With the rapid progressive efforts from modern drug discovery and development, numerous promising paradigms for disease treatment are emerging from the wealth of medicinal chemistry and biology. A comprehensive understanding of the molecular mechanisms underlying the function of targeted biological systems expedites the drug development process. Among these innovative approaches, PROteolysis TArgeting Chimeras (PROTACs), also known as hetero-bifunctional degraders, stand out as a revolutionary paradigm for selectively degrading a diverse array of disease-associated targets (Dale et al., 2021; Békés et al., 2022; Ramachandran and Ciulli, 2021). The formation of a ternary complex involving a PROTAC binding to an E3 ligase and its neo-substrate, commonly referred to as the protein of interest (POI), triggers the cell’s native protein degradation machinery (e.g. ubiquitination), which marks the target protein for proteasomal degradation. Previous studies have highlighted the importance of forming a high binding-affinity, stable and long-lived E3-PROTAC-POI complex for achieving potent degradation (Bondeson et al., 2018). However, it is noteworthy that the formation of a stable ternary complex does not always correlate with the degradation potency of the POI (Bondeson et al., 2018; Huang et al., 2018; Nowak et al., 2018).

The final degradation of a target protein, involving the three-step ubiquitination process (activation-E1, conjugation-E2, and ligation-E3) is facilitated by the binding a ternary complex induced by a PROTAC. Recent studies used E3 ligases such as MDM2, IAP, RNF4, and βTRCP to degrade various biological targets (Zhao et al., 2019; Tinworth et al., 2019; Ward et al., 2019; Ottis et al., 2017). While stable ternary complexes are often associated with high POI degradation efficiency, discrepancy between stable ternary complexes and degradation have also been observed. For instance, certain von Hippel Lindau (VHL)-based PROTACs exhibit high binding affinity to POIs but fail to induce degradation despite forming stable ternary complexes. Conversely, proteins like p38α, which exhibit low binding affinity with VHL-based PROTACs, undergo rapid degradation within a short timeframe (Bondeson et al., 2018; Smith et al., 2019; Lai et al., 2016). Therefore, it is evident that the formation of a stable ternary complex does not guarantee high degradation efficiency. Recent studies have suggested that the accessibility of lysine (Lys) residues of the POI after the formation of stable ternary complexes is critical for degradation and requires further research (Bondeson et al., 2018; Han, 2020; Bai et al., 2022).

The formation of a stable E3-PROTAC-POI complex represents the initial crucial step for POI degradation. Several modeling-based approaches have been developed to optimize and rationalize the PROTAC-mediated ternary complex. For example, several studies have employed protein–protein docking, linker conformational searching (Drummond and Williams, 2019; Drummond et al., 2020; Zaidman et al., 2020; Bai et al., 2021), and machine learning approaches (Rao et al., 2023; Zheng et al., 2022) to generate ensembles of E3-PROTAC-POI complexes in silico. Recent advancement in docking algorithms aimed to provide higher accuracy and energetically favorable ternary complexes (Ignatov et al., 2023; Pereira et al., 2023). Notably, recent studies have suggested that, in addition to forming a ternary complex, the orientation of E3-PROTAC-POI allowing accessibility of the surface Lys residues of the POI are important for ubiquitination. Because the subsequent ubiquitination process occurs in a ligase complex comprising multiple proteins, recent research efforts have begun to investigate the assembly of the E3-PROTAC-POI complex with the entire ligase complex and key enzymes E2/ubiquitin (E2/Ub) for ubiquitination. For example, two studies applied docking and structure-based modeling approaches to model CDKs-PROTACs and BCL-xl-PROTACs in a Cereblon (CRBN)-based Culling-Ring Ligase 4 A (CRL4A) complex and a VHL-based CRL complex, respectively (Bai et al., 2022; Lv et al., 2021). These studies integrated experimentally determined protein structures to construct a degradation complex using protein structure alignment approaches, and the distance measurements between Lys residues and Ub to show potential ubiquitination. Moreover, enhanced sampling molecular dynamics (MD) techniques have also been employed to investigate the spatial organization of POI within the context of a ligase complex for ubiquitination (Dixon et al., 2022). Nevertheless, challenges remain in accurately modelling the structural dynamics of the degradation machinery complex, and gaining a deeper understanding of the steps involved in POI ubiquitination would aid PROTAC design.

Several existing studies have examined a set of PROTACs targeting the same E3 ligase and POI for structure–activity relationships (Liu et al., 2022; Bricelj et al., 2021; Lee et al., 2022). Many of these studies have maintained consistent binders for the E3 and POI while varying the linker regions. For example, studies of several PROTACs (i.e. small-molecule degrader of bromodomain and extra-terminal domain [dBET] family) designed for degrading bromodomain-containing protein 4 (BRD4^BD1) utilized the same BRD inhibitor (i.e. thieno-triazolo-1,4-diazepine [JO1]) and E3 ligase inhibitor, pomalidomide (or lenalidomide and thalidomide) while varying the linkers (Nowak et al., 2018; Zhou et al., 2018; Lu et al., 2015; Filippakopoulos et al., 2010). Although the crystal structures of these CRBN-dBETs-BRD4^BD1 ternary complexes have highly similar CRBN-BRD4^BD1 contacts, significant differences in degradation capabilities (DC_50/5h) have been observed (Nowak et al., 2018; Zhou et al., 2018; Winter et al., 2015; Qin et al., 2018). These findings underscored the need to further understanding of how the linkers and the formation of stable ternary complex may influence degradation capability. Although the linker region may result in minor differences in ligand solubility and/or membrane permeability, the proximity and accessibility of surface Lys residue(s) of the POI to the E2/Ub cascade in the ubiquitination site can be due to different linkers (Smith et al., 2019; Zhang et al., 2022). Thus, investigating the structures of the degradation complexes with PROTAC-induced ubiquitination while considering protein structural dynamics provides valuable insights into PROTAC degradability.

In this study, we employed protein–protein docking, structural alignment, atomistic MD simulations, and post-analysis to model a series of CRBN-dBET-BRD4^BD1 ternary complexes and the entire degradation machinery complex consisting of BRD4^BD1 and a CRL4A E3 ligase scaffold (Figure 1). These degraders, with different linker properties, were all capable of forming stable ternary complexes, but exhibited different degradation capabilities (Nowak et al., 2018; Zhou et al., 2018; Winter et al., 2015; Qin et al., 2018). The best degrader, dBET70, had a DC_50/5h of about 5 nM, followed by dBET23 (DC_50/5h~50 nM) (Figure 1D and Figure 1—figure supplement 1). Although utilizing the exact same warheads, other degraders such as dBET1 and dBET57 had a DC_50/5h of about 500 nM. Our studies identified structural features of the dBETs that contribute to large-scale protein motions and explained the connection between the cellular activities and degradability reported for the dBETs with atomics details. Because finding energetically stable ternary conformations is a critical first step for protein degradation (Nowak et al., 2018; Zaidman et al., 2020; Bai et al., 2021), in addition to protein–protein docking, we also performed MD simulations to thoroughly cover the conformational space and performed energy calculations to ensure that our modeled ternary complexes were thermodynamically stable.

Figure 1 with 2 supplements see all

Download asset Open asset

Assembling these stable CRBN-dBETx-BRD4^BD1 complexes into multiple modeled CRL4A E3 ligase-based scaffold conformations, we first observed that no surface Lys residue(s) of BRD4^BD1 was ready for the next ubiquitination step in the modeled degradation machinery complexes. Nevertheless, our unbiased MD simulations illustrated protein structural dynamics of the entire complex and local sidechain arrangements to bring Lys residue(s) to the catalytic pocket of E2/Ub for reactions. Post-analysis revealed the essential motions of the degradation complex and interactions crucial for ubiquitination. Our results relate the structural motion to potential ubiquitination and explain how the linker property affecting the degradation potency. Our results show the importance of the dynamic features in protein structure and how the linker region of a PROTAC may contribute to protein motions to achieve PROTAC-mediated POI degradation.

To understand why PROTACs, although forming similar stable E3-PROTAC-POI ternary complexes, induce different degradation efficacy of POIs, we used an integrative approach that combines docking, structural alignment and atomistic MD simulations. Due to the limited availability of experimental determined CRBN-dBETx-BRD4^BD1 conformations, we initially employed protein–protein docking to generate numerous CRBN-dBETx-BRD4^BD1 ternary complexes for four degraders with different degradation efficacies (Appendix 1—table 1 and Figure 1—figure supplement 1). Subsequently, we constructed degradation machinery complexes by assembling various ternary complexes into the CRL4A E3 ligase scaffold. MD simulations and subsequent post-analysis were employed to quantify protein dynamics and to identify crucial hinge regions governing the structure–dynamics–function relationship within the degradation complexes. With quantitative data, we revealed the importance of the structural dynamics of dBETx-induced motions, which arrange positions of the surface lysine residues of BRD4^BD1 and the entire degradation machinery.

Conformational ensembles of CRBN-dBETx-BRD4^BD1 ternary complexes

An effective design of PROTACs relies on a comprehensive understanding, how a degrader, such as dBET, yields different conformations of CRBN-dBETx-BRD4^BD1 ternary complexes, ultimately contributing to degradation efficiency. To generate the conformation ensemble, we first constructed the ternary complexes through protein–protein docking using the Molecular Operating Environment (MOE) program and removed complexes with atomic clashes. Figure 1—figure supplement 1 illustrates the docked CRBN-dBETx-BRD4^BD1 conformations, with all conformations, except CRBN-dBET57-BRD4^BD1, presenting over 160 conformations and displaying various inter-molecular orientations between CRBN and BRD4^BD1. Notably, PROTAC dBET57, characterized by a shortest linker, exhibited a constrained protein–protein orientation, resulting in only 24 distinct ternary conformations (Figure 1—figure supplement 1C and Appendix 1—table 1). To validate the modelled conformations, we superimposed the docking results of CRBN-dBET23-BRD4^BD1 and CRBN-dBET70-BRD4^BD1 ternary complexes onto available crystal structures, revealing highly similar conformations. The computed smallest C$\alpha$-root mean square deviation (RMSD) values were 1.99 Å, and 2.60 Å, respectively (Figure 1—figure supplement 2). These ternary ensembles, as depicted in Figure 1—figure supplement 1 were subsequently utilized to construct degradation machinery complexes, as detailed in the Materials and methods subsection (Figure 2).

Figure 2 with 6 supplements see all

Download asset Open asset

Furthermore, we conducted multiple MD simulations in explicit solvent for five CRBN-dBETx-BRD4^BD1 complexes, encompassing four initial conformations derived from our docking models and one from an existing crystal structure (Table 1, total of 15 runs for ternary complexes). Only dBET23 crystal structure is available with the PROTAC and both proteins, while the experimentally determined ternary complexes of dBET1, dBET57 and dBET70 are not available. During 400-ns simulations, the two warheads of a PROTAC bound tightly to CRBN and BRD4^BD1, while the linker displayed high flexibility by adopting different conformations and facilitating the interaction between the two proteins across different protein–protein contact surfaces (Figure 3). The residue contact map throughout the 400-ns MD simulation also showed different patterns of protein-protein interactions, indicating that the linkers were able to adopt different conformations (Figure 3—figure supplement 1). Interestingly, despite the molecular docking results suggested limited fluctuation in CRBN-dBET57-BRD4^BD1, our MD simulations revealed considerable variation in ternary conformations. The observed large-scale motions in all CRBN-dBETx-BRD4^BD1 complexes, as predicted by both docking and MD simulations, underscored the importance of CRBN and BRD4^BD1 orientations influenced by the presence of dBETs as key factors of contributing to degradation efficiency.

Figure 3 with 2 supplements see all

Download asset Open asset

Table 1

MD index	Run	MD index	Run	MD index	Run
CRBN-dBET1_#35-BRD4BD1	3	A1_dBET1_#35	1	B1_dBET1_md3	2
CRBN-dBET23_#14-BRD4BD1	3	A1_dBET23_#14	1	B1_dBET23_#14	2
CRBN-dBET57_#9-BRD4BD1	3	A1_dBET57_#9	1	B1_dBET57_#9_MD1	2
				B1_dBET57_#9_MD2	1
CRBN-dBET70_#91-BRD4BD1	3	A1_dBET70_#91	1	B1_dBET70_#91	2
CRBN-dBET23_xray-BRD4BD1	3	A1_dBET23_xray	1	B1_dBET23_xray	1

Protein structural dynamics in degradation machinery complex

Among the 1226 distinct degradation machinery complexes constructed, none showed the structure of a Lys residue of BRD4^BD1 was positioned for ubiquitination. Therefore, for each PROTAC, we selected a degradation machinery complex conformation from clusters A and B with a top docking score CRBN-dBETx-BRD4^BD1 ternary conformation for MD simulations (Table 1). To further evaluate the docking scores, we also performed protein–protein interaction energy calculations using MD runs initiated by these ternary conformations with top docking scores. The energy calculations confirmed the thermodynamic stability of top-ranked ternary complexes obtained through protein–protein docking, highlighting the robust binding of their associated dBETs (Figure 3—figure supplement 2). Moreover, they remained structurally stable throughout the MD runs, as evidenced by RMSD plots (see RMSD plots in Figure 2—figure supplement 6).

After confirming that each CRBN-dBETx-BRD4^BD1 used for constructing the degradation complexes represented popular ternary conformations, we conducted classical MD simulations for these degradation complexes to observe structural dynamics and local interactions that lead to successful ubiquitination. Notably, to ensure that the compatibility of each CRBN-dBETx-BRD4^BD1 with both clusters A and B of CRL4A E3 ligase scaffolds, we also selected a few ternary complexes from our MD runs (Table 1). As depicted in Figure 1C, a successful ubiquitination reaction requires protein conformational arrangements that bring a surface Lys of BRD4^BD1 close to Gly of Ub, where at least one Asp of E2 positioned in proximity to the same Gly to facilitate charge transfer (Appendix 1—table 3). Therefore, we employed two criteria: (1) ensuring that the distance between Lys (N atom) and Gly (C atom) was less than 10 Å; and (2) confirming that distance between Asp (O atom) of E2 and the same Gly (O atom) was less than 6 Å to determine the likelihood of ubiquitination.

Although none of the initial modelled machinery complexes exhibited a Lys residue of BRD4^BD1 being ready for ubiquitination, our MD sampling facilitated the movement of different Lys residues of BRD4^BD1 from distance greater than 16 Å of the catalytic pocket on E2 to establish close contacts with Gly of Ub for each dBET. Specifically, we observed significant large-scale motions within degradation complex B1_dBET1__#35 (cluster B, docked conformation #35), which recruited three Lys residues K72, K76 and K111 at distances greater than 16 Å, positioning them in the catalytic pocket on E2. Concurrently, a conserved charged residue Asp of E2 approached the C-terminus G75 of Ub, potentially facilitating isopeptide bond formation (Figure 4A and B). Similarly, degradation complex B1_dBET23_{_#14} (cluster B, docked conformation #14) and B1_dBET70_{_#91} (cluster B, docked conformation #91) also brought K99 of BRD4^BD1 into proximity of the catalytic cavity of E2, where one Asp of E2 approached the the C-terminus G75 of Ub as well (Figure 4C, D, G and H).

Figure 4 with 3 supplements see all

Download asset Open asset

In contrast, degradation complex B1_dBET57_{_#9_MD1} (cluster B, conformation from MD run 1 initiated with docked conformation #9) positioned K102 of BRD4^BD1 in close proximity to the catalytic pocket on E2 (Figures 4E and 3F). Notably, BRD4^BD1 of dBET1, dBET23 and dBET70 exhibited a closer proximity to Ub (~6 Å) as compared to dBET57 (~10 Å; Figure 4A, C, E and G). Despite degradation complex B1_dBET1__#35 positioned three Lys residues of BRD4^BD1 to form stable interactions with the Gly of Ub, the surrounding Asp of E2 was unstable which hindered a successful ubiquitination process (Figure 4A). Further analysis revealed that dBET23 (~85%) and dBET70 (~78%) exhibited a higher probability of fulfilling both criteria for potential chemical reactions, whereas dBET1 (~40%) and dBET57 (~22%) had a lower likelihood (Figure 4—figure supplement 2D). This analysis revealed how dBET23 and dBET70 achieved high degradation efficiency, while dBET1 and dBET57 induced protein motions for bond formation, without achieving all required arrangements simultaneously, resulting in lower degradation efficiency. Our findings are consistent with experimental measurements indicating that all dBETs could degrade BRD4^BD1 but with varying degradation potency. By considering the Lys–Gly and Asp-Gly distances, our simulations underscored that dBET23 and dBET70 (with DC_50/5h values of ~5 nM and ~50 nM, respectively) are more efficient degraders compared to dBET1 and dBET57 (with DC_50/5h values of ~500 nM).

To gain further insights into the overall motion of the degradation complexes, we employed principal component analysis (PCA) to extract essential motions within the complex; Notably, the first two PC modes, PC1 and PC2, accounted over 60% of the overall motions in most MD runs (Appendix 1—table 4). While the specific nature of these motions varied depending on the system, the dBETs induced an inter-protein orientation between CRBN and BRD4^BD1, with two hinge regions: loops in Rbx1 and DDB1 proteins contributing prominently to the essential motions of the complex (Figure 5—figure supplement 1). These two hinge regions exhibited twisted and opposite direction motions, facilitating the rearrangement and bringing BRD4^BD1 closer to the catalytic cavity of E2, which is a critical step for successful ubiquitination. Indeed, the essential motions revealed by the first two PC modes clearly revealed the plasticity between the protein–protein interfaces in the degradation machinery. Notably, the most significant motion in PC1 shifted E2/Ub/Rbx1/NEDD8 closer to BRD4^BD1 (Figure 5—figure supplement 2). Furthermore, the pairwise force matrix between individual residues within the degradation complex offered insights into the non-covalent interactions governing the dynamic behavior of the complex. This interaction network, starting from the dBETs’ linker and extending to the hinge loop of DDB1 and Rbx1 (Figure 5—figure supplement 3), indicated a clear correlation between linker rotation and the movement of the entire degradation complex. Our study highlights the importance of structural dynamics of the degradation complex during the ubiquitination processes, which is influenced by a dBET ligand.

To quantify the correlation between linker rotation and the movement of the degradation complex, we selected four atoms shown in Figure 5A to construct a pseudo dihedral angle and used histograms to depict the population of these pseudo angles. Deviations between the histograms from different time periods indicated the motion of the degradation complex during the MD simulations. Across the first and last 100 ns of an MD run, the degradation complexes large motions with the presence of each dBET (Figure 5B). Given that the only distinction among the four dBETs was the linker, we observed that the complex motions correlated to the C-C-N3-C dihedral angles in the linkers of dBET1, dBET23, dBET70, and dBET57 (Figure 6A). Notably, all four dBETs exhibited substantial shifts of BRD4^BD1 of approximately 10 and 15 degrees, which underscores their flexibility in promoting extensive motion of the BRD4^BD1 protein for Lys–Gly interaction. The various motions from different degradation complexes also resulted in different sets of Lys residues potentially forming contacts with E2 and Ub. For instance, the degradation complexes dBET23_{_#14} and dBET70_{_#91} brought K99 of BRD4^BD1 closer to G75 of Ub (Figure 6B), while dBET57_{_#9_MD1} placed K102 in proximity to G75 of Ub. Despite allowing for different Lys–Gly interactions, dBET57 exhibited weaker interactions. The pairwise force analysis revealed a strong attraction between K99 and G75 of dBET23 and dBET70, and a weaker attraction between K102 and G75 of dBET57_{_#9_MD1} (Figure 5—figure supplement 4).

Figure 5 with 5 supplements see all

Download asset Open asset

Figure 6

Download asset Open asset

To further examine the correlation between PROTAC rotation and the Lys-Gly interaction, we performed a time-dependent correlation analysis. This analysis showed that PROTAC rotation translates motion over time, leading to the Lys-Gly interaction, with a correlation peak around 60–85 ns, marking the time of the interaction (Figure 7 and Figure 7—figure supplement 1). In addition, the pseudo dihedral angles also showed a high correlation (0.85 in the case of dBET1) with Lys-Gly distance. This indicated that degradation complex undergoes structural rearrangement and drives the Lys-Gly interaction.

Figure 7 with 1 supplement see all

Download asset Open asset

Given the presence of numerous local energy minima conformations in the degradation complex, MD simulations could not sample all conformations comprehensively. Furthermore, as evidenced by the protein–protein docking results (Figure 1—figure supplement 1C), the short linker of dBET57 yielded a less flexible ternary complex, suggesting challenges in sampling various conformations using MD runs. Our dihedral entropies analysis showed that dBET57 has ~0.3 kcal/mol lower entropies than the other dBETs with three different linkers, suggesting that dBET57 is less flexible than other PROTACs (Figure 8).

Figure 8

Download asset Open asset

To address this, we selected another distinctly different ternary complex of dBET57 (Figure 5—figure supplement 5B) to construct another degradation complex, B1_dBET57_{#9_MD2}, for MD simulation. Unlike B1_dBET57_{#9_MD1,} B1_dBET57_{_#9_MD2} exhibited significantly more rigidity, with only minimal rotation (<5 degrees) in the pseudo dihedral angle. This rigidity limits the structural dynamics of the whole complex to bring BRD4^BD1 closer to Ub for Lys–Gly interactions (Figure 5—figure supplement 5). This observation underscores the crucial role of the linker in guiding the motion of the degradation complex, thereby influencing its capacity for ubiquitination. In summary, the linker induces conformations of ternary complexes but also guides the motion of the degradation complex, which allows an effective Lys–Gly interaction for ubiquitination. Our studies highlight the critical role of linker flexibility in facilitating protein structural dynamics necessary for ubiquitination.

As bifunctional small-molecule ligands, PROTACs play a pivotal role in recruiting a neo-substrate of E3 ligase, namely the target protein POI, to form a ternary E3-PROTAC-POI complex by stabilizing the protein–protein interactions. By forming an E3-PROTAC-POI complex, E3 ligase brings the POI close to an E2 enzyme for protein ubiquitination, with both E2 and E3 proteins belonging to a larger Ub ligase machinery complex. While forming a stable E3-PROTAC-POI complex represents a critical step in ubiquitination, it alone is insufficient for subsequent ubiquitination and degradation by proteasomes. Several studies suggest that positioning the POI’s Lys residues close to the E2 for ubiquitination is crucial for successful POI degradation (Smith et al., 2019; Dixon et al., 2022; Petzold et al., 2016). However, due to the lack of experimental determined structures for POI-bound degradation complexes, directly observing the proximity of Ub to POI’s Lys residues remains challenging. In addition, proteins exhibit dynamic behaviors, and the fluctuation in the ternary E3-PROTAC-POI complex and the degradation machine can either promote or impede catalytic activity. To investigate the degradability of PROTACs, we employed a strategy combining protein docking, MD simulations, and structural and energy analysis tools. Our focus was on a series of dBET degraders, differing in their linkers, which all induced stable CRBN-dBETx-BRD4^BD1 complexes different degradation efficiencies.

One straightforward approach to promote ubiquitination is to obtain a stable E3-PROTAC-POI ternary complex to allow pre-organized orientation placing a Lys residue(s) of the POI close to the E2/Ub catalytic site in the degradation machine. However, experimental determined structures for degradation complexes are scarce. Protein–protein docking can reveal low energy ternary complexes and popular conformational ensembles, providing insights into potential orientations for ubiquitination. Although protein–protein docking is usually quick (i.e. <50 hr to sample 5000 protein–protein complexes using one CPU in computation), the docked E3-PROTAC-POI conformations may not be the functionally active form to induce successful ubiquitination. Nevertheless, results from protein–protein docking are highly informative and suggest potential flexibility of a ternary complex, as shown in Figure 1—figure supplement 1. However, MD simulations or other conformational search methods can further generate more stable ternary conformations, and the sampling from protein–protein docking may not be thorough. As illustrated by dBET57, protein–protein docking generated only similar ternary complex conformations (Figure 1—figure supplement 1C). In contrast, classical MD simulations using all atoms and explicit water molecules sampled additional low energy conformations (Figure 3C).

Using the protein–protein docking results, multiple thermodynamically stable ternary E3-PROTAC-POI conformations were placed in the degradation machine to access whether surface Lys residues of POI were appropriately orientated for ubiquitination. However, relying solely on a handful of static conformations for predicting POI degradability is not insufficient. For example, our docking results showed that none of the Lys residues of BRD4^BD1 was ready for ubiquitination. Notably, our modelled CRL4A E3 ligase scaffolds in Figure 2—figure supplement 2 show two distinct conformations: a ring-forming conformation (cluster A) with a gap distance approximately 1.0 Å to 10.0 Å, and a ring-open conformation (cluster B) with a much larger gap distance, approximately 14 Å to 35 Å. When docking the ternary complex into the ring-forming scaffold (Figure 2—figure supplement 2A), the orientations of BRD4^BD1 tended to be sterically confined, limiting their range of motion to bring one Lys residue of BRD4^BD1 closer to the Gly of Ub. In contrast, using MD to simulate protein dynamics, CRL4A E3 ligase scaffolds with ring-open conformations provided a larger space that may allow an assembled BRD4^BD1 to properly rearrange its conformations for successful ubiquitination, even though the initial distance between BRD4^BD1 and Ub is far. Our studies demonstrated that using CRBN-dBETs-BRD4^BD1 conformations sampled by protein–protein docking and/or MD simulations provided reasonably good initial structures to model the structural dynamics of the degradation machinery. The essential motions were captured by PCA, which illustrated that all dBETs studied can produce motions to bring BRD4^BD1 closer to Ub, thus supporting their degradability. Atomistic simulations revealed that specific Lys residue(s) of BRD4^BD1 (i.e. K99, K102 and K72/76/111 of dBET23 and dBET70, dBET57 and dBET1, respectively) were attracted by polar residues around Ub and E2.

The analysis of how a PROTAC, particularly the linker region, affects the overall dynamics and specific arrangements to bring Lys close to the catalytic site offers further information on linker design. The slight variations in linker length and composition that may significantly alter the degradability of a PROTAC was puzzling. Even though PROTACs can strengthen interactions between E3 and POI, protein degradation is not always associated with forming stable ternary complexes. One hypothesis is that some ternary complexes have a preferred E3/POI orientation in the degradation machinery complex to result in efficient ubiquitination. Here, we demonstrated that by using accurately modeled initial conformations, performing a few hundred nanosecond classical MD simulations could illustrate how protein structural dynamics orient surface Lys residues of BRD4^BD1 and Asp of E2 close to the Gly of Ub, thus offering information for predicting protein degradation. We examined the hinges and used pseudo-dihedral angles of the degradation complexes to describe the large-scale swinging motions near the end of the MD runs (300–400 ns). To understand the role of the linker of different dBETs, we compared conformation ensembles from the first and last 100 ns of MD simulations. The movement of the degradation machinery correlated with rotations of specific dihedrals of the linker region in dBETs (Figure 6). In addition, the, dihedral rotation of PROTACs shows correlation with the Lys-Gly interaction (Figure 7 and Figure 7—figure supplement 1). This phenomenon shows that dBETs contribute to the structural dynamics of degradation complexes and suggests that linkers with different properties may lead to various large-scale motions.

The dynamics of the degradation machinery, induced by PROTACs, has implications in linker design and degradation prediction. Because the ‘warhead’ (thieno-triazolo-1,4-diazepine [JQ1]) used to bind to an E3 ligase and the ‘warhead’ (pomalidomide) used to bind to a POI are tightly bound ligands to each protein, the two warheads typically have limited motions when they are in the binding site of E3 or POI. Clearly, the length and composition of the linker are critical in degradation efficiency, and sites of conjugation to each warhead also affect the stability of the ternary complexes. Despite no general rules that apply to every system for PROTAC linker design, guidelines used in designing small-molecule drugs binding to their target proteins still hold. First, the linker should have sufficient length to form a ternary complex. Although longer and more flexible linkers typically provide enhanced opportunities for E3-PROTAC-POI complexes to orient suitable conformations for ubiquitination, shorter and/or rigid linkers may provide a preorganized E3-PROTAC-POI for efficient catalysis. However, knowing in advance the best related orientations is challenging, but the shorter linker in our systems, dBET57, had disadvantages in accommodating a Lys residue in the E2/Ub catalytic region for forming stable interactions for catalysis. In contrast, dBET23 and dBET70, with greater flexibility, allowed BRD4^BD1 to adopt diverse conformations for catalysis, ultimately leading to more efficient degradation. Notably, dBET1 recruited multiple Lys residues in the E2/Ub catalytic region and K111 was highly stable. However, the unstable Asp from the surrounding prevented charge transfer, and thus hindered the isopeptide bond formation for degradation. Notably, dBET1 and dBET57 still induce stable CRBN-dBET57-BRD4^BD1 and CRBN-dBET1-BRD4^BD1 complexes, which effectively assemble the neo-substrate BRD4^BD1 with the degradation machine. However, the either Lys–Gly or Asp-Gly attractive force was unstable, which reduced the chances for successful ubiquitination.

Revealing structural dynamics of the entire degradation complex that bring surface Lys residue(s) of BRD4^BD1 for target ubiquitination explains the degradability of these dBETs. Notably, the classical MD trajectories with atomistic details offer information about overall protein motions and specific Lys residues critical for the degradability of BRD4^BD1 for further investigation. Theoretically speaking, excessively long MD simulations should be able to sample a wide range of protein conformations. However, long MD simulations can still stick in local minima and are computationally expensive. We demonstrated the use of the most stable ternary CRBN-dBETx-BRD4^BD1 conformations from protein–protein docking to facilitate sampling the correct orientation of the surface Lys residue(s) of BRD4^BD1 for ubiquitination. Of note, CRBN-dBET57-BRD4^BD1 formed by the more rigid linker of dBET57 could be more easily trapped in a local energy minimum, and we need to select conformations from both protein-docking (dBET57_#9) and MD-sampling results (CRBN-dBET57_#9-BRD4^BD1) to construct the degradation machinery complex for structural dynamics studies. Different E3 ligases may also result in significantly different related orientation when bound to the same POI. Although static conformations from experimentally determined structures may suggest whether a dBET orients CRBN-dBETx-BRD4^BD1 conformations suitable for degradation, real motions cannot be predicted directly from static conformations. Examining the dynamic nature of the degradation complex provides a more complete picture of the conformations of both ternary CRBN-dBETx-BRD4^BD1 and machinery complexes that determine Lys accessibility for POI ubiquitination, an important factor for consideration when optimizing a PROTAC to improve its potency.

The PDB format of the four modeled degradation complex structures are provided in supplementary files. Other input/output and trajectory files and in-house codes for data analysis are available on Zenodo.

1. 1. Ta H

   (2025) Zenodo

   PROTAC-induced Protein Structural Dynamics in Targeted Protein Degradation.

   https://doi.org/10.5281/zenodo.14791107

1. 1. Ai R
   2. Qaiser Fatmi M
   3. Chang C-EA

   (2010) T-Analyst: a program for efficient analysis of protein conformational changes by torsion angles

   Journal of Computer-Aided Molecular Design 24:819–827.

   https://doi.org/10.1007/s10822-010-9376-y

   • PubMed
   • Google Scholar
2. 1. Bai N
   2. Miller SA
   3. Andrianov GV
   4. Yates M
   5. Kirubakaran P
   6. Karanicolas J

   (2021) Rationalizing PROTAC-mediated ternary complex formation using rosetta

   Journal of Chemical Information and Modeling 61:1368–1382.

   https://doi.org/10.1021/acs.jcim.0c01451

   • PubMed
   • Google Scholar
3. 1. Bai N
   2. Riching KM
   3. Makaju A
   4. Wu H
   5. Acker TM
   6. Ou S-C
   7. Zhang Y
   8. Shen X
   9. Bulloch DN
   10. Rui H
   11. Gibson BW
   12. Daniels DL
   13. Urh M
   14. Rock BM
   15. Humphreys SC

   (2022) Modeling the CRL4A ligase complex to predict target protein ubiquitination induced by cereblon-recruiting PROTACs

   The Journal of Biological Chemistry 298:101653.

   https://doi.org/10.1016/j.jbc.2022.101653

   • PubMed
   • Google Scholar
4. 1. Békés M
   2. Langley DR
   3. Crews CM

   (2022) PROTAC targeted protein degraders: the past is prologue

   Nature Reviews. Drug Discovery 21:181–200.

   https://doi.org/10.1038/s41573-021-00371-6

   • PubMed
   • Google Scholar
5. 1. Bondeson DP
   2. Smith BE
   3. Burslem GM
   4. Buhimschi AD
   5. Hines J
   6. Jaime-Figueroa S
   7. Wang J
   8. Hamman BD
   9. Ishchenko A
   10. Crews CM

   (2018) Lessons in PROTAC design from selective degradation with a promiscuous warhead

   Cell Chemical Biology 25:78–87.

   https://doi.org/10.1016/j.chembiol.2017.09.010

   • PubMed
   • Google Scholar
6. 1. Bricelj A
   2. Steinebach C
   3. Kuchta R
   4. Gütschow M
   5. Sosič I

   (2021) E3 ligase ligands in successful PROTACs: an overview of syntheses and linker attachment points

   Frontiers in Chemistry 9:707317.

   https://doi.org/10.3389/fchem.2021.707317

   • PubMed
   • Google Scholar
7. 1. Dale B
   2. Cheng M
   3. Park KS
   4. Kaniskan HÜ
   5. Xiong Y
   6. Jin J

   (2021) Advancing targeted protein degradation for cancer therapy

   Nature Reviews. Cancer 21:638–654.

   https://doi.org/10.1038/s41568-021-00365-x

   • PubMed
   • Google Scholar
8. 1. Dixon T
   2. MacPherson D
   3. Mostofian B
   4. Dauzhenka T
   5. Lotz S
   6. McGee D
   7. Shechter S
   8. Shrestha UR
   9. Wiewiora R
   10. McDargh ZA
   11. Pei F
   12. Pal R
   13. Ribeiro JV
   14. Wilkerson T
   15. Sachdeva V
   16. Gao N
   17. Jain S
   18. Sparks S
   19. Li Y
   20. Vinitsky A
   21. Zhang X
   22. Razavi AM
   23. Kolossváry I
   24. Imbriglio J
   25. Evdokimov A
   26. Bergeron L
   27. Zhou W
   28. Adhikari J
   29. Ruprecht B
   30. Dickson A
   31. Xu H
   32. Sherman W
   33. Izaguirre JA

   (2022) Predicting the structural basis of targeted protein degradation by integrating molecular dynamics simulations with structural mass spectrometry

   Nature Communications 13:5884.

   https://doi.org/10.1038/s41467-022-33575-4

   • PubMed
   • Google Scholar
9. 1. Drummond ML
   2. Williams CI

   (2019) In silico modeling of PROTAC-mediated ternary complexes: validation and application

   Journal of Chemical Information and Modeling 59:1634–1644.

   https://doi.org/10.1021/acs.jcim.8b00872

   • PubMed
   • Google Scholar
10. 1. Drummond ML
    2. Henry A
    3. Li H
    4. Williams CI

    (2020) Improved accuracy for modeling protac-mediated ternary complex formation and targeted protein degradation via new In Silico methodologies

    Journal of Chemical Information and Modeling 60:5234–5254.

    https://doi.org/10.1021/acs.jcim.0c00897

    • PubMed
    • Google Scholar
11. 1. Essmann U
    2. Perera L
    3. Berkowitz ML
    4. Darden T
    5. Lee H
    6. Pedersen LG

    (1995) A smooth particle mesh Ewald method

    The Journal of Chemical Physics 103:8577–8593.

    https://doi.org/10.1063/1.470117

    • Google Scholar
12. 1. Filippakopoulos P
    2. Qi J
    3. Picaud S
    4. Shen Y
    5. Smith WB
    6. Fedorov O
    7. Morse EM
    8. Keates T
    9. Hickman TT
    10. Felletar I
    11. Philpott M
    12. Munro S
    13. McKeown MR
    14. Wang Y
    15. Christie AL
    16. West N
    17. Cameron MJ
    18. Schwartz B
    19. Heightman TD
    20. La Thangue N
    21. French CA
    22. Wiest O
    23. Kung AL
    24. Knapp S
    25. Bradner JE

    (2010) Selective inhibition of BET bromodomains

    Nature 468:1067–1073.

    https://doi.org/10.1038/nature09504

    • PubMed
    • Google Scholar
13. 1. Fischer ES
    2. Böhm K
    3. Lydeard JR
    4. Yang H
    5. Stadler MB
    6. Cavadini S
    7. Nagel J
    8. Serluca F
    9. Acker V
    10. Lingaraju GM
    11. Tichkule RB
    12. Schebesta M
    13. Forrester WC
    14. Schirle M
    15. Hassiepen U
    16. Ottl J
    17. Hild M
    18. Beckwith REJ
    19. Harper JW
    20. Jenkins JL
    21. Thomä NH

    (2014) Structure of the DDB1-CRBN E3 ubiquitin ligase in complex with thalidomide

    Nature 512:49–53.

    https://doi.org/10.1038/nature13527

    • PubMed
    • Google Scholar
14. 1. Han B

    (2020) A suite of mathematical solutions to describe ternary complex formation and their application to targeted protein degradation by heterobifunctional ligands

    The Journal of Biological Chemistry 295:15280–15291.

    https://doi.org/10.1074/jbc.RA120.014715

    • PubMed
    • Google Scholar
15. 1. He X
    2. Man VH
    3. Yang W
    4. Lee TS
    5. Wang J

    (2020) A fast and high-quality charge model for the next generation general AMBER force field

    The Journal of Chemical Physics 153:114502.

    https://doi.org/10.1063/5.0019056

    • PubMed
    • Google Scholar
16. 1. Hotelling H

    (1933) Analysis of a complex of statistical variables into principal components

    Journal of Educational Psychology 24:417–441.

    https://doi.org/10.1037/h0071325

    • Google Scholar
17. 1. Huang H-T
    2. Dobrovolsky D
    3. Paulk J
    4. Yang G
    5. Weisberg EL
    6. Doctor ZM
    7. Buckley DL
    8. Cho J-H
    9. Ko E
    10. Jang J
    11. Shi K
    12. Choi HG
    13. Griffin JD
    14. Li Y
    15. Treon SP
    16. Fischer ES
    17. Bradner JE
    18. Tan L
    19. Gray NS

    (2018) A chemoproteomic approach to query the degradable kinome using a multi-kinase degrader

    Cell Chemical Biology 25:88–99.

    https://doi.org/10.1016/j.chembiol.2017.10.005

    • PubMed
    • Google Scholar
18. Preprint

    1. Hung TI
    2. Hsieh YJ
    3. Lu WL
    4. Wu KP
    5. Chang CA

    (2023) What Strengthens Protein-Protein Interactions: Analysis and Applications of Residue Correlation Networks

    bioRxiv.

    https://doi.org/10.1101/2023.03.15.532709

    • Google Scholar
19. 1. Ignatov M
    2. Jindal A
    3. Kotelnikov S
    4. Beglov D
    5. Posternak G
    6. Tang X
    7. Maisonneuve P
    8. Poda G
    9. Batey RA
    10. Sicheri F
    11. Whitty A
    12. Tonge PJ
    13. Vajda S
    14. Kozakov D

    (2023) High accuracy prediction of PROTAC complex structures

    Journal of the American Chemical Society 145:7123–7135.

    https://doi.org/10.1021/jacs.2c09387

    • PubMed
    • Google Scholar
20. 1. Jakalian A
    2. Jack DB
    3. Bayly CIF

    (2002) Fast, efficient generation of high-quality atomic charges: AM1-BCC model: II. Parameterization and validation

    J Comput Chem 23:1623–1641.

    https://doi.org/10.1002/jcc.10128

    • PubMed
    • Google Scholar
21. 1. Jorgensen WL
    2. Chandrasekhar J
    3. Madura JD
    4. Impey RW
    5. Klein ML

    (1983) Comparison of simple potential functions for simulating liquid water

    The Journal of Chemical Physics 79:926–935.

    https://doi.org/10.1063/1.445869

    • Google Scholar
22. 1. Lai AC
    2. Toure M
    3. Hellerschmied D
    4. Salami J
    5. Jaime-Figueroa S
    6. Ko E
    7. Hines J
    8. Crews CM

    (2016) Modular PROTAC design for the degradation of oncogenic BCR-ABL

    Angewandte Chemie 55:807–810.

    https://doi.org/10.1002/anie.201507634

    • PubMed
    • Google Scholar
23. 1. Lee J
    2. Lee Y
    3. Jung YM
    4. Park JH
    5. Yoo HS
    6. Park J

    (2022) Discovery of E3 ligase ligands for target protein degradation

    Molecules 27:6515.

    https://doi.org/10.3390/molecules27196515

    • PubMed
    • Google Scholar
24. 1. Liu Z
    2. Hu M
    3. Yang Y
    4. Du C
    5. Zhou H
    6. Liu C
    7. Chen Y
    8. Fan L
    9. Ma H
    10. Gong Y
    11. Xie Y

    (2022) An overview of PROTACs: a promising drug discovery paradigm

    Molecular Biomedicine 3:46.

    https://doi.org/10.1186/s43556-022-00112-0

    • PubMed
    • Google Scholar
25. 1. Lu J
    2. Qian Y
    3. Altieri M
    4. Dong H
    5. Wang J
    6. Raina K
    7. Hines J
    8. Winkler JD
    9. Crew AP
    10. Coleman K
    11. Crews CM

    (2015) Hijacking the E3 ubiquitin ligase cereblon to efficiently target BRD4

    Chemistry & Biology 22:755–763.

    https://doi.org/10.1016/j.chembiol.2015.05.009

    • PubMed
    • Google Scholar
26. 1. Lv D
    2. Pal P
    3. Liu X
    4. Jia Y
    5. Thummuri D
    6. Zhang P
    7. Hu W
    8. Pei J
    9. Zhang Q
    10. Zhou S
    11. Khan S
    12. Zhang X
    13. Hua N
    14. Yang Q
    15. Arango S
    16. Zhang W
    17. Nayak D
    18. Olsen SK
    19. Weintraub ST
    20. Hromas R
    21. Konopleva M
    22. Yuan Y
    23. Zheng G
    24. Zhou D

    (2021) Development of a BCL-XL and BCL-2 dual degrader with improved anti-leukemic activity

    Nature Communications 12:6896.

    https://doi.org/10.1038/s41467-021-27210-x

    • PubMed
    • Google Scholar
27. 1. Maier JA
    2. Martinez C
    3. Kasavajhala K
    4. Wickstrom L
    5. Hauser KE
    6. Simmerling C

    (2015) ff14SB: improving the accuracy of protein side chain and backbone parameters from ff99SB

    Journal of Chemical Theory and Computation 11:3696–3713.

    https://doi.org/10.1021/acs.jctc.5b00255

    • PubMed
    • Google Scholar
28. 1. Miller BR
    2. McGee TD
    3. Swails JM
    4. Homeyer N
    5. Gohlke H
    6. Roitberg AE

    (2012) MMPBSA.py: an efficient program for end-state free energy calculations

    Journal of Chemical Theory and Computation 8:3314–3321.

    https://doi.org/10.1021/ct300418h

    • PubMed
    • Google Scholar
29. 1. Nowak RP
    2. DeAngelo SL
    3. Buckley D
    4. He Z
    5. Donovan KA
    6. An J
    7. Safaee N
    8. Jedrychowski MP
    9. Ponthier CM
    10. Ishoey M
    11. Zhang T
    12. Mancias JD
    13. Gray NS
    14. Bradner JE
    15. Fischer ES

    (2018) Plasticity in binding confers selectivity in ligand-induced protein degradation

    Nature Chemical Biology 14:706–714.

    https://doi.org/10.1038/s41589-018-0055-y

    • PubMed
    • Google Scholar
30. 1. Ottis P
    2. Toure M
    3. Cromm PM
    4. Ko E
    5. Gustafson JL
    6. Crews CM

    (2017) Assessing different e3 ligases for small molecule induced protein ubiquitination and degradation

    ACS Chemical Biology 12:2570–2578.

    https://doi.org/10.1021/acschembio.7b00485

    • PubMed
    • Google Scholar
31. 1. Pereira GP
    2. Jiménez-García B
    3. Pellarin R
    4. Launay G
    5. Wu S
    6. Martin J
    7. Souza PCT

    (2023) Rational prediction of PROTAC-compatible protein-protein interfaces by molecular docking

    Journal of Chemical Information and Modeling 63:6823–6833.

    https://doi.org/10.1021/acs.jcim.3c01154

    • PubMed
    • Google Scholar
32. 1. Petzold G
    2. Fischer ES
    3. Thomä NH

    (2016) Structural basis of lenalidomide-induced CK1α degradation by the CRL4(CRBN) ubiquitin ligase

    Nature 532:127–130.

    https://doi.org/10.1038/nature16979

    • PubMed
    • Google Scholar
33. 1. Qin C
    2. Hu Y
    3. Zhou B
    4. Fernandez-Salas E
    5. Yang C-Y
    6. Liu L
    7. McEachern D
    8. Przybranowski S
    9. Wang M
    10. Stuckey J
    11. Meagher J
    12. Bai L
    13. Chen Z
    14. Lin M
    15. Yang J
    16. Ziazadeh DN
    17. Xu F
    18. Hu J
    19. Xiang W
    20. Huang L
    21. Li S
    22. Wen B
    23. Sun D
    24. Wang S

    (2018) Discovery of QCA570 as an exceptionally potent and efficacious proteolysis targeting chimera (protac) degrader of the bromodomain and extra-terminal (bet) proteins capable of inducing complete and durable tumor regression

    Journal of Medicinal Chemistry 61:6685–6704.

    https://doi.org/10.1021/acs.jmedchem.8b00506

    • PubMed
    • Google Scholar
34. 1. Ramachandran S
    2. Ciulli A

    (2021) Building ubiquitination machineries: E3 ligase multi-subunit assembly and substrate targeting by PROTACs and molecular glues

    Current Opinion in Structural Biology 67:110–119.

    https://doi.org/10.1016/j.sbi.2020.10.009

    • PubMed
    • Google Scholar
35. 1. Rao A
    2. Tunjic TM
    3. Brunsteiner M
    4. Müller M
    5. Fooladi H
    6. Gasbarri C
    7. Weber N

    (2023) Bayesian optimization for ternary complex prediction (BOTCP)

    Artificial Intelligence in the Life Sciences 3:100072.

    https://doi.org/10.1016/j.ailsci.2023.100072

    • Google Scholar
36. 1. Roe DR
    2. Cheatham TE

    (2013) PTRAJ and CPPTRAJ: software for processing and analysis of molecular dynamics trajectory data

    Journal of Chemical Theory and Computation 9:3084–3095.

    https://doi.org/10.1021/ct400341p

    • PubMed
    • Google Scholar
37. 1. Ryckaert JP
    2. Ciccotti G
    3. Berendsen HJC

    (1977) Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes

    Journal of Computational Physics 23:327–341.

    https://doi.org/10.1016/0021-9991(77)90098-5

    • Google Scholar
38. 1. Smith BE
    2. Wang SL
    3. Jaime-Figueroa S
    4. Harbin A
    5. Wang J
    6. Hamman BD
    7. Crews CM

    (2019) Differential PROTAC substrate specificity dictated by orientation of recruited E3 ligase

    Nature Communications 10:131.

    https://doi.org/10.1038/s41467-018-08027-7

    • PubMed
    • Google Scholar
39. 1. Stacklies W
    2. Seifert C
    3. Graeter F

    (2011) Implementation of force distribution analysis for molecular dynamics simulations

    BMC Bioinformatics 12:101.

    https://doi.org/10.1186/1471-2105-12-101

    • PubMed
    • Google Scholar
40. Website

    1. Swart P
    2. Hagberg A
    3. Schult DA

    (2008) Exploring Network Structure, Dynamics, and Function Using NetworkX

    Accessed June 28, 2008.

    http://conference.scipy.org/proceedings/SciPy2008/paper_2
41. 1. Tang Z
    2. Chang C-EA

    (2017) Systematic dissociation pathway searches guided by principal component modes

    Journal of Chemical Theory and Computation 13:2230–2244.

    https://doi.org/10.1021/acs.jctc.6b01204

    • PubMed
    • Google Scholar
42. 1. Tinworth CP
    2. Lithgow H
    3. Dittus L
    4. Bassi ZI
    5. Hughes SE
    6. Muelbaier M
    7. Dai H
    8. Smith IED
    9. Kerr WJ
    10. Burley GA
    11. Bantscheff M
    12. Harling JD

    (2019) PROTAC-mediated degradation of bruton’s tyrosine kinase is inhibited by covalent binding

    ACS Chemical Biology 14:342–347.

    https://doi.org/10.1021/acschembio.8b01094

    • PubMed
    • Google Scholar
43. 1. Valimberti I
    2. Tiberti M
    3. Lambrughi M
    4. Sarcevic B
    5. Papaleo E

    (2015) E2 superfamily of ubiquitin-conjugating enzymes: constitutively active or activated through phosphorylation in the catalytic cleft

    Scientific Reports 5:14849.

    https://doi.org/10.1038/srep14849

    • PubMed
    • Google Scholar
44. 1. Ward CC
    2. Kleinman JI
    3. Brittain SM
    4. Lee PS
    5. Chung CYS
    6. Kim K
    7. Petri Y
    8. Thomas JR
    9. Tallarico JA
    10. McKenna JM
    11. Schirle M
    12. Nomura DK

    (2019) Covalent ligand screening uncovers a rnf4 e3 ligase recruiter for targeted protein degradation applications

    ACS Chemical Biology 14:2430–2440.

    https://doi.org/10.1021/acschembio.8b01083

    • PubMed
    • Google Scholar
45. 1. Waterhouse A
    2. Bertoni M
    3. Bienert S
    4. Studer G
    5. Tauriello G
    6. Gumienny R
    7. Heer FT
    8. de Beer TAP
    9. Rempfer C
    10. Bordoli L
    11. Lepore R
    12. Schwede T

    (2018) SWISS-MODEL: homology modelling of protein structures and complexes

    Nucleic Acids Research 46:W296–W303.

    https://doi.org/10.1093/nar/gky427

    • PubMed
    • Google Scholar
46. 1. Winter GE
    2. Buckley DL
    3. Paulk J
    4. Roberts JM
    5. Souza A
    6. Dhe-Paganon S
    7. Bradner JE

    (2015) Phthalimide conjugation as a strategy for in vivo target protein degradation

    Science 348:1376–1381.

    https://doi.org/10.1126/science.aab1433

    • Google Scholar
47. 1. Zaidman D
    2. Prilusky J
    3. London N

    (2020) PRosettaC: rosetta based modeling of protac mediated ternary complexes

    Journal of Chemical Information and Modeling 60:4894–4903.

    https://doi.org/10.1021/acs.jcim.0c00589

    • PubMed
    • Google Scholar
48. Book

    1. Zhang W
    2. Burman SSR
    3. Chen J
    4. Donovan KA
    5. Cao Y
    6. Shu C
    7. Zhang B
    8. Zeng Z
    9. Gu S
    10. Zhang Y

    (2022)

    Machine Learning Modeling of Protein-Intrinsic Features Predicts Tractability of Targeted Protein Degradation

    Genomics Proteomics Bioinformatics.

    • Google Scholar
49. 1. Zhao Q
    2. Lan T
    3. Su S
    4. Rao Y

    (2019) Induction of apoptosis in MDA-MB-231 breast cancer cells by a PARP1-targeting PROTAC small molecule

    Chemical Communications 55:369–372.

    https://doi.org/10.1039/c8cc07813k

    • PubMed
    • Google Scholar
50. 1. Zheng S
    2. Tan Y
    3. Wang Z
    4. Li C
    5. Zhang Z
    6. Sang X
    7. Chen H
    8. Yang Y

    (2022) Accelerated rational PROTAC design via deep learning and molecular simulations

    Nature Machine Intelligence 4:739–748.

    https://doi.org/10.1038/s42256-022-00527-y

    • PubMed
    • Google Scholar
51. 1. Zhou B
    2. Hu J
    3. Xu F
    4. Chen Z
    5. Bai L
    6. Fernandez-Salas E
    7. Lin M
    8. Liu L
    9. Yang C-Y
    10. Zhao Y
    11. McEachern D
    12. Przybranowski S
    13. Wen B
    14. Sun D
    15. Wang S

    (2018) Discovery of a small-molecule degrader of bromodomain and extra-terminal (bet) proteins with picomolar cellular potencies and capable of achieving tumor regression

    Journal of Medicinal Chemistry 61:462–481.

    https://doi.org/10.1021/acs.jmedchem.6b01816

    • PubMed
    • Google Scholar

Author details

1. Kingsley Y Wu

   Department of Chemistry, University of California, Riverside, Riverside, United States

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing

   Contributed equally with

   Ta I Hung

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-7989-1584

2. Ta I Hung

   1. Department of Chemistry, University of California, Riverside, Riverside, United States
   2. Department of Bioengineering, University of California, Riverside, United States

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing

   Contributed equally with

   Kingsley Y Wu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0009-0008-9702-3829

3. Chia-en A Chang

   Department of Chemistry, University of California, Riverside, Riverside, United States

   Contribution

   Conceptualization, Resources, Software, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing – original draft, Project administration, Writing – review and editing

   For correspondence

   chiaenc@ucr.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6504-8529

• 619

  views

• 49

  downloads

• 0

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.