Small Molecule Screening Identifies Inhibitors of the Epstein-Barr Virus Deubiquitinating Enzyme, BPLF1

Abstract

Herpesviral deubiquitinating enzymes (DUBs) were discovered in 2005, are highly conserved across the family, and are proving to be increasingly important players in herpesviral infection. EBV’s DUB, BPLF1, is known to regulate both cellular and viral target activities, yet remains largely unstudied. Our work has implicated BPLF1 in a wide range of processes including infectivity, viral DNA replication, and DNA repair. Additionally, knockout of BPLF1 delays and reduces human B-cell immortalization and lymphoma formation in humanized mice. These findings underscore the importance of BPLF1 in viral infectivity and pathogenesis and suggest that inhibition of EBV’s DUB activity may offer a new approach to specific therapy for EBV infections. We set out to discover and characterize small molecule inhibitors of BPLF1 deubiquitinating activity, through high-throughput screening. An initial small pilot screen resulted in discovery of 10 compounds yielding >80% decrease in BPLF1 DUB activity at a 10 μM concentration. Follow-up dose response curves of top hits identified several compounds with an IC₅₀ in the low micromolar range. Four of these hits were tested for their ability to cleave ubiquitin chains as well as their effects on viral infectivity and cell viability. Further characterization of the top hit, commonly known as suramin was found to not be selective yet decreased viral infectivity by approximately 90% with no apparent effects on cell viability. Due to the conserved nature of Herpesviral deubiquitinating enzymes, identification of an inhibitor of BPLF1 may prove to be an effective and promising new avenue of therapy for EBV and other herpesviral family members.

1. Introduction

Epstein Barr Virus (EBV), a gammaherpesvirus, was the first human tumor virus discovered and is a member of the Herpesviridae. EBV infects approximately 90% of the population worldwide. It is the causative agent of infectious mononucleosis, and is a pathogenic driver of Burkitt’s lymphoma, Hodgkin’s lymphoma and nasopharyngeal carcinoma (NPC) (1–3). EBV also causes lethal immunoblastic lymphomas in persons with acquired and innate immune disorders (4, 5). Patients who are immunosuppressed and have undergone bone marrow or solid organ transplant are at a significant risk of post-transplant lymphoproliferative disorder (PTLD). One estimation suggests that EBV is responsible for up to 200,000 cases of cancer world-wide per year (6). Additionally, there are recent, multiple lines of evidence that implicate EBV in the pathogenesis of multiple sclerosis (7–9) and systemic lupus erythematosus (10, 11).

BPLF1, EBV’s largest protein (3149 amino acids), is located in the viral tegument and is also expressed as a late lytic cycle protein; thus it can function during initial and late stages of infection. Transcripts are detected 8h after viral reactivation and reach maximum levels around 24h (12); endogenous BPLF1 protein is detected as early as 8h after induction (13). BPLF1’s most notable function is its deubiquitinating activity which is encoded within the first 205 amino acids, centered on a catalytic active site composed of a His-Asp-Cys triad (14). Mutation of the active site cysteine 61 abolishes deubiquitinating activity (15). Using an N–terminal fragment containing the DUB activity (BPLF1 1–246) it was shown that BPLF1 DUB activity cleaves both K63 and K48 polyubiquitin chains, in addition to removing monoubiquitin from protein substrates (15–17), suggesting it may exert both regulatory functions as well as a role in rescuing proteins from degradation (15). Proteins that are tagged with K48-linked ubiquitin chains are typically targeted for proteasome-mediated degradation, whereas K63 ubiquitin-linked proteins are involved in diverse proteasome-independent regulatory and signaling roles (18). The DUB activity is conserved across the Herpesviral family suggesting that it provides essential functions in the viral life cycle and that novel targets of BPLF1 are likely also targets by other family members.

BPLF1 and herpesviral homologs are involved in production of viral genomes and modulate viral infectivity shown by use of siRNA and knockout virus, all of which result in decreased production of infectious virus (19–21). BPLF1 knockout virus (22) specifically was found to decrease production of infectious virus ~90% in comparison to WT virus, demonstrating that BPLF1 is a significant determinant of viral infectivity (23). A central EBV hallmark is its ability to establish latency and to transform human cells: infection with BPLF1 knockout virus results in delayed and reduced B-cell outgrowth (23). BPLF1 knockout virus also delays and reduces formation of lymphomas in a humanized mouse model (23).

Several specific targets of BPLF1 1–246 have now been identified, including the EBV encoded ribonucleotide reductase (RR), where deubiquitination results in down-regulation of RR activity (15). Several members of the Trans-Lesion Synthesis DNA repair pathway (PCNA (17), Rad18 (16) and pol eta (24)) are deubiquitinated by BPLF1 1–246 and were found to be important for infectious virus production. BPLF1 has been shown to be involved in various roles of immune evasion and regulating innate immune response. BPLF1 interacts with and deubiquitinates TRAF6 which inhibited NF-κB signaling during lytic infection (22). BPLF1 DUB activity was found to suppress TLR-mediated activation of NF-κB (13). Both BPLF1 and its Kaposi’s Sarcoma Herpesvirus (KSHV) DUB homolog, ORF64, interact with and decrease RIG-I ubiquitination resulting in decreased RIG-I-mediated IFN signaling (25, 26). BPLF1 also contains deneddylating activity within the same active site and targets cullin-RING ligases that serve as scaffolds for E3 ubiquitin ligases in the process of proteasomal degradation (27–29). BPLF1 and herpesviral homologs function both in initial and late stages during infection and are involved in virus entry, transport, replication, assembly, infectivity, and lymphomagenesis (12, 15, 22, 23, 30–34).

Specific small molecule therapies targeting EBV for all related indications remains an unmet medical need. Previous findings implicate BPLF1 in basic aspects of EBV virology and pathogenesis and attest to its importance for understanding the roles of herpesviral deubiquitinating activity in both lytic and persistent infections. Inhibition of EBV’s DUB activity by specifically targeting BPLF1 offers a new approach to specific therapy of EBV infections (35) and consequently we set out to identify small molecule inhibitors of BPLF1s DUB activity. The conserved nature of deubiquitinating enzymes across Herpesviridae suggests that potential targets identified may ultimately be effective toward other family members.

Here we developed a robust, high throughput screening (HTS) assay and utilized the Library of Pharmacologically Active Compounds (LOPAC 1280; Sigma) to identify small molecule compounds capable of inhibiting BPLF1 DUB activity. The LOPAC compound set is composed of 1280 compounds that target a wide variety of major classes of drug targets. We identified several molecules capable of inhibiting BPLF1 1–246 DUB activity (IC₅₀) in the low micromolar range. The top inhibitory compound, suramin was not selective but was non-toxic to cells, inhibited ubiquitin chain cleavage, and resulted in decreased infectious virus production. Additionally, suramin inhibited the DUB activity of its KSHV homolog demonstrating a broader role toward another Herpesvirus.

2. Materials and Methods

2.1. HTS Assay

Enzymatically active BPLF1 1–246 (the N –terminal fragment composed of amino acids 1–246) was overexpressed and purified in large quantities (approximately 200 mg/L of culture) in E. Coli using standard techniques as previously described (15). An HTS-assay that was developed and conditions include: 25 nM BPLF1 1–246, 250 nM Ub-Rhodamine 110 in 10 μl final volume of reaction buffer (50 mM Hepes pH 7.4, 0.5 mM EDTA, 100 mM NaCl, 1 mM DTT, 0.1 mg/ml BSA, and 0.01% Tween 20). The screen was developed with liquid handling automation (Mosquito nanodispense HTS instrument (TTP LabTech) and Combi Multidrop) where compounds (in DMSO) are stamped (100 nL) in 384-well assay ready plates with a final DMSO concentration of 1%. BPLF1 is diluted in reaction buffer and is incubated with the test compound for 30 min. Ub-Rhodamine 110 in reaction buffer is added, mixed, and the reaction is allowed to proceed for an additional 40 minutes. 2.5 ul of guanidine-HCl is added to quench the reaction and plates are read using an excitation wavelength of 485 nm and an emission wavelength of 535 nm (PerkinElmer Envision). Each screening plate contains 16 replicates of high (1% DMSO) and low controls (both no enzyme and 5 μM Ub-aldehyde) columns for normalizing test compounds. All compounds and screening plates were barcoded and all data was uploaded and analyzed in the CICBDD database (ScreenAble Solutions). Compounds inhibiting at over 70% were reordered, diluted in DMSO and re-screened in duplicate over a 10-point dilution range. IC₅₀s were determined with Prism GraphPad software. Interference assays were performed with the standard Ub-Rhodamine 110 assay in the absence of inhibitors, and then plates were read using an excitation wavelength of 485 nm and an emission wavelength of 535 nm. Inhibitors were then added and the plates were re-read to detect loss of signal due to the presence of the small molecule.

2.2. Ub-chain cleavage assay

Purified K63-linked ubiquitin chains were purchased from BioMol. Inhibitors were used at 10 μM concentration in the presence of 25 nM BPLF1 1–246 in the same reaction buffer used for the Ub-Rhodamine 110 assay. 2 μg of K63-linked chains were added to the reaction and samples were incubated at 37 degrees C overnight. Samples were run on SDS gel and visualized with Sypro Ruby stain (Invitrogen) according to manufacturer’s protocol.

2.3. Cell lines, cell viability and transfections

HEK293 and 293EBV+ cells (36) and cells were maintained in DMEM supplemented with 10% FBS, 1% penicillin and streptomycin, and 1% L-glutamine (Corning). iSLK.219 (doxycycline-inducible SLK cells harboring latent rKSHV.219) (a kind gift from D. Ganem) were maintained in DMEM (Corning) supplemented with 10% FBS (Sigma), 1% penicillin and streptomycin (Corning), G418 (250 μg/ml) (Sigma), hygromycin (400 μg/ml) (Corning), and puromycin (10 μg/ml) (Corning). rKSHV.219 is a recombinant virus that expresses GFP from the constitutively active EF-1α promoter, expresses RFP from the KSHV PAN lytic promoter, and expresses a puromycin resistance gene as a selectable marker (37).

To monitor cell viability cells were seeded at 2 × 10⁵ cells in 60 mm plates. Small molecule compounds were added at the concentrations indicated and placed at 37 degrees for 48h, then viable cells were counted using trypan blue exclusion. Transfections were performed using Lipofectamine 2000 (Invitrogen) or Effectene (Qiagen) according to manufacturer’s suggestions.

2.4. EBV Infectivity

293 EBV+ cells (36) or 293 BPLF1 KO cells (BPLF1-deficient EBV-bacmid) (22) were induced into lytic replication by transfection with BZLF1 and gp110 (23) in the presence of the small molecule compound for 24h. After 24h the media was replaced and new inhibitor was added for an additional 24 h. Supernatant fluids containing infectious viral particles were collected and cleared of cellular debris by centrifugation for 5 minutes at 500 x g. Supernatant fluids collected from 293 BPLF1 KO cells were concentrated 10X using 100K MW Amicon centrifugal filters. 100 ul of cleared lysate was placed on Raji cells for determination of infectious virus. Raji cells were treated with 50 ng/ml phorbol-12-myristate-3-acetate and 3 mM sodium butyrate 24 h after infection (40). 48h after infection cells were analyzed for by flow cytometry for the presence of GFP (the EBV bacmid encodes for GFP) and titers were determined.

2.5. Suramin selectivity

Suramin selectivity was determined by Life Sensors. Suramin was screened against 10 DUBs in the Ub-Rhodamine 110 assay. Each DUB concentration was selected such that Ub-Rhodamine consumption was linear with respect to enzyme concentration for at least 40 minutes. Ub-Rhodamine concentrations for each individual DUB were selected such that the initial reaction rate was linear with respect to Ub-Rhodamine concentration (i.e. the Ub-Rhodamine concentration was below K_M). Suramin was screened against each DUB at 14 concentrations in quadruplicate, ranging from 100 μM to 12 nM in a 2-fold dilution series. All assays were performed using the following buffer conditions: 50 mM Hepes pH 7.4, 0.5 mM EDTA, 100 mM NaCl, 1 mM DTT, 0.1 mg/ml BSA, 0.01% Tween 20, 1% DMSO (from compound addition). Reactions were monitored every 5 minutes using an EnVision plate reader (485 nM excitation, 535 nM absorption). The 40-minute time point was used for determining the IC₅₀ of suramin against each DUB. Wells containing DUB and Ub-Rhodamine with no compound (1% DMSO), and wells containing Ub-Rhodamine only (1% DMSO) were used to normalize relative fluorescence units (RFUs) to percent inhibition. A four-parameter logistic fit was applied using Prism (GraphPad) to interpolate the IC₅₀s based on the background-subtracted, normalized data for suramin inhibition of each DUB.

2.6. KSHV assays

To induce KSHV reactivation, cells were treated with 0.2 μg/mL Dox and 10, 5, or 1μM suramin. Supernatants from reactivated cells were collected at 72 hours post dox treatment and filtered through a sterile 0.45-μm filter. Naïve 293 cells were infected by spinoculation with filtered supernatant at 2500rpm for 90 min at 30°C in the presence of 8 μg/ml of Polybrene.

For viral genome copy quantitation, supernatants from reactivated cells were diluted in sterile water and boiled for 10 min at 95°C. To detect viral genomes, SYBR green qPCR was performed with KSHV ORF57 primers (F: 5’- TGGACATTATGAAGGGCATCCTA; R: 5’- CGGGTTCGGACAATTGCT), and genome copies were quantitated by comparison to an ORF57 plasmid-derived standard curve.

For FACS analysis of viral reactivation or infection, cells were fixed in 4% methanol-free formaldehyde (Polysciences, Inc) in PBS for 10 minutes at room temperature. Flow cytometry was performed using a MACSQuant VYB cytometer (Miltenyi Biotec), and analysis was performed using FlowJo software (Tree Star).

3. Results

3.1. Development and Optimization of the HTS Assay

A Ubiquitin(Ub)-Rhodamine 110 assay was developed to measure BPLF1 enzymatic activity. An enzymatically active BPLF1 fragment composed of the first 246 amino acids (DUB activity is conferred in the first 205 amino acids) (BPLF1 1–246) was expressed and purified as previously described (15). Ub-Rhodamine 110 is a sensitive fluorogenic deubiquitinating enzyme substrate. Purified BPLF1 1–246, isolated from overexpression in E. Coli, was incubated with Ub-Rhodamine 110 whereupon BPLF1’s DUB activity catalyzes the release of the Rhodamine 110 moiety, which is directly attached to the C-terminus of ubiquitin by an amide bond. Cleavage of the fluorophore results in de-quenching of the fluorescent signal and serves as the measurement of BPLF1 1–246 activity. The increase in fluorescence can be measured dynamically in a plate-based format, allowing high-throughput kinetic and end-point assays. Utilizing this assay, BPLF1 1–246 titration from 25 to 200 nM yielded ideal enzymatic behavior showing a linear increase in initial velocity with increasing enzyme concentration. 25 nM BPLF1 1–246 was selected as the enzyme concentration for the screening assay since it produced a reproducible linear signal for over an hour and provides a low tight binding limit for follow up assays (Figure 1A). The K_M of BPLF1 1–246 for Ub-Rhodamine 110 was determined to be approximately 2 μM (Figure 1B). For the screening assay, a Ub-Rhodamine 110 substrate concentration of 250 nM (8-fold lower than the K_M) was selected, which is preferential for identifying competitive inhibitors. Ubiquitin-aldehyde, a general inhibitor of deubiquitinating enzymes, served as a positive control for inhibition of BPLF1 1–246 (Figure 1C). 5 μM Ub-aldehyde was chosen as the low signal control for screening plates, which inhibits BPLF1 1–246 DUB activity by approximately 95%. Overall, this assay development effort has produced a robust and reproducible HTS-compatible assay with excellent statistics (Z’ factor > 0.8).

Figure 1.

Optimization of the BPLF1 HTS Assay. A. BPLF1 1–246 titration shows ideal enzyme behavior with initial velocity proportional to enzyme concentration and linear range beyond 60 minutes for [BPLF1 1–246] at 25 nM (inset) Ub-Rhodamine 110 concentration was held constant at 25 nM. B. K_M for Ub-Rhodamine 110 was determined to be approximately 2 μM. BPLF1 concentration was held constant at 25 nM. Fitted line shows Michaelis-Menten analysis. C. Optimal concentration of Ub-aldehyde for inhibition of BPLF1 1–246 was determined. Ub-aldehyde was used at 5 μM in the low control column for screening plates. BPLF1 1–246: 25 nM. Ub-Rhodamine 110: 250 nM.

3.2. Screening of the LOPAC compound set

The HTS-compatible assay developed as described above meets the industry standards for a screening campaign. Conditions include: 25 nM BPLF1 1–246, 250 nM Ub-Rhodamine 110 in 10 μl final volume of reaction buffer with LOPAC compounds at 10 μM concentration. Compounds were stamped in 384-well assay ready plates with a final DMSO concentration of 1%, well below the determined tolerance limit (3%). BPLF1 1–246 in reaction buffer was added for pre-incubation with test compounds for 30 min. The reaction was then incubated with Ub-Rhodamine 110 for 40 minutes (initial rate, linear conditions). The reaction was stopped by addition of guanidine-HCl (determined to have no effect on fluorescence signal), and plates were read using an excitation wavelength of 485 nm and an emission wavelength of 535 nm. Each screening plate contained high (1% DMSO) and low controls (both no enzyme and 5 μM Ub-aldehyde) columns for normalizing test compounds (Figure 2). The screen performed well (average Z’ = 0.88) and 10 compounds yielding >80% decrease in DUB activity were identified (0.78% hit rate).

Figure 2.

LOPAC set pilot screen data for BPLF1 1–246. Shown is a scatter plot for screening results of 1280 compounds for four 384-well plates: test compounds (gray), DMSO control wells (red), Ub-aldehyde control wells (blue), and no enzyme control wells (green). Each plate contained a full column of positive and negative controls; test compounds were normalized to controls (ScreenAble, Spotfire). Highlighted compounds (dark gray, labeled with UNC compound number) are compounds yielding >70% inhibition at 10 μM concentration.

To validate initial findings, compounds inhibiting at greater than 70% were repurchased from commercial sources, serially diluted in DMSO and screened in duplicate over a 10-point dose curve to determine the IC₅₀ using a three parameter fit (Table 1) (one compound was no longer commercially available and was excluded from the study). Four compounds with IC₅₀s in the low micromolar range were chosen for further study; suramin (0.98 μM), β-lapachone (2.5 μM), aurothioglucose (2.5 μM) and NSC95397 (4.8 μM). The top compounds were monitored through an interference assay to determine if the compound alone could interfere with the fluorescence signal generated by the cleavage of Ub-Rhodamine 110. To this end a Ub-Rhodamine 110/BPLF1 1–246 mixture was allowed to proceed for 40 min and then the plate was read. Compounds were then added and the plate was re-read to see if signal quenching was detected. None of the compounds tested resulted reduced signal intensity indicating there was no signal interference due to the compound structure.

Table 1.

IC₅₀ determination of top compound hits.

UNC Number	Compound	IC50 (μM)
UNC10101238A	Suramin sodium salt	0.98
UNC10100775A	beta-lapachone	2.5
UNC10100743A	Aurothioglucose	2.5
UNC10100904A	Gossypol	3.0
UNC10100114A	Aurintricarboxylic acid	3.3
UNC10101270A	U-74389G maleate	3.4
UNC10100687A	NSC 95397	4.8
UNC10100868A	NF 023	6.1
UNC10100691A	6-Hyd roxy- DL-DOPA	>10
UNC10100993A	PPNDS tetrasodium	>10
UNC10100928A	Me-3,4-dephostatin	>10
UNC10101110A	Basilen blue E-3G	>10

3.3. Suramin, β-lapachone, NSC95397, and aurothioglucose inhibit K63-linked ubiquitin chain cleavage

A secondary assay to confirm inhibition of BPLF1 enzymatic activity was be performed by measuring BPLF1’s well know ability to cleave K63-linked ubiquitin chains (15). Purified BPLF1 1–246 was incubated in the presence of K63 linked ubiquitin chains (Ub_1–7) and 10 μM inhibitor. Figure 3 demonstrates that suramin, β-lapachone, NSC95397, and aurothioglucose all inhibited K63-linked chain cleavage and verifies that these compounds do inhibit BPLF1 DUB activity. The negative control on the right shows the banding pattern of the 1–7 linked ubiquitin molecules and the BPLF1 only lane shows the cleavage of the high molecular weight forms. It is observed that the inhibition of K63-linked chain cleavage appears to be weaker with suramin than with the other inhibitors tested even though suramin was found to have a lower IC₅₀ as determined through the Ub-Rhodamine 100 assay. It is unclear exactly why this occurs. Perhaps suramin is less effective against K63 linked-cleavage but yet is more effective at inhibiting a monoubiquitinated substrate. Different deubiquitinating enzymes have varying affinities toward K48-linked, K63-linked and monoubiquitinated substrates and therefore interactions between the DUB and inhibitory compound may result in variable inhibition between differentially ubiquitinated substrates.

Figure 3.

Top compounds inhibit BPLF1 1–246 K63-linked chain cleavage. Purified BPLF1 1–246(25 nM) was incubated with 2 μg K63-linked_(1–7) ubiquitin chains (BioMol) in the presence of 10 μM inhibitor overnight at 37 °C. Samples were run on 4–15% SDS gel and visualized with Sypro Ruby stain. Negative control represents a reaction without BPLF1 1–246. BPLF1 labels refer to BPLF1 1–246 fragment.

3.4. Cell Viability and Infectious Virus Production

Once it was confirmed that suramin, β-lapachone, NSC95397, and aurothioglucose inhibit BPLF1 1–246 DUB activity these compounds were examined for cellular toxicity and effects on infectious virus production. To monitor cell viability 4 × 10⁵ 293 EBV+ cells were plated and incubated with 0 (DMSO), 1, 5, and 10 μM concentrations of the selected drug in growth media. At 24 h fresh media and compound were replaced and living cells were counted 48 h after initial compound treatment as determined by trypan blue exclusion. Aurothioglucose and suramin were found to not inhibit cell growth or viability at concentrations up to 10 μM, whereas β-lapachone and NSC95397 resulted in significantly reduced cell viability at 1 and 5 μM respectively. (Figure 4A).

Figure 4.

Cell viability and infectious virus production in the presence of BPLF1 inhibitors. A 293 EBV+ cells were incubated with selected compounds for 48 h and living cells were determined by trypan blue exclusion. B. Infectious virus production for suramin and aurothioglucose. Lytic replication was induced by transfection of Z and allowed to proceed in the presence of suramin for 48h and then titered on Raji cells. Error bars represent standard error of mean.

Our previous work has shown that BPLF1 knockout virus decreases infectious virus production by approximately 90% (23). To investigate if inhibition of BPLF1 by these compounds could result in a similar decrease in infectious virus production, cells undergoing lytic reactivation were treated with aurothioglucose and suramin since they did not decrease cell viability. β-laphacone and NSC95397 were removed from further study due to cellular toxicity findings. 293 EBV+ cells were induced into lytic reactivation by transfection with BZLF1 which results in the full complement of lytic gene expression and production of GFP-encoding infectious virions (36). Supernatant fluids containing released infectious virions were incubated with Raji cells, and infectivity was measured by flow cytometry for GFP-expressing cells (23) (Figure 4B). Suramin decreased infectious virus production in a dose dependent fashion up to approximately 90% at 10 μM – similar to levels observed with BPLF1 knockout virus. While aurothioglucose was found to directly inhibit BPLF1 1–246 DUB activity in vitro, it did not result in decreased virus production at the concentrations tested.

3.5. Suramin inhibits DUB activity of KSHV ORF64

Herpesviral deubiquitinating activity is conserved across the Herpesviridae. It was previously shown by siRNA knockdown targeting the KSHV ORF64 homolog, that reduction of ORF64 resulted in decreased viral lytic transcription and protein expression (38). ORF64 DUB activity was monitored in the presence of suramin utilizing the Ub-Rhodamine 110 assay (Figure 5A). FLAG-tagged ORF64 (ORF46_N), consisting of amino acids 1–205 was purified from 293T cells and incubated for 30 minutes with Ub-Rhodamine 110. In the presence of 10 μM suramin an >80% decrease in DUB activity was observed. Results demonstrate that inhibitory effects of suramin are not limited to EBV and may function against other herpesviral members.

Figure 5.

Suramin inhibits DUB activity of KSHV ORF64. A. ORF64 1–205 N terminal fragment was affinity purified and incubated with ubiquitin Rhodamine 110 and 10 μM suramin for 30 minutes at RT, and activity was monitored by fluorescence using an excitation wavelength of 485 nm and an emission wavelength of 535 nm. Ubiquitin aldehyde (Ub-Al) (inhibitor control) is a potent inhibitor of DUB activity and pcDNA3 serves as the empty vector control. B.Suramin does not affect reactivation of KSHV. % reactivation of latently infected cells C. Suramin does not inhibit genome copy production. Quantification of KSHV genome copies in supernatant of reactivated cells: D. Suramin does not reduce KSHV infectious virus production. Infection of naïve cells using supernatants from reactivated cells. Error bars represent standard deviation.

The effects of suramin on KSHV reactivation, replication and infectious virus production were evaluated (Figure 5 B–D). Suramin at concentrations up to 10 μM were found to have no significant effect on infectious virus production, genome copy number, or the ability of KSHV to reactivate from latency. It should be noted that the overall impact of ORF64 on infectious virus production and genome replication has not been determined and a knockout virus of ORF64 has not been reported.

3.6. Suramin Selectivity

Over 100 cellular DUBs are encoded for by mammalian cells. In an effort to measure suramin specificity against other cellular DUBs, lysate from 293T cells were incubated in the presence of suramin using Ub-Rhodamine 110 as a substrate. (Figure 6A). 10 μM suramin was found to inhibit the pool of cellular DUBs by about 50% indicating low specificity in the 10 μM range. In comparison the non-specific DUB inhibitor Ub-aldehyde reduced the cellular DUB activity by >90%.

Figure 6.

Suramin specificity. A. Suramin was tested for inhibition of other cellular DUBs. Total cell lysates (represents a pool of over 100 cellular DUBs) were examined for DUB activity with Rhodamine 110. Suramin reduced cellular DUB activity approximately 50% showing lack of specificity. Ub-aldehyde served as a control and showed nearly complete inhibition of cellular DUBS. Error bars represent standard error of mean. B. Infectious virus production from 293 BPLF1 KO cells in the presence of suramin. Lytic replication was induced by transfection of Z in the presence of suramin for 48h. Supernatant fluids were collected, concentrated 10X, and titered on Raji cells. Error bars represent standard error of mean.

Due to the nonspecific nature of suramin, the observed loss of EBV infectious virus production (Figure 4B) may be, at least in part, due to effects exerted on cellular enzymes by suramin. To determine if the observed loss in infectivity is due to the presence of BPLF1, we utilized a BPLF1 knockout virus (22), where BPLF1 is not expressed, and measured infectious virus production in the presence suramin as performed with WT EBV (Figure 6B). The presence of suramin did not result in a decrease in infectious virus production with BPLF1 KO virus, in contrast to the WT virus, suggesting that the loss of infectivity observed for the WT EBV (Figure 4B) is due to the inhibition of BPLF1 and not to off-target effects on cellular substrates.

To further analyze selectively, suramin was screened against a panel of cellular DUBs using Ub-Rhodamine 110 in collaboration with LifeSensors. Suramin was screened against 10 DUBs across a range of enzyme families. Each DUB concentration was selected such that Ub-Rhodamine consumption was linear with respect to enzyme concentration for at least 40 minutes. Ub-Rhodamine concentrations for each individual DUB were selected such that the initial reaction rate was linear with respect to Ub-Rhodamine concentration (i.e. the Ub-Rhodamine concentration was below K_M). DUB and Ub-Rhodamine concentrations are listed in Table 2. Where possible, conditions were selected to match those utilized for the initial BPLF1 LOPAC screen. Of the 10 DUBs chosen for selectivity profiling, suramin inhibited 9 with an IC₅₀ of 1–4 μM, indicating that this compound has relatively poor selectivity and may be acting via a nonspecific mechanism (Table 2). The exception was JosD2, against which suramin exhibited an IC₅₀ of 28 μM. While the mechanism of suramin inhibition is beyond the scope of this study, the IC₅₀ results indicate a possibility that suramin is inhibiting BPLF1 and other DUBs in a nonspecific manner. However, the results confirm the activity of suramin against BPLF1, and provide a framework for the future study of this and other prospective BPLF1 inhibitors.

Table 2.

The IC₅₀s of suramin against selected DUBs.

DUB	DUB family	[DUB] (nM)	[Ub-Rho] (nM)	Suramin IC50 (μM)
BPLF1 1–246	Viral	0.5	250	3.3
PLPro	Viral	5	250	3.2
JosD2	MJD	1	250	28
USP8c	USP	1	250	3.0
SseL	Bacterial	25	250	2.0
Atax3L	MJD	25	250	2.6
Bap1	UCH	0.25	250	3.4
Cezanne	OTU	1	25	1.6
UCH-L5	UCH	0.1	25	1.5
USP7	USP	0.1	25	1.8
UCH-L1	UCH	0.5	5	3.6

4. Discussion

All members of the Herpesviridae encode for DUB activity within the N-terminal region of the large tegument protein. The crystal structure of a herpesviral homolog, the murine cytomegalovirus, M48, was solved and it revealed that the structure of the interface with ubiquitin and the arrangement of the active site residues are unique and represents a new and distinct class of DUB (39). While the catalytic triad is conserved in all the herpesviruses, the sequence similarity of the DUBs is low at approximately 12% (39). Secondary structure among the herpesviral DUBs is predicted to be relatively high (14), but sequence and structural similarity prediction searches found no cellular or other viral family homologs (39). Therefore, herpesvirus DUBs represent a unique class and may be specifically targetable. BPLF1 is an unexplored avenue of study for EBV therapy and differs from previous EBV anti-viral attempts that were largely directed toward EBV replication machinery and the viral protein kinase (40, 41). While many of the previous anti-viral drugs did demonstrate effective inhibition of replication and infectivity in vitro, these drugs were found to have only limited use in vivo for EBV treatment (41). Targeting BPLF1 DUB activity may provide a more specific and effective strategy. An inhibitor of BPLF1 may prove to be an effective treatment during lytic replication, such as in cases of infectious mononucleosis or chronic active EBV, as well as in inhibiting initial steps in infection (since BPLF1 is packaged in the viral tegument).

Development of a robust, simple assay in a high throughput system to measure DUB activity of BPLF1 is essential for embarking on an HTS campaign. Concentrations of enzyme and substrate were optimized to produce ideal linear enzyme kinetics which progressed for over an hour (Figure 1). Hit compounds inhibiting signal by >70% were analyzed for signal interference to rule out false positives, and new compounds were ordered and used in a dose-response format to determine IC₅₀. Verifying results with the new compounds eliminates effects due to possible degradation and lot variation. Figure 3 demonstrates direct inhibition of BPLF1 1–246 DUB activity via a secondary assay with the compounds with the lowest IC₅₀s adding validity to the observed screen results.

To examine effects of the small molecule compounds on the EBV life cycle we first set out to ensure the compounds were not toxic in the inhibitory range (Figure 4), and compounds with no observed effects of cell viability were advanced to monitor production of infectious virus. The cell death observed with NSC95397 and β-lapachone may be due to non-specific inhibition of critical cellular processes at the concentrations tested. NSC95397 has been shown to inhibit many cellular kinases including AKT, IKKα/β, MKK7, and TBK1 (42). β-lapachone has been shown to induce cell cycle arrest and apoptosis in colon cancer cells independent of the p53 pathway (43).

Interestingly suramin inhibited infectious virus production to similar levels as observed with BPLF1 knockout virus, whereas treatment with aurothioglucose resulted in no observed loss of virus production while neither decreased in cell viability. The observed difference in infectious virus production may be due to additional non-specific interactions as these compounds have been shown to inhibit other enzymatic activities. Moreover, it should be noted that in vitro assays were performed with the N-terminal fragment whereas the infectious studies contain full length BPLF1. Aurothioglucose, also called gold thioglucose, was originally used to treat rheumatoid arthritis but has more recently been replaced by more effective anti-rheumatic drugs (44). Aurothioglucose has been shown to inhibit adenylyl cyclase activity in human lymphocyte membranes (45).

Suramin, a polysulfonated polyaromatic with symmetry around a central urea moiety, was synthesized by Bayer chemists; Oskar Dressel, Richard Kothe and Bernhard Heymann in 1917 (46). The compound has a very interesting history and was even used as a bargaining chip in negotiations for lost German territories (see Steverding, D. 2010 for a historical review (47)). Suramin was found to cure trypanosomiasis in experimental animals and in humans (46), and the Bayer company offered the drug formulation to the British government in return for Germany’s lost African territories (48). The British refused the offer and Bayer withheld the structure, but it was eventually elucidated and published in 1924 by French pharmacist Ernest Fourneau (49).

Suramin is used as an effective treatment of African trypanosomiasis (African sleeping sickness), an insect-borne parasitic disease transmitted by Trypanosoma brucei gambiense and Trypanosoma brucei rhodesiense (50), and river blindness, which is caused by infection with the parasitic worm Onchocerca volvulus. Additionally suramin has been used in clinical trials targeting autism, prostate cancer, and other forms of cancer including bladder, breast, renal cell and non-small cell lung carcinoma (51–53). The mechanism of action is unknown but suramin has been shown to combine with trypanosomal glycolytic enzymes and inhibit energy metabolism (54). Multiple targets of suramin have been described: P2Y purinoceptor 2, NAD-dependent protein deacylase sirtuin-5, follicle-stimulating hormone receptor, ryanodine receptor 1, prothrombin, and phospholipase A2 (55–60). No activity toward deubiquitinating enzymes has been described previously.

Suramin has been evaluated as an antiviral agent among other viral families and has been shown to be effective in inhibiting entry and reducing infectious virus production. Suramin inhibits entry of Chikungunya, Zika, and Ebola virus by disrupting attachment to the host cell and reduces release of infectious virus (61, 62). Suramin was also inhibits Chikungunya, Sindbis, Dengue, and Semliki Forest virus replication (63, 64).

While suramin targets many different proteins, treatment with it did result in decreased virus production. Additionally, it was shown to inhibit the KSHV homolog ORF64, suggesting it may have effects toward the other Herpesviruses. Suramin was found to have inhibitory effects toward many deubiquitinating enzymes in the same μM range as BPLF1 1–246 (Figure 6 and Table 2) making it somewhat surprising that this nonspecific molecule is not toxic to cells. Perhaps in vitro effects observed in the purified system do not correlate with the cell based assays due to differing endogenous protein expression (full length vs. the N-terminal fragment) and regulation, and efficiency of drug-uptake. At this point it is unclear mechanistically how a molecule known to have promiscuous activity in vitro may still result in loss of infectious EBV production. While suramin is non-specific it does have many analogs (65–68); one of which was included in the study (NF 023) and resulted in inhibition of BPLF1 1–246 activity with an IC₅₀ in the low micromolar range (Table 1). Suramin analogs are being investigated now and may prove to be more specific. Further study is necessary to determine if any could serve as an effective antiviral for EBV and other Herpesviruses. Additional screening for Herpesviral DUB inhibitors is ongoing with the goal of identifying inhibitors more amenable for lead optimization via medicinal chemistry.

Highlights.

• A small pilot screen identified 10 compounds yielding >80% decrease in EBV’s BPLF1 1–246 deubiquitinating activity.

• Dose response assays identified 4 compounds capable of inhibiting BPLF1 1–246 in the low micromolar range.

• Top 4 compounds inhibited BPLF1 1–246 ubiquitin chain cleavage in vitro.

• The top inhibitor identified, suramin, decreased EBV infectivity by approximately 90% and was non-toxic to cells.

• Suramin also inhibited the deubiquitinating activity of the KSHV homolog, ORF64.