Selective Targeting of a Disease-Related Conformational Isoform of Macrophage Migration Inhibitory Factor Ameliorates Inflammatory Conditions

Macrophage migration inhibitory factor (MIF) is a proinflammatory cytokine with a pleiotropic spectrum of biological functions. One particular hallmark of MIF is its ability to override immunosuppressive effects of glucocorticoids, thereby promoting and aggravating local and systemic inflammatory responses mediated by macrophages and monocytes (1–3). Furthermore, MIF upregulates inflammatory responses by inducing the expression of other proinflammatory mediators, such as TNF-α (4), NO (5), and PGE₂ (3), and by enhancing chemotaxis and macrophage infiltration (6). In addition, MIF inhibits tumor suppressor p53 (7) and activates the central kinases Akt/PKB and Erk that positively affect survival of immune cells, therefore contributing to the maintenance of inflammatory processes (3, 8). Levels of MIF are elevated in many inflammatory conditions, such as rheumatoid arthritis (9), asthma (10, 11), sepsis (12), psoriasis (13), inflammatory bowel disease (IBD) (14, 15), systemic lupus erythematosus (SLE) (16), and cancer (17). Experiments showed that MIF-deficient mice have reduced susceptibility to experimental sepsis (18), arthritis (19), IBD (15), and organ injury caused by SLE (20), suggesting that MIF is a key player in the development of specific diseases. Moreover, animals benefited from treatment with neutralizing polyclonal or monoclonal anti-MIF Abs in models of experimental sepsis or septic shock (21), delayed-type hypersensitivity (22), arthritis (23), IBD (15), and other diseases (24).

In contrast with other proinflammatory cytokines, MIF is constitutively expressed by numerous cell types, has a wide tissue distribution (25–27), and is present in the circulation of healthy subjects in the range of 1–15 ng/ml (21, 28, 29). MIF presents a ubiquitous but challenging therapeutic target because it requires the functional neutralization of a protein that is constantly present in healthy and diseased subjects.

We recently described the fully human mAbs BaxB01, BaxG03, and BaxM159 that effectively inhibit the biological activity of MIF in vivo and in vitro (30, 31). These Abs are specific for two distinct epitopes (aa 50–68 or 86–102), which form a β-barrel structure containing the highly conserved ₅₇Cys-Ala-Leu-Cys₆₀-motif (30); this forms the catalytic motif of the thiol protein oxidoreductase linked to MIF’s biologic function (32–34). After more in-depth investigation, we found that this set of Abs is specific for a previously unrecognized, redox-dependent isoform of MIF that we designated as oxidized MIF (oxMIF). OxMIF is selectively expressed under proinflammatory conditions in patients and animals. The outcomes of our in vivo studies indicate that oxMIF represents a disease-related form of MIF that can be specifically targeted for anti-inflammatory therapy by mAbs.

BaxB01, BaxG03, BaxM159, BaxH02, and an isotype-matched human control Ab were produced as described previously (30). Polyclonal rabbit anti-MIF Abs were generated by immunization with recombinant MIF and purified from serum over protein A columns (GE Healthcare) and immobilized recombinant MIF. Recombinant MIF was expressed in Escherichia coli and purified as described previously (30). OxMIF was generated either by adding saturated l-cystine solution (Sigma) to recombinant MIF and subsequent dialysis against PBS or by adding oxidized glutathione (Sigma).

Plasma samples from healthy subjects (control donors) were collected from volunteers in a normal state of health, with no apparent signs of disease. Plasma samples from sepsis and asthma patients were purchased from Tissue Solutions. Blood samples for FACS analysis were collected from patients with bacteremia at the general hospital (Allgemeines Krankenhaus, Vienna, Austria). Plasma samples from psoriasis patients were purchased from Asterand. Urine and plasma samples from SLE patients were collected at the Wilhelminenspital (Vienna, Austria). Plasma samples from IBD patients and biopsies for colonic explant cultures were collected from ulcerative colitis (UC) and Crohn disease patients at the Medical University Innsbruck. Biopsies were taken from areas showing endoscopic signs of high inflammatory activity (lesional) and from endoscopically normal-appearing, noninflamed mucosa (nonlesional) (35). The control biopsies were obtained from patients without signs of intestinal inflammation. Colonic biopsies were cultivated for 24 h, and supernatants were harvested for MIF measurements. All patients and control donors signed an informed consent for sample collection. Blood sample collection from patients with bacteremia was approved by the Ethics Committee from the Medical University Vienna, Austria. Urine and plasma sample collection from SLE patients was approved by the hospital ethics committee of the Wilhelminenspital. The IBD study was approved by the Ethics Committee from the Medical University Innsbruck, Austria.

To induce acute colitis, we injected i.p. an agonistic mAb directed against mouse CD40 (FGK45 IgG2a; BioXCell) or an isotype control into recombination activating gene (Rag)-1^−/− C57BL/6 mice (7–9 wk, male; Taconic). Anti-oxMIF Abs (or isotype control) were administered at the day of disease induction by anti-CD40 Abs and at day 1 post anti-CD40 treatment. Mice were culled on day 7.

Spontaneous chronic enterocolitis of C57BL/6 IL-10^−/− mice (11 wk, male; Jackson Laboratories) was synchronized by placing the mice on a piroxicam-containing diet for 2 wk followed by a piroxicam-free diet for an additional 2 wk. On day −1 and twice per week for 4 wk, mice were injected i.p. with anti-oxMIF Abs. All mice were culled for tissue harvest after 4 wk. Disease activity indices and histology index were assessed by established scoring systems; disease activity index parameters: hunching, wasting, stool consistency, colon thickening; and histology index parameters: inflammation, hyperplasia/goblet cell depletion, epithelial damage/crypt destruction, crypt abscess. Levels of murine MCP-1, TNF-α, and IL-6 were determined by cytometric bead array (BD Biosciences).

Mouse colitis models were conducted according to U.S. National Institutes of Health (NIH) standards (Guide for the Care and Use of Laboratory Animals, Department of Health and Human Services publication no. NIH 78-232, revised 1985). NIH “Principles for the Use of Animals” were mandatorily accepted. The studies were conducted in complete compliance with the “Public Health Service Policy on Humane Care and Use of Laboratory Animals.”

To induce crescentic glomerulonephritis in rats, we prepared nephrotoxic serum targeting the rat glomerular basement membrane. In brief, New Zealand White rabbits were immunized s.c. with 1 mg rat glomerular basement membrane in 1 ml CFA (Sigma). The rabbits were boosted every 2–3 wk with incomplete Freund’s adjuvant. Serum was collected and complement was inactivated by heating to 56°C for 30 min. Anti-rat erythrocyte Abs were removed by absorption with normal rat erythrocytes at 4°C overnight. The nephrotoxic serum was used to induce glomerulonephritis in male WKY rats (Charles River) as described previously (36). Treatment of the animals with anti-oxMIF Abs was done on days 4 and 6 after disease induction. Animals were culled at day 8 and kidneys were taken for histological analysis. For combination treatment of anti-oxMIF Abs with glucocorticoids, subtherapeutic doses of dexamethasone were injected i.p. on day 4 after disease induction. Urine was collected at day 7. Urinary proteins were analyzed using the Quick Start Bradford Protein Assay Kit-1 (Bio-Rad), creatinine (Cr) using the M L-Type Assay (Wako Chemicals), and TNF-α and IL-1β using a Bio-Plex Rat Cytokine Multiplex Kit (Bio-Rad). Morphology of renal injury was assessed in H&E and periodic acid–Schiff–stained renal tissue sections. Macrophages were detected by immunoperoxidase technique (37). Experimental rat models of glomerulonephritis were performed in accordance with the United Kingdom Home Office Animals (Scientific Procedure) Act and the Institutional Guidelines.

For determination of oxMIF specificity, Abs were coated into microplates and incubated with recombinant MIF, oxMIF, or oxMIF reduced with DTT. Captured MIF or oxMIF was detected with rabbit anti-MIF Abs and a goat anti-rabbit-IgG-HRP conjugate (Thermo Scientific). Plates were stained with 3,3′,5,5′-tetramethylbenzidine (Sigma). For quantitative determination of oxMIF, microplates were coated with BaxM159 that selectively captured oxMIF from the samples applied. OxMIF standard was generated by treating recombinant MIF with 0.2% Proclin300 (active component 5-chloro-2-methyl-4-isothiazolin-3-one; Sigma) that transformed recombinant MIF into an oxMIF surrogate. Captured oxMIF was detected as described earlier. For quantitative determination of total human MIF, 96-well microplates were coated with BaxM159. Samples were treated with 0.2% Proclin300 to transform MIF into oxMIF and detected as described earlier. Total MIF in rodent samples was determined by coating rabbit anti-MIF Abs and detection of captured reduced MIF and oxMIF by a biotinylated rabbit anti-MIF Ab and streptavidin-HRP conjugate (GE Healthcare). Total MIF and oxMIF have been shown to be stable in plasma when stored at 4°C or −80°C (Supplemental Fig. 3).

Abs were immobilized onto a CM5 sensor chip (GE Healthcare). Reduced glutathione (Sigma) and oxidized glutathione were added to HEPES-buffered saline containing EDTA and Surfactant P20 (GE Healthcare) at different ratios. Recombinant MIF (10 μm) was diluted in these buffers and injected into a BIAcore T200 (GE Healthcare). Redox potentials (mV) of the GSH/GSSG buffers were measured using an InLab Redox Microelectrode (Mettler Toledo).

Samples were boiled in NuPAGE LDS-Sample Buffer containing 100 mM DTT and proteins were separated on NuPAGE 4–12% Bis-Tris Gels (Life Technologies). Proteins were transferred on nitrocellulose, and MIF was detected by polyclonal rabbit anti-MIF (FL115; Santa Cruz) and anti-rabbit-IgG-HRP conjugate (Thermo Scientific) using SuperSignal Chemiluminescent Substrate (Thermo Scientific).

Heparinized blood from mice was diluted in Cell Staining Buffer (Biolegend) containing either Alexa Fluor 700–labeled anti-CD3ε and PerCP-Cy5.5–labeled anti-Ly6G or allophycocyanin-labeled anti-CD14 and PE-Cy7–labeled anti-CD19 Abs (all from eBioscience). Blood was also incubated with BaxB01, rabbit anti-MIF Ab (FL115; Santa Cruz), or isotype controls. Unlabeled Abs were detected by R-PE–labeled goat anti-human IgG Ab or FITC-labeled anti-rabbit IgG (both from Sigma). RBCs were lysed with FACS Lysing solution (BD Biosciences), and data were acquired on a FACS Canto II (BD Biosciences).

Buffy coat of blood samples from human donors were diluted in Cell Staining Buffer (Biolegend) containing Fc-receptor blocking mixture (BD Biosciences). Cells were incubated with allophycocyanin-labeled anti-CD19, allophycocyanin-labeled anti-CD3, allophycocyanin-labeled anti-CD14, allophycocyanin-labeled anti-CD15, allophycocyanin-labeled anti-CD56 (all from BD Biosciences), and Pacific blue–labeled anti-CD45 (Life Technologies). Cells were also incubated with BaxB01, polyclonal rabbit anti-MIF Ab (FL 115; Santa Cruz), or isotype controls. Detection of unlabeled Abs, lysis of RBCs, and FACS analysis was done as described earlier.

Mouse spleens or leukocyte subpopulations have been prepared by Ficoll gradient centrifugation (granulocytes and PBMCs) and monocyte separation (lymphocytes and monocytes) by cell culture plate attachment, and 1–6 × 10⁶ cells were lysed in cell lysis buffer (Cell Signaling Technology). Cleared lysates were incubated with Ab and immunoprecipitated with Protein G Dynabeads (Life Technologies). Immunoprecipitated MIF was analyzed by Western blot as described earlier.

Rat kidney cryosections were fixed in 4% formaldehyde and incubated with goat anti-human IgG directly labeled with HRP (Sigma) to detect anti-oxMIF Ab or isotype control Ab. Anti-oxMIF Ab BaxB01 was used to detect total MIF in renal tissue as MIF was denatured during fixation of tissue (Fig. 1F). After Avidin/Biotin block (Vector Laboratories), biotinylated BaxB01 and R.T.U. VECTASTAIN ABC reagent (Vector Laboratories) were used to detect total MIF. The nuclei were counterstained with hematoxylin, and the slides were dried in ascendant series auf ethanol and xylene. After mounting with VECTAMOUNT Permamount and coverslipping, full slide scans were acquired using an Olympus VS120 slide scanning microscope at 20-fold magnification (UPLANSAPO 20×, numerical aperture 0.75; PIKE F505/C Camera; Allied Vision Technologies). Images are presented at 10-fold magnification.

Distribution was tested with Kolmogorov–Smirnov tests. If normal distribution was confirmed, data were evaluated by one-way ANOVA followed by Dunnett’s multiple comparison test; otherwise, data were evaluated by Kruskal–Wallis test followed by Dunn’s multiple correction test. Two groups were compared by unpaired two-tailed Student t test (normal distribution) or Mann–Whitney U test.

Human monoclonal anti-MIF Abs BaxB01, BaxG03, and BaxM159 were shown to neutralize MIF functional activity in cell-based assays and to be protective in animal models (30, 31). However, surface plasmon resonance (SPR; Fig. 1A, 1B) and ELISA studies (Fig. 1C–E) revealed that binding of these Abs to MIF is redox dependent. In vitro oxidation of recombinant human MIF with either l-cystine or oxidized glutathione induced a structural rearrangement and generated a conformational isoform (oxMIF) that was detected by these anti-MIF Abs with nanomolar affinities, whereas recombinant MIF in its reduced state was not detected. The half-maximum redox potential at which MIF underwent a structural rearrangement that enabled Ab binding occurred at Eh −94 mV (Fig. 1A, 1B). ELISA studies confirmed the specificity of BaxB01, BaxG03, and BaxM159 for oxMIF (Fig. 1C, 1D), and reduction of oxMIF with DTT abrogated Ab binding (Fig. 1E). BaxB01, BaxG03, and BaxM159 bound with high specificity to recombinant mouse and rat oxMIF (data not shown). Rat and mouse MIF share ∼90% homology with human MIF and contain the highly conserved ₅₇Cys-Ala-Leu-Cys₆₀ motif (5, 33, 38).

The reversible and redox-dependent transition of MIF to oxMIF is the only way to achieve binding to Abs BaxB01, BaxG03, or BaxM159 under physiological conditions. However, under nonphysiological conditions, such as immobilizing recombinant MIF on solid surfaces (e.g., plastic or glass) or treatment with denaturing and/or cross-linking agents (e.g., formaldehyde), MIF irreversibly changes its structure, which also leads to Ab binding (30) (Fig. 1F). In addition, several small molecules that oxidize MIF, for example, 5-chloro-2-methyl-4-isothiazolin-3-one, can be used to efficiently convert MIF into an oxMIF surrogate, and hence lead to Ab binding. Taken together, our data demonstrate the discovery of a redox-dependent isoform of MIF and a strict specificity of our mAbs for this isoform.

The biological relevance of the two immunologically distinct, redox-dependent conformational isoforms of MIF became evident when plasma samples from patients with different acute and chronic inflammatory diseases were compared with those from healthy control volunteers using ELISA. Two ELISA methods were developed: one that quantitatively detected total MIF, that is, the sum of oxMIF and reduced MIF, and a second that specifically quantitated oxMIF. Whereas the plasma samples of 60 of 69 healthy controls did not contain detectable oxMIF, we found small amounts of oxMIF (up to 7.9 ng/ml) in samples of 9 individuals (9/69, 13% oxMIF⁺; median 0.0 ng/ml). The plasma levels of oxMIF were significantly elevated in patients with septicemia (40/40, 100% oxMIF⁺; median 7.2 ng/ml), psoriasis (43/43, 100% oxMIF⁺; median 3.8 ng/ml), asthma (63/63, 100% oxMIF⁺; median 9.4 ng/ml), UC (15/20, 75% oxMIF⁺; median 3.7 ng/ml), and Crohn disease (31/39, 79% oxMIF⁺; median 3.7 ng/ml; Fig. 2A). Elevated plasma levels of oxMIF were also determined in a subpopulation of SLE patients, namely, patients with systemic exacerbations of SLE without renal involvement (8/9, 89% oxMIF⁺; median 8.3 ng/ml; Fig. 2A). In contrast, oxMIF was not significantly elevated in the plasma of SLE patients in remission (n = 43; median 0 ng/ml) or in patients with acute lupus nephritis (LN; n = 21; median 0; data not shown).

The same plasma samples from healthy controls showed a baseline level of total MIF (median 5.1 ng/ml), consistent with MIF levels published in the literature (21, 28). In this sample collection, total MIF was significantly elevated only in patients with septicemia (median 13.3 ng/ml) and asthma (median 12.7 ng/ml), but not in patients with psoriasis, IBD, or SLE (Fig. 2B). The degree of statistical significance was markedly higher for elevation of oxMIF levels than for elevation of total MIF levels, as the baseline level for oxMIF in control samples was 0 ng/ml. Interestingly, correlation plots demonstrated that plasma levels of oxMIF and total MIF correlated in these different inflammatory diseases (Fig. 2C). On average, total MIF levels were ∼50% higher than oxMIF levels, indicating that oxMIF comprises two thirds of MIF in the circulation of patients. However, neither oxMIF levels nor total MIF levels followed a standard distribution, and both show strong variation with levels of up to 250 ng/ml.

Immune cells are known to store high levels of MIF intracellularly (25–27). We analyzed lysates from granulocytes, monocytes, and lymphocytes derived from healthy control donors by immunoprecipitation to assess whether oxMIF was present in the cytosol of leukocytes. As expected, we detected MIF in each cell type but could not precipitate oxMIF (Fig. 3A). Our studies also revealed that MIF can be detected on the surface of immune cells. We established flow-cytometry methods to analyze membrane-bound total MIF and oxMIF. In samples of human leukocytes isolated from healthy controls, MIF was expressed on the surface of monocytes/macrophages and granulocytes. Within the different lymphocyte subsets, MIF was expressed on the surface of B cells, but barely on T or NK cells. However, oxMIF was never detected on the surface of leukocytes from healthy controls (Fig. 3B). To assess whether oxMIF is expressed on the surface of human leukocytes under disease conditions, we analyzed blood samples from bacteremic patients. Whereas total MIF was detected on the surface of all immune cells (Fig. 3C, right column), oxMIF was detected on the surface of granulocytes, macrophages, and NK cells, but not on the surface of B or T cells (Fig. 3C, left column).

We also investigated immune cells from an experimental sepsis model in mice to support the observation that oxMIF is expressed on the surface of leukocytes under diseased conditions. Total MIF was detected on the cell surface of all immune cell populations (granulocytes, monocytes, B cells, and T cells) from challenged and nonchallenged mice. OxMIF was expressed only on granulocytes and monocytes from infected mice and could not be detected on the cell surface of any leukocytes from the healthy control group (Supplemental Fig. 1A, 1B), which was in accordance with the results obtained from analyzing human leukocytes (Fig. 3B, 3C). In addition, we tested for the presence of oxMIF in spleen lysates. Our results show that the cytoplasm of immune cells and other spleen cells does not contain oxMIF, regardless of whether the spleen was derived from healthy mice or mice with peritonitis (Supplemental Fig. 1C). Thus, our data show that during the course of an inflammatory response, increased levels of oxMIF could be detected in the plasma and the cell surface of immune cells.

We analyzed the urine from patients diagnosed with SLE, with or without renal flares, and from healthy controls to gain further insights into oxMIF expression in different pathologic conditions. SLE samples were classified into three different groups depending on the patient’s diagnosis at the time of sampling, that is, remission (without signs of acute kidney inflammation), systemic flare of SLE (without renal manifestation), or acute LN (characterized by acute renal flares). We found a clear correlation between disease severity and the oxMIF/Cr ratio in the urine in patients with acute LN (Fig. 4A). We did not see any elevation of urinary oxMIF/Cr ratio in patients with systemic SLE without renal manifestation, or in patients in remission (Fig. 4A). Total MIF/Cr ratios indicated that all MIF measured in the urine was oxMIF (Fig. 4B). In plasma, oxMIF was significantly elevated during systemic exacerbations of SLE without renal involvement, when compared with healthy controls (Fig. 2A). In contrast, oxMIF was not elevated in the plasma of patients in remission or in patients with LN (data not shown). Thus, an acute renal flare coincided with local expression of oxMIF in the kidney, which could be detected primarily in the urine, whereas systemic exacerbation of SLE resulted in elevated oxMIF in the circulation.

We analyzed supernatants from colonic explant cultures to address the presence of oxMIF in the colon of patients with IBD (35). We showed that biopsies taken from foci of acute inflammation from patients with UC and Crohn disease contained significantly higher amounts of oxMIF and MIF than those taken from nonlesional sites or biopsies from healthy controls (Fig. 4C, 4D). Levels of TNF-α (Fig. 4E), IL-6 (Fig. 4F), and IL-8 (data not shown) paralleled the presence of oxMIF in colonic explant supernatants. We conclude that oxMIF is expressed in inflamed tissues and subsequently released to, for example, the plasma or urine during acute phases of inflammatory diseases.

We addressed whether neutralization of oxMIF is sufficient to reduce inflammation in vivo in animal models of chronic inflammation. First, we assessed the efficacy of oxMIF-specific Abs in two IBD models with different pathogeneses: the anti-CD40–induced acute colitis model (39) and the IL-10^−/− piroxicam model of chronic colitis (40, 41). Injection of agonistic anti-CD40 mAb to T cell– and B cell–deficient Rag-1^−/− mice induced pathogenic symptoms of acute colitis and intestinal innate inflammatory response that functionally depend on TNF-α, IFN-γ, and IL-12p40 and IL-23p40 pathways (39). Treatment with either BaxB01 or BaxG03 reduced disease activity scores (Fig. 5A) and histological signs of colitis (Fig. 5B, 5C), characterized by little crypt elongation, goblet cell depletion, and few inflammatory cells in anti-oxMIF–treated mice.

We synchronized development of chronic enterocolitis in IL-10^−/− mice by feeding with piroxicam, a nonsteroidal anti-inflammatory drug (42, 43). Repeated injections of anti-oxMIF Abs BaxB01 and BaxG03 effectively alleviated signs of chronic enterocolitis as demonstrated by the decrease of the disease activity scores (Fig. 5D) and histology index (Fig. 5E), and reduced thickening of the colon (Fig. 5F) in the treated groups. We also measured a decreased secretion of the proinflammatory cytokines TNF-α, MCP-1, and IL-6 after anti-oxMIF treatment (Fig. 5G–I).

We assessed the efficacy of anti-oxMIF treatment in a rat model of crescentic glomerulonephritis. After injection of nephrotoxic serum targeting the glomerular basement membrane, the animals developed glomerular crescents from tissue injury driven by intraglomerular macrophages, leading to proteinuria and finally to kidney failure (36, 44). We detected urinary oxMIF 4 d after disease induction. Treatment of the animals with BaxB01 starting on day 4 after disease induction significantly reduced proteinuria (Fig. 6A), glomerular macrophage infiltrates (Fig. 6B), and glomerular crescents (Fig. 6C). Urinary oxMIF levels increased significantly during disease progression in the control Ab group but remained 10-fold lower in the BaxB01-treated group (Fig. 6D). In addition, urinary levels of the proinflammatory cytokines TNF-α and IL-1β decreased (Supplemental Fig. 2). Immunohistochemistry staining of the rat kidney showed that the BaxB01 Ab was localized in the glomeruli and the adjacent tubular regions, exactly those regions where inflammation and expression of oxMIF was expected. In contrast, staining for total MIF showed a more uniform distribution pattern that could not be linked to foci of inflammation (Fig. 6E).

The outcomes of these studies demonstrate the anti-inflammatory effect of Abs BaxB01 and BaxG03 in rodent models of IBD and crescentic glomerulonephritis. In previous studies we showed that Abs BaxB01, BaxG03, and BaxM159 have beneficial effects in models of experimental sepsis and contact hypersensitivity (30). Collectively, these data strongly support the concept that mAbs directed against oxMIF alleviate inflammation in vivo.

One specific property of MIF is its ability to override immunosuppressive effects of glucocorticoids (1, 3, 27). Thus, we used the rat model of crescentic glomerulonephritis to assess the efficacy of combination treatment with anti-oxMIF Abs and glucocorticoids. We treated rats on day 4 after disease induction with 0.025, 0.075, or 0.25 mg/kg dexamethasone to determine a subtherapeutic dose. We observed a clear dose-dependent reduction in urinary proteinuria (Fig. 7A), glomerular macrophage infiltrates (Fig. 7B), and glomerular crescents (Fig. 7C). We selected the subtherapeutic dose of 0.025 mg/kg dexamethasone for combination with different doses of BaxB01. Rats received a single dose of dexamethasone on day 4 and treatment with BaxB01 on days 4 and 6 after disease induction. We treated control rats with either dexamethasone or BaxB01 alone. Dose–response curves demonstrated an amplified reduction of proteinuria (Fig. 7D), macrophage infiltration (Fig. 7E), and generation of glomerular crescents (Fig. 7F) when BaxB01 and a subtherapeutic dose of dexamethasone are combined. Furthermore, the results indicated a greater reduction of disease parameters with the combination than expected with the sum of BaxB01 and dexamethasone monotherapy. The effects of anti-oxMIF combined with 0.025 mg/kg dexamethasone were more pronounced than achieved with a 10-fold higher dose of dexamethasone monotherapy, implying that BaxB01/dexamethasone synergy leads to a >10-fold higher apparent activity of dexamethasone. These data demonstrate that an anti-oxMIF Ab treatment potentiates glucocorticosteroids in alleviating crescentic glomerulonephritis.

Our data show that MIF occurs in two immunologically distinct conformational isoforms, reduced MIF and oxMIF, which reversibly interconvert, dependent on the local redox potential. Whereas MIF is abundantly expressed in the body and can be detected in the circulation, on the cell surface, and in the cytoplasm of cells isolated from healthy controls, we found oxMIF predominantly expressed in patients with inflammatory diseases. Moreover, we found that oxMIF is produced specifically during acute inflammatory phases and its presence could be linked to sites of inflammation.

MIF is a proinflammatory cytokine that uses different mechanisms, which include: 1) overriding glucocorticoids, 2) triggering the release of other proinflammatory cytokines from monocytes/macrophages, and 3) chemokine functions postulated to play a role in the recruitment of leukocytes into inflamed tissues (1–3, 6, 37, 45, 46). We showed that anti-oxMIF treatment interfered with these three mechanisms. In the mouse colitis models and the rat glomerulonephritis model, the beneficial effects of anti-oxMIF treatment were linked to extenuated immune responses characterized by a reduction of Th1-biased inflammatory cytokines. In the rat glomerulonephritis model, treatment with anti-oxMIF Abs reduced the number of infiltrating macrophages, thus establishing a direct link to the inhibition of MIF chemokine functions. Because combination of dexamethasone with anti-oxMIF Abs not only resulted in additive but also clear amplified treatment effects in the rat glomerulonephritis model, we conclude that anti-oxMIF Abs inhibit inflammatory triggers of the MIF-glucocorticoid axis.

In vitro conversion of reduced MIF to oxMIF is not associated with posttranslational modifications, as described previously (47, 48) (data not shown). Indeed, the molecular basis of the structural rearrangement of MIF under proinflammatory conditions is unknown. Several redox-regulated proteins are known to interact with MIF (33). Other studies have identified hepatopoietin (49), tumor suppressor p53 (50), and peroxiredoxin (51) as intracellular targets for MIF, whereas binding of thioredoxin in the extracellular space regulates internalization and proinflammatory signaling of MIF in vitro (52). However, we suspect that the conversion of abundantly expressed MIF to a disease-related isoform (oxMIF) is a tightly regulated mechanism that stimulates disease progression by activating and triggering proinflammatory signals.

We have previously reported the generation of a highly diverse panel of 145 unique fully human anti-MIF Abs derived from a phage display library (30). This panel has been screened for Abs that neutralize MIF biologic activity in vitro and in vivo. In this previous report (30), we showed that in vivo protective anti-MIF Abs are specific for two distinct epitopes (aa 50–68 or 86–102), which form a β-barrel structure containing the functionally important thiol protein oxidoreductase motif. Exactly the same Abs turned out to be specific for oxMIF. We repeatedly tested anti-MIF Abs from this panel that were not able to discriminate between oxMIF and reduced MIF in animal models, but these Abs failed our screening because we never observed a beneficial therapeutic effect. Our data indicate a remarkable correlation among epitope specificity, specificity for oxMIF, and in vivo protection in models of inflammation. We conclude that specific inhibition of oxMIF is a potent method of interfering with the biological activity of MIF. OxMIF is therefore a relevant new target in inflammation. Furthermore, the detection of oxMIF in urine and plasma samples from patients with inflammatory diseases and the correlation of oxMIF with disease stages illustrate its potential as a biomarker for initial diagnostic screening and for differentiation between disease stages.

M.T., R.J.K., D.V., P.D., A.S., H.K., M.R., G.C.M., H.E., and F.S. are full-time employees of Baxter; R.J.K., H.E., and F.S. have ownership interest (including patents) in Baxter Healthcare; F.W.K.T., B.J., H.T., and C.T. have a research grant from Baxter and are consultants of the same; and J.K. is a consultant for Baxter. The remaining authors have no financial conflicts of interest.