A Novel, Macrophage Migration Inhibitory Factor Suicide Substrate Inhibits Motility and Growth of Lung Cancer Cells

Macrophage migration inhibitory factor (MIF) is a pleiotropic cytokine/growth factor that contributes to inflammatory, autoimmune, and malignant disease pathologies. MIF exerts at least some of its biological actions through binding to its extracellular cognate receptor complex consisting of CD74 and CD44 (1, 2). Studies from this laboratory have revealed a requirement for extracellular MIF in the steady-state activation of Rho GTPase family members, leading to cell growth and migratory phenotypes (3, 4). These MIF-mediated effects are thought to be linked to CD74/CD44-initiated signaling given the fact that CD44-dependent signaling has been shown to promote Rho GTPase family member activation (5).

MIF possesses the unusual ability to catalyze the tautomerization of the nonphysiologic substrates D-dopachrome and L-dopachrome methyl ester into their corresponding indole derivatives (6). The NH₂-terminal proline of MIF (Pro-1) seems to be a critical residue for enzymatic activity, as site-directed mutagenesis that substitutes a serine or glycine for this proline (P1G) is devoid of tautomerase activity (7). The physiologic role of MIFs catalytic activity remains in question, however, because of conflicting reports of loss or gain of biological activity with various proline mutants (7, 8).

Of the MIF enzyme/biological antagonists reported, ISO-1 [S,R-3-(4-hydroxyphenyl)-4,5-dihydro-5-isoxazole acetic acid methyl ester], is the best characterized (9, 10). Although other compounds have been shown to have improved biological inhibitory activities against MIF (11, 12), ISO-1 is considered to be the prototypical MIF antagonist. We and others have reported that this compound is as effective in blocking MIF-dependent malignant phenotypes as MIF knockdown by siRNA (4, 13).

We now report on a unique virtual screening strategy that identified a novel suicide substrate for MIF. Our studies indicate that this compound is ∼5× to 10× more potent than the best known MIF inhibitor in blocking MIF-dependent catalysis, migration and anchorage-independent growth.

Cells and reagents. Human A549 lung adenocarcinoma (American Type Culture Collection) cells were grown in DMEM containing 10% FCS, 2 mmol/L glutamine, and 50 μg/mL gentamicin (Invitrogen). RNA interference transfections, migration assays, and soft agar colony formation experiments were performed as previously described (4). Candidate MIF inhibitor compounds were purchased from Chemnavigator, 4-iodo-6-phenylpyrimidine (4-IPP) was purchased from Specs, 6-phenylpyrimidine was from Sigma-Aldrich, and ISO-1 was a generous gift from Dr. Yousef Al-Abed (Feinstein Institute for Medical Research, Manhasset, NY).

Virtual screening. SDF-formatted files from the ACD (MDL) library were transformed using Genfra (Accelrys), obtaining a library of 343,802 structures suitable for the virtual screening software Ludi (Accelrys). Ludi considers hydrogen bond donors and acceptors and aliphatic nature (variables: 8-Å radius with no linkage, or rotatable bonds, 5,000 fits, 300 lowest hits to be stored, aliphatic_aromatic, and electrostatic_check options). The first three energy functions were used and a manual consensus score ranked the lists.

Enzymatic analyses. MIF enzymatic activity in cell lysates and liver tissue was carried out by lysing 1 × 10⁷ A549 cells or ∼1 gram pieces of liver in PBS containing 1 mmol/L NaVO₄, 2 mmol/L NaF, and a protease inhibitor cocktail (Roche Biochemical) using dounce homogenization on ice. rMIF (14), cell lysate, or liver lysate was added to a final volume of 700 μL PBS in plastic cuvettes. For rMIF inhibition assays, compounds were added to the rMIF/PBS solution and preincubated for 5 min. L-3,4-dihydroxyphenylalanine methyl ester (4 mmol/L) and 8 mmol/L sodium periodate (Sigma-Aldrich) were combined in a 3:2 ratio to form L-dopachrome methyl ester. L-dopachrome methyl ester (300 μL) was then immediately added to the cuvettes, incubated for an additional 5 min to allow color change, which was measured at OD ₄₇₅ nm.

Crystallization and structure determination. Recombinant MIF protein (0.1 mmol/L; ref. 9) was mixed with a 10-fold molar excess of 4-IPP, incubated over night, and concentrated to 1 mmol/L protein with a Centricon 3000 M.W. centrifugal filter unit (Millipore). The complex was crystallized using the hanging drop vapor-diffusion method at 20°C. Equal volumes of the MIF inhibitor complex and reservoir solution [2 mol/L ammonium sulfate, 4% isopropanol, and 0.1 mol/L tris (pH 7.5)] were mixed to form 2 μL drops. Crystals belonging to space group P2₁2₁2₁ formed within 2 wk. Data were collected in house using a Rigaku RAXIS-IV imaging plate area detector. The crystal was cryoprotected in the reservoir solution containing 25% glycerol. Diffraction data were collected at −180°C, and HKL2000 (15) was used to integrate and scale the diffraction data to 1.8 Å. The structure was solved by molecular replacement with Phaser (16) using the structure of MIF in complex with Oxim-11 as search model (11). Refinement was performed using CNS and Refmac (17, 18), and manual manipulation of the model was done using MIFit 4.1. The structure is deposited with the RCSB Protein Data Bank and has been assigned the RCSB ID code RCSB045257 and PDB ID code 3B9S.

Mass spectroscopy. MIF/4-IPP adducts were analyzed by liquid chromatography/mass spectrometry (LC-MS). Samples were run on a C18 (Waters) column attached to an HP1100 LC with auto sampler and Diode Array UV detector (Agilent Technologies) coupled to a Micromass Quattro LCZ (Waters Corp.) triple quadrupole mass spectrometer, using the electrospray ion source. The LC method used a reversed phase gradient, 10% to 60% acetonitrile, and the mass spectrometer was set up in the electrospray positive-ion mode to scan over the range M/z 200 to 1,000. LC peaks containing consecutive M/z scans were combined to produce mass spectra of the analyte, consisting of a series of multiply-charged ions of the intact protein. Spectra were further analyzed using MaxEnt1 feature of MassLynx (Waters Corp.) to determine mass of the proteins from the multiply-charged series of ions.

Virtual screening against Met-2 of MIF reveals a novel small molecule inhibitor of MIF. As opposed to prior virtual screens run against MIF that targeted the NH₂-terminal proline, our computational screening strategy focused on targeting methionine at position A2 (A2 refers to monomer A, position 2 from the trimeric crystal structure of MIF). We chose this strategy because (a) Met-2 resides at the base of the hydrophobic binding pocket, whereas the NH₂-terminal proline is situated on the side of the pocket (19); and (b) prior studies have shown that disrupting this hydrophobic substrate–binding pocket by insertion of a single amino acid residue adjacent to Met-2 leads to a complete loss of enzymatic and biological activity (9).

Our initial virtual screen used the crystal structure of MIF (PDB-1MIF) and the Available Chemicals Directory library, the results of which provided a ranking list of the top 100 compounds. Of the 76 commercially available compounds obtained from this list, only 41 were found to be soluble in aqueous solutions at 100 μmol/L concentrations. Of these, 9 were found to be inhibitory at concentrations of 50 μmol/L or less to the catalytic activity of purified, recombinant MIF (22.0% success rate; 9 of 41). When these compounds were tested against MIF-dependent catalytic activity in whole cell lysates, we found that only three compounds remained inhibitory at similar IC₅₀s. Of these three remaining compounds, only one was found to exhibit superior MIF catalytic inhibitory activity versus ISO-1: 4-IPP (Fig. 1A). 4-IPP and ISO-1 were tested for their relative inhibitory effects against the catalytic activity of recombinant human MIF. As shown in Fig. 1B, the IC₅₀ of 4-IPP is 10 times lower than that of ISO-1.

4-IPP is a suicide substrate for MIF. Because the kinetics observed with 4-IPP against MIF catalysis were reminiscent of an irreversible inhibitor (data not shown), we next tested MIF/4-IPP admixtures to determine whether any shifts in the molecular weight of MIF were observable. As shown in Fig. 2A (top), LC-MS analysis revealed that MIF coincubated with 4-IPP resulted in a mass increase of 154 units. Coincubation of 4-IPP with a catalytically inactive mutant of MIF (point mutation of Pro-1 to Gly-1; ref. 7), resulted in no covalent modification as assessed by mass spectroscopy (Fig. 2A , bottom).

Cocrystallization of MIF and 4-IPP reveals 6-phenylpyrimidine covalently modified catalytic proline. Because 4-IPP is M.W. 282 and the iodine at position 4 of the pyrimidine ring can act as a leaving group, we suspected that the increase in mass of 154 units observed with wild-type rMIF/4-IPP (but not catalytically inactive mutants) was due to a covalent modification of rMIF by deiodinated 4-IPP giving 6-phenylpyrimidine–modified MIF (M.W. 156; with covalent attachment, M.W. 154). To test this, we performed cocrystalization of MIF/4-IPP admixtures. As shown in Fig. 2B, 4-IPP binds covalently to MIF at the nitrogen of the NH₂-terminal proline by substitution of the iodo group at the 4-position of 4-IPP (data collection details are found in Supplementary Table S1). The phenyl substituent is oriented deep in the MIF catalytic pocket near Met-2 (Fig. 2C). This is similar to existing crystal structures of MIF with antagonists 7-hydroxy-2-oxo-chromene-3-carboxylic acid ethyl ester (20) and OXIM6 (PDB entries 1GCZ and 2OOZ, respectively; ref. 11), although the phenyl substituent of bound 4-IPP is located slightly deeper into the binding pocket and in a different orientation. There is some apparent freedom of rotation of the phenyl ring as two different conformations exist in the trisubstituted trimeric MIF crystal structure. The major interaction partners for bound 4-IPP within 2.5 Å are MET A2,TYR A36, HIS A62, VAL A106, PHE A113, TYR B95, and ASN B97.

Inhibition of MIF-dependent migration and anchorage independence by 4-IPP. Studies from this (4) and other laboratories (13) reveal that ISO-1 effectively inhibits MIF-dependent cell migration and anchorage-independent growth in human cancer cell lines. As such, we tested 4-IPP against ISO-1 and determined the relative inhibitory effects on the migration of A549 lung adenocarcinoma cells on collagen substrates. Consistent with our recent findings, MIF siRNA strongly reduced the migration of A549 cells (Fig. 3A; ref. 4). Similar to MIF siRNA, pretreatment of A549 cells with either ISO-1 or 4-IPP at 100 μmol/L reduced A549 motility by >70%. In contrast, 10 μmol/L (Fig. 3A) and 50 μmol/L (data not shown) ISO-1 had no effect on A549 migration, whereas 10 μmol/L 4-IPP attenuated migration by >50% that of vehicle alone (Fig. 3A). To ensure that this effect was not restricted to A549 cells, H1299 and H23 NSCLC cell lines showed a similar effect with 4-IPP albeit more pronounced (Supplementary Fig. S1). Interestingly, both of these cell lines were largely resistant to ISO-1 at 100 μmol/L, whereas 4-IPP was maximally effective at this concentration. As MIF is thought to promote non–small cell lung cancer (NSCLC) motility in a Rac1-dependent fashion, 4-IPP was tested for its ability to block Rac1 binding to the PAK-binding domain of Pak (PBD; ref. 4). As shown in Fig. 3B, 4-IPP reduced Rac1 binding to PBD similar to what we have observed with siRNA in the same cells.

We next tested and compared the relative efficacies of ISO-1 and 4-IPP to inhibit lung adenocarcinoma anchorage-independent growth. As shown in Fig. 3C, both ISO-1 and 4-IPP inhibited A549 soft agar growth, but 4-IPP was between 5- to 10-fold more potent (similar to IC₅₀ data from Fig. 1A) than ISO-1.

We then tested in vivo inhibition of tissue-associated MIF by 4-IPP. Injections of 4-IPP (1 mg; ∼50 mg/kg) inhibited liver MIF enzyme tautomerase activity at 6 hours or for 3 and 7 days (daily injections of 1 mg 4-IPP) by >50% (Fig. 3D). Importantly, we observed no weight loss or obvious toxicity with 1, 2, or 4 mg daily i.p. injections (50, 100, and 200 mg/kg, respectively) of 4-IPP over the course of 7 days (Fig. 3D; data not shown). Although longer treatment regimens remain to be tested, acute toxicity of 4-IPP seems to be nominal.

To identify more potent analogues of our lead MIF antagonist, we performed an in silico combinatorial docking screen using the cocrystal structure of the MIF:4-IPP complex defining MIF as the receptor and the phenylpyrimidine rings of 4-IPP as the core. Four unique congeners of 4-IPP were predicted to have substantially increased binding potential within the MIF substrate binding pocket. We subsequently synthesized the four analogues and, as predicted, the IC₅₀s of all four individual analogues were between ∼10× and 20× more active (IC₅₀s: A1, ∼200 nmol/L; A2, ∼275 nmol/L; A3, ∼400 nmol/L; A4, ∼475 nmol/L) than parental 4-IPP in blocking MIF-dependent catalysis (Fig. 4A). Finally, to determine if enhanced enzyme inhibitory activity correlates with enhanced biological inhibitory activity, we tested the most potent of the four analogues against 4-IPP in an in vitro migration assay. As shown in Fig. 4B, A1-4-IPP was substantially more potent in blocking migration than our lead parent compound, 4-IPP.

Using a novel virtual screening strategy, we have discovered a family of suicide substrates of MIF that covalently modify the NH₂-terminal, catalytically active proline of MIF. Our results indicate that 4-IPP is dehalogenated upon contact with enzymatically active, but not inactive, MIF, and a subsequent covalent bond then forms between C-4 of pyrimidine and the NH₂-terminal nitrogen of Proline at position 1. Importantly, this covalently modified MIF is rendered inactive in both its catalytic and biological functions by the suicide substrate, 4-IPP. Our studies further indicate that 4-IPP treatment inhibits the migration and anchorage independence of human lung adenocarcinoma cell lines in vitro and MIF-associated liver enzyme activity when administered in vivo.

Several small molecule inhibitors of MIF have been described since enzymatic activity of MIF was first discovered (9, 11, 12, 20). Orita and colleagues (20) first described virtual screening as a tool to identify novel small molecule antagonists of MIF. Using the catalytically active NH₂-terminal proline as a target, the authors discovered that SBDD-identified MIF inhibitory compounds could be classified into four independent groups. Despite excellent IC₅₀ activities against MIF catalysis of lead compounds identified from this study, no functional testing of these inhibitors has been reported to date.

Our group and others have recently reported that ISO-1 blocks the migration, invasion, and anchorage-independent growth of human cancer cell lines (4, 13). As the MIF receptor CD74 has been found to be necessary for antitumor activities of ISO-1, it is likely that 4-IPP and its more active analogues similarly block MIF/receptor interactions albeit more efficiently than ISO-1.

In summary, we have identified a unique family of small compound inhibitors of MIF that act as suicide substrates to irreversibly inhibit MIF-dependent tumor cell invasive and anchorage-independent properties. Although further studies are needed to elucidate the antitumor efficacy of this novel lead compound, we propose that 4-IPP represents a unique class of anti-MIF therapeutic antagonists.

R.A. Mitchell, J.O. Trent, J. Meier, and R. Riggs are co-inventors on patents and patent applications describing the therapeutic value of 4-IPP and 4-IPP analog MIF antagonists. The other authors disclosed no potential conflicts of interest.

Grant support: NIH 5P20RR018733 (R.A. Mitchell, J.O. Trent), NIH CA 102285 (R.A. Mitchell), and a grant from Philip Morris USA Inc. and Philip Morris International (R.A. Mitchell).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.