General Stepwise Approach to Optimize a TR-FRET Assay for Characterizing the BRD/PROTAC/CRBN Ternary Complex

Abstract

Proteolysis-targeting chimeras (PROTACs) degrade target proteins by engaging the ubiquitin-proteasome system. Assays detecting target–PROTAC-E3 ligase ternary complexes are critical for PROTAC development. Both time-resolved fluorescence energy transfer (TR-FRET) assays and amplified luminescent proximity homogeneous assays can characterize ternary complexes and assess PROTAC efficacy; stepwise optimization protocols for these assays are lacking. To identify assay conditions that can be applied to various targets and PROTACs, we used a stepwise approach to optimize a TR-FRET assay of BRD2(BD1)/PROTACs/CRBN ternary complexes. This assay is sensitive and specific and responds to the bivalent PROTACs dBET1, PROTAC BET Degrader-1, and PROTAC BET Degrader-2 but not to non-PROTAC ligands of BRD2(BD1) or CRBN. The activity rank order of dBET1, PROTAC BET Degrader-1, and PROTAC BET Degrader-2 in the TR-FRET assay corresponded with previously reported cell growth inhibition assays, indicating the effectiveness of our assay for predicting PROTAC cellular activity. The TR-FRET ternary complex formation assay for BRD2(BD1)/PROTAC/CRBN can be configured to characterize the binding activities of BRD2(BD1) and CRBN ligands with the same compound activity rank order as that of previously reported binary binding assays for individual targets but with the advantage of simultaneously assessing the ligand activities for both targets. Our assay is modular in nature, as BRD2(BD1) can be replaced with other BRDs and successfully detect ternary complexes without modifying other assay conditions. Therefore, the TR-FRET ternary complex assay for BRDs provides a general assay protocol for establishing assays for other targets and bivalent molecules.

Proteolysis-targeting chimeras (PROTACs) engage the ubiquitin-proteasome system to degrade proteins of interest with a bifunctional molecule that binds to both the protein of interest and an E3 ubiquitin ligase.¹ PROTACs can degrade nuclear receptors, protein kinases, transcriptional regulators,² and conventionally difficult-to-target proteins, such as oncogenic KRAS^G12C.³ Since their inception,⁴ PROTACs have provided a novel mode of action and generated novel chemical tools with great potential for drug development.²

To develop effective PROTACs, assays that reliably detect the binding of PROTACs to both proteins of interest and E3 ligases are critical. Reported assays to monitor PROTAC binding include biochemical assays, such as fluorescence polarization (FP) assays, time-resolved fluorescence energy transfer (TR-FRET) assays, and amplified luminescent proximity homogeneous assays (ALPHAs); biophysical assays, such as isothermal titration calorimetry, surface plasmon resonance, and biolayer interferometry assays; cell-based assays, such as nanoBRET assays; analytical methods, such as size-exclusion chromatography and mass spectrometry; and other methods, such as crystallography, coimmunoprecipitation, and NanoBiT assays.⁵

Unlike traditional small-molecule inhibitors that function via binary interactions between small molecules and their targets, effective PROTAC molecules recruit both the protein to be degraded and the E3 ligase to form a ternary complex. The TR-FRET and ALPHA methods are particularly useful for assessing PROTAC efficacy because they can effectively detect ternary complex formation among the protein of interest, PROTAC molecule, and E3 ligase. Dose–response curves generated from these assays are bell-shaped, with the peak of the curves corresponding with the concentration of the PROTAC at its maximal efficacy (i.e., maximal PROTAC efficacy concentration).^5,6 Because PROTACs that produce maximal efficacy at lower concentrations are more effective, ternary complex formation assays are used to rank PROTACs in structure–activity relationship studies. However, previously reported ALPHA or TR-FRET PROTAC ternary complex formation assays have been historically presented as optimized protocols without detailed information on assay optimization, making them less adoptable for future studies.

The well-established pan-bromodomain (BRD) PROTAC dBET1 (Figure 1) has high affinities to BRD4(BD1), BRD2(BD1), BRD2(BD2), BRD3(BD1), BDR3(BD2), BRD4(BD2), BRDT(BD1), BRDT(BD2), and cereblon (CRBN).⁷ Therefore, we used dBET1 as a representative PROTAC and developed a general TR-FRET PROTAC ternary complex formation assay. We first performed a stepwise optimization approach to detect formation of GST-BRD2(BD1)/dBET1/His-CRBN(DDB1) ternary complexes. We used terbium-labeled anti-GST antibody (Tb-anti-GST) (Tb as the donor fluorophore) and AF488-anti-His (AF488 as the acceptor fluorophore) to detect GST-BRD2(BD1) and His-CRBN(DDB1), respectively. We then confirmed that the established TR-FRET PROTAC ternary complex formation assay specifically detects complexes formed only with appropriate bivalent PROTACs, such as dBET1,⁷ PROTAC BET Degrader-1, and PROTAC BET Degrader-2⁸ (Figure 1) but not with monovalent components of the PROTACs, such as the BRD ligands (+)-JQ1⁹ and HJB97⁸ or the CRBN ligands thalidomide¹⁰ and lenalidomide (Figure 1).¹¹ Among the three PROTACs we evaluated, PROTAC BET Degrader-1 had the highest PROTAC ternary complex formation efficacy, with a maximal PROTAC efficacy concentration of 4.1 nM. The PROTAC BET Degrader-2 and the dBET1 ranked second and third, with respective maximal PROTAC efficacy concentrations of 12.3 and 412 nM. The rank order of maximal PROTAC efficacy concentration of the three BRD-CRBN PROTACs we tested were consistent with that of their previously reported cell growth inhibition activity,⁸ demonstrating that the biochemical TR-FRET BRD/PROTAC/CRBN ternary complex formation assay can effectively predict the cellular efficacy of these PROTACs.

Figure 1.

Chemical structures of tested compounds. The compounds tested included pan-BRD PROTAC dBET1, PROTAC BET Degrader-1, and PROTAC BET Degrader-2; the pan-BRD ligands (+)-JQ1 and HJB97; the cereblon ligands thalidomide and lenalidomide; the BRD/BD2 selective ligand RVX-208; and the VHL ligand VH032.

In our TR-FRET ternary complex formation assay, the formation of the BRD2(BD1)/PROTAC BET Degrader-1/CRBN(DDB1) ternary complex was blocked by either of the BRD2(BD1) ligands (+)-JQ1 and HJB97 or the CRBN ligands thalidomide and lenalidomide in a dose-dependent manner but not by compounds irrelevant to the complex, such as the BRD(BD2) selective ligand RVX-208¹² or the VHL ligand VH032 (Figure 1),^13,14 further confirming the specificity of the assay. This indicates that the assay can be configured to characterize BRD or CRBN ligands for their specific inhibitory binding activities. Ternary complex-based ligand inhibitory assays are advantageous over conventional binary affinity assays because they can be used to evaluate ligands for both the proteins to be degraded and the E3 ligases, whereas conventional binary affinity assays can only measure ligand binding for one of the two proteins involved.⁵

We used the same assay conditions established for BRD2(BD1) to test other BRDs, such as BRD2(BD2), BRD3(BD1), BDR3(BD2), BRD4(BD1), BRD4(BD2), or BRDT(BD1), and observed similar bell-shaped PROTAC dose–response curves from the formation of their respective ternary complexes. Therefore, the assay conditions established for BRD/PROTAC/CRBN may be applicable to targets other than the BRD protein family, although further optimization is needed for each specific target.

Results and Discussion

To establish a successful TR-FRET assay, a pair of appropriate donor and acceptor fluorophores must come in close proximity to allow TR-FRET to occur. We chose the His-CRBN(DDB1) protein, which we previously used to characterize CRBN ligands for their binding affinities in a TR-FRET binding assay,¹⁵ and GST-tagged BRD2(BD1) as the protein to be degraded. We used a Tb-anti-GST antibody (Tb as the donor fluorophore) to detect GST-BRD2(BD1) and an AF488-anti-His antibody (AF488 as the acceptor fluorophore) to detect His-CRBN(DDB1). The PROTAC molecule dBET1⁷ binds to both GST-BRD2(BD1) and His-CRBN(DDB1). In the presence of Tb-anti-GST and AF488-anti-His and under appropriate assay conditions, the formation of a five-component TR-FRET complex (i.e., Tb-anti-GST/GST-BRD2(BD1)/dBET1/His-CRBN(DDB1)/AF488-anti-His) brought the donor and acceptor fluorophores into close proximity to allow TR-FRET to occur when Tb was properly excited. The TR-FRET signal represents the amount of the ternary complex formed, which was controlled by the concentrations of dBET1. At an optimal PROTAC concentration, a maximal amount of ternary complexes are formed. PROTAC concentrations higher or lower than the optimal PROTAC concentration reduce the amount of ternary complexes formed, leading to a bell-shaped dose–response curve. The peak of the bell-shaped curve corresponded to the concentration of PROTAC with maximal efficacy (i.e., maximal PROTAC efficacy concentration). A schematic of the correlation between the TR-FRET signal from the ternary complex and PROTAC concentration is depicted in Figure 2.

Figure 2.

Schematic depicting the TR-FRET ternary complex formation assay. Complexes are formed among GST-BRD2(BD1)/dBET1/His-CRBN(DDB1), with Tb as the donor fluorophore and AF488 as the acceptor fluorophore. A typical bell-shaped dose–response curve is shown.

Stepwise Optimization to Achieve a Sensitive and Stable TR-FRET Assay that Detects the GST-BRD2(BD1)/dBET1/His-CRBN(DDB1) Ternary Complex

To establish a TR-FRET assay to detect BRD/PROTAC/CRBN ternary complexes, we used GST-BRD2(BD1), His-CRBN(DDB1), Tb-anti-GST, and AF488-anti-His with dilutions of dBET1 (1–3 dilutions, 0.57 nM to 100 μM). To optimize the TR-FRET ternary complex formation assay, we used six different concentration combinations of Tb-anti-GST, GST-BRD2(BD1), His-CRBN(DDB1), and AF488-anti-His (conditions 1–6, Table 1) according to our previous study on ligand binding of His-CRBN(DDB1).¹⁵ Our previous study revealed high-affinity binding of His-CRBN(DDB1) to its ligands lenalidomide, thalidomide, and pomalidomide, which allows a very low protein concentration of His-CRBN(DDB1) (2 nM) to be used in the assay to achieve high assay sensitivity. Condition 1 used the lowest amount of all four protein components (2 nM each), and condition 6 used the highest amount of His-CRBN(DDB1) and AF488-anti-His proteins (8 nM each).

Table 1. Protein Concentrations Used to Optimize a TR-FRET Assay to Detect GST-BRD2(BD1)/dBET1/His-CRBN(DDB1) Ternary Complexes.

	protein concentration (nM)
condition	Tb-anti-GST	GST-BRD2(BD1)	His-CRBN(DDB1)	AF488-anti-His
1	2	2	2	2
2	2	4	4	2
3	2	2	4	4
4	4	4	2	2
5	2	2	8	4
6	2	2	8	8

Bell-shaped dose–response curves of dBET1 occurred for all six conditions, with conditions 5 and 6 having the highest maximal height of the curve, followed by conditions 2 and 3 and conditions 1 and 4 (Figure 3). Note that the data shown were generated from the 180 min incubation period, and detailed incubation time optimization is described below.

Figure 3.

Optimization of the TR-FRET assay to detect GST-BRD2(BD1)/dBET1/His-CRBN(DDB1) ternary complexes. Tb-anti-GST (Tb as the donor fluorophore) and AF488-anti-His (AF488 as the acceptor fluorophore) were used. (a) The dBET1 TR-FRET dose–response curves from conditions 1–6 at a 180 min incubation period with signals expressed in RTU (10 000 × 520 nm/490 nm). (b) The dBET1 TR-FRET dose–response curves from conditions 1–6 at a 180 min incubation period with signals expressed as fold change to DMSO. (c) The dBET1 TR-FRET dose–response curves from condition 5 with signals measured every 30 min from 30–300 min and expressed in RTU (10 000 × 520 nm/490 nm). (d) The dBET1 TR-FRET dose–response curves from condition 5 with signals measured every 30 min from 30–300 min and expressed as fold change to DMSO. The DMSO TR-FRET signals are displayed to the left of the curves.

When we examined the TR-FRET signals without normalizing them to DMSO controls, the higher concentrations of His-CRBN(DDB1) coupled with higher concentrations of AF488-anti-His yielded higher relative TR-FRET units (RTU) for all of the dBET1 concentrations tested, as well as for the DMSO control group (Figure 3a). We observed that the high to low relative TR-FRET signals rank order is as follows: conditions 6 and 5 [8 nM His-CRBN(DDB1)], conditions 3 and 2 [4 nM His-CRBN(DDB1)], and conditions 1 and 4 [2 nM His-CRBN(DDB1)]. We also observed that condition 6 (8 nM AF488-anti-His) exhibited a higher TR-FRET signal than did condition 5 (4 nM AF488-anti-His), whereas condition 3 (4 nM AF488-anti-His) exhibited a higher TR-FRET signal than did condition 2 (2 nM AF488-anti-His). Interestingly, higher concentrations of GST-BRD2(BD1) and Tb-anti-GST did not increase the TR-FRET signals. Condition 1 demonstrated slightly higher TR-FRET signals than did condition 4, although condition 1 contained 2 nM of His-CRBN(DDB1) and AF488-anti-His and condition 4 contained 4 nM of GST-BRD2(BD1) and Tb-anti-GST. Therefore, the combined concentrations of His-CRBN(DDB1) and AF488-anti-His apparently confer a larger effect than does the combined concentrations of Tb-anti-GST and GST-BRD2(BD1) on TR-FRET ternary complex formation.

We next divided the TR-FRET signal from dBET1 by that from DMSO to obtain the normalized TR-FRET signal because the signal fold change to DMSO (i.e., signal-to-background ratio) is a suitable assay quality indicator, with higher signal fold changes indicating increased assay robustness.¹⁶ Among the six conditions tested, the TR-FRET maximal PROTAC efficacy concentration of dBET1 was 412 nM for all conditions, regardless of whether RTU (Figure 3a) or the normalized TR-FRET signal (i.e., fold change to DMSO; Figure 3b) was used to generate the dose–response curves. Condition 5, which had the second highest relative TR-FRET signal peak (Figure 3a), demonstrated the highest signal fold change to DMSO, with a 4.59-fold changed peak ratio at 412 nM (Figure 3b). In contrast, condition 6, which had the highest relative TR-FRET signal peak, exhibited the second highest signal fold change to DMSO, with a 4.29-fold changed peak ratio at 412 nM. The only difference between conditions 5 and 6 was the AF488-anti-His concentration (i.e., 4 nM in condition 5 and 8 nM in condition 6). The increased concentration of AF488-anti-His in condition 6 increased the relative TR-FRET signal for both dBET1 and DMSO, but the signal increase over DMSO (i.e., 274 RTU for condition 5 and 393 RTU for condition 6) was higher than that over 412 nM of dBET1 (i.e., 1259 peak RTU for condition 5 and 1689 peak RTU for condition 6), resulting in a slightly lower peak signal fold change to DMSO. For the other conditions tested, the peak signal fold changes to DMSO were 3.42-fold change (condition 2), 2.91-fold change (condition 3), 2.25-fold change (condition 1), and 2.15-fold change (condition 4) at the same maximal PROTAC efficacy concentration. Because condition 5 exhibited the highest signal fold change to DMSO and the second highest RTU, we used condition 5 at a wide range of dBET1 concentrations for further tests, including replacing BRD2(BD1) with other BRDs to expand the applicability of the assay.

PROTAC ternary complex formation assays based on TR-FRET and ALPHA technologies are reported. In previously reported TR-FRET-based assays, one or both proteins are typically labeled after expression and purification,^6,17,18 thereby requiring an extra purification step after labeling, which may introduce variation. In previously reported ALPHA-based methods, most^7,19,20 but not all²¹ use purified proteins without labeling. The TR-FRET ternary complex formation assay we developed for BRD/PROTAC/CRBN does not require a postexpression-labeling step, eliminating any potential labeling variations.

Incubation Time Optimization of the TR-FRET Assay to Detect GST-BRD2(BD1)/dBET1/His-CRBN(DDB1) Ternary Complexes

Using condition 5, we investigated the time dependency of the TR-FRET signals in the ternary complex formation assay with dilutions of dBET1 (1–3 dilutions, 0.57 nM to 100 μM) or DMSO, with the signals collected every 30 min from 30 to 300 min (Figure 3c). Bell-shaped PROTAC ternary complex formation dose–response curves were observed, and the dBET1 maximal PROTAC efficacy concentration was 412 nM for all of the incubation time points tested, consistent with our observation during optimization of the protein concentrations (Figure 3a). The peak of the curves increased slightly from 30 to 300 min, with respective relative TR-FRET signals of 1096, 1098, 1169, 1199, 1241, 1259, 1250, 1258, 1246, and 1332 for 30, 60, 90, 120, 150, 180, 210, 240, 270, and 300 min. The 180 min incubation period was the shortest time to reach a stable peak signal. In contrast, the relative TR-FRET signals for the DMSO controls were 253, 252, 256, 262, 271, 274, 277, 281, 281, and 286 for incubation times from 30 to 300 min in 30 min intervals, respectively. As shown in Figure 3d, the signal fold change to DMSO at different dBET1 concentrations for each incubation time was stable for all of the time points tested. The highest signal fold changes to DMSO observed at 412 nM dBET1 were 4.3-, 4.4-, 4.6-, 4.6-, 4.6-, 4.6-, 4.5-, 4.5-, 4.4-, and 4.6-fold change at every 30 min interval from 30 to 300 min. These observations indicate that our TR-FRET ternary complex formation assay is robust and stable. After considering the highest TR-FRET signals (both RTU and fold change to DMSO), we selected the 180 min incubation period as the optimal time for the TR-FRET ternary complex formation assay.

CRBN-Based BRD2(BD1) PROTACs form Ternary Complexes in the TR-FRET Assay

We applied our TR-FRET ternary complex formation assay with condition 5 and a 180 min incubation to a panel of representative CRBN-based PROTACs for BRD2(BD1) (i.e., dBET1,⁷ PROTAC BET Degrader-1, and PROTAC BET Degrader-2),⁸ BRD2(BD1) ligands (+)-JQ1⁹ and HJB97,⁸ and CRBN ligands thalidomide¹⁰ and lenalidomide¹¹ for their ability to promote PROTAC ternary complex formation between BRD2(BD1) and CRBN. We used a 1–3 dilution pattern for all of the chemicals tested. The optimized chemical concentration range was 0.57 nM to 100 μM for dBET1, (+)-JQ1, thalidomide, and lenalidomide; 0.017 nM to 3 μM for BET Degrader-1 and HJB97; and 5.6 pM to 1 μM for BET Degrader-2. The dose–response curves for complex formation were plotted by using either RTU (Figure 4a) or fold change to DMSO (Figure 4b). We observed similar bell-shaped curves for dBET1, PROTAC BET Degrader-1, and PROTAC BET Degrader-2.

Figure 4.

TR-FRET dose–response curves of PROTAC ternary complex formation. The compounds tested include dBET1, PROTAC BET Degrader-1, PROATC BET Degrader-2, (+)-JQ1, HJB97, thalidomide, and lenalidomide. The assays were performed under condition 5 with a 180 min incubation with Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM). (a) Relative TR-FRET signals in RTU (10 000 × 520 nm/490 nm). (b) Normalized TR-FRET signals as fold change to DMSO.

Only the bivalent BRD/CRBN PROTACs dBET1, PROTAC BET Degrader-1, and PROTAC BET Degrader-2 were able to bring GST-BRD2(BD1) and His-CRBN(DDB1) into close proximity to form GST-BRD2(BD1)/PROTAC/His-CRBN(DDB1) ternary complexes (Figure 4). The respective maximal PROTAC efficacy concentrations for the PROTAC molecule dBET1, PROTAC BET Degrader-1, and PROTAC BET Degrader-2 were 412, 4.1, and 12.3 nM, with the corresponding peak relative TR-FRET signals of 1277, 2339, and 775 RTU and corresponding peak signal fold changes to DMSO of 4.7, 8.7, and 2.9.

As expected, the non-PROTAC molecules, (+)-JQ1, HJB97, thalidomide, and lenalidomide, that bind only to either BRD2(BD1) or CRBN(DDB1) failed to induce ternary complex formation, which was evidenced by flat binding curves. These results clearly indicate that the established TR-FRET ternary complex formation assay for BRD2(BD1) and CRBN(DDB1) is specific for bivalent PROTAC molecules.

The maximal PROTAC efficacy concentrations for the three PROTACs tested were 4.1, 12.3, and 412 nM for PROTAC BET Degrader-1, PROTAC BET Degrader-2, and dBET1, respectively. In cell growth inhibition assays, PROTAC BET Degrader-1, PROTAC BET Degrader-2, and dBET1 have previously reported IC₅₀ values of 4.3, 9.6, and 78.8 nM in RS4;11 cells, respectively, and 45.5, 72.3, and 657 nM in MOLM-13 cells, respectively.⁸ Therefore, the compound rank order of maximal PROTAC efficacy concentrations from the TR-FRET PROTAC ternary complex formation assay is consistent with the IC₅₀ values from previously reported cell growth inhibition assays⁸ (Table 2), demonstrating the effectiveness of the biochemical TR-FRET BRD/PROTAC/CRBN ternary complex formation assay for predicting the cellular activities of PROTACs.

Table 2. Comparison of PROTAC BET Degrader-1, PROTAC BET Degrader-2, and dBET1 for their Previously Reported IC₅₀ Values and TR-FRET Maximal PROTAC Efficacy Concentrations.

	cell growth inhibition (IC50, nM)8
compound	RS4; 11 cells	MOLM-13 cells	TR-FRET maximal efficacy concentration (nM)
PROTAC BET Degrader-1	4.3	45.5	4.1
PROTAC BET Degrader-2	9.6	72.3	12.3
dBET1	78.8	657	412

In addition to different maximal PROTAC efficacy concentrations, dBET1, PROTAC BET Degrader-1, and PROTAC BET Degrader-2 also exhibited different peak heights (Figure 4). This difference may be due to the nature of the TR-FRET phenomenon, which is unique to each PROTAC ternary complex that affects the proximity of the donor and acceptor fluorophores and therefore their energy transfer efficiency.²² The data shown in Figure 3 indicate that although the assay conditions affected the height of the PROTAC peaks, the PROTAC concentrations corresponding to the peaks were constant. Therefore, when comparing compounds, the PROTAC concentration corresponding to the peak, and not the height of the peak itself, relates to the maximal PROTAC efficacy.

Disruption of the BRD2(BD1)/PROTAC BET Degrader 1/CRBN(DDB1) Ternary Complex by Binders of BRD2(BD1) or CRBN

To further investigate the specificity of the PROTAC BET Degrader 1, which was the most effective PROTAC tested, for inducing formation of the ternary complex and the ability of the assay to distinguish compounds that bind either BRD2(BD1) or CRBN from nonbinders, we used PROTAC BET Degrader-1 at 4.1 nM under condition 5 with a 180 min incubation to test a panel of compounds for their ability to disrupt the complex. The panel of compounds included the pan-BRD ligands (+)-JQ1⁹ and HJB97,⁸ CRBN ligands thalidomide¹⁰ and lenalidomide,¹¹ and the irrelevant compounds RVX-208 [a BRD(BD2) selective ligand]¹² and VH032 (a VHL ligand; Figure 5).^13,14 We included HJB97 and thalidomide because they are components of the bivalent PROTAC BET Degrader-1 that binds to BRD and CRBN, respectively. All chemicals were tested in 1–3 dilutions with respective concentration ranges of 0.56 nM to 100 μM for (+)-JQ1, RVX-208, and VH032 and 0.017 nM to 3 μM for HJB97, thalidomide, and lenalidomide. The PROTAC BET Degrader-1 in the absence of a competing compound and DMSO without BET Degrader-1 or a competing compound served as negative (highest signal) and positive (lowest signal) controls, respectively.

Figure 5.

Dose–response curves of competing compounds in the TR-FRET GST-BRD2(BD1)/PROTAC BET Degrader 1/His-CRBN(DDB1) ternary complex formation assay. The compounds tested included (+)-JQ1, HJB97, thalidomide, lenalidomide, RVX-208, and VH032. PROTAC BET Degrader 1 (4.1 nM) was used under condition 5 with Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM) with a 180 min incubation.

Compounds that bind BRD2(BD1) [i.e., (+)-JQ1 and HJB97] or CRBN (thalidomide and lenalidomide) inhibited the PROTAC ternary complex formation in a dose-dependent manner with IC₅₀ values of 239.8, 27.1, 34.7, and 13.2 nM for (+)-JQ1, HJB97, thalidomide, and lenalidomide, respectively. The relative TR-FRET signal (345 ± 15 RTU) of the compounds at high concentrations was similar to that of the DMSO control (304 ± 18 RTU), indicating that (+)-JQ1, HJB97, thalidomide, and lenalidomide abolished ternary complex formation at high concentrations. In contrast, VH032 failed to inhibit ternary complex formation, and RVX-208 only slightly disrupted ternary complex formation at high concentrations. This indicates that PROTAC BET Degrader-1 specifically induces the formation of the GST-BRD2(BD1)/PROTAC BET Degrader-1/His-CRBN(DDB1) ternary complex. Therefore, our TR-FRET PROTAC ternary complex formation assay can be used to identify compounds that disrupt complex formation by binding to one of the two protein targets in the complex. The highest signal from the PROTAC BET Degrader-1 and lowest signal from the DMSO control were 1960 and 304 RTU, respectively. Thus, the highest signal/lowest signal ratio was 6.4, which is a reasonably high assay signal window.

The BRD2(BD1) ligands (+)-JQ1 and HJB97 have reported FP binding inhibitory IC₅₀ values of 36.1 and 3.1 nM, respectively, with an 11.6-fold difference.⁸ In the TR-FRET PROTAC ternary complex formation assay, (+)-JQ1 and HJB97 exhibited corresponding IC₅₀ values of 239.8 and 27.1 nM, with an 8.8-fold difference, which is close to that of the previously reported FP assay. The CRBN ligands thalidomide and lenalidomide have reported TR-FRET binding inhibitory IC₅₀ values of 22.4 and 8.9 nM, respectively, with a 2.5-fold difference.¹⁵ In the TR-FRET PROTAC ternary complex assay, thalidomide and lenalidomide demonstrated corresponding IC₅₀ values of 34.7 and 13.2 nM, with a 2.6-fold difference, which is close to that of the previously reported assay. Comparisons of the compound activities are summarized in Table 3. Therefore, the TR-FRET PROTAC ternary complex formation assay can be properly configured to characterize compounds for their binding inhibitory activities against both protein targets of the ternary complex, which is advantageous over binary binding assays that measure the binding activities of ligands against only one target.⁵

Table 3. IC₅₀ Values of (+)-JQ1, HJB97, Thalidomide, and Lenalidomide in Previous Reports and the TR-FRET PROTAC Ternary Complex Formation Assay.

	BRD2(BD1) IC50 (nM)		CRBN IC50 (nM)
compound	previously reported FP assay8	TR-FRET PROTAC assaya	previously reported TR-FRET assay15	TR-FRET PROTAC assaya
(+)-JQ1	36.1	239.8	NAb	NA
HJB97	3.1	27.1	NA	NA
thalidomide	NA	NA	22.4	34.7
lenalidomide	NA	NA	8.9	13.2

^aIC₅₀ value measured with our TR-FRET ternary complex assay [Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), PROTAC BET Degrader-1 (4.1 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM) with a 180 min incubation.

^bNotation: NA = not applicable.

Applying TR-FRET Ternary Complex Formation Assay Conditions to Other BRDs without Modification

To investigate whether the TR-FRET ternary complex formation assay conditions established for BRD2(BD1)/dBET1/CRBN(DDB1) can be applied to other BRDs without modification, we replaced BRD2(BD1) with other BRDs: BRD2(BD2), BRD3(BD1), BRD3(BD2), BRD4(BD1), BRD4(BD2), BRDT(BD1), and BRDT(BD2). We used the same assay conditions as described above: Tb-anti-GST (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM) together with 2 nM of the GST-BRD to be tested with a 180 min incubation. Because dBET1 is known to interact with all of the BRDs above,⁷ we used a concentration range of 0.17 nM to 30 μM in a 1–3 dilution pattern. We used samples with dBET1 replaced by DMSO as the background controls. As shown in Figure 6, dBET1 bell-shaped dose–response curves occurred for BRD2(BD1), BRD2(BD2), BRD3(BD1), BRD3(BD2), BRD4(BD1), BRD4(BD2), and BRDT(BD1) but not for BRDT(BD2).

Figure 6.

dBET1 dose–response curves from the TR-FRET PROTAC ternary complex formation assay. The reagents included Tb-anti-GST (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM) together with 2 nM of one of the following GST-BRDs: GST-BRD2(BD1), GST-BRD2(BD2), GST-BRD3(BD1), GST-BRD3(BD2), GST-BRD4(BD1), GST-BRD4(BD2), GST-BRDT(BD1), or GST-BRDT(BD2). The incubation time was 180 min. (a) Relative TR-FRET signals in RTU (10 000 × 520 nm/490 nm). (b) Normalized TR-FRET signals as fold change to DMSO. TR-FRET signals from the DMSO controls are displayed to the left of the curves.

On the basis of the curve heights, we divided the eight tested BRDs into four groups (Figure 6). Group 1 (with high PROTAC efficacy) included BRD2(BD1) and BRDT(BD1), with a peak relative TR-FRET signal (RTU), DMSO background signal (RTU), fold change to DMSO, and maximal PROTAC efficacy concentration of 1258, 285, 4.4, and 370 nM for BRD2(BD1) and 1302, 379, 3.4, and 370 nM for BRDT(BD1), respectively. Group 2 (with medium PROTAC efficacy) included BRD2(BD2), BRD3(BD1), and BRD4(BD2), with a peak relative TR-FRET signal (RTU), DMSO background signal (RTU), fold change to DMSO, and maximal PROTAC efficacy concentration of 792, 287, 2.8, and 370 nM for BRD2(BD2); 711, 266, 2.7, and 370 nM for BRD3(BD1); and 804, 299, 2.7, and 123 nM for BRD4(BD2), respectively. Group 3 (with low PROTAC efficacy) included BRD3(BD2) and BRD4(BD1), with a peak relative TR-FRET signal (RTU), DMSO background signal (RTU), fold change to DMSO, and maximal PROTAC efficacy concentration of 449, 292, 1.5, and 123 nM for BRD3(BD2) and 475, 292, 1.6, and 370 nM for BRD4(BD1), respectively. Group 4 included BRDT(BD2), which had a flat curve that was indicative of no ternary complex formation.

The successful application of the optimized TR-FRET PROTAC ternary complex formation assay conditions for BRD2(BD2), BRD3(BD1), BRD3(BD2), BRD4(BD1), BRD4(BD2), and BRDT(BD1) demonstrated the modular nature of our assay, which can serve as a starting point to optimize assays for other targets and PROTACs. However, the same assay conditions were not successfully applied to BRDT(BD2). Because dBET1 reportedly binds to both BRDT(BD2) and CRBN,⁷ additional assay optimization is needed.

DMSO Tolerance Test of the TR-FRET Assay for the BRD2(BD1)/PROTAC BET Degrader-1/CRBN Ternary Complex

DMSO is a commonly used solvent for chemicals, and we used DMSO to prepare all of our stock chemical solutions. At high concentrations, DMSO can affect TR-FRET assays.^{14,15,23−25} We used the TR-FRET assay for the BRD2(BD1)/PROTAC BET Degrader-1/CRBN ternary complex as an example to evaluate the assay’s DMSO tolerance. We selected PROTAC BET Degrader-1 as a representative PROTAC because of its highest signal window (Figure 4). The specific assay conditions were as follows: PROTAC BET Degrader-1 (1–3 dilutions, 0.017 nM to 3 μM), Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM) at a 180 min incubation and final DMSO concentrations of 0.1%, 0.5%, 1%, 2%, 5%, and 10% (Figure 7). We observed bell-shaped dose–response curves for PROTAC BET Degrader-1 at all of the DMSO concentrations tested. However, when the DMSO concentration increased from 0.1%, 0.5%, 1%, 2%, or 5% to 10%, the TR-FRET signals decreased at all of the PROTAC BET Degrader-1 concentrations tested, with a respective peak relative TR-FRET signal (RTU), DMSO background signal (RTU), fold change to DMSO, and maximal PROTAC efficacy concentration of 3363, 267, 12.6, and 4.1 nM, for 0.1% DMSO; 3185, 269, 11.8, and 4.1 nM for 0.5% DMSO; 2810, 274, 10.3, and 4.1 nM for 1% DMSO; 2362, 278, 8.5, and 4.1 nM for 2% DMSO; 1489, 281, 5.3, and 12.3 nM for 5% DMSO; and 689, 336, 2.1, and 37.0 nM for 10% DMSO (Figure 7a and b; Table 4). The increased DMSO concentrations not only decreased the TR-FRET signals but also shifted the maximal PROTAC efficacy concentration of PROTAC BET Degrader-1 from low (i.e., 4.1 nM at 0.1%, 0.5%, 1%, and 2% DMSO) to high (i.e., 12.3 nM at 5% and 37.0 nM at 10% DMSO), representing a 9-fold change. The peak TR-FRET signals of the PROTAC ternary complex at DMSO concentrations of 0.5%, 1%, 2%, 5%, and 10% were 94.7%, 83.6%, 70.2%, 44.3%, and 20.5% of that of 0.1% DMSO, respectively. From 0.1% to 2% DMSO, the maximal PROTAC efficacy concentration of PROTAC BET Degrader-1 remained at 4.1 nM. Therefore, the DMSO concentration should not be higher than 2% to maintain maximal PROTAC efficacy. Consequently, we maintained the DMSO concentration at 1% or 1.1% for all assays except for the DMSO tolerance test to ensure maximal ternary complex formation and accommodate the various solubilities of the tested chemicals.

Figure 7.

DMSO tolerance test of the PROTAC BET Degrader-1 TR-FRET ternary complex formation assay. Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM) were used with a 180 min incubation. (a) Relative TR-FRET signals (10 000 × 520 nm/490 nm, RTU). (b) Normalized TR-FRET signals (as fold change to DMSO). TR-FRET signals from the DMSO control are displayed to the left of the curves.

Table 4. Effect of DMSO Concentration on the Peak Relative TR-FRET Signal and the Corresponding Maximal PROTAC Efficacy Concentration for PROTAC BET Degrader-1 in the TR-FRET Complex Formation Assay.

DMSO concentration	peak relative TR-FRET signal (RTU)	maximal PROTAC efficacy concentration (nM)
0.1%	3363	4.1
0.5%	3185	4.1
1%	2810	4.1
2%	2362	4.1
5%	1489	12.3
10%	689	37.0

In conclusion, we used a stepwise optimization approach to develop a TR-FRET ternary complex formation assay for BRD2(BD1)/CRBN and their PROTACs. Only bivalent CRBN-based PROTACs for BRDs, such as dBET1, PROTAC BET Degrader-1, and PROTAC BET Degrader-2, but not the non-PROTAC monovalent ligands of BRD2(BD1) or CRBN, induced formation of the ternary complexes, indicating the specificity of the assay. Furthermore, we used Tb-anti-GST to detect GST-BRD2(BD1) and AF488-anti-His to detect His-CRBN(DDB1), thereby avoiding postpurification protein labeling and the possible variation this introduces. The rank order of activity for dBET1, PROTAC BET Degrader-1, and PROTAC BET Degrader-2 from the TR-FRET PROTAC ternary complex formation assay matches that of previously reported cell growth inhibition assays, indicating the effectiveness of the optimized TR-FRET PROTAC ternary complex formation assay for predicting the cellular activities of the PROTACs. The TR-FRET PROTAC ternary complex formation assay for BRD2(BD1)/CRBN can be configured to characterize BRD2(BD1) and CRBN ligands for their binding inhibitory activities against respective targets with the same activity rank order as that previously reported in binary binding assays for individual targets. However, our TR-FRET PROTAC ternary complex formation assay is advantageous over monotargeted-based binary assays because it can assess ligand inhibitory activities for both relevant targets. PROTAC ternary complex formation was inhibited only by binders and not by nonbinders of BRDs or CRBN, further confirming the specificity of the assay. The optimized TR-FRET PROTAC ternary complex formation assay is also modular in nature because BRD2(BD1) can be replaced with another BRD, such as BRD2(BD2), BRD3(BD1), BRD3(BD2), BRD4(BD1), BRD4(BD2), and BRDT(BD1), to detect ternary complex formation without the need to change other assay conditions. Thus, the optimized TR-FRET PROTAC ternary complex formation assay for BRD2(BD1)/PROTAC/CRBN may provide a general assay protocol and be used to establish assays that detect ternary complexes for other bivalent compounds and their targets.

Materials and Methods

Reagents

We purchased dBET1, PROTAC BET Degrader-1, PROTAC BET Degrader-2, HJB97, thalidomide, and lenalidomide from MedChemExpress USA (Monmouth Junction, NJ). RVX-208 was purchased from MedKoo Biosciences, Inc. (Morrisville, NC). We purchased (+)-JQ1 from Cayman Chemical (Ann Arbor, MI), and VH032 was prepared in house.¹⁴ Triton X-100 and bovine serum albumin were purchased from Sigma (St. Louis, MO). Tb-anti-GST and AF488-anti-His, Tris (1 M, pH 7.5), DTT (1 M), and DMSO were purchased from Fisher Scientific (Pittsburgh, PA). GST-BRD2(BD1), GST-BRD2(BD2), GST-BRD3(BD1), GST-BRD3(BD2), GST-BRD4(BD1), GST-BRD4(BD2), GST-BRDT(BD1), and GST-BRDT(BD2) were purchased from BPS Bioscience (San Diego, CA). His-CRBN(DDB1) protein was custom prepared by GenScript U.S.A., Inc. (Piscataway, NJ)¹⁵ and by the Protein Production Facility at St. Jude Children’s Research Hospital. Black 384-well, low-volume assay plates were purchased from Corning Incorporated Life Sciences (Tewksbury, MA).

General Assay Conditions

All chemicals were prepared as stock solutions in DMSO and dispensed with an Echo 555 acoustic liquid handler (Labcyte Inc., San Jose, CA) to yield the final indicated concentrations. The final DMSO concentration in the assays was 1%, unless otherwise specified. The assay buffer was composed of 50 mM Tris (pH 7.5), 0.01% Triton X-100, 0.01% bovine serum albumin, and 1 mM DTT and was freshly prepared before each experiment. After each reagent addition step, the assay plates were shaken at 900 rpm for 1 min on an IKA MTS 2/4 digital microtiter plate shaker (IKA Works, Wilmington, NC, USA) and centrifuged at 1000 rpm for 1 min in an Eppendorf 5810 centrifuge with an A-4-62 swing-bucket rotor (Eppendorf AG, Hamburg, Germany). The relative TR-FRET signals (10 000 × 520 nm/490 nm, in RTU) were measured with a PHERAstar FS plate reader (BMG Labtech; Durham, NC).

Protein Concentration Optimization

We incubated dBET1 (1–3 dilutions) or DMSO with a mixture of Tb-anti-GST and GST-BRD2(BD1) (10 μL/well, Conditions 1–6) in black 384-well, low-volume assay plates for 30 min. A mixture of His-CRBN(DDB1) and AF488-anti-His (10 μL/well, Conditions 1–6) was then dispensed. The TR-FRET signal of each well was measured after the plates were incubated in the dark for 180 min. The final dBET1 concentrations ranged from 0.57 nM to 100 μM. The final protein concentrations for condition 1 were as follows: Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (2 nM), and AF488-anti-His (2 nM). The final protein concentrations for condition 2 were as follows: Tb-anti-GST (2 nM), GST-BRD2(BD1) (4 nM), His-CRBN(DDB1) (4 nM), and AF488-anti-His (2 nM). The final protein concentrations for condition 3 were as follows: Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (4 nM), and AF488-anti-His (4 nM). The final protein concentrations for condition 4 were as follows: Tb-anti-GST (4 nM), GST-BRD2(BD1) (4 nM), His-CRBN(DDB1) (2 nM), and AF488-anti-His (2 nM). The final protein concentrations for condition 5 were as follows: Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM). The final protein concentrations for condition 6 were as follows: Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (8 nM).

Incubation Time Optimization

We incubated dBET1 (1–3 dilutions) or DMSO with a mixture (10 μL/well) of Tb-anti-GST (4 nM) and GST-BRD2(BD1) (4 nM) in black 384-well, low-volume assay plates for 30 min. A mixture (10 μL/well) of His-CRBN(DDB1) (16 nM) and AF488-anti-His (8 nM) was then dispensed. The TR-FRET signal of each well was measured every 30 min from 30–300 min as the plates were incubated in the dark. The final dBET1 concentrations ranged from 0.57 nM to 100 μM. The final protein concentrations were as follows: Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM), as described for optimized condition 5.

TR-FRET Ternary Complex Formation with Selected Ligands

Dilutions (1–3 dilutions) of dBET1, PROTAC BET Degrader-1, PROTAC BET Degrader-2, (+)-JQ1, HJB97, thalidomide, lenalidomide, or DMSO were incubated with a mixture (10 μL/well) of Tb-anti-GST (4 nM) and GST-BRD2(BD1) (4 nM) in black 384-well, low-volume assay plates for 30 min. A mixture (10 μL/well) of His-CRBN(DDB1) (16 nM) and AF488-anti-His (8 nM) was then dispensed. The TR-FRET signal of each well was measured after the plates were incubated in the dark for 180 min. The final ligand concentration ranges were as follows: dBET1 (0.57 nM to 100 μM), PROTAC BET Degrader-1 (0.017 nM to 3 μM), PROTAC BET Degrader-2 (5.6 pM to 1 μM), (+)-JQ1 (0.57 nM to 100 μM), HJB97 (0.017 nM to 3 μM), thalidomide (0.57 nM to 100 μM), and lenalidomide (0.57 nM to 100 μM). The final protein concentrations were as follows: Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM), as described for optimized condition 5.

Inhibitory Activities of Selected Ligands

Dilutions (1–3 dilutions) of (+)-JQ1, HJB97, thalidomide, lenalidomide, RVX-208, VH032, or DMSO were incubated with a mixture (10 μL/well) of Tb-anti-GST (4 nM) and GST-BRD2(BD1) (4 nM) with PROTAC BET Degrader-1 (8.2 nM) in black 384-well, low-volume assay plates for 30 min. A mixture (10 μL/well) of His-CRBN(DDB1) (16 nM) and AF488-anti-His (8 nM) was then dispensed. The TR-FRET signal of each well was measured after the plates were incubated in the dark for 180 min. The final tested ligands and their concentrations were as follows: (+)-JQ1 (0.57 nM to 100 μM), HJB97 (0.017 nM to 3 μM), thalidomide (0.017 nM to 3 μM), lenalidomide (0.017 nM to 3 μM), RVX-208 (0.57 nM to 100 μM), or VH032 (0.57 nM to 100 μM). All ligand dilutions were incubated with PROTAC BET Degrader-1 (4.1 nM). The final DMSO concentration was 1.1%, with 1% contributed from the ligand dilutions and 0.1% from PROTAC BET Degrader-1. The final protein concentrations were Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM), as described for optimized condition 5. In addition, PROTAC BET Degrader-1 (4.1 nM) or DMSO with proteins under condition 5 was included in each plate to serve as references or controls, and the signals were measured after a 180 min incubation.

TR-FRET Ternary Complex Formation Among GST-BRDs/dBET1/His-CRBN(DDB1)

We incubated dBET1 (1–3 dilutions) or DMSO with a mixture (10 μL/well) of Tb-anti-GST (4 nM) and GST-BRDx (4 nM) in black 384-well, low-volume assay plates for 30 min. A mixture (10 μL/well) of His-CRBN(DDB1) (16 nM) and AF488-anti-His (8 nM) was then dispensed. The TR-FRET signals in each well were measured after the plates were incubated in the dark for various times (every 30 min, 30–300 min). The final dBET1 concentrations ranged from 0.17 nM to 30 μM. The final protein concentrations were as follows: Tb-anti-GST (2 nM), GST-BRDx (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM), as described for the optimized condition 5 for GST-BRD2(BD1). GST-BRDx represents GST-BRD2(BD1), GST-BRD2(BD2), GST-BRD3(BD1), GST-BRD3(BD2), GST-BRD4(BD1), GST-BRD4(BD2), GST-BRDT(BD1), or GST-BRDT(BD2).

DMSO Tolerance Test

PROTAC BET Degrader-1 (1–3 dilutions) or DMSO was incubated with a mixture (10 μL/well) of Tb-anti-GST (4 nM) and GST-BRD2(BD1) (4 nM) in black 384-well, low-volume assay plates for 30 min. A mixture (10 μL/well) of His-CRBN(DDB1) (16 nM) and AF488-anti-His (8 nM) was then dispensed. The TR-FRET signals in each well were measured after the plates were incubated in the dark for 180 min. The final PROTAC BET Degrader-1 concentration ranged from 0.017 nM to 3 μM. The final DMSO concentrations were 0.1%, 0.5%, 1%, 2%, 5%, and 10%. The final protein concentrations were as follows: Tb-anti-GST (2 nM), GST-BRD2(BD1) (2 nM), His-CRBN(DDB1) (8 nM), and AF488-anti-His (4 nM), as described for optimized condition 5.

Data Analysis

All assays were tested in quadruplicate at least three times, and each data point represents the mean ± standard deviation. Data were analyzed with GraphPad PRISM software (version 9.0.0; GraphPad Software, San Diego, CA). The signal fold change to DMSO was calculated by dividing the relative TR-FRET signal of a compound with that of DMSO. For the TR-FRET ternary complex formation assay, dots representing the relative TR-FRET signals or signal fold changes to DMSO from various concentrations of each compound were connected to illustrate the PROTAC efficacy dose–response curves. For ligand binding inhibitory tests using the TR-FRET ternary complex formation assay, the TR-FRET signals of each ligand at their corresponding concentrations were analyzed with a built-in sigmoidal dose–response fitting equation to derive the dose–response curves and corresponding IC₅₀ values, if applicable.

Glossary

Abbreviations

TR-FRET

time-resolved fluorescence resonance energy transfer

PROTAC

proteolysis-targeting chimera

BRD

bromodomain

CRBN

cereblon

ALPHA

amplified luminescent proximity homogeneous assay

BRD2(BD1)

bromodomain-containing protein 2 bromodomain 1

BRD2(BD2)

bromodomain-containing protein 2 bromodomain 2

BRD3(BD1)

bromodomain-containing protein 3 bromodomain 1

BRD3(BD2)

bromodomain-containing protein 3 bromodomain 2

BRD4(BD1)

bromodomain-containing protein 4 bromodomain 1

BRD4(BD2)

bromodomain-containing protein 4 bromodomain 2

BRDT(BD1)

bromodomain testis-specific protein bromodomain 1

BRDT(BD2)

bromodomain testis-specific protein bromodomain 2

KRAS

Kirsten rat sarcoma viral oncogene homologue protein

FP

fluorescence polarization

HTS

high throughput screening

GST

glutathione S-transferase

His

6×polyhistidine tag

DDB1

damage-specific DNA binding protein 1

Tb-anti-GST

terbium-labeled anti-GST tag antibody

AF488-anti-His

alexa fluor 488 dye-labeled anti-His tag antibody

VHL

von Hippel–Lindau

RTU

relative TR-FRET unit

IC₅₀

half maximal inhibitory concentration

Author Contributions

W.L. designed and conducted the experiments and performed data analysis. W.L. and T.C. wrote the manuscript.

The authors declare no competing financial interest.