Interaction between parasite-encoded JAB1/CSN5 and macrophage migration inhibitory factor proteins attenuates its proinflammatory function

Multiple protozoans produce homologs of the cytokine MIF which play a role in immune evasion, invasion and pathogenesis. However, how parasite-encoded MIF activity is controlled remains poorly understood. Cytokine activity can be inhibited by intracellular binding partners that are released in the extracellular space during cell death. We investigated the presence of an endogenous parasite protein that was capable of interacting and interfering with MIF activity. A screen for protein-protein interaction was performed using immunoaffinity purification of amebic cell lysate with specific anti-Entamoeba histolytica MIF (EhMIF) antibody followed by mass spectrometry analysis, which revealed an E. histolytica-produced JAB1 protein (EhJAB1) as a potential binding partner. JAB1 was found to be highly conserved in protozoans. Direct interaction between the EhMIF and EhJAB1 was confirmed by several independent approaches with GST pull-down, co-immunoprecipitation, and Biolayer interferometry (BLI) assays. Furthermore, the C-terminal region outside the functional JAMM deneddylase motif was required for EhMIF binding, which was consistent with the top in silico predictions. In addition, EhJAB1 binding blocked EhMIF-induced IL-8 production by human epithelial cells. We report the initial characterization of a parasite-encoded JAB1 and uncover a new binding partner for a protozoan-produced MIF protein, acting as a possible negative regulator of EhMIF.


Results
Identification and characterization of a parasite-produced JAB1. Co-immunoprecipitation (co-IP) of proteins followed by mass spectrometric identification is a standard approach for identifying novel protein-protein interactions 20,21 . Parasite cell lysates were incubated with a specific antibody against Entamoeba histolytica MIF (EhMIF) or IgG control antibody. The immunoprecipitates were subjected to mass spectrometry analysis. An Entamoeba histolytica-encoded JAB1 (EhJAB1, EHI_050500) was identified as a potential EhMIF binding partner (Fig. 1A,B). While a number of putative interacting proteins were co-precipitated with anti-Eh-MIF antibody, EhJAB1 was selected for further analysis given: (i) EhJAB1 was detected in the anti-EhMIF co-IP sample and not detected in the control IP; (ii) EhJAB1 was among the highest percent coverage (Supp. Fig. S1); and (iii) previous report of the intracellular interaction between human MIF and human JAB1 (HuJAB1) [22][23][24] . EhJAB1 gene consists of 957 base pairs with no intron and a GC content of 30%. The gene encodes for a protein of a predicted molecular weight of 36.5 kDa.EhJAB1 has 39% identity and 61% similarity with HuJAB1 and appears structurally conserved (Figs 1C and S2). JAB1 is conserved in multiple parasites that cause disease in humans, all containing the MPN domain with the JAMM (JAB1/MPN/Mov34 metalloenzyme) motif (Fig. 1D). JAMM motif is a metalloprotease motif, consisting of the amino acid sequence EX n HXHX 10 D, present in the deubiquitinating enzymes RPN11 and JAB1 25 . RPN11 was also found is these parasites (Supp. Fig. S3).

EhMIF directly interacts with EhJAB1. We investigated whether the interaction between EhMIF and
EhJAB1 was direct or indirect. First, we performed GST-pulldown assays ( Fig. 2A). EhMIF and EhJAB1 fused to GST were expressed and purified (Fig. 2B). GST-EhJAB1 immobilized on magnetic beads was incubated with EhMIF. GST alone immobilized on magnetic beads and incubated with the same concentration of EhMIF was used as control. As shown in Fig. 2C, the GST-EhJAB1 fusion protein readily pulled down EhMIF relative to the GST control, demonstrating direct interaction. In an independent approach, we used co-immunoprecipitation experiments to further verify direct interaction. EhJAB1 immunoprecipitated with anti-EhMIF antibody, but not (A) Schematic of procedure for identifying novel protein-protein interaction with co-immunoprecipitation followed by mass spectrometric analysis. (B) E. histolytica JAB1 (EhJAB1) amino acid sequence. Peptides unique to the EhJAB1 protein identified by mass spectrometry are highlighted (blue). JAMM motif (EX n HXHX 10 D) shown in red box. (C) Structural homology between Human JAB1 (HuJAB1) and EhJAB1. HuJAB1 (orange) was superimposed with the predicted structure of EhJAB1 (blue). (D) Multiple sequence alignment of the conserved JAMM motif of JAB1 from pathogenic parasites. Identical (green), conserved (blue), semi-conserved (pink), and non-conserved residues (red).
with the IgG antibody control (Fig. 2D-F). Together, these findings suggest a direct binding between EhMIF and EhJAB1 proteins.

High-affinity binding between EhMIF and EhJAB1 proteins. Biolayer interferometry (BLI) is a useful
technique for measuring interactions between proteins in real time 26 . We determined the equilibrium dissociation constant for EhMIF binding to EhJAB1 using BLI. GST-tagged EhJAB1 was coupled to the surface of anti-GST antibody coated BLI sensors, followed by binding measurements in different concentrations of EhMIF. Analysis revealed a dissociation constant of K D of 3.86 × 10 −8 M (Fig. 3A). GST alone coupled to anti-GST antibody coated BLI sensors was used as control. BLI measurements demonstrate that EhMIF did not bind to the GST control, K D not applicable (Fig. 3B). These findings suggest EhMIF binds to EhJAB1 within the range considered biologically relevant 27,28 . EhJAB1 interacts with EhMIF via its C-terminal domain. We investigated the binding domain of EhJAB1 responsible for interacting with EhMIF. First, an in silico approach using pyDockWEB was applied to assess the binding of EhJAB1 to EhMIF. pyDockWEB allows the best rigid-body docking orientations generated by pyDock scoring function, which consists of electrostatics, desolvation energy and limited van der Waals contribution 29 . The C-terminal domain of EhJAB1 was revealed to be the EhMIF binding region in all top ten predictions (Figs 4A and S6). Next, we further examine the region of EhJAB1 that is responsible for the formation of the complex with EhMIF using deletion mutant analysis. GST-tagged full-length EhJAB1 and various GST-tagged deletion mutants of EhJAB1, with either N-terminal, MPN, JAMM or C-terminal domain deletion (Fig. 4B,C), were immobilized on magnetic beads and incubated with EhMIF protein. EhMIF specifically interacted with GST-EhJAB1, except for the mutant lacking the C-terminal domain. Interaction was regained when the 187-246 amino acid sequence of the C-terminal was restored (Fig. 4D). These results indicate binding of EhJAB1 to EhMIF is mediated through the C-terminal region outside of the MPN domain.
EhJAB1 inhibits EhMIF interaction with the CD74 receptor and EhMIF-induced cytokine production. CD74 is a cell surface receptor for human MIF, which mediates many of its inflammatory effects 30 .
Several parasite MIF homologs, including EhMIF, were previously shown to interact with the CD74 receptor 13,17,18,31 . Here, we were able to reproduce this finding and found that EhMIF-CD74 interaction was blocked by preincubating EhMIF with EhJAB1 (Fig. 5A). The epithelial surfaces of the skin, nasal, intestinal, respiratory, and genitourinary tracts are the first points of contact for many protozoans. Epithelial cells express CD74 and are a rich source of IL-8 32 . Stimulation of IL-8 production has been the most reproducible activity of protozoan MIF homologs. EhMIF was recently shown to stimulate IL-8 secretion from human intestinal epithelial cells 19 . IL-8 is a potent neutrophil chemoattractant that contributes to inflammation in various infectious and inflammatory diseases. We proceeded to examine if EhJAB1 binding to EhMIF attenuates its proinflammatory function using two different colonic epithelial cell lines. EhJAB1 blocked EhMIF-induced IL-8 secretion by HCT116 and Caco2 colonic cells (Fig. 5B,C). These data indicate that when bound to EhJAB1, EhMIF is no longer capable of carrying out its inflammatory functions.

Discussion
Parasites can exploit the host inflammatory response to promote tissue invasion 1 . EhMIF-stimulated human intestinal epithelial cells secrete IL-8, a potent promoter of inflammation 19 . One of the downstream effects of E. histolytica MIF-induced inflammation is an increase in matrix metalloproteinases (MMPs) production, which was shown recently to promote E. histolytica tissue invasion in human colon 19,33 . That said, parasites are likely to have developed mechanisms to regulate their MIF actions especially in situations where they fail to evade the host inflammatory response induced by producing such a molecule.
Identifying a binding partner that inhibits MIF function provides insight into how MIF's actions are regulated 27 . In this study, a proteomics approach was used to identify parasite-encoded protein interaction partners of a homolog of MIF from Entameoba histolytica. We found that EhJAB1 protein interacts with EhMIF. This interaction was validated by multiple independent approaches with GST-pulldowns, co-immunoprecipitation, in silico experiments and Biolayer interferometry. JAB1, also known as COP9 signalosome subunit 5 (CSN5), is well characterized in non-parasitic eukaryotes 34 . It constitutes the catalytic center of the large multi-protein COP9 signalosome complex, since it harbors the JAMM/MPN + metalloprotease motif [35][36][37] . It carries out a Zn dependent reaction called deneddylation, equivalent to de-ubiquitylation, where it removes a ubiquitin (Ub) like protein Nedd8 from the Cullin-RING E3 Ub-ligases (CRL) 35,38,39 . Regulation of CRLs, such as the Skp/Cullin/F-box (SCF) complex, by CSN mediated deneddylation is critical for proper cell division, cell cycle control and DNA damage response 40,41 . Here, we report the initial characterization of a parasite-encoded JAB1. Given parasites including E. histolytica express cullins and Nedd8 proteins [42][43][44][45] , it is plausible that JAB1 in parasites regulate cullins by deneddylation and modulate ubiquitin proteasomal system (UPS) activity, however, further studies to confirm this are needed. Surprisingly, we did not find a JAB1/CSN5 homolog in Plasmodium genome. However, the incomplete genome assembly and annotation in Plasmodium spp. could limit our in-silico analysis and explain why JAB1 was not identified in Plasmodium spp.
JAB1 function is not limited to CSN dependent deneddylation. JAB1 is also stable and functional in a free, monomeric, CSN independent form [46][47][48] . JAB1 monomers are catalytically inactive, but bind to certain proteins and alter their activity 34,48 . We showed that the JAMM catalytic motif of EhJAB1 was not required for its interaction with EhMIF, which is consistent with the previous report on mammalian proteins 23 . The dissociation constant for EhJAB1-EhMIF complex was well within the range considered physiologically relevant, suggesting a biologically significant interaction 27,28 .
Intracellular molecules released from damaged cells can have profound effects on the immune response. There is growing evidence that not all of these immunomodulatory molecules are pro-inflammatory, and interestingly, some have immunosuppressive or regulatory effects. These anti-inflammatory actions occur either directly or indirectly. For example, in the case of human MIF, an endogenous binding partner that is released in the extracellular space during cell death interacts and negatively interferes with MIF activity 27,49-51 . JAB1 is an intracellular protein 22 , which is consistent with our mass spectrometric analysis of the parasite cytosolic fraction; while MIF homologs, such as EhMIF are secreted proteins 18,19 . E. histolytica parasite has developed a number of mechanisms to evade the host immune response 1,52 . However, amebic parasites could become damaged if they fail to evade the inflammatory response triggered by EhMIF. It would be reasonable to speculate that the free monomeric EhJAB1 released from injured cells into the extracellular environment would then form complexes with EhMIF, preventing interaction with the host receptor CD74 and reducing EhMIF-induced inflammation, as a negative feedback mechanism. In addition, given the structural and functional similarity between human and E. histolytica MIF and JAB1 proteins, we do postulate that EhJAB1 could also interact with human MIF. It would be interesting to determine whether an EhJAB1/human MIF interaction functions to evade the host immune response in future studies.
In conclusion, our data suggest that E. histolytica homolog of JAB1 interacts with the cytokine EhMIF and negatively regulates its pro-inflammatory function. Our study also generates the hypothesis that targeting EhMIF and its interaction with EhJAB1 may disrupt the parasite's ability to exploit the host immune response and ultimately serve as a therapeutic target against this devastating parasitic disease.

Methods
Plasmids, cloning and PCR. The EhJAB1 gene, codon optimized for expression in E.coli BL21(DE3) cells, was cloned within pDEST15 vector. Deletion constructs were prepared by inverse PCR on the pDEST15-full length JAB1 clone using the primers listed in the Supp. Fig. S4. A schematic of the mutation strategy is provided in the Supp. Fig. S5. Clones were screened by PCR across the gene boundaries within the vector followed by confirmation with sequencing. The EhJAB1 gene was amplified from the pDEST15 vector with primers carrying 5′BamHI and 3′XhoI sites and sub-cloned within BamHI and XhoI sites of pGEX-4T1 vector to utilize the thrombin site for cleaving off the GST tag from the GST-EhJAB1 protein. Also, the previously described codon optimized EhMIF gene cloned within pJexpress414 vector (DNA2.0) 17   The purification of GST-fusion protein and His-tagged fusion protein was done as previously described 17 using glutathione-sepharose (GE-Healthcare) and Ni-NTA agarose beads (Qiagen) respectively. Polymyxin B (Sigma) was used in the purification procedures for the removal of endotoxin. On-column cleavage of the GST-EhJAB1 protein was done with the Thrombin Cleavage Capture Kit (Millipore) that utilizes biotinylated thrombin for its streptavidin agarose based removal.
GST-Pulldown Assay. GST pull down assays were done using the MagneGST TM Protein Purification System kit (Promega). Bacterial lysate expressing GST fused-full length or deletion mutant EhJAB1 protein, diluted in MagneGST binding/wash buffer to a concentration 500 ng/ml in 500 µl, was mixed with 25 µl magnetic beads and was incubated overnight at 4 °C with rotation. Beads were washed 3 times with 500 µl binding/wash buffer followed by incubation with 1 µg purified EhMIF at 4 °C for overnight. The GST-CD74 pulldown assays were performed as previously described 17,31 . Briefly, EhMIF protein was pre-incubated with or without 1X, 2X or 5X concentration of EhJAB1 protein for 30 minutes in 500 µl binding/wash buffer prior to incubation with the GST-CD74 bound magnetic beads at 4 °C for overnight. The bound complexes were eluted in 40 µl elution buffer after washing the beads 5 times with 500 µl binding/wash buffer. In a control experiment equivalent amount of GST protein was used in place of GST-EhJAB1 or GST-CD74 proteins. Twenty five percent of the eluted samples were assayed by immunoblots. Immunoprecipitation. Approximately 1.25 × 10 7 trophozoites of E. histolytica, grown at 37 °C in TYI-S-33 medium, were harvested in 3 ml non-denaturing lysis buffer (20 mM Tris HCl pH 8, 137 mM NaCl, 1% TritonX-100, 2 mM EDTA, 1x Protease inhibitor freshly added). Lysis was done with sonication followed by centrifugation at high speed for 10 minutes at 4 °C. Immunoprecipitations were performed with anti-EhMIF rabbit serum 17 and rabbit IgG (Santa Cruz Biotechnology). For each, 100 µl Dynabeads TM Protein A (Invitrogen) was chemically conjugated with 10 µg antibodies using 5 mM Bis[Sulfosuccinimidyl] suberate or BS 3 (Thermo Scientific) following the manufacturer's manual. Antibody conjugated beads were added to 500 µl amebic lysate and incubated overnight at 4 °C with rotation. Following binding, beads were washed 5 times with 300 µl wash buffer (10 mM Tris HCl pH 7.4, 150 mM NaCl, 1% TritonX-100, 1 mM EDTA, 1x Protease inhibitor freshly added). The bound complexes were eluted in 40 µl 1x Laemmli sample buffer (Bio-Rad) by boiling. Co-precipitated endogenous proteins were analyzed by Mass-Spectrometry. For direct binding experiment, tag free EhJAB1protein was biotinylated using the EZ-Link R Sulfo-NHS-LC-Biotinylation kit (Thermo Scientific). Immunoprecipitation was performed with 50 µl dynabeads bound to 10 µg anti-EhMIF rabbit serum or rabbit IgG. For each set, 1.5 µg of EhMIF mixed with 500 ng biotinylated EhJAB1protein was incubated with the antibody coated beads in 300 µl binding buffer (as described above) for overnight at 4 °C with rotation. Wash and elution of the bound complexes were done the same way as above. The immunoprecipitation was repeated in the presence of 5 µg and 10 µg non-biotinylated tag free EhJAB1 protein. Twenty-five percent of the eluted samples were assayed by immunoblots.
Mass Spectrometry. The entire eluted fraction from anti-EhMIF and rabbit IgG immunoprecipitations were subjected to electrophoresis on SDS polyacrylamide gel. Proteins were separated via SDS-PAGE for a length of 1 cm 53 . About 1 cm × 1 cm section of the gel, spanning all of the resolved proteins in each well, were excised. The gel samples were submitted to the W. M. Keck Biomedical Mass Spectrometry Laboratory for mass spectrometry analysis.

Immunoblotting. Protein samples obtained from GST-pulldown and immunoprecipitation experiments
were resolved by SDS-PAGE followed by transfer onto polyvinylidene difluoride membranes (Millipore). The membranes were incubated overnight with primary antibodies at 4 °C. For EhMIF protein detection, rabbit anti-EhMIF antibody was used followed by anti-rabbit IgG HRP conjugate (Sigma) secondary antibody. For biotinylated EhJAB1, goat anti-biotin HRP conjugated antibody (Cell Signaling Technology) was used. Enhanced chemiluminescence (Thermo scientific) based substrates were used to detect antibody conjugated peroxidase activity.

IL8 secretion assay by ELISA. The human colonic epithelial cells (HCT116 & Caco2, American Type
Culture Collection) with densities 10 6 cells/ml were cultured in 48 well plate (Corning) with 100 µL complete media (Dulbecco's Modified Eagle Medium, Gibco) for 12 hours followed by gentle washing and incubation with 100 µl serum free media for 12 hours. After washing the plates with 200 µl media, cells were treated with 0.5 µg/ml EhMIF in presence or absence of 0.5, 1 and 2.5 µg/ml EhJAB1 protein for 8 hours. IL-8 in cell culture supernatant was measured by enzyme linked immunosorbent assay (ELISA, eBioscience).
Binding kinetics using BLI assay. The binding affinities between GST-EhJAB1 and EhMIF proteins were measured using the Blitz System (Octet R Red 96 system, ForteBio). Briefly, Anti-GST Dip and Read TM Biosensors (ForteBio) were hydrated for 10 minutes in the sample dilution buffer (1x DPBS, 0.1% BSA, 0.02% Tween 20) followed by 3 cycles of priming and neutralization, 20 seconds each, in regeneration buffer (10 mM Glycine, pH 1.7) and the sample dilution buffer, respectively. Next, baseline stabilization of the primed biosensors was done in the dilution buffer for 5 minutes. Then, 10 µg GST-JAB1 or GST proteins diluted in sample dilution buffer was loaded onto the biosensors for 5 minutes. After washing the loaded biosensors in sample dilution buffer for 5 minutes, they were exposed for 10 minutes to 5, 2-fold dilution series of EhMIF protein starting at 80 nM concentration. Post-binding dissociation of EhMIF was done for 10 minutes in the sample dilution buffer. Binding affinities (K D ) were calculated using the Blitz system software (ForteBio).

Bioinformatics.
Orthologues of JAMM/MPN + motif of the EhJAB1 protein from different protozoan parasites were aligned by Multiple Sequence Comparison by Log Expectation (MUSCLE) software 54 . 3D structure of EhJAB1 was constructed by Protein Homology/Analogy Recognition Engine v 2.0 (PHYRE 2 ) 55 . The predicted EhJAB1 structure was then compared with that of the Human JAB1 protein using the UCSF Chimera software v. 1.10.2. The protein-protein interaction between EhJAB1 and EhMIF was examined and potential docking interfaces were predicted using pyDockWeb that applies a rigid body protein-protein docking prediction model by electrostatic and desolvation scoring 29 . The predictions were viewed using the UCSF Chimera software v. 1.10.2.
Statistical Tests. Statistical differences were determined using ANOVA followed by Dunnett's post-hoc test.
A p value less than 0.05 was considered statistically significant.