to J.P) funded by NIH of Korea for the research of this work. Data availability The authors declare that the data supporting the findings of this study are available within the article, the source data, and the Supplementary Information. All data supporting the findings of this study are provided in the Source Data file. Source data are provided with this paper. The mass spectrometry data generated in this study have been deposited in the MassIVE repository under accession code MSV000096752 and ProreomeXchange repository under accession code PXD059370. Source data are provided with this paper. Competing interests Juyong Lee is serving as the Chief Technology Officer of Arontier and holds stock options in the company. The remaining authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Junhyeong Yim, Jaeseok Lee, Solbi Kim, Jieun Choi. Contributor Information Hankum Park, Email: hanpark@snu.ac.kr. Juyong Lee, Email: nicole23@snu.ac.kr. Jongmin Park, Email: jpark@kangwon.ac.kr. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-73252-4. References Associated Data Supplementary Materials Supplementary Information Reporting Summary Transparent Peer Review file Source Data Data Availability Statement The authors declare that the data supporting the findings of this study are available within the article, the source data, and the Supplementary Information. All data supporting the findings of this study are provided in the Source Data file. Source data are provided with this paper. The mass spectrometry data generated in this study have been deposited in the MassIVE repository under accession code MSV000096752 and ProreomeXchange repository under accession code PXD059370. Source data are provided with this paper.