Development of a Fluorescence Polarization Based High-Throughput Assay to Identify Casitas B-Lineage Lymphoma RING Domain Regulators

The E3 ubiquitin protein ligase Casitas B-lineage Lymphoma (Cbl) proteins and their binding partners play an important role in regulating signal transduction pathways. It is important to utilize regulators to study the protein-protein interactions (PPIs) between these proteins. However, finding specific small-molecule regulators of PPIs remains a significant challenge due to the fact that the interfaces involved in PPIs are not well suited for effective small molecule binding. We report the development of a competitive, homogeneous, high-throughput fluorescence polarization (FP) assay to identify small molecule regulators of Cbl (RING) domain. The FP assay was used to measure binding affinities and inhibition constants of UbCH7 peptides and small molecule regulators of Cbl (RING) domains, respectively. In order to rule out promiscuous, aggregation-based inhibition, two assay conditions were developed and compared side by side. Under optimized conditions, we screened a 10,000 natural compound library in detergent-free and detergent-present (0.01% Triton X-100) systems. The results indicate that the detergent-present system is more suitable for high-throughput screens. Three potential compounds, methylprotodioscin, leonuride and catalpol, have been identified that bind to Cbl (RING) domain and interfere with the Cbl (RING)-UbCH7 protein-protein interaction.


Introduction
The casitas B-lineage Lymphoma (Cbl) proteins have created much interest in recent years. So far seven members of Cbl-family proteins have been characterized including three mammalian isoforms (c-Cbl, Cbl-b, Cbl-c/Cbl-3), two isoforms in Drosophila melanogaster (Fly Cbl, short and long), Caenorhabditis elegans (SLI-1), and chick Cbl [1,2]. In addition, there are four known oncogenic Cbl proteins including v-Cbl, 70Z-Cbl, DY368-Cbl and DY371-Cbl [1] (Fig. S1). The oncogenic protein v-Cbl, the first member of the Cbl-family proteins, was cloned from the Cas NS-1 retrovirus in 1989 [3]. Later, it was subsequently discovered that v-Cbl is a truncated form of a larger cellular homologue known as c-Cbl [4]. c-Cbl is a ubiquitously expressed, primarily cytosolic protein which has 906 and 913 amino acids in humans and mice, respectively [5]. The full length of c-Cbl consists of an N-terminal tyrosine-kinase-binding (TKB) domain, a RING Finger motif, a proline-rich region and a C-terminal ubiquitin-associated (UBA) domain that overlaps with a LEUCINE ZIPPER (LZ) motif [1].
By virtue of its E3 ubiquitin ligase activity, c-Cbl is widely known as a negative regulator of growth factor action. Signaling proteins known to be targeted for ubiquitination by c-Cbl proteins include EGF receptor, PDGF receptor, Macrophage Colony-Stimulating Factor (M-CSF) receptor, Neu receptor, Epidermal Growth Factor-2 (ErbB-2) receptor, c-Src, c-Cbl, and T-cell and B-cell antigen receptor [6,7]. Inhibition of this process results in enhanced signal transduction [1]. In addition to the E3 ubiquitin ligase activity, c-Cbl has been reported to act as an adaptor protein in a range of intracellular signaling cascades [1,8].
The Cbl family of ubiquitin ligases, along with a set of phosphotyrosine-binding adaptors, integrates receptor endocytosis into the densely wired networks of signal transduction, which plays a vital role in health and disease. However, the mechanism is not well understood due to limited techniques to dissect Cbl protein functions. Here we designed a competitive, homogeneous, highthroughput FP assay to identify small molecule regulators that binds to Cbl (RING) domain.

Peptides and Proteins
Peptides ( Table 1) were purchased from Pharmanic (Chengu, Sihucna, China) with a purity $98%. The peptide concentrations were determined by amino acid analysis at the protein core facility (Chengdu Medical College). The recombinant Cbl protein (Fig.  S2) was produced by the protein core at Chengdu Medical College by the PCR technique as described by Yokouchi et al [9]. c-Cbl (RING) protein including the linker and RING domain (codon 358-457) was tagged with glutathione S-transferase (GST) and purified with GSH-Sepharose (Shanghai Biopharma).

Compound Library
The natural product library contains 10,000 natural products with a minimum of 98% purity confirmed by NMR and HPLC (Pharmanic, Chengdu, Sichuan, China). Compounds are selected according to Pharmacopoeia of the People's Republic of China. All compounds are approved by China Food and Drug Administration as Chinese Traditional Medicine. Compounds were present at 10 mmol/L in DMSO. Compounds were seeded from A2 to H11 in 96-well plates.

Reagents
All the chemicals are purchased from Kepeng Ltd (Beijing, China) except specific indication.

FP Assay Development
Assays were carried out in detergent-free buffer system (PBS, pH = 7.4) or detergent-present buffer system (0.01% Triton X-100 in PBS, pH = 7.4). All FP measurements were made on 384-well, low-volume, black, v-bottom polystyrene microplates (Corning, Shanghai, China) using the ZS-2 plate reader. The binding affinities of fluorescein-labeled peptides/regulators binding to c-Cbl (RING) protein were verified by titrating an increasing concentration of protein into a constant concentration of labeled peptides. The polarization values were measured at an excitation wavelength of 485 nm and an emission wavelength of 538 nm. The total fluorescence (TF) values were also measured. The change in polarization was graphed as a function of the log of the protein concentration, and the dissociation constant (K d ) and the change in mP (DmP) were obtained from the resulting sigmoidal curve. The concentrations of fluorescein-labeled peptide and proteins were optimized.

FP Assay Stability Assessment
Unlabeled short peptide was titrated against the FITC-labeled peptide (100 nmol/L) and protein (3 mmol/L) mixture. Plates were incubated at room temperature for up to 3 hours or 30 min at different temperatures. FP values were taken and IC 50 values, concentrations giving 50% inhibition of protein-protein interaction with respect to control, and dissociation constant (K d ) values were calculated.

DMSO Tolerance Assay
Increasing concentrations of DMSO at 1-20% of the assay volume (20 ml) were added to the FITC-labeled peptide and protein mixture. FP and TF measurements were taken at room temperature (25uC) after 30 min incubation.

FP Based Screen
Master mixtures containing fluorescein-labeled probe and Cbl (RING) protein Cbl (RING) protein, 3 mmol/L and probe 1 at 100 nmol/L) in a reaction volume of 19 ml were prepared and loaded to plates. The library compounds (10 mmol/L -0.078 mmol/L, half dilutions) were prepared separately and transferred into the assay plates. The first and last columns of each plate were used as positive and negative control wells. Wells containing the drug only were used as controls to detect autofluorescence of the compounds in the total fluorescence assay. After subtraction of the background signal, data was processed and the IC 50 values were determined by non-linear least square fitting (four parameters) using SigmaPlot 11.0.

Hits Validation
Hits were validated using the primary screening assay conditions in triplicate.

Statistical Analysis
Data were reported as mean 6 SD. Each experiment was performed 3-5 times, each treatment performed in duplicate or triplicate.

Screen Assay Development
We hypothesized that unlabeled or fluorescein-labeled peptides (Peptide 1, Probe 1 and 2, Table 1) derived from UbCH7 structure bind to the Cbl (RING) domain and can be used to identify compounds/chemical regulators that competitively bind to the Cbl (RING) domain.
The FP assay utilizing fluorescein-labeled peptides binding to the Cbl (RING) domain was developed and optimized in a 384well format. Two probes were designed to determine the effect of FITC on the assay. Binding experiments were performed and K d values were calculated. Titration of c-Cbl (RING) protein with these probes causes the FP and the fluorescence anisotropy (FA) to increase 4 to 6-fold. Binding affinities to probe 1 or probe 2 were evaluated ( Fig. 1C&D). Probe 2 binds to c-Cbl (RING) protein with a higher affinity (K d = 0.1460.08 mmol/L, Fig. 1D) than probe 1 (K d = 1.6760.09 mmol/L, Fig. 1C). To study competitive inhibition using the FP binding assay, the concentration ratio between the receptors and the K d of the labeled probes should be at least 1 in order to generate IC 50 curve [10]. Therefore, probe 1 was chosen for the following validation and screen assays. In addition, the receptor concentration needed for the FP assay should be $ K d and should have a reasonable DmP. The DmP value was about 100 units when 3 mmol/L c-Cbl (RING) protein was titrated. Therefore, 3 mmol/L was used as the c-Cbl (RING) protein concentration for the following screen.
To determine whether probe 1 binds to the active site of c-Cbl(RING) protein and to choose a positive control, competition assays were carried out using peptide 1 (Ac-PFKPP-NH 2 ) or mutant peptide 2 (Ac-PAKPP-NH 2 ) (Fig. 2). We predicted that peptide 1 will be able to replace probe 1 and peptide 2, which lacks the active phenylalanine residue, will not be able to do so. The addition of peptide 1 was shown to cause the return of the FP values to the level of free probes ( Fig. 2A) which indicates the ability of peptide 1 to replace probe 1. The IC50 value for probe 1 binding to c-Cbl(RING) protein in detergent-free system is 1.0560.09 mmol/L. However, the addition of mutant peptide 2 does not alter the FP values, which is consistent with the prediction ( Fig. 2A). To determine whether adding detergent to the assay buffer would change the dynamic stability, assays conducted in Figure 2A have been repeated under same conditions except the buffer contains 0.01% Triton X-100. The IC 50 value for probe 1 binding to c-Cbl(RING) protein in the presence of detergent is 1.0660.09 mmol/L (Fig. 2B) which is comparable to the IC 50 value in detergent-free system. Probe 2 does not bind to c-Cbl(RING) protein under detergent-present condition (Fig. 2B).
To optimize the FP assay properly, it was necessary to assess the stability of the assay. The conditions for the FP assay were established by testing various parameters including incubation temperature, time, TF values, DmP values, DMSO tolerance, and buffer components. The K d values increased and DmP values decreased significantly when the plates were heated over 35uC, indicating the probe-protein system started to dissociate (Fig. 2C  & D). Our data show that at room temperature (25uC) the systems are stable up to 3 h (Fig. 2E). Therefore, the incubation time was optimized to 30 min for rapid screening. The DMSO tolerance for the FP assay was also tested, and stable FP values (Fig. 2F) and TF values (data not shown) were observed at up to 15% DMSO in PBS buffer (v/v). In summary, the competitive FP binding assay conditions ideal for identifying c-Cbl (RING) domain regulator are: (1) 3 mmol/L c-Cbl (RING) protein, (2) 100 nmol/L probe 1, (3) 5% final DMSO concentration (v/v), (4) incubation for 30 min at 25uC before reading the plate, (5) with or without 0.01% Triton X-100.

Pilot Screens and Assay Performance Measurements
Pilot screens against the natural product library of 10,000 compounds described above was performed at screening concentrations of 10 mmol/L to 0.078 mmol/L (half serial dilutions) in 5% DMSO (v/v). The screens were completed in the detergentfree assay first and the detergent-present assay later. Controls present in each assay plate consisted of 5% DMSO (negative control) or peptide 1 (positive control). The assay was semiautomated and the steps are listed in Table 2. The performance of the FP assays and post-screen analysis were summarized in Table 3. Assay performance was assessed using Z' factor generated from whole DMSO plates.   Table 4) were identified as hit compounds in the detergent-free system. There are only 12 compounds showed inhibitory activity at the concentration of 10 mmol/L. Three candidates, methylprotodioscin, leonuride, and Catalpol, were identified as hit compounds in the detergent-present system. All compounds have been tested in more extensive completive binding assays. Two of the compounds from the detergent-free screen, tehmaglutin and quercetin, were verified as false positive hits possibly due to the aggregation ( Table 4).

Discussion
Design and development of c-Cbl regulators is an active field of study because c-Cbl is associated with multiple diseases including obesity, diabetes and cancer [8,[11][12][13]. There are reports of assays being developed to screen the c-Cbl TKB domain indictors [14,15]. Studies show that the majority of the binding energy of protein-protein interactions is contributed by a few amino acids at the interface [16]. Therefore, molecules derived from natural or bioactive peptides produced by plants, animals or humans are good candidates for screening. The potency of these peptides can be increased by enhancing hydrophobic, electrostatic and hydrogen-bonding interactions between the peptide and protein target. Crystallization studies show that c-Cbl binds to UbcH7 with its RING domain and its linker helix [17]. The RING domain provides a shallow groove which interacts with the L1 and L2 loops of UbcH7. This c-Cbl groove provides contacts for UbcH7 which include Phe-63 of the UbcH7 L1 loop as well as Pro-97 and Ala-98 of the UbcH7 L2 loop [17]. The Phe-63 (blue) makes multiple van der Waals contacts with Ile-383 on the c-Cbl RING domain. It also makes contacts with Trp-408, Cys-404, Ser-407, and Ser-411 from the a helix of the c-Cbl RING domain (Fig. 1A). The Pro-97 colored in blue of UbcH7 L2 loop makes Table 2. Screen assay protocol.
Step Parameter Value Description  Table 3. FP assay performance and post-screen analysis summary.

Category Parameters Descriptions
Nature of the assay Cell-free multicomponent competitive assay Assay Assay strategy Detection of c-Cbl (RING) domain regulator using fluorescence polarization assay

Reagents and sources See materials and Methods
Assay protocol Key steps outlined in Table 2 Nature of the library Fractions derived from natural Chinese medicine Size of the library 10,000 natural compounds arrayed in 96-well plates as single compounds at 10 mM in DMSO contact with Ile-383, Trp-408, and Pro-417 of c-Cbl. The carbonyl group of Ala-98 of UbcH7 L2 loop makes a hydrogen bond with amide group on Pro-417 of c-Cbl which is another potential binding interface for chemical regulators (Fig. 1B).
Based on these studies, fluorescein isothiocyanate (FITC) was used to label peptides derived from UbcH7 L1 loop or L2 loop structure which mimic the binding sequence ( Table 1).
FP is a solution-based, homogeneous technique requiring no immobilization or separation of reaction components and therefore is suitable for HTS assays. The principle of the screen is to identify the compounds that displace probe 1 from the c-Cbl (RING) protein by detecting the resulting decrease in FP. The fluorescent labeled peptide (probe 1) bound to c-Cbl (RING) domain with a sub-micromolar dissociation constant was inhibited by peptide 1 (Fig. 2A &2B). Therefore, probe 1 can be used as the fluorescent labeled probe for the FP assay.
It is widely accepted that aggregation, autofluorescence, and reactivity of artifacts in a HTS for inhibitors are potential causes for false positives [18][19][20]. It is also recognized that a substantial fraction of small molecules or natural compounds exhibit aggregating behavior leading to false positive results in screening assays [18]. This nonspecific attachment to target protein could be dramatically reduced by adding small amount of detergent to the assay buffer [19]. Therefore, we compared two screen conditions: detergent-free and detergent-present (0.01% Triton X-100) to assess the best conditions for HTS. The assay stabilities including temperature, time, and DMSO tolerance were tested and compared. Both detergent-free and detergent-present systems are stable under the indicated conditions (Fig. 2C, 2D, 2E & 2F).
To determine which condition is suitable for HTS, two pilot screens have been done and compared side by side. There are 422 compounds showed activity at the concentration of 10 mmol/L in the detergent-free screen which is an unexpected high hit ratio. One of the hit compounds identified from the detergent-free screen, quercetin, has been reported multiple times as a false positive hit [19,20]. There are only 3 compounds showed activities in the detergent-present system. Tehnaglutin and quercetin, two hit compounds identified from the detergent-free screen, turned out negative in the detergent-present screen which indicates the detergent-present condition is more efficient for identifying true hit compounds. It is very interesting that methylprotodioscin, catalpol, and quercetin have been reported to be effective for diabetes or cancer treatments [21][22][23][24][25]. These natural products have been long used in traditional Chinese medicine. However, the mechanisms of these drug actions are still largely unknown. Our data shed some lights on the possible mechanism of these compounds. These compounds could be studied in future experiments in order to identify their specific mechanisms.
In all, we report on the development of a fluorescence polarization based high-throughput assay to identify potential c-Cbl (RING) domain regulators. Based on the pilot screens and validation assays, we have confirmed the binding activities of the three hit compounds. Our data have proven this screen can be used as a fast, robust, and quantitative measure of binding affinity for a wide range of c-Cbl (RING) domain regulators. The quantitative nature of the data significantly aids the efforts to choose the hit compounds. The Z' factor of 0.73 indicates this assay could be easily adapted for HTS. We predict that this assay will be suitable for identifying small molecule regulators of the c-Cbl (RING) domain.