In Vitro Ubiquitination Platform Identifies Methyl Ellipticiniums as Ubiquitin Ligase Inhibitors

Abstract

The transfer of the small protein ubiquitin to a target protein is an intricately orchestrated process called ubiquitination that results in modulation of protein function or stability. Proper regulation of ubiquitination is essential, and dysregulation of this process is implicated in several human diseases. An example of a ubiquitination cascade that is a central signaling node in important disease-associated pathways is that of CBLB [a human homolog of a viral oncogene Casitas B-lineage lymphoma (CBL) from the Cas NS-1 murine retrovirus], a RING finger ubiquitin ligase (E3) whose substrates include a number of important cell-signaling kinases. These include kinases important in immune function that act in the T cell receptor and costimulatory pathways, the Tyro/Axl/MerTK (TAM) receptor family in natural killer (NK) cells, as well as growth factor receptor kinases like epidermal growth factor receptor (EGFR). Loss of CBLB has been shown to increase innate and adaptive antitumor immunity. This suggests that small-molecule modulation of CBLB E3 activity could enhance antitumor immunity in patients. To explore the hypothesis that enzymatic inhibition of E3s may result in modulation of disease-related signaling pathways, we established a high-throughput screen of >70,000 chemical entities for inhibition of CBLB activity. Although CBLB was chosen as a proof-of-principle target for inhibitor discovery, we demonstrate that our assay is generalizable to monitoring the activity of other ubiquitin ligases. We further extended our observed in vitro inhibition with additional cell-based models of CBLB activity. From these studies, we demonstrate that a class of natural product–based alkaloids, known as methyl ellipticiniums (MEs), is capable of inhibiting ubiquitin ligases intracellularly.

Keywords

CBL ubiquitin ligase inhibitor high throughput E3

Introduction

Protein homeostasis is a tightly regulated and essential component of cellular viability. In mammalian cells, this process is partially regulated by posttranslational modification of proteins by the addition of the small protein ubiquitin. The successful transfer of ubiquitin to a target protein is part of a complex poly-enzymatic process referred to as ubiquitination. Successful completion of this modification requires the presence of a ubiquitin-activating enzyme, commonly called an E1; a ubiquitin-conjugating enzyme (E2); a ubiquitin ligase (E3); the target protein; ubiquitin; and the cofactors adenosine triphosphate (ATP) and magnesium. While ubiquitination requires the presence of all components, it is the E3 that confers target specificity toward the final acceptor of ubiquitin.¹ This specificity makes the E3 an attractive therapeutic target. The discovery of an E3-specific inhibitor is well suited for creatively designed and robustly implemented high-throughput screening (HTS).² To this end, we developed, optimized, and implemented a HTS platform for the discovery of modulators of E3 activity. As a proof of principle, we have used this assay to discover inhibitors of the Casitas B-lineage lymphoma proto-oncogene B (CBLB) ubiquitin ligase.³ We subsequently applied our in vitro ubiquitination assay (IVUA) to other ubiquitin ligases to both assess the specificity of our putative CBLB inhibitors and demonstrate the broad applicability of our screening platform to other E3s.

The mammalian CBL family (CBL, CBLB, and CBLC) of enzymes plays an integral role in the regulation of receptor tyrosine kinase (RTK) signaling by coordinating the ubiquitination of activated RTKs, ultimately resulting in their internalization and lysosomal degradation.⁴ Modulation of epidermal growth factor receptor (EGFR) signaling through CBL-mediated ubiquitination has been widely studied and well characterized.^5,6 As members of the Really Interesting New Gene (RING) finger ubiquitin ligase superfamily, CBLs all contain the catalytic RING finger (RF) domain, a tyrosine kinase binding (TKB) domain, a proline-rich SH3-interacting domain, and, in CBL and CBLB, a ubiquitin-associated (UBA) domain.⁴ Uniquely among the CBL family, CBLB negatively regulates adaptive (T cell) and innate (NK cell) antitumor immunity. In T cells, CBLB inhibits the costimulatory pathway by ubiquitinating the p85 subunit of PI3 kinase (PI3K) and preventing the recruitment of PI3K to the CD28 costimulatory receptor.⁷ In NK cells, CBLB inhibits NK cell activity downstream of the inhibitory Tyro/Axl/MerTK (TAM) receptor activation.^7,8 Genetic ablation of CBLB in mice results in increased CD8 T cell– and NK cell–mediated killing of transplanted and spontaneous tumors.^7,9,10 Importantly, reconstitution experiments of either wild-type or RF mutant CBLB into CBLB knockout mice demonstrated that the loss of CBLB catalytic E3 activity is essential to tumor suppression in CBLB-null mice.^8,11

Although we demonstrate that the platform we have developed can be applied broadly to other E3s, the potential translational applications led us to choose CBLB as our first target of interest for a 384-well formatted HTS for inhibitors of its ubiquitin ligase activity. From a screen of >70,000 small molecules, including natural products, we discovered a variety of pharmacophores, with a range of potencies, capable of inhibiting CBLB activity in in vitro biochemical assays. One structural class, natural product–derived alkaloids named methyl ellipticiniums (MEs), inhibited CBLB at low-micromolar concentrations. Herein, we describe the first CBLB-specific HTS and identify the chemical structure of an intracellularly active small-molecule inhibitor of CBLB. MEs represent a new pharmacophore for ubiquitin ligase inhibition with the potential to improve our understanding of the function of CBLB and therapeutic modulation of this target.

Materials and Methods

Recombinant Protein Production

Experimental procedures are described in greater detail in the Supplemental Material. Briefly, the UBA domains of CBLB (aa 914–982) and Cbl (aa 839–906) were cloned into pGEX 2TK or pGEX 5X-1 (GE Healthcare Biosciences, Pittsburgh, PA), respectively, as previously described.¹² The N-terminal regions containing the tyrosine-binding domain, linker, and RF of CBLB (aa 2–448) and Cbl (aa 2–469) were cloned into pGEX 6P-1 as previously described.^4,13 The bacterial expression construct for Ube2d2, pETb-UbcH5b, was kindly provided by Dr. Allan Weissman. Recombinant GST proteins were produced as previously described.¹³ Recombinant Ube2d2 was produced and purified as previously described.¹⁴ Recombinant HRD1 (HMG-CoA reductase degradation 1) protein was kindly provided by Allan Weissman.

Large-scale preparation of recombinant proteins was performed by the Protein Expression Laboratory, Frederick National Laboratory for Cancer Research, by modification of the above procedures.

UBA-Dependent Polyubiquitin Enrichment Experiments

This experimental design is a modification of previously published protocols and is described in greater detail in the Supplemental Material.^12,15,16

Establishing a Plate-Based IVUA

This assay is a modification of a previously published assay, with minor changes.¹⁵ Most notably, we have previously used radiolabeled ligand to follow ubiquitin transfer; however, we have modernized the assay to instead use biotinylated ubiquitin, whose presence can be detected through the use of avidin conjugated to horseradish peroxidase (HRP). Complete descriptions of methods are provided in the Supplemental Material.

IVUA HTS Development and Implementation

Following comprehensive assay optimization (Supplemental Materials), 384-well HTS was executed in the following manner. Black Fluotrac 600 high protein-binding plates (subsequently referred to as reaction plates; Greiner Bio-One, Monroe, NC) were coated overnight in a solution of 10 µg/mL UBA-GST in 1× phosphate-buffered saline (PBS) pH 7.4 supplemented with 1 mM ZnCl at a volume of 50 µL per well. On the following day, plates were emptied and manually filled with 100 µL of blocking solution [1× Tris-buffered saline with Tween 20 (TBST) supplemented with 1% bovine serum albumin (BSA)] using a Microfill device (BioTeK, Winooski, VT). Following a 1 h blocking period, plates were then washed thrice using an automated plate washer (ELx405 equipped with a plate stacker, BioTek) with 1× TBST and left filled with 1× PBS while the reaction solution was prepared. The reaction solution (1.2×) is composed of 1.2× CBLB reaction buffer [60 mM Tris pH 7.5, 0.12 mM dithiothreitol (DTT), 0.6 mM MgCl₂, 0.012% Triton X-100, and 0.6 mg/mL bovine gelatin] with 12 nM E1, 60 nM E2, 90 nM E3, 600 nM unlabeled ubiquitin, and 60 nM biotinylated ubiquitin. The reaction solution is prepared and kept on ice immediately before use. Substances are diluted from stock plates into dilution plates containing a solution whose final concentration after compound addition is sixfold higher than the screening concentration. The composition of substance dilution buffer is 100 mM Tris pH 7.5, 600 µM ATP, 3 mM MgCl₂, 2.4% DMSO, and 60 µM library substance. Substance dilution plates also contain low controls in column 1 (100 mM Tris pH 7.5 and 2.4% DMSO) and high controls in column 2 (100 mM Tris pH 7.5, 2.4% DMSO, 3 mM MgCl₂, and 600 µM ATP). A Biomek FX-2 liquid handler (Beckman Coulter, Pasadena, CA) is used to assemble the complete substance dilution plates. UBA-coated, blocked, and washed reaction plates are emptied of 1× PBS and refilled with 25 µL per well of reaction solution. Filled reaction plates are then placed on the FX-2, and the reaction is initiated by the 384-well-head mediated transfer of 5 µL from the substance dilution well to its respective well of the reaction plate. This brings the final screening conditions to 30 µL at 1× concentration of all reaction components and initiates the in vitro ubiquitination reaction. Reactions are then allowed to proceed at room temperature for 1 h prior to quenching with the addition of 20 µL of 6 mM ZnCl in 100 mM Tris pH 7. Plates are then sealed and left on a plate rocker overnight. On the following day, plates are washed thrice with 1× TBST before the addition of 100 µL per well of a 1:10,000 dilution of Avidin-HRP in 1× TBST supplemented with 1% BSA. Incubation with Avidin-HRP proceeds for 1 h, during which time the working solution of QuantaBlu (QB) reagent is generated according to the manufacturer’s protocol and allowed to come to room temperature. Following another cycle of triple washing with 1× TBST, 50 µL of QB working solution is added to each well, and HRP-dependent fluorophore generation is allowed to occur. After 1 h, 50 µL of QB stop solution is added to each well, and plates are then shaken on a plate shaker, equilibrated, and read for fluorescence [excitation (ex): 320 nm; emission (em): 420 nm] on a Tecan M1000 Infinite plate reader (Tecan Group, Männedorf, Switzerland). Substance wells on each plate are normalized to their plate-based controls based on the following formula:

% N o r m a l i z e d A c t i v i t y = \frac{(\begin{array}{l} R F U_{S u b s t a n c e} - \\ A v g . R F U_{L o w C o n t r o l s} \end{array})}{(\begin{array}{l} A v g . R F U_{H i g h C o n t r o l s} - \\ A v g . R F U_{L o w C o n t r o l s} \end{array})} * 100

(1)

In addition, an assessment of data quality (Z-factor) for each plate was calculated using the following formula:

Z - F a c t o r = 1 - \frac{3 * (S t d . D e v ._{H i g h C o n t r o l s} + S t d . D e v ._{L o w C o n t r o l s})}{| A v g . R F U_{H i g h C o n t r o l s} - A v g . R F U_{L o w C o n t r o l s} |}

(2)

Quadruplicate reconfirmation and 5-point dose response were executed identically to the primary screening except that the substance dilution plates were assembled such that substances were formatted in quadruplicate (2×2 pattern) or 5-point duplicate [2 (duplicate) × 5 (doses) pattern across a row] dose response. Buffer composition was identical to that of primary screening. p38 kinase dose–response testing was carried out as previously published.¹⁷

Primary lead (and derivatives) compound IC₅₀ determination was carried out manually in a similar manner to that of primary screening. Dry resupplied compounds were resuspended in DMSO at a concentration of 100 mM. From this stock solution, a 6× high test concentration solution was prepared with the following composition: 600 µM substance, 600 µM ATP, 3 mM MgCl₂, 3% DMSO, and 100 mM Tris pH 7.5. This solution was then used to generate a 10-point 6× dose–response solution set from 600 to 0.5 µM (varying on the substance concentration). Vehicle control (no substance) and “No ATP” control solutions were also assembled. A 1.2× reaction solution was assembled as described above. Using 96-well plates (Thomas Scientific, Swedesboro, NJ), 12 reactions (10 substance and 2 control) were set up and initiated by the addition of substance solutions to reaction solutions using a 12-channel pipettor; initiated reactions were then transferred to a UBA-coated, blocked, and washed reaction plate. All substance concentrations were screened in quadruplicate and executed as for the primary screen once the initiated reactions were transferred to the reaction plates. Following processing and normalization as above, IC₅₀ values were calculated by fitting the normalized data to the following formula using Prism Software (GraphPad Software, San Diego, CA) using a variable slope and a least-squares fit:

% N o r m a l i z e d A c t i v i t y = \frac{100}{(1 + 10^{((L o g I C_{50} - L o g [S u b s t a n c e]) * (H i l l S l o p e))})}

(3)

IC₅₀ determination using E6AP was carried out as above for CBLB using the following final ubiquitination conditions: 30 nM E1, 50 nM E2 (UbcH7/UBE2L3, Boston Biochem, Cambridge, MA), 100 nM E3 (E6AP/UBE3A, Boston Biochem), 500 nM ubiquitin, and 50 nM biotinylated ubiquitin.

Methods for the assay optimization experiments shown in Supplemental Figure S1 are included in the Supplemental Materials.

Gel-Based Assay for E1/E2 Activity

A polyacrylamide gel electrophoresis (PAGE) assay was designed to assess the ubiquitination cascade upstream of the E3 ligase activity. Assay reactions, as described above for establishing the IVUA, were performed in Eppendorf tubes containing 28 µL of master mix containing 50 mM Tris-HCl pH 7.5, 0.5 mM MgCl₂, 0.2 mM ATP, 0.1 mM DTT, 500 nM ubiquitin, and 85 nM biotinylated ubiquitin with or without 1 nM E1. To test the activity of compounds, 1 µL of 30× prepared dilutions of each stock compound to be tested was added to achieve the final desired concentrations of 0.1, 1, or 10 µM. As a negative control, 1nM E1 was used in the absence of ubiquitin and ATP. 0.5 mM MgCl₂ without the other assay reagents (e.g., no ubiquitin or ATP). The reaction was initiated by the addition of 1 µL of E2 (0.016 µg) and incubated at RT for 2 h. The reaction was stopped with the addition of 8 µL of 4× Laemmli loading buffer, and the samples were immediately loaded onto Tris-Glycine4-20% polyacrylamide gradient precast gels (Bio-Rad, Hercules, CA) without boiling and without addition of reducing agent. The proteins were resolved by sodium dodecyl sulfate (SDS)-PAGE, transferred to nitrocellulose membranes using an iBlot2 transfer apparatus (Life Technologies, Waltham, MA), and immunoblotted with primary antibodies overnight. Goat anti-rabbit immunoglobulin G (IgG) HRP conjugate (cat. no. 1721019, BioRad, Hercules, CA) or goat anti-mouse IgG HRP conjugate antibodies (cat. no. 1721011, BioRad) were used with SuperSignal West Pico Plus detection reagent (Thermo Fisher Scientific, Waltham, MA) to visualize the blots on a Li-Cor Odyssey Imager (Li-Cor Biotechnology, Lincoln, NE).

Cell-Based Assay for CBL Family–Dependent EGFR Ubiquitination

HeLa cells (CCL2) were obtained from ATCC (American Type Culture Collection, Manassas, VA) and maintained in culture in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Small interfering RNA (siRNA) transfection was performed with Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher Scientific). 2×10⁶ cells were plated in 10 cm dishes and allowed to adhere overnight. On the following day, cells were transfected with the following siRNAs: Cbl (cat. no. s2476), CBLB (cat. no. s2479), or a negative control (cat. no. 4390844) (all Thermo Fisher Scientific) according to the manufacturer’s protocol for 24 h, and then trypsinized, pooled, and replated into two 10 cm plates to make the transfections equal among sets. The cells were further incubated for 24–48 h and then starved in DMEM without serum for 3 h before stimulation with 25 ng/ml epidermal growth factor (EGF). Immunoblotting and immunoprecipitation are described below.

To develop HeLa cells with a genomic knockout of either CBL, oligos targeting CBL (sense: AAACGCATCTCCGTACTATCTTGTG; antisense: CACCGACAAGATAGTACGGAGATGC) were cloned into the PX459 vector. The HF-PX459 (V2) was a gift from Mike McGrew (Addgene plasmid no. 118632, RRID:Addgene_118632, Addgene, Watertown, MA, http://n2t.net/addgene:118632).¹⁸ These vectors contained the Cas9 gene, a single guide RNA, and puromycin resistance. Cells were selected in puromycin, single-cell clones expanded, and knockout clones were identified by immunoblotting and confirmed by sequencing the genomic DNA. These stable knockouts were then used in subsequent experiments, in which they were pretreated with the indicated compounds for 18 to 48 h prior to a 10 min EGF (100 ng/mL) stimulation and lysate creation, as described below.

To harvest proteins, cells were washed twice in ice-cold Dulbecco’s phosphate-buffered saline (DPBS) and then lysed in ice-cold Nonidet P40 (NP40) cell lysis buffer [50 mM Tris pH 7.4, 250 mM NaCl, 5 mM ethylenediaminetetraacetic acid (EDTA), 50 mM NaF, 1 mM sodium orthovanadate (Na₃VO₄), 1% NP40, and 0.02% NaN₃] supplemented with 1 mM Na₃VO₄ and protease inhibitors (cOmplete Protease Inhibitor Cocktail Tablets, Millipore Sigma, St. Louis, MO). The lysates were cleared of debris by centrifugation at 13,000 g for 10 min at 4 °C. The supernatants were collected, and protein concentrations were determined using a BioRad protein assay (BioRad). For immunoblotting, lysates were heated at 99 °C in 2× Laemmli sample buffer for 7 min. For immunoprecipitation, lysates containing 1 mg protein were incubated with anti-EGFR antibody 199.12 (cat. no. MA5-13319, Invitrogen, Rockford, IL) and Protein A/G+ agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4 °C with tumbling. Immune complexes were collected by centrifugation and washed five times in cold lysis buffer, resuspended in 2× Laemmli sample buffer, and heated at 99 °C for 7 min. The proteins were resolved by SDS-PAGE, transferred to nitrocellulose membranes (Protran BA85, Whatman, Sanford, MA) or nitrocellulose iBlot2 membranes (Thermo Fisher Scientific), and incubated with primary antibody overnight at 4 °C. HRP-linked secondary antibodies donkey anti-rabbit IgG (cat. no. NA934V, GE Healthcare, Piscataway, NJ), HRP-linked donkey anti-mouse IgG (cat. no. NA931, GE Healthcare), goat anti-rabbit IgG HRP conjugate (cat. no. 1721019, BioRad), or goat anti-mouse IgG HRP conjugate (cat. no. 1721011, BioRad) were used with ECL SuperSignal substrates (Pierce Biotechnology, Rockford, IL) to visualize target proteins, using either a film processor or Li-Cor Odyssey Imager. Alternatively, the KwikQuant Western Blot Detection Kit (cat. no. R1004) and KwickQuant Imager (cat. no. D1001) from Kindle Biosciences (Greenwich, CT) were used to detect proteins on the nitrocellulose membrane. Each experiment was repeated at least three times. Densitometric analysis of immunoblot band intensities was performed using Adobe Photoshop software (version CC 2017, Adobe Systems, San Jose, CA) or Image Studio Software (Li-Cor Biotechnology).

Reagents and Chemicals

These are comprehensively described in the Supplemental Material.

Results

E3-Specific Product Detection as Proof of Principle toward High-Throughput Assay Development

Ubiquitination is an iterative cycle of isopeptide bond formation; E1 covalently accepts the ubiquitin before covalently transferring it to E2, RF E3s then orient the E2–ubiquitin complex to facilitate ubiquitin transfer to the target, and the process begins again with either a new non-ubiquitinated target or a mono-ubiquitinated target from the first cycle.¹ As such, this process generates a distribution of ubiquitinated products. Each step of this process can be visualized by low-throughput techniques, like Western blots, or by using chemically tagged ubiquitin (e.g., biotinylation) and simply monitoring molecular weight changes as ubiquitination cycles proceed.¹⁵ To establish an E3-specific assay, however, we needed to develop a way to capture only those products that are directly derived from the activity of the CBLB ligase, not from upstream E1 and E2 activity. To do this, we took advantage of two features of CBLB biochemistry: CBLB autoubiquitination and the ability of the UBA domain of CBLB to specifically bind polyubiquitinated products. CBLB autoubiquitination is believed to be a means of negative feedback to regulate the intracellular activity of CBLB.^6,12,16 Incorporating this feature of CBLB into the assay allows us to reduce the total number of proteins in the assay while maintaining an appreciable signal-to-noise ratio. More importantly, we have previously shown that the UBA domain of CBLB (UBAb) preferentially binds to polyubiquitin chains and only weakly to monoubiquitin.^12,16 In contrast, the UBA domain of CBL (UBAc) does not bind to either polyubiquitin chains or monoubiquitin.

As shown in Figure 1A , by precoating a high-protein-binding 96-well plate with recombinantly expressed CBLB UBA domain protein, we have developed a capture plate capable of enriching polyubiquitinated proteins over monoubiquitinated ones. Plates coated with the UBAb bound tetraubiquitin but not monoubiquitin (Fig. 1A and 1B). Plates coated with the UBAc or left uncoated did not bind either mono- or tetraubiquitin. Importantly, there was a linear concentration–dependent increase in the signal with increasing amounts of tetraubiquitin ( Fig. 1B ).

Figure 1.

Recombinantly expressed ubiquitin-associated (UBA) domain of Casitas B-lineage lymphoma proto-oncogene B (CBLB) specifically captures polyubiquitin chains (TetraUb) over monoubiquitin (MonoUb). (A) 96-well enzyme-linked immunosorbent assay (ELISA) plates were coated with the UBA domain of Cbl-b (UBAb) or UBA domain of Cbl (UBAc), or were left uncoated. The plates were incubated with biotinylated mono- or tetraubiquitin, as indicated. (B) UBAb demonstrates a ubiquitin concentration–dependent increase in signal. Binding of the ubiquitin was detected using horseradish peroxidase (HRP)-conjugated streptavidin. Data represent the mean +/− standard deviation (SD) for three replicates for each point.

After establishing a possible means for capturing polyubiquitin chains from a complex mixture, we then set about applying this technique to an IVUA in a 384-well format ( Fig. 2 ). Since full-length CBLB (FL-CBLB) contains its own UBA domains, it would be possible for FL-CBLB, in solution, to capture polyubiquitinated reaction products and thus compete with the plate-bound UBA for product binding and enrichment. To eliminate this possible complication, we used the previously published enzymatically active CBLBN1/2 truncated construct as the source of our CBLB ubiquitin ligase activity.^6,14 The specificity of this approach is demonstrated in Figure 2A , where we have shown that in the presence of the catalytically active CBLB (RF^WT), there is an appreciable increase in signal relative to background. When E2 is present in the absence of the active CBLB E3 (condition 2) or in the presence of a catalytically inactive CBLB (RF^mut condition 6), there is a weak signal that is statistically increased over E1 alone. This is consistent with reports that E2s, including Ube2d2 used here, can form ubiquitin chains on their active-site cysteine in the absence of an E3.^19,20 Orthogonal validation of the UBA polyubiquitin capture approach is shown in Figure 2B , where IVUA reactions identical to those in Figure 2A were carried out, but rather than capturing the reaction products on the UBA-coated plate, they were used to generate a Western blot, which was probed for the presence and molecular weight distribution of biotinylated ubiquitin. Under the reducing conditions of the Western blot, the covalent bond between an E2 active-site cysteine and the carboxy terminus of ubiquitin is lost, accounting for the absence of signal in lanes 2 and 6 (equivalent to conditions 2 and 6 in Fig. 2A ).^19,20 In contrast, the amide bond formed following CBLB autoubiquitination (ubiquitin carboxy terminus to a CBLB lysine ε-amino group) is resistant to denaturation during Western blot execution and is preserved as an indication of CBLB-dependent ubiquitination. As anticipated, lane 4 of Figure 2B is the only lane in which there are detectable polyubiquitinated products, confirming that the signal detected in the plate-based capture assay is indeed due to capture of poly-autoubiquitinated CBLB-dependent reaction products.

Figure 2.

In vitro plate-based E3 assay specifically detects Casitas B-lineage lymphoma proto-oncogene B (CBLB) E3 autoubiquitination and can be adapted for detection of other E3 ligase products. (A) An in vitro ubiquitination assay (IVUA) was monitored using an active wild-type (WT) (RF^WT) or a catalytically inactive (RF^mut) form of the CBLB E3 enzyme. Only in the presence of ubiquitination-competent conditions (column 4) is an E3-dependent ubiquitination signal detected. (B) An identical assay to (A) was performed and visualized via Western blot; only in lane 4 is ubiquitination activity observed using an antibiotin antibody. All assays included recombinant E1 and adenosine triphosphate (ATP). (C) The E3 plate assay was performed in the presence or absence of 5 µM of the E1 inhibitor (TAK243), as indicated. The data in the plate-based assays represent the mean ± standard deviation (SD) for three replicates. The p values were calculated using a two-tailed Student’s t test. * ≤ 0.05; ** ≤ 0.005; *** ≤ 0.0005; **** ≤ 0.00005. Plate assays were performed as in (A) with varying concentrations of (D) Cbl-b, (E) Cbl, and (F) HMG-CoA reductase degradation (HRD). The data in the plate-based assays represent the mean for three replicates.

Sensitivity to Ubiquitination Inhibition and CBLB Specificity Controls

To ensure that the assay we are using is aligned with the central objective of inhibitor discovery, it was necessary to demonstrate that we could detect the presence of a ubiquitination cascade inhibitor. Using the well-described E1 inhibitor TAK243, Figure 2C clearly demonstrates that we are able to detect inhibition of the ubiquitination cascade.²¹ The E3-dependent signal is significantly reduced in the presence of the E1 inhibitor ( Fig. 2C , conditions 6 vs. 7). Also, the activity observed in the presence of E1 and E2 without E3 was reduced by the E1 inhibitor ( Fig. 2C , conditions 4 and 5), consistent with the interpretation that E1+E2 can form detectable ubiquitin chains. E1 alone gave a detectable signal over background ( Fig. 2C , conditions 1 and 2), which was also reduced in the presence of the E1 inhibitor ( Fig. 2C , conditions 2 and 3). To our knowledge, formation of ubiquitin chains by E1 alone has not been reported, and this observed signal may represent the low background affinity of UBA domains for the capture of monoubiquitinated E1. Although our goal is the discovery of an E3-specific inhibitor, this platform is capable of detecting inhibition upstream of the E3 ( Fig. 2C , conditions 4 vs. 5). This allows us to not only discover a wider array of potential inhibitor classes and chemotypes, but also rapidly triage verified primary-screening leads by likely target (discussed in detail below).

In addition to being able to determine where along the ubiquitination cascade inhibition may occur [either by plate-based methods ( Fig. 2C ) or by Western blot ( Fig. 2B )], we also designed secondary and tertiary assays to assess whether discovered inhibitors are specific for CBLB E3 activity or whether they can inhibit other members of the RF E3 ligase family.^4,13,22 As shown in Figure 2D–2F, using our plate-based assay, it is possible to replace the CBLB ( Fig. 2D ) E3 with other E3s such as CBL ( Fig. 2E ) or HRD1 (an unrelated RF E3; Fig. 2F ), and then monitor concentration-dependent enzymatic activity. This flexibility in our platform allows us to rapidly assess E3 specificity, allowing for specificity-based hit triage. In addition, the generalizability of our assay to other E3s allows future screens for other translationally relevant E3 targets.

HTS Optimization

After establishing that our UBA capture strategy was adaptable to a 384-well format, it was then necessary to fully optimize the assay for cost-effective reagent use and automated liquid handling. The assay relies on the UBA domain–dependent capture of CBLB polyubiquitinated reaction products, and as such, a titration response curve was experimentally generated to establish the relationship between the observed signal and the concentration of UBA domain used to precoat assay plates ( Suppl. Fig. S1A ). Following this step, components of the IVUA were varied to determine their appropriate assay concentration (Suppl. Fig. S1B–1D). Following reagent concentration optimization, assay linearity studies were undertaken ( Suppl. Fig. S2 ). Linearity throughout the assay activity window is important because rapid substrate depletion can bias an assay toward discovery of only those modulators capable of rapid and complete enzymatic inhibition prior to significant enzymatic turnover.²³ In our experience, small molecules in this class are often covalent inactivators that may lack target specificity. The final optimized assay maintains excellent linearity, reagent stability, and signal-to-background ratio throughout our chosen endpoint window (60 min) ( Suppl. Fig. S2 ).

The drug discovery literature is filled with reports of compounds whose discovery initially appeared to be target specific yet on further experimentation proved to be active in a variety of assays of widely divergent formats.^24,25 These compounds have come to be colloquially known as pan-assay interfering substances (PAINS).²⁴ While there is no universal way to eliminate their discovery during screening, there are several ways to diminish the likelihood of their appearance. A common mechanism of action for PAINS is nonspecific protein binding, often resulting in localized denaturation and thus apparent enzyme inhibition.²⁶ One way to reduce the appearance of this behavior is through the inclusion of an excess amount of excipient protein to essentially act as a “molecular sponge,” thus reducing nonspecific denaturation of the targeted enzyme.²⁵ Because different assays have different tolerances for both the concentration of excipient protein and its identity, we tested the inclusion of both casein and bovine gelatin for the effect on the assay over a range of concentrations ( Suppl. Fig. S3 ) before ultimately choosing to use gelatin, largely for its better solubility at physiologic pH. In addition, we determined the effect of nonionic detergents (which can reduce nonspecific compound-induced aggregation) on the assay to further increase assay stringency and robustness ( Suppl. Fig. S3 ). A final important consideration is the effect of DMSO on the assay signal. As shown in Supplemental Figure S3 , our assay has excellent DMSO tolerance, which is essential because DMSO is the storage solvent for most chemical libraries.

Our IVUA can be thought of as a modified sandwich ELISA (enzyme-linked immunosorbent assay) in that the UBA domain–coated plates capture polyubiquitinated products that contain modified biotinylated ubiquitin. These captured products are then probed for the presence of biotin by using a HRP-linked avidin (HRP-avidin) molecule. This provides two additional vectors on which the assay must be optimized, biotinylated ubiquitin concentration as well as the HRP-avidin concentration. Both of these variables were examined in a concentration-dependent manner ( Suppl. Fig. S4 ) before arriving at the final screening conditions.

HTS Outcomes

Following complete assay optimization, we executed an HTS of the Molecular Targets Program [Center for Cancer Research, National Cancer Institute (NCI), Bethesda, MD] pure compound library. This screen encompassed 71,201 discrete substances distributed among 236 384-well plates. A common HTS measure of robustness is the Z-factor, a weighted ratio of the difference in standard deviations to the difference between average readings for high and low controls, with a theoretical maximum value of 1 and a minimum acceptable threshold of 0.5.²⁷ Using this metric, our assay performed well, with an average Z-factor of 0.73 and only six of 236 plates failing to meet the threshold cutoff of 0.5 ( Suppl. Fig. S5A ). Failing plates were rescreened and subsequently had acceptable Z-factor readings. In addition, the average reading among all substance wells was 92% of vehicle controls, and 71% of all screened substances were within one standard deviation of the average reading, indicating a narrow, largely symmetrical distribution of substance reads around the vehicle controls (100% activity, no inhibition). Using a “hit cutoff” of percentage of normalized activity < 50%, 418 substances were chosen for quadruplicate reconfirmation screening at the primary-screening concentration (10 µM) from screening stocks, leading to a primary-screening hit rate of 0.6% ( Suppl. Fig. S5B ). Following quadruplicate confirmation and querying the primary-screening hits against a database of previous screens run in our group, 80 compounds remained as reconfirmed (% activity < 50%) and specific (no previously observed activity in other screening assays) leads. The remaining 80 compounds were considered confirmed primary-screening actives and were subjected to secondary and tertiary evaluation.

Primary-Screening Lead Evaluation

The 80 confirmed primary-screening actives were then subjected to 5-point duplicate dose–response screening over a concentration range of 2.5 to 40 µM in the CBLB screening assay. To further ensure that these compounds were CBLB-specific inhibitors, these confirmed screening leads were also screened in tandem against the serine–threonine kinase p38 in a previously published HTS format (data not shown).¹⁷ This tandem screening and availability of resupply further reduced our confirmed primary-screening actives to a pool of eight distinct chemical scaffolds capable of reducing the CBLB-dependent screening signal by at least 50% at 40 µM while not reducing the p38 signal by more than 10% ( Fig. 3 ).

Figure 3.

Verified screening leads. Structures, identifiers, and screening assay IC₅₀ values of lead molecules discovered through high-throughput screening (HTS) for inhibitors of Casitas B-lineage lymphoma proto-oncogene B (CBLB) E3 activity. IC₅₀ curves are provided as Supplemental Figure S6 .

As shown in Figure 3 , these compounds represent several different chemotypes and have a range of measured potencies; complete IC₅₀ curves are provided as Supplemental Figure S6 . Compound 1, NSC638255, contains a quaternary nitrogen, distinguishing it from other verified leads. This structural class has previously been reported in the literature, including a structural analog evaluated in a human clinical trial due to its apparent cancer-specific cytotoxicity.^28,29 Compounds 2 (NSC66175) and 3 (NSC650438) are most similar to one another, broadly belonging to a class of hydroxylated benzyl derivatives. Compound 2 (a cinnamic acid derivative) has been published extensively in a wide array of research contexts, suggesting that it may be nonspecifically bioactive.³⁰ Intriguingly, compound 3 belongs to the chemical family of tyrphostins and is the inactive stereoisomer of the tyrosine kinase inhibitor cis-erbstatin, which, given its well-described behavior in a number of disease models, may bode well for rapid translation.³¹ Similarly, compound 4, NSC401005, is a heretofore inactive structural isomer of the well-described natural product pleurotin(e).³² Compound 5, Immlg-6815, has a single previously ascribed cellular activity, cytotoxicity, but no clearly defined intracellular drug target.³³ Compound 6, NSC106995, is the trihydroxyphenazine compound lomofungin, a well-annotated natural product with cell type–independent cytotoxicity.^34,35 Compound 7, NSC109266, is a naphthalene diol derivative with no published bioactivities, and compound 8 belongs to the benzofuran family of compounds and has been previously characterized as an inhibitor of the anti-apoptotic protein myeloid cell leukemia sequence-1 (MCL1, available via https://pubchem.ncbi.nlm.nih.gov).

While the primary leads do not have novel chemical composition of matter, the IVUA platform that we have developed is unique from other published CBL family–oriented HTS campaigns in that our platform is built on enzymatic activity rather than fluorescence anisotropy changes upon protein–protein interactions.^36,37 In addition, to the best of our knowledge, our campaign is the first published inhibitor screen specifically targeting CBLB. Of the previously published CBL-directed screens, three compounds were identified as disrupting the interaction of CBL and a peptide derived from an E2 binding partner, catapol, methylprotodioscin, and leonuride.³⁷ These three were tested in our IVUA screening platform and were found to be inactive ( Suppl. Fig. S7 ), further highlighting the importance of our assay and the first identified CBLB enzymatic inhibitors.

After obtaining sufficient resupply of verified primary-screening leads, their activity was first assessed in an E3 specificity assay similar to that used to generate Figure 2E and 2F. In the specificity assay, inhibition of the CBLB E3 enzyme is tested simultaneously with both the closely related E3 CBL as well as the unrelated RF E3 HRD1. Testing of freshly sourced primary-screening leads among all three E3 enzymes revealed that only compound 1 demonstrated appreciable CBL family specificity, equivalently inhibiting both CBL and CBLB at 10 µM, while reducing HRD1 activity to a lesser extent ( Fig. 4 ).

Figure 4.

Compound 1 exhibits dose-dependent inhibition and relative specificity for Casitas B-lineage lymphoma (CBL) family members. Compound 1 was tested for E3 inhibition specificity up to 10 µM against three different ubiquitin ligases [two closely related: CBL and Casitas B-lineage lymphoma proto-oncogene B (CBLB); and 1 unrelated: HMG-CoA reductase degradation 1 (HRD1)]. p value calculated using Student’s t test.

Although the lack of specificity for CBLB over CBL is disappointing, it is perhaps not surprising because the CBLBN1/2 enzyme used in our assays shares 76% sequence identity to the homologous portion of the CBL protein and the catalytic RF domains share 95% identity.³⁸ Encouraged by the specificity of compound 1 for the CBL family relative to HRD1, we prioritized our efforts on further characterization of the specificity and potency of compound 1 and related ME analogs.

Although the CBL family E3 specificity assay suggests that compound 1 is functioning at the E3 stage of the ubiquitination cascade (otherwise, all three E3 conditions would be equally inhibited since they share common E1 and E2 elements), it was necessary to definitively demonstrate this selectivity. To determine if compound 1 is interfering with either the E1 ubiquitin activation step or the E2 ubiquitin conjugation step, we carried out IVUAs using only reagents necessary for each step (E1 or E1/E2) and determined what impact, if any, the presence of compound 1 had on these processes. As shown in Figure 5 , compound 1 did not inhibit the monoubiquitination of E1 (as indicated by the molecular weight shift seen in lanes 2–5 compared to lane 1 when ATP and ubiquitin are omitted from the reaction).

Figure 5.

Compound 1 does not inhibit E1 activation of E2 ubiquitin conjugation. Assays were performed in tubes, as in Figure 2E–2F, in the absence of E3. The reactions were run on sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), unreduced without boiling denaturation, and then immunoblotted as indicated. Compound 1 was incubated in the E1/E2 assay in dose response up to 10 µM, and the samples run as in Figure 2E–2F. Molecular weight (MW) in kDa is shown to the left of the panels. Equivalent increases in MW of the E2 ± compound 1 indicate that compound 1 is not inhibiting either E1 ubiquitin activation or E2 ubiquitin conjugation. TAK243 at 1 µM is included as a positive control for inhibition. The arrow indicates the MW shift of monoubiquitinated E2, which is present in lanes 10–13 and is not reduced in the presence of compound 1. The arrowhead (lower panel) indicates a higher-MW ubiquitinated product that is observed only in the presence of the active E1/E2 complex (lanes 10–13), but is not immunoreactive to the anti-E2 primary antibody (middle panel, no arrowhead).

Compound 1 also did not inhibit the monoubiquitination of the E2, as measured by both size shift and ubiquitin immunoblotting when E1 is present ( Fig. 5 , compare lanes 10–13 to lanes 6–9). As a control, the TAK243 E1 inhibitor blocked monoubiquitination of the E1 (shown by a lower size migration) and E2 (shown by the lower size and loss of ubiquitination) ( Fig. 5 , lane 13). This is further evidence for CBL family E3 specificity of this compound.

Although compound 1 did show modest specificity for the CBL family of E3 ligases over HRD1, it unfortunately also demonstrated cytotoxicity toward the HeLa cervical cancer cell line ( Suppl. Table S1 ) over a concentration range that overlaps that of the biochemical IC₅₀ for CBLB inhibition. While CBL and CBLB are expressed in HeLa cells, their activity is not required for viability, suggesting that the cytotoxicity observed is likely not due to CBL family inhibition. This observation is not without precedent, because the NCI Developmental Therapeutics Program (DTP) has previously determined that MEs like compound 1 exhibit cytotoxicity toward cell culture models of glioblastoma multiforme.^28,39 One of the objectives of the DTP (https://dtp.cancer.gov/organization/dscb/) is to serve as a resource for chemical diversity in the chemotherapeutic research community; therefore, we were able to request, receive, and test a number of ME derivatives for increased potency and lower cellular cytotoxicity ( Suppl. Table S1 ). Through the combination of several experiments, we were able to generate a ratio of HeLa cell cytotoxicity (LD₅₀) and plate-based IVUA IC₅₀ determination for CBLB inhibition, which allowed us to uncover a number of MEs that had both greater potency for CBLB inhibition and reduced apparent off-target cytotoxicity in HeLa cells ( Suppl. Table S1 , last column, and Fig. 6 ).

Figure 6.

Methyl ellipticinium (ME) derivatives with improved Casitas B-lineage lymphoma proto-oncogene B (CBLB) potency (IC₅₀) and greatly reduced cytotoxicity (LD₅₀). A higher toxicity–potency ratio indicates higher likelihood for a therapeutic window before off-target specificity is observed.

This toxicity-to-potency ratio provides insight into which compounds may be best suited to further characterization. In addition, several of the tested derivatives also had increased CBLB E3 specificity, using a ratio of the IC₅₀s for HRD1 to CBLB. The ratio for compound 1 was 3.47, while the ratio for some of the tested derivatives exceeded 10 to 1 ( Suppl. Table S1 ). This suggests that certain MEs can improve potency and specificity while also potentially reducing off-target cytotoxicity. To assess the possibility that there was no inhibition of the upstream E1/E2 ubiquitination cascade using compounds 9–12, these were tested in the E1/E2-only gel-based assay, as in Figure 5 . As shown in Supplemental Figure S8 , we did not observe inhibition of ubiquitination upstream of the E3 ligase activity. The three distinct E3s incorporated into our IVUA platform up to this point are all RF ubiquitin ligases.¹ The discovery of MEs as a broadly inhibitory class of the RF family created opportunities for us to verify that our IVUA platform was applicable to the other family of ubiquitin ligases, the HECT (homologous to E6-AP carboxy terminus) family, and to evaluate the effect of MEs on HECT-dependent ubiquitination.⁴⁰ As a proof of principle, we chose to focus on the E6AP ubiquitin ligase. E6AP is itself important in a number of disease states, including both cervical cancer and Angelman syndrome, and has recently been structurally determined by nuclear magnetic resonance (NMR) spectroscopy.⁴¹ To assess the activity of compound 11 on E6AP activity, we replaced both the E2 and the E3 in our assay with those necessary to carry out an IVUA for E6AP activity (UbcH7/UBE2L3 and E6AP/UBE3A, respectively). As shown in Supplemental Figure S9 , compound 11 demonstrated equivalent inhibition of both CBLB (RF) and E6AP (HECT) autoubiquitination. The ability to integrate both RF and HECT ubiquitin ligases into our assay allows for the rapid assessment of target specificity and expands the platform utility across the spectrum of ubiquitin ligase enzymes.

To extend the in vitro observations obtained using our IVUA platform, it was necessary to generate a cell-based model in which to evaluate the activity of our discovered inhibitors against single E3s. Due to the high homology between CBL proteins, it is possible for CBL family members to show substrate redundancy when coexpressed.^4,7,13 For example, in HeLa cells, which normally express CBL and CBLB but not CBLC, we have previously shown that the siRNA-mediated reduction of CBL or CBLB leads to decreased ubiquitination of activated EGFR, and the simultaneous siRNA-mediated knockdown of CBL and CBLB leads to almost a complete loss of ubiquitination of the activated EGFR (Ref. ¹³ and Fig. 7A , lanes 1–4). To generate a stable HeLa cell line that only expresses CBLB, we used CRISPR/Cas9 (clustered regularly interspaced short palindromic repeats/CRISPR-associated protein-9) gene editing to genetically ablate the CBL gene, CBL–CRISPR ( Fig. 7A , lanes 5–8). On EGFR activation in CBL–CRISPR cells, there is robust ubiquitination of the EGFR that is almost completely abrogated by siRNA-mediated loss of CBLB ( Fig. 7A , compare lanes 6 and 8). Thus, in CBL–CRISPR HeLa cells, ubiquitination of the activated EGFR is mediated primarily by CBLB. Using this model, we have demonstrated that several ME derivatives are capable of reducing CBLB-mediated EGFR ubiquitination, as evidenced by the reduction in ubiquitinated EGFR ( Fig. 7B ). Figure 7B (upper) demonstrates a 70% reduction in EGFR ubiquitination in the presence of 5 µM of compound 1, and Figure 7B (lower) demonstrates that 10 µM of compound 11 can reduce CBLB-dependent EGFR ubiquitination to near-background levels. In addition to more complete inhibition of CBLB activity, the ME derivatives, like compound 11, were also less toxic to the cells, allowing for a longer experimental window of sustained CBLB inhibition and further highlighting the potential utility of these compounds.

Figure 7.

Casitas B-lineage lymphoma proto-oncogene B (CBLB) inhibitors demonstrate intracellular activity in a model of CBLB-dependent epidermal growth factor receptor (EGFR) ubiquitination. (A) The loss of both CBLB and Casitas B-lineage lymphoma (Cbl), via either small interfering RNA (siRNA) knockdown (lanes 1–4) or clustered regularly interspaced short palindromic repeats (CRISPR)-mediated gene ablation (lanes 5–8), leads to a nearly complete loss of epidermal growth factor (EGF)-induced EGFR ubiquitination (lanes 4 and 8). CRISPR/Cas9 (CRISPR-associated protein-9) Cbl knockout in HeLa cells (CBL–CRISPR cells) leaves only CBLB in these cells. EGF stimulation of these cells results in ubiquitination of the EGFR (lane 6). siRNA-mediated knockdown of CBLB in CBL–CRISPR cells reduces ubiquitination of the EGFR to background levels (lane 8). (B) Following serum starvation, CBL–CRISPR cells were incubated with or without 100 ng/ml of EGF for 10 min in the presence of compounds as indicated. (Upper) Compound 1 was added at indicated concentrations 18 h prior to addition of EGF. (Lower) Compound 11 was added at 10 µM to the Cbl knockout (KO) HeLa cells for 48 h prior to addition of EGF. TAK243 at 1 µM was included as a positive control.

Discussion

The importance of ubiquitination is apparent when one considers how many disease processes are associated with its dysregulation.^1,7,41–43 Despite this fact, there are few published screens looking for modulators of this process and no clinically approved drugs targeted at these pathways. While there has been clinical success inhibiting proteasomal degradation of ubiquitinated products (i.e., bortezomib and carfilzomib) and new excitement at the prospect of redirecting ubiquitin machinery through the use of proteolysis-targeting chimeras (PROTACs), clinical development of ubiquitination inhibitors has remained largely restricted to inhibitors of either E1 enzymes or the E3 MDM2 (www.clinicaltrials.gov). One challenge to progress in drug discovery and development is that the multienzyme (E1/E2/E3) and multi-cofactor (ATP/magnesium/ubiquitin) ubiquitination process itself is very difficult to incorporate into a robust HTS. To address the challenge of E3-targeted drug discovery, we have established a ubiquitination screening platform that we have demonstrated can be applied to both RF and HECT family ubiquitin ligases.

Using this IVUA platform, we have determined that MEs appear to broadly inhibit E3-dependent ubiquitin ligation. While the exploration of ME analogs does not represent a systematic attempt at a structure–activity relationship study, the data support several conclusions. The structural diversity tested largely consisted of modifications of the substituents of the quaternary nitrogen of the A ring and changing the identity but not the location of the polar substituent in the D ring. Our experimental evidence ( Suppl. Table S1 ) suggests that the quaternary nitrogen is required for activity. We have tested the chemically related compound with an unsubstituted nitrogen, NSC98927, and found it to be significantly less active than the other analogs ( Suppl. Table S1 ). We have also tested a series of unsubstituted carbazoles and found them inactive as well (data not shown). Increasing the bulkiness of the alkyl group at the quaternary nitrogen does not appear to affect either inhibition of or the specificity for the CBLB E3. In terms of the effect of substituting a hydroxyl for a chloride in the D ring, in our limited dataset this change was also accompanied by changes in the quaternary nitrogen substituent that complicate the interpretation of the data. Comparing the observed behavior of NSC311152 versus NSC359449 ( Suppl. Table S1 ), however, suggests that the presence of a hydroxyl substitution does not affect specificity or cytotoxicity because the behavior of these two compounds is virtually identical (overlapping standard errors for each measured parameter). In contrast, when comparing compounds 1, 12, and 20, with chloro, methoxy, and methyl substituents, respectively, in the same location of the D ring, it appears that the larger methoxy substitution reduces the observed cytotoxicity while maintaining potency. This suggests that further chemical elaboration at this position, and more broadly across the molecule, may be warranted and could yield molecules with a better potential therapeutic index. In addition to our work on the ellipticiniums, we have also recently used our IVUA platform to enable the discovery of structurally novel natural products, extending the utility of this platform to a novel area of research, natural products discovery.⁴⁴

As a RF ubiquitin ligase, the biological significance of CBLB enzymatic activity is a rapidly evolving area of research, with the discovery of its role in immunomodulation of the innate immune system published in 2014.^7,8,45–49 This research activity highlights the need for small-molecule modulators of CBLB activity that can be used to better understand both the biology of this target in a research setting as well as those that can be translated into a clinical setting to improve disease outcomes. The experiments described in this article are the first published cell-free screen for modulators of CBLB enzymatic activity. The assay that we have developed offers several potential advantages over previously published ubiquitin ligation screening assays.⁵⁰ First, our assay does not rely on proprietary reagents for either the ubiquitin capture or the signal development, increasing its broad potential use. Second, the use of a non-homogenous assay allows us to build in analyte wash steps prior to signal development, which, depending on the screening library composition, may be important to avoid spectrophotometric interference due to the presence of photometrically active library components. Finally, we have shown that this assay can be readily adapted to other ubiquitin ligases, including both RF and HECT E3s, broadening its appeal to the ubiquitin-related research community. We have used our robust screening assay to identify and characterize several classes of chemical compounds that can be used to advance the understanding of both CBLB activity and ubiquitination more broadly.

Prior to this work, several groups have described inhibitors of CBL or CBLB activity. Two groups have reported peptidomimetic compounds that bind to the TKB domain of either CBL or CBLB, and block interaction with a specific substrate.^{36,48,51–53} These compounds are not catalytic inhibitors of E3 activity but rather allosteric agents that prevent specific substrates from being targeted by CBL or CBLB. Similarly to substrate peptidomimetics, Wu et al. developed modified peptides based on the structure of the CBL RF interface with the E2 UbcH7.^37,49 They showed that intraperitoneal (IP) injection of the peptides could protect mice against high-fat diet-induced obesity and insulin resistance, phenocopying the effects of CBL gene knockouts.⁴⁹ The same group also identified several small molecules that compete for UbcH7 binding to CBL at this site.³⁷ In our autoubiquitination assay, these compounds do not inhibit CBLB E3 activity ( Suppl. Fig. S7 ). UbcH7 has been crystallized in complex with CBL but does not function with CBL proteins in E3 assays, possibly explaining their lack of activity in our hands.^13,54–57 In addition, a pharmaceutical company has described a CBLB inhibitor in a conference abstract in which no chemical or screening assay information was provided.⁴⁵ While the work of other groups reinforces the translational importance of CBL family modulators, to our knowledge the compounds reported here are the first publicly disclosed pharmacophores for CBLB E3 inhibition discovered through HTS. Furthermore, we have demonstrated that some of these compounds not only inhibit CBLB catalytic activity in vitro but also are active against the enzyme within cells. That MEs appear broadly active against both RF and HECT E3s suggests that these compounds are best suited for use as bioprobes in nonclinical settings. With additional chemical optimization and further study, however, it may be possible to advance these pharmacophores for testing in higher-order systems of CBLB dysregulation such as T cells, NK cells, and animals. As importantly, the screening platform we have developed can be used to screen additional chemical libraries in other drug discovery settings, across a range of possible ubiquitination cascades.

Supplemental Material

sj-pdf-1-jbx-10.1177_24725552211000675 – Supplemental material for In Vitro Ubiquitination Platform Identifies Methyl Ellipticiniums as Ubiquitin Ligase Inhibitors

Supplemental material, sj-pdf-1-jbx-10.1177_24725552211000675 for In Vitro Ubiquitination Platform Identifies Methyl Ellipticiniums as Ubiquitin Ligase Inhibitors by Brice A. P. Wilson, Donna Voeller, Emily A. Smith, Antony Wamiru, Gang Liu, Stanley Lipkowitz and Barry R. O’Keefe in SLAS Discovery

Footnotes

Acknowledgements

The authors would like to acknowledge the efforts of Lauren Procter, Morgan Pagonis, and Jane Jones of the Protein Expression Laboratory, Frederick National Laboratory for Cancer Research, for their work scaling up the production of various recombinant proteins.

Supplemental material is available online with this article.

Authors’ Note

The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. government.

Author Contributions

B.A.P.W., B.R.O., and S.L. planned the research objectives; D.V. established pre-HTS assays and executed post-HTS lead evaluation, including all cell-based workups; B.A.P.W. established and optimized HTS parameters; E.A.S. and A.W. executed HTS; E.I.G. performed data analysis, and B.A.P.W., D.V., B.R.O., and S.L. composed the manuscript. All authors reviewed and edited the manuscript. B.A.P.W. and D.V. contributed equally to this manuscript.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research has been supported by the Intramural Research Program of the National Institutes of Health (NIH), National Cancer Institute, Center for Cancer Research. This project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under contract HHSN261200800001E.

ORCID iD

Brice. A. P. Wilson

References

Lipkowitz

Weissman

A. M.

RINGs of Good and Evil: RING Finger Ubiquitin Ligases at the Crossroads of Tumour Suppression and Oncogenesis. Nat. Rev. Cancer 2011, 11, 629–643.

Deshaies

R. J.

Joazeiro

C. A. P.

RING Domain E3 Ubiquitin Ligases. Annu. Rev. Biochem. 2009, 78, 399–434.

Bachmaier

Krawczyk

Kozieradzki

, et al. Negative Regulation of Lymphocyte Activation and Autoimmunity by the Molecular Adaptor Cbl-b. Nature 2000, 403, 211–216.

Kales

S. C.

Ryan

P. E.

Nau

M. M.

, et al. Cbl and Human Myeloid Neoplasms: The Cbl Oncogene Comes of Age. Cancer Res. 2010, 70, 4789–4794.

Peschard

Park

Escape from Cbl-Mediated Downregulation: A Recurrent Theme for Oncogenic Deregulation of Receptor Tyrosine Kinases. Cancer Cell 2003, 3, 519–523.

Ettenberg

S. A.

Magnifico

Cuello

, et al. Cbl-b-Dependent Coordinated Degradation of the Epidermal Growth Factor Receptor Signaling Complex. J. Biol. Chem. 2001, 276, 27677–27684.

Liyasova

M. S.

Lipkowitz

Molecular Pathways: Cbl Proteins in Tumorigenesis and Antitumor Immunity: Opportunities for Cancer Treatment. Clin. Cancer Res. 2015, 21, 1789–1794.

Paolino

Choidas

Wallner

, et al. The E3 Ligase Cbl-b and TAM Receptors Regulate Cancer Metastasis via Natural Killer Cells. Nature 2014, 507, 508–512.

Chiang

J. Y.

Jang

I. K.

Hodes

, et al. Ablation of Cbl-b Provides Protection against Transplanted and Spontaneous Tumors. J. Clin. Invest. 2007, 117, 1029–1036.

10.

Loeser

Loser

Bijker

M. S.

, et al. Spontaneous Tumor Rejection by cbl-b-Deficient CD8+ T Cells. J. Exp. Med. 2007, 204, 879–891.

11.

Paolino

Thien

C. B.

Gruber

, et al. Essential Role of E3 Ubiquitin Ligase Activity in Cbl-b-Regulated T Cell Functions. J. Immunol. 2011, 186, 2138–2147.

12.

Davies

G. C.

Ettenberg

S. A.

Coats

A. O.

, et al. Cbl-b Interacts with Ubiquitinated Proteins; Differential Functions of the UBA Domains of c-Cbl and Cbl-b. Oncogene 2004, 23, 7104–7115.

13.

Liyasova

M. S.

Voeller

, et al. Cbl Interacts with Multiple E2s In Vitro and in Cells. PLoS ONE 2019, 14, e0216967.

14.

Ryan

P. E.

Sivadasan-Nair

Nau

M. M.

, et al. The N Terminus of Cbl-c Regulates Ubiquitin Ligase Activity by Modulating Affinity for the Ubiquitin-Conjugating Enzyme. J. Biol. Chem. 2010, 285, 23687–23698.

15.

Lorick

K. L.

Jensen

J. P.

Fang

, et al. RING Fingers Mediate Ubiquitin-Conjugating Enzyme (E2)-Dependent Ubiquitination. Proc. Natl. Acad. Sci. USA 1999, 96, 11364–11369.

16.

Peschard

Kozlov

Lin

, et al. Structural Basis for Ubiquitin-Mediated Dimerization and Activation of the Ubiquitin Protein Ligase Cbl-b. Mol. Cell 2007, 27, 474–485.

17.

Wilson

B. A. P.

Alam

M. S.

Guszczynski

, et al. Discovery and Characterization of a Biologically Active Non–ATP-Competitive p38 MAP Kinase Inhibitor. J. Biomol. Screen. 2016, 21, 277–289.

18.

Idoko-Akoh

Taylor

Sang

H. M.

, et al. High Fidelity CRISPR/Cas9 Increases Precise Monoallelic and Biallelic Editing Events in Primordial Germ Cells. Sci. Rep. 2018, 8, 15126.

19.

Chen

Pickart

C. M.

A 25-Kilodalton Ubiquitin Carrier Protein (E2) Catalyzes Multi-Ubiquitin Chain Synthesis via Lysine 48 of Ubiquitin. J. Biol. Chem. 1990, 265, 21835–21842.

20.

David

Ziv

Admon

, et al. The E2 Ubiquitin-Conjugating Enzymes Direct Polyubiquitination to Preferred Lysines. J. Biol. Chem. 2010, 285, 8595–8604.

21.

Hyer

M. L.

Milhollen

M. A.

Ciavarri

, et al. A Small-Molecule Inhibitor of the Ubiquitin Activating Enzyme for Cancer Treatment. Nat. Med. 2018, 24, 186–193.

22.

Metzger

M. B.

Hristova

V. A.

Weissman

A. M.

HECT and RING Finger Families of E3 Ubiquitin Ligases at a Glance. J. Cell Sci. 2012, 125, 531.

23.

Copeland

R. A.

Assay Considerations for Compound Library Screening. In Evaluation of Enzyme Inhibitors in Drug Discovery, 2nd ed. Wiley-Interscience: Hoboken (NJ), 2013; pp 123–168.

24.

Baell

J. B.

Nissink

J. W. M.

Seven Year Itch: Pan-Assay Interference Compounds (PAINS) in 2017—Utility and Limitations. ACS Chem. Biol. 2018, 13, 36–44.

25.

Wilson

B. A. P.

Thornburg

C. C.

Henrich

C. J.

, et al. Creating and Screening Natural Product Libraries. Nat. Prod. Rep. 2020, 37, 893–918.

26.

McGovern

S. L.

Caselli

Grigorieff

, et al. A Common Mechanism Underlying Promiscuous Inhibitors from Virtual and High-Throughput Screening. J. Med. Chem. 2002, 45, 1712–1722.

27.

Zhang

J. H.

Chung

T. D.

Oldenburg

K. R.

A Simple Statistical Parameter for Use in Evaluation and Validation of High Throughput Screening Assays. J. Biomol. Screen. 1999, 4, 67–73.

28.

Shoemaker

Balaschak

Alexander

, et al. Therapeutic Activity of 9-Chloro-2-Methylellipticinium Acetate in an Orthotopic Model of Human Brain Cancer. Oncol. Rep. 1995, 2, 663–667.

29.

Khayat

Borel

Azab

, et al. Phase I Study of Datelliptium Chloride, Hydrochloride Given by 24-h Continuous Intravenous Infusion. Cancer Chemother. Pharmacol. 1992, 30, 226–228.

30.

Zhao

Song

Xie

, et al. Research Progress in the Biological Activities of 3,4,5-Trimethoxycinnamic Acid (TMCA) Derivatives. Eur. J. Med. Chem. 2019, 173, 213–227.

31.

Levitzki

Mishani

Tyrphostins and Other Tyrosine Kinase Inhibitors. Annu. Rev. Biochem. 2006, 75, 93–109.

32.

Sznarkowska

Kostecka

Meller

, et al. Inhibition of Cancer Antioxidant Defense by Natural Compounds. Oncotarget 2017, 8, 15996–16016.

33.

Xie

Yan

Zhu

, et al. Benzothiazoles Exhibit Broad-Spectrum Antitumor Activity: Their Potency, Structure–Activity and Structure–Metabolism Relationships. Eur. J. Med. Chem. 2014, 76, 67–78.

34.

Hoskins

J. W.

Ofori

L. O.

Chen

C. Z.

, et al. Lomofungin and Dilomofungin: Inhibitors of MBNL1-CUG RNA Binding with Distinct Cellular Effects. Nucleic Acids Res. 2014, 42, 6591–6602.

35.

Zhang

Sheng

Wang

, et al. Identification of the Lomofungin Biosynthesis Gene Cluster and Associated Flavin-Dependent Monooxygenase Gene in Streptomyces lomondensis S015. PLoS ONE 2015, 10, e0136228.

36.

Kumar

E. A.

Charvet

C. D.

Lokesh

G. L.

, et al. High-Throughput Fluorescence Polarization Assay to Identify Inhibitors of Cbl(TKB)–Protein Tyrosine Kinase Interactions. Anal. Biochem. 2011, 411, 254–260.

37.

Xie

Sun

Pessetto

Z. Y.

, et al. Development of a Fluorescence Polarization Based High-Throughput Assay to Identify Casitas B-Lineage Lymphoma RING Domain Regulators. PLoS ONE 2013, 8, e78042.

38.

Keane

M. M.

Rivero-Lezcano

O. M.

Mitchell

J. A.

, et al. Cloning and Characterization of cbl-b: A SH3 Binding Protein with Homology to the c-cbl Proto-Oncogene. Oncogene 1995, 10, 2367–2377.

39.

Acton

E. M.

Narayanan

V. L.

Risbood

P. A.

, et al. Anticancer Specificity of Some Ellipticinium Salts against Human Brain Tumors In Vitro. J. Med. Chem. 1994, 37, 2185–2189.

40.

Wang

Argiles-Castillo

Kane

E. I.

, et al. HECT E3 Ubiquitin Ligases—Emerging Insights into Their Biological Roles and Disease Relevance. J. Cell Sci. 2020, 133, jcs228072.

41.

Buel

G. R.

Chen

Chari

, et al. Structure of E3 Ligase E6AP with a Proteasome-Binding Site Provided by Substrate Receptor hRpn10. Nat. Commun. 2020, 11, 1291.

42.

Sasiela

C. A.

Stewart

D. H.

Kitagaki

, et al. Identification of Inhibitors for MDM2 Ubiquitin Ligase Activity from Natural Product Extracts by a Novel High-Throughput Electrochemiluminescent Screen. J. Biomol. Screen. 2008, 13, 229–237.

43.

Weissman

A. M.

Shabek

Ciechanover

The Predator Becomes the Prey: Regulating the Ubiquitin System by Ubiquitylation and Degradation. Nat. Rev. Mol. Cell Biol. 2011, 12, 605–620.

44.

Kim

C. K.

Wang

Wilson

B. A. P.

, et al. Suberitamides A-C, Aryl Alkaloids from a Pseudosuberites Sp. Marine Sponge That Inhibit Cbl-b Ubiquitin Ligase Activity. Marine Drugs 2020, 18, 536.

45.

Agarwal

Riling

, et al. Abstract 2228: Cbl-b Inhibitors as Novel Intra-Cellular Checkpoint Inhibitors for Cancer Immunotherapy. Cancer Res. 2016, 76, 2228.

46.

Gosling

Rountree

Taherbhoy

, et al. Abstract 2696: Genetic and Pharmacologic Evaluation of the Ubiquitin Ligase CBL-B as a Small-Molecule, Tumor Immunotherapy Target. Cancer Res. 2019, 79, 2696.

47.

Liyasova

M. S.

Lipkowitz

Abstract 3031: Bufalin Regulates Cbl-b Levels in HeLa Cells through Protein Synthesis Inhibition. Cancer Res. 2016, 76, 3031.

48.

Ohno

Ochi

Maita

, et al. Structural Analysis of the TKB Domain of Ubiquitin Ligase Cbl-b Complexed with Its Small Inhibitory Peptide, Cblin. Arch. Biochem. Biophys. 2016, 594, 1–7.

49.

Sun

Pessetto

Z. Y.

, et al. Casitas B-Lineage Lymphoma RING Domain Inhibitors Protect Mice against High-Fat Diet-Induced Obesity and Insulin Resistance. PLoS ONE 2015, 10, e0135916.

50.

Schneider

Chen

Tang

, et al. Development of a Homogeneous AlphaLISA Ubiquitination Assay Using Ubiquitin Binding Matrices as Universal Components for the Detection of Ubiquitinated Proteins. Biochim. Biophys. Acta—Mol. Cell Res. 2012, 1823, 2038–2045.

51.

Kumar

E. A.

Yuan

Palermo

N. Y.

, et al. Peptide Truncation Leads to a Twist and an Unusual Increase in Affinity for Casitas B-Lineage Lymphoma Tyrosine Kinase Binding Domain. J. Med. Chem. 2012, 55, 3583–3587.

52.

Kawai

Hirasaka

Maeda

, et al. Prevention of Skeletal Muscle Atrophy In Vitro Using Anti-Ubiquitination Oligopeptide Carried by Atelocollagen. Biochim. Biophys. Acta 2015, 1853, 873–880.

53.

Ochi

Abe

Nakao

, et al. N-Myristoylated Ubiquitin Ligase Cbl-b Inhibitor Prevents on Glucocorticoid-Induced Atrophy in Mouse Skeletal Muscle. Arch. Biochem. Biophys. 2015, 570, 23–31.

54.

Zheng

Wang

Jeffrey

P. D.

, et al. Structure of a c-Cbl-UbcH7 Complex: RING Domain Function in Ubiquitin-Protein Ligases. Cell 2000, 102, 533–539.

55.

Markson

Kiel

Hyde

, et al. Analysis of the Human E2 Ubiquitin Conjugating Enzyme Protein Interaction Network. Genome Res. 2009, 19, 1905–1911.

56.

Wenzel

D. M.

Lissounov

Brzovic

P. S.

, et al. UBCH7 Reactivity Profile Reveals Parkin and HHARI to Be RING/HECT Hybrids. Nature 2011, 474, 105–108.

57.

Huang

de Jong

R. N.

Wienk

, et al. E2-c-Cbl Recognition Is Necessary but Not Sufficient for Ubiquitination Activity. J. Mol. Biol. 2009, 385, 507–519.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.42 MB