COP9 signalosome is an essential and druggable parasite target that regulates protein degradation

Understanding how the protozoan protein degradation pathway is regulated could uncover new parasite biology for drug discovery. We found the COP9 signalosome (CSN) conserved in multiple pathogens such as Leishmania, Trypanosoma, Toxoplasma, and used the severe diarrhea-causing Entamoeba histolytica to study its function in medically significant protozoa. We show that CSN is an essential upstream regulator of parasite protein degradation. Genetic disruption of E. histolytica CSN by two distinct approaches inhibited cell proliferation and viability. Both CSN5 knockdown and dominant negative mutation trapped cullin in a neddylated state, disrupting UPS activity and protein degradation. In addition, zinc ditiocarb (ZnDTC), a main metabolite of the inexpensive FDA-approved globally-available drug disulfiram, was active against parasites acting in a COP9-dependent manner. ZnDTC, given as disulfiram-zinc, had oral efficacy in clearing parasites in vivo. Our findings provide insights into the regulation of parasite protein degradation, and supports the significant therapeutic potential of COP9 inhibition.


INTRODUCTION
Protein turnover, which is the balance between protein synthesis and protein degradation, is essential for life. The Ubiquitin-Proteasomal System (UPS) is conserved in all eukaryotes including protozoans, and is responsible for the vast majority of protein degradation within the cell. Cullin-based ubiquitin ligases catalyze the ubiquitination of proteins destined for proteasomal degradation. Disruption of proteasomal activity results in accumulation of unwanted and toxic proteins, ultimately leading to cell death (1).
Protozoan parasites, such as Trypanosoma, Entamoeba, Leishmania, and Toxoplasma, present a major threat to global public health, and contribute significantly to morbidity and mortality worldwide. Antibiotic treatment is essential for managing patients infected with these parasites.
That said, new therapies are urgently needed given the lack of effective vaccines, drug resistance, limited efficacy and toxicity associated with current treatment (2)(3)(4). Given that the UPS pathway is essential for cell survival, inhibition of the proteasome has emerged as an attractive anti-parasitic target (5)(6)(7)(8)(9). However, the regulation of UPS-mediated protein degradation in clinically important protozoans remains poorly understood. Understanding the regulatory pathways involved in proteasomal degradation could, therefore, lead to new therapeutic opportunities for patients with these difficult-to-treat infections, either as monotherapy or in combination with established parasite proteasomal inhibitors.
Here, we investigated COP9 (Constitutive photomorphogenesis 9) signalosome function in disease-relevant protozoa using Entamoeba histolytica as the model parasite. E. histolytica is a protozoan parasite that is a leading cause of severe diarrhea in the world's poorest communities. There is no vaccine and only one class of drugs (nitroimidazoles) available to effectively treat invasive forms of disease. This is a major concern as we are ill-prepared and left with no option if resistance or intolerable side effects develops (4,(10)(11)(12). COP9 signalosome was found to be encoded by other protozoans such as Leishmania, Trypanosoma, and Toxoplasma, in addition to E. histolytica. We uncover E. histolytica COP9 signalosome as an essential upstream regulator of the parasite UPS protein degradation pathway. The zincditiocarb complex, a major metabolite of disulfiram, inhibited the E. histolytica COP9 activity, highlighting the potential for repurposing disulfiram as an anti-parasite agent.
Next we investigated whether the CSN5 protein was a true component of the parasite COP9 signalosome complex by determining the interaction between CSN5 and other subunits. Among the subunits that make up the COP9 signalosome complex, CSN2 and CSN5 are the most conserved subunits (15,17). Therefore, we investigated the interaction between the parasite CSN2 and CSN5. We identified the CSN2 gene and created a Flag tagged CSN2 expressing ameba cell line, followed by affinity pulldown with anti-Flag antibody ( fig. S1 and Fig. 1B). The immunoprecipitated endogenous proteins were first analyzed by quantitative massspectrometry, which revealed CSN2 and CSN5 subunits, along with other CSN subunits as the most enriched co-purified proteins (Fig. 1, B and C). In addition, protein-protein interaction screening using immunoaffinity purification of amebic cell lysate with specific anti-CSN5 antibody followed by mass spectrometry analysis revealed similar results ( fig. S2). Although we did not co-purify CSN4 and 8, we identified the putative genes in the E. histolytica genome.
CSN5 and CNS2 protein-protein interaction was confirmed by reciprocal co-immunoprecipitation  non-conserved residues (red). (B) Experimental approach for identifying CSN2 interacting proteins from E. histolytica cells by co-immunoprecipitation (Co-IP) followed by mass spectrometry. (C) Scatter plot and Pearson's correlation analysis of log2 ratios label-free quantification intensities for proteins identified by mass spectrometry in anti-Flag coimmunoprecipitations from E. histolytica cells expressing Flag-CSN2 compared to the empty vector control. CSN2 and CSN5 subunits (red), putative COP9 subunits (blue), and non-COP9 subunits (grey). (D) Diagram of experimental procedure for validating CSN5 and CSN2 interaction. (E and F) Reciprocal co-immunoprecipitation with anti-Flag (E) and anti-E.

COP9 signalosome is necessary for parasite protein degradation
To characterize the role of COP9 signalosome in E. histolytica biology, we first used a genetic knockdown approach. Because subunit 5 forms the catalytic center of the COP9 signalosome (CSN5) (14), we evaluated the function of the parasite COP9 by CSN5 gene silencing using an established high-efficiency inducible RNA interference technology (18,19). CSN5 knockdown was validated by immunoblotting with anti-E. histolytica CSN5 antibody, and a marked decrease in the CSN5 proteins level was observed in the knockdown cells compared to the controls (fig .   S3A). Next we performed a cell viability assay and found that CSN5 knockdown resulted in a proliferation defect and eventual cell death within 36 hours when compared with the empty vector control ( Fig. 2A).
Deneddylation involves the removal of the ubiquitin-like protein Nedd8 from Nedd8-conjugated (neddylated) proteins (14,20). Given that COP9 complex has deneddylation activity in nonparasitic eukaryotes and Nedd8 proteins are conserved in parasites (21,22), we hypothesized that the parasite COP9 has deneddylating function. First, we examined the effect of CSN5 depletion on global protein neddylation and found that knocking down CSN5 resulted in an increase in neddylated proteins compared to the controls (Fig. 2B), which suggested that the COP9 deneddylation activity is inhibited upon CSN5 knockdown.
Cullins are the best characterized neddylated COP9 substrates in other eukaryotes (23). A conserved cullin lysine residue forms an isopeptide bond with the carboxy-terminal Gly-76 of the Nedd8 protein (24). We have previously identified an E. histolytica cullin1 protein (16) which has the conserved neddylation site ( fig. S4). Immunoblot analysis revealed the most intense neddylated band after CSN5 disruption being above the 75 KDa mark, which includes neddylated cullin1 (Fig. 2B, fig. S3E), which can sometimes run lower than its predicted molecular weight (25). Therefore, we tested if cullin1 neddylation is altered upon CSN5 knockdown, and found that knockdown cells accumulated neddylated cullin1 by immunoblot analysis using antibodies specific for E. histolytica cullin1 and Nedd8 (Fig. 2C). This is consistent with a recent cancer biology study that showed a specific COP9 inhibitor caused accumulation of neddylated cullin prior to cancer cell proliferation defects (26). Thus, disrupting E. histolytica COP9 by CSN5 knockdown resulted in accumulation of cullin1 in its neddylated state.
In addition to the knockdown approach, we utilized dominant negative mutant expression as an independent strategy to inhibit endogenous protein function. Overexpression of enzymatically dead mutant proteins result in dominant negative effect. This method has been successfully used to explore gene function in E. histolytica biology (27). A single amino acid substitution, Asp Cullin based RING E3 ubiquitin ligases (CRLs) tag cellular proteins with ubiquitin (Ub) chains for their subsequent degradation by the proteasome, and CRL activity is dependent on deneddylation of cullins (13,15,28). Similar to the CSN5 knockdown, CSN5 mutant expression resulted in accumulation of cullin1 in its neddylated form (Fig. 2E, fig. S3C). Next, we investigated the effect of parasite COP9 disruption on UPS-dependent protein turnover using an E. histolytica strain that constitutively express a GFP reporter substrate of UPS-mediated protein degradation, ubiquitin-arginine-GFP (Ub-R-GFP). We found that mutant CSN5 overexpression disrupted protein degradation resulting in accumulation of GFP as measured by quantitative confocal imaging (Fig. 2, F and G) and fluorometric assay (Fig. 2H). Taken together, our data indicate that the parasite COP9 signalosome is essential for protein degradation. Data represent mean ± SD of quintuples from one experiment and are representative of three independent experiments. *P < .05, ** P < 0.01, *** P < 0.001, two-tailed t test.

CSN5 genetic disruption
Our finding that COP9 signalosome function was essential for E. histolytica cell survival led us to test its druggability. Since COP9 essential deneddylation activity is embedded within the metalloprotease site of the CSN5 subunit, we screened known metalloprotease inhibitors and chelating agents for their ability to dock onto CSN5 using virtual screening by molecular docking (29). Among the hits of potential inhibitors was zinc-ditiocarb (ZnDTC) ( fig. S5A). ZnDTC was of particular interest because it is a metabolite of the drug disulfiram (Fig. 3A). Disulfiram has been approved by the FDA since 1951 to treat alcohol dependence with safety and pharmacokinetics profile already established (30).
To evaluate whether ZnDTC had anti-amebic activity, we determined the dose response curve for the effect of ZnDTC on parasite viability. The compound was effective against parasites at nanomolar concentration with an EC50 of 11.21 ± 6.39 nM (Fig. 3A), which is below the serum and tissue level achieved in individuals on disulfiram therapy at recommended doses (31,32).
The EC50 of metronidazole, the current drug of choice to treat amebiasis, was found to be 10.26 ± 6.06 µM (Fig. 3A). This is in keeping with previous reports for metronidazole EC50 ranging from 5 to 20 µM (11,12).
Drug resistance by gene overexpression is a useful approach for target identification and validation (33,34).Therefore, to test if CSN5 is the target of ZnDTC we examined the susceptibility of cells overexpressing CSN5. Overexpression of CSN5 resulted in increased resistance to ZnDTC treatment ( fig. S3B, Fig. 3B, fig. S5B), providing evidence that CSN5 is targeted by ZnDTC in E. histolytica parasite. As expected, concentrations of 100 nM and greater resulted in significant growth defect ( fig. S5B), which occurs when drug concentrations exceed overexpression capacity (33). In support of these findings, CSN5 knockdown rendered cells more sensitive to ZnDTC ( fig. S3A, Fig. 3B).
To further evaluate whether ZnDTC acts in a COP9 dependent manner, we examined cullin1 deneddylation and UPS activity in drug-treated cells compared to controls. We found that cells treated with ZnDTC phenocopy genetic disruption of CSN5 by trapping cullin1 in a neddylated state (Fig. 3C, fig. S3D, fig. S3F), and inhibited protein degradation (Fig 3, D and E).
Collectively, these findings imply that ZnDTC exerts anti-amebic activity by disrupting E. histolytica COP9 signalosome-regulated proteolysis. Fluorometric assay of Ub-R-GFP degradation. ZnDTC inhibits Ub-R-GFP degradation in a dose dependent manner. Data represent mean ± SD of quintuples from one experiment and are representative of three independent experiments. ***P < .001, two-tailed t test.

ZnDTC has potent activity against E. histolytica in a mouse model that mirrors human infection
We next tested the therapeutic effects of zinc ditiocarb in vivo. In the body, disulfiram is rapidly metabolized to ditiocarb (DTC) which in the presence of metal ions such as zinc, forms zincditiocarb complex (ZnDTC). DTC-metal complexes have a relatively long half-life and are widely distributed throughout the body, including the gastrointestinal tract. Therefore, in order to achieve adequate levels, ZnDTC is safely given in vivo as oral disulfiram plus zinc gluconate (31,32,(35)(36)(37)(38). Entamoeba histolytica causes an inflammatory diarrhea termed amebic colitis (10). We used a mouse model that simulates human amebic colitis for in vivo studies. Mice were infected with luciferase-expressing parasites and infection was monitored by live bioluminescence imaging. Infected mice received a 5 day treatment course based on the minimum recommended treatment duration for human amebiasis (10). Similar to the antiamebic drug metronidazole (39,40), we started to observe parasite clearance after 2 days of ZnDTC therapy (Fig. 4A). Consistent with our live imaging findings, all ZnDTC-treated mice were culture negative at the end of the 5-day treatment course, compared to none in the untreated group (Fig. 4, A and B, fig. S5C). Histopathological examination and immunohistochemical staining using specific anti-E. histolytica antibodies revealed numerous parasites in untreated mice, absent in the treated mice (Fig. 4C). In addition, ZnDTC significantly reduced the destructive inflammatory response, and tissue damage as measured by tissue myeloperoxidase (MPO) levels and histological score (Fig. 4, C to E). These findings support an in vivo antiparasitic effect. **P < .01, ***P < .001, Fisher's exact test and Mann-Whitney U test.

DISCUSSION
Protozoan parasites continue to pose serious threats to health and contribute significantly to morbidity and mortality worldwide. Those living in poverty-stricken settings are disproportionately affected. Limited treatment options, poor efficacy, drug resistance, toxicity and expense contribute to the poor health outcomes associated with current therapies. Parasitic protozoan diseases are in dire need of new drugs. The ubiquitin proteasome pathway is the main engine for parasite protein degradation and therefore regulates multiple essential biological processes. Hence, the ubiquitin proteasome pathway has received tremendous attention recently in the parasitology field as an attractive drug target (1). Yet, our understating of how the ubiquitin proteasome pathway is regulated in these disease-relevant parasites remain limited. Here, E. histolytica was used as a prototype to study the role of COP9 in protein degradation in a medically significant protozoan. We found that the COP9 signalosome was essential for parasite biology, as disruption of the parasite COP9 led to dysregulation of the ubiquitin proteasome pathway which impaired protein degradation and led to cell death.
All pathogenic protozoans express cullins and Nedd8 proteins. While the majority encode COP9, it appears some do not, including Plasmodium parasites. Plasmodium encodes hydrolases that are capable of cleaving Nedd8, and while important, these enzymes appear to lack the ability to deneddylate cullin (22). It remains possible that these parasites produce proteins with CSN5-like characteristics and activity, and we should be encouraged that with continued research we might be able to identify these proteins.
Disulfiram, also known as Antabuse, is an FDA-approved drug that has been used to treat alcohol dependence for several decades, with well-established pharmacokinetic properties, safety and tolerance (30). Disulfiram has shown varying activity against parasites, with an even greater antiparasitic potency observed when complexed to zinc (ZnDTC) (41). However, these observations were based mainly on in vitro studies, and the underlying mechanisms that explain the enhanced potency remain poorly understood. In this study, we showed that the disulfiram metabolite, ZnDTC, has anti-parasite effects in vitro and in vivo, and identified the COP9 signalosome as a drug target. These findings are attractive for several reasons. Firstly, disulfiram penetrates a wide range of tissues including the blood-brain barrier, and so holds potential use in the treatment of infections of the central nervous system caused by COP9 producing protozoans. Disulfiram is well-tolerated and has excellent bioavailability, therefore the drug can be given orally for extended periods. For example, the combination of oral zinc gluconate and disulfiram was well-tolerated for 53 continuous months with negligible side effects in a patient with metastatic cancer (35). Furthermore, disulfiram is a globally available, economical and low-cost drug, which could make it an affordable option for patients in low income countries (37). Our findings provide mechanistic insight into the antiparasitic activity of disulfiram, and establish in vivo efficacy, suggesting that disulfiram may be suitably poised for further drug development under repurposed indications. Established homologs present in humans can be an issue in drug development. That said, the ZnDTC safety profile plus success targeting parasite proteasome (1) provide reasons to be optimistic. In addition, COP9 signalosome inhibitors are being optimized to treat cancer (26), hence one could envision repurposing these drugs for use against parasitic protozoal diseases. The possibility of developing a disulfiram reaction is not anticipated to deter treatment adherence under potential repurposed indication, as there are many other safe and well-established, FDA-approved medications that interact with alcohol, and hence also require abstinence while on treatment without affecting compliance. The commonly prescribed antibiotic metronidazole, for example, which coincidentally belongs to the only drug class available for the treatment of invasive amebic disease, has been reported to lead to the development of an unpleasant and even lifethreatening disulfiram-like reaction when alcohol is concomitantly consumed (42,43). Despite this, metronidazole has become the treatment of choice for several parasitic and anaerobic infections and is used worldwide with good adherence (43).
In conclusion, we provide genetic and chemical evidence for COP9 signalosome as an antiprotozoan drug target. These findings advance in our understanding of parasite protein turnover, and the results can be potentially leveraged to create new therapeutic opportunities to address the unmet medical needs for individuals suffering from these neglected diseases.

Study approval
All animal procedures were approved by the University of Virginia Institutional Animal Care and RRID:Addgene_105768), using BamhI and XhoI sites. The Flag-CSN2 was then sub-cloned into the NheI and SmaI sites of pKT3 vector for its constitutive expression in amoeba. The UPS reporter construct Ub-R-GFP was obtained as a gift from Nico Dantuma (Addgene plasmid # 11938; http://n2t.net/addgene:11938; RRID:Addgene_11938) and sub-cloned into the pKT3 plasmid using the NheI and SmaI sites. The luciferase plasmid pHTP.luc was used to generate the luciferase expressing ameba.
Parasites were transfected using Attractene (Qiagen). The transfectants were selected with hygromycin B for pEhHYG-tetR-O-CAT based plasmids and G418/neomycin for pHTP.luc and pKT3 based plasmids. The initial selections were done at a concentration of 9µg/ml antibiotic and gradually increased up to 20µg/ml except for the pHTP.luc plasmid which was finally selected with 50µg/ml G418. The parasites transfected with pEhHYG-tetR-O-CAT empty or CSN5 constructs (overexpression/sense, knockdown/antisense and D147N mutant) were transfected again with the pKT3-Ub-R-GFP plasmid and dual transfectants were selected with 9µg/ml hygromycin B and G418. The concentration was gradually increased to 40µg/ml for hygromycin B and 50µg/ml G418. Same procedure was followed to generate ameba dually transfected with pKT3-Flag-CSN2 and pEhHYG-tetR-O-CSN5 overexpression plasmids.

Cell viability assays
Parasite growth was measured using CellTiter-Glo® Luminescent Cell Viability Assay kit The effect of CSN5 overexpression and knockdown on drug susceptibility was measured and compared with the empty vector control cells. All three genotypes were induced with 10µg/ml tetracycline 6 hours before plating them for the assay (10,000 cells per well) in media containing 10µg/ml tetracycline and increasing doses of ZnDTC (0, 12.5, 25 and 50 nM, 0.1% DMSO).

Protein degradation studies
Parasites transfected with the UPS substrate Ub-R-GFP protein degradation reporter (44)

Antibody purification
Purified GST-CSN5, GST-cullin1 and MBP-Nedd8 were used to immunize chicken, rabbit and guinea pigs (Cocalico Biologicals), respectively, and pre-immune serum, test bleeds, and the final bleed were received and tested by Western blotting. Sera were cleared of anti-GST or anti-MBP and non-specific antibodies prior to affinity purification with respective antigens. Affinity pull down of specific antibodies were done at 4֯ C overnight, using antigens conjugated to NHSactivated agarose beads (Pierce™ NHS-Activated Agarose Spin Columns, Thermo Fisher Scientific). Antibodies were eluted in 0.1M glycine, pH 2.8 and neutralized with 1M Tris, pH 9.0, and buffer exchanged into PBS with a 30kDa MWCO Centrifugal Filter (EMD Millipore).

Mass spectrometry analysis
Immunoprecipitated proteins from the anti-Flag pulldown and anti-E. histolytica CSN5 were analyzed by mass spectrometry as described previously (16). The samples were processed by the W. M. Keck Biomedical Mass Spectrometry Laboratory at the University of Virginia. Labelfree quantification was performed using MaxQuant (45).

Parasite infection and treatment
Wild-type CBA/J mice at 8-10 weeks of age were obtained from the Jackson Laboratory. Mice were infected by intracecal injection of pHTP.luc plasmid transfected E. histolytica trophozoites (40), and vector expression was maintained with 50 µg/mL neomycin in drinking water .
Treatment began 24 hours after infection (11), confirmed first by in vivo bioluminescent imaging and continued for total 5 days. The treated group were given 50 mg/kg disulfiram (Sigma Aldrich) and 1 mg/kg zinc gluconate (Alfa Aesar) suspended in Ora-plus (Paddock laboratories) via oral gavage. Untreated control groups were given Ora-plus only. Experiments were repeated with disulfiram and zinc gluconate from MedChemExpress and Pure Encapsulations respectively. Zinc gluconate was chosen over other formulations because of its safety profile.
Dosing was based on previous animal studies that simulate the drug concentrations and pharmacokinetics of persons on FDA-recommended doses (32,35,46).

In vivo bioluminescent imaging
Mice infected with luciferase-expressing parasites were injected IP with 150 µL Rediject Luciferin (Perkin Elmer) and imaged by the Xenogen IVIS II System.

Histology and Immunohistochemistry
Mouse cecal tissue, fixed in Bouin's solution (Sigma) and stored in 70% ethanol, was processed and stained with hematoxylin and eosin by the University of Virginia Research Histology Core.
Histological scoring was performed by 2 independent blinded scorers and carried out as previously described (40). Mouse immunohistochemical staining was done using the DAKO Autostainer Universal Staining System with antibody against E. histolytica MIF protein (10).

ELISA
Intestinal tissue lysates were evaluated by ELISA for myeloperoxidase (MPO) (R&D Systems) (40), total protein concentration was measured using the Pierce TM BCA Protein Assay Kit (Thermo Scientific).

Bioinformatic analyses
Amino acids from protozoan CSN5 metalloprotease site and CSN2 PCI domains were aligned by Multiple Sequence Comparison by Log Expectation (MUSCLE) software. Structural modelling of E. histolytica CSN5 protein was done by Protein Homology/Analogy Recognition Engine v 2.0 (PHYRE 2 ).The structures were visualized and analyzed using the UCSF Chimera software v. 1.10.2. Virtual screening was performed with drug molecules for their ability to dock with the CSN5 catalytic site using Autodock 4.2 (47,48).

Statistical analyses
Statistical differences were determined using Fisher's exact test, Mann-Whitney U test and two tailed t test. Pearson's correlation was used for correlation analysis. Nonlinear robust regression analysis was performed on the drug dose response curves. A P value less than .05 was considered statistically significant.