The Golgi-Associated Protein p115 Mediates the Secretion of Macrophage Migration Inhibitory Factor¹

Macrophage migration inhibitory factor (MIF)⁴ is a 12.5-kDa cytokine and an important regulator of innate immunity that is produced in response to proinflammatory, mitogenic, or hormonal stimuli. MIF plays a pivotal, upstream role in the host antimicrobial responses, the development of autoimmune inflammatory diseases, and in tumor progression (1). Activating stimuli induce the rapid release of MIF from preformed, cytoplasmic pools; this is followed by an up-regulation in MIF mRNA expression and the replenishment of intracellular protein content (2, 3). Notably, MIF lacks a signal sequence and is secreted by an unconventional route for protein export. MIF undergoes no Golgi complex-dependent posttranslational modifications and inhibitors of the endoplasmic reticulum (ER)/Golgi transport do not affect LPS-stimulated MIF secretion from monocytes/macrophages (4, 5).

In contrast to the “classical” pathway for secretion, in which a hydrophobic leader sequence targets proteins for export via the endoplasmic reticulum and Golgi (6, 7, 8), nonclassical protein secretion is less well understood (9, 10, 11). Several distinct mechanisms have been revealed for particular proteins such as vesicular transport via lysosomes or exosomes, plasma membrane shedding, and direct translocation across the membrane by specialized transporters. Potential secretory pathways that have been described for IL-1β, for instance, include cell surface blebbing and the formation of microvesicles that lyse in the extracellular space (12) and the fusion of endolysosomal vesicles with the cell surface plasma membrane (13). Fibroblast growth factor (FGF) 2, which is another unconventionally secreted cytokine, appears to cross the membrane in an ATP- and membrane-potential independent manner (10, 14, 15) and requires the action of phosphatidylinositol phosphate kinases (PIP-K) (16, 17). There is evidence that MIF may be associated with vesicles in particular cell types (18, 19), and a recent report has shown that MIF secretion is reduced by glyburide (5). Glyburide inhibits ATP-binding cassette transporters; it targets numerous proteins such as ion channels, enzymes, and other transporters (20, 21).

To gain insight into the protein components and the mechanisms responsible for MIF release, we performed a yeast two-hybrid interaction study to identify intracellular proteins that might bind MIF and mediate its export. We report herein the identification of the Golgi-associated protein p115 (22) as a binding partner for MIF, and we provide evidence for the p115-dependent release of MIF from human monocytes/macrophages. We further show that a recently described, small molecule MIF inhibitor, 4-iodo-6-phenylpyrimidine (4-IPP) (23), targets MIF secretion by influencing the interaction between MIF and p115.

A human pituitary cDNA library and the CytoTrap yeast two-hybrid system were purchased from Stratagene and used according to the manufacturer’s instructions. Briefly, cdc25H yeast cells were cotransformed with a MIF-bait plasmid (pSOS-MIF) and a library of human pituitary cDNA (pMyr). The expression of library cDNA was controlled by a galactose-inducible promoter. Transformants were grown on selectable minimal glucose plates for 4–6 days at 25°C and colonies were replica plated onto minimal galactose plates for 5–7 days at 37°C. Positive colonies exhibiting efficient growth on galactose plates were isolated and tested for galactose-dependent growth at 37°C. Library plasmids were recovered and the specificity of interactions was tested by retransformation of cdc25H cells with pSOS-MIF, pSOS-ColI, or pSOS-MAFB. Library plasmids that specifically interacted with MIF were identified by DNA sequencing.

A truncated p115 protein (p115^702–962) representing the clone identified in the yeast two-hybrid screen was subcloned into the pET22b vector, which contained a carboxy-terminal histidine tag (Novagen) and was expressed in Escherichia coli BL21(DE3). The crude bacterial extract was purified using a HiTrap chelating HP column (Amersham Biosciences) with a linear gradient of 20–400 mM imidazole in 50 mM Tris (pH 8.0) and 200 mM NaCl. The peak fractions were subjected to gel filtration chromatography (Superdex-75, HiLoad 16/60; Amersham Biosciences). Protein purity was verified by SDS-PAGE/Coomassie staining and Western blotting. For the p115 capture assay, recombinant p115^702–962 was biotinylated using EZ-LinkTM-sulfo-NHS-LC biotin (Pierce). Human MIF was prepared as described previously (4) and biotinylated using a Biotin Protein Labeling Kit from Roche. The chemical modification of MIF by N-acetyl-p-benzoquinone imine (NAPQI) (24) and 4-iodo-6-phenylpyrimidine (4-IPP) was achieved by incubation of MIF with a 10× molar excess of compound at room temperature overnight (23). The modified MIF was dialyzed against PBS. For expression in mammalian cells, full-length p115 was cloned into pcDNA3.1 (Invitrogen).

Ninety-six well Nunc Immuno-Module plates were coated with 1 μM p115^702–962/well. The plates were washed with TBST and blocked with Superblock (Pierce). Biotinylated MIF (150 nM) was added in duplicate to wells along with increasing concentrations of unlabeled MIF or lysozyme as control. Incubation was continued overnight at 4°C, followed by washing with TBST. The bound, biotinylated MIF was detected by adding streptavidin-conjugated alkaline phosphatase (R&D Systems) for 1 h at room temperature, followed by washing and detection of the alkaline phosphatase with p-nitrophenyl phosphate (Sigma-Aldrich). For a reverse binding assay, 1 μM MIF was coated onto wells and 150 nM biotinylated p115^702–962 was added along with increasing concentrations of unlabeled p115^702–962. Values are plotted as percent absorbance at 405 nm relative to wells containing biotinylated MIF or p115^702–962 alone.

For measurement of MIF’s immunoreactivity in an ELISA system, the two mAb clones MAB289 (5) and 3H2F (25) were used as capture Abs.

Cell lines were obtained from American Type Culture Collection, with the exception of HEK293A cells, which were purchased from Invitrogen. For differentiation of THP-1 monocytes into macrophages, 50 ng/ml PMA (Sigma-Aldrich) was added for 24 h, after which the medium was changed and the cells were cultured for an additional 48 h (26).

The MIF secretion assay was performed as described previously (5). Briefly, cells first were synchronized for 3 h in medium containing 1% FCS. THP-1 monocytes or primary human PBMC-derived macrophages (1 × 10⁶/ml) were stimulated with 10 μg/ml LPS 0111:B4 (Sigma- Aldrich) and PMA-differentiated THP-1 macrophages were stimulated with 0.1 μg/ml LPS. For studies of MIF inhibitors, the test compounds were added 1 h before LPS stimulation. Cytokine levels in the supernatants were analyzed by ELISA (eBioscience), with the exception of the MIF ELISA, which was performed as described previously (25). Unless stated otherwise, background immunoreactivity was subtracted from the stimulated cytokine level. Cell viability was verified using an lactate dehydrogenase (LDH) assay (Roche).

RNA was isolated using RNeasy from Qiagen. Reverse transcription was performed using StrataScript according to the manufacturer’s instructions (Stratagene). For quantitative PCR, the following primers were used: MIF forward, 5′-CGGACAGGGTCTACATCAA-3′ and MIF reverse, 5′-CTTAGGCGAA-GGTGGAGTT-3′ and β-actin forward, 5′-GGATGCAGAAGGAGATCA-CTG-3′ and β-actin reverse: 5′-CGATCCACACGGAGTACT TG-3′. The amplification profile included denaturation at 95°C for 5 min followed by annealing and elongation for 1 min at 60°C for a total of 40 cycles. For the p115 quantitative PCR, primers for p115 and β-actin were designed and synthesized by SuperArray. The quantitative PCR profile was done according to SuperArray’s manual and specific product amplification was verified with a melting curve.

COS-7 cells were transiently transfected with a full-length p115 expression vector (pcDNA3.1/V5) using FuGENE (Roche). Cells were lysed in radioimmunoprecipitation buffer (50 mM Tris-HCl (pH 7.4),150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 2 mM EDTA, and protease inhibitors) and then incubated with 1 μg of recombinant MIF or 1 μg of GST-MIF beads. Monoclonal anti-MIF Ab (R&D Systems) was used to immunoprecipitate MIF. Precipitates were run on a 4–12% Bis-Tris NuPage gel (Invitrogen) and transferred onto polyvinylidene difluoride membrane. The interaction between MIF and p115 was visualized using an anti-V5 Ab to detect the V5 tag on the p115 C terminus. For endogenous coimmunoprecipitation, LNCap cell lysates were prepared using radioimmunoprecipitation buffer and used as a source of endogenous p115 and MIF (27). Endogenous MIF was immunoprecipitated and protein complexes were resolved by SDS-PAGE and visualized with anti-p115 Ab (CB1009; Calbiochem).

PBMCs (2.5 × 10⁶/ml) were cultured and differentiated on glass slides (BD Falcon) washed in PBS, and fixed and permeabilized using 3.7% formaldehyde and 0.1% Triton-X-100 in PBS (30 min at 37°C). After blocking in 5% goat serum (Sigma-Aldrich) for 20 min, the slides were incubated overnight (at 4°C) with anti-p115 and anti-MIF Abs at a dilution of 1/200. After washing with PBS, the cells were stained with a mixture of Alexa Fluor 488-conjugated goat anti-rabbit IgG and Alexa Fluor 568-conjugated goat anti-mouse IgG (Invitrogen) at a dilution of 1/200 for 1 h at room temperature; To-Pro3 was used as nuclear dye (Invitrogen). Slides were mounted with slowfade (Invitrogen) and the cells were visualized using a confocal laser-scanning microscope (LSM510; Zeiss). To confirm that RNA interference (RNAi)- mediated knockdown of p115 did not result in fragmentation of the Golgi, THP-1 monocytes were differentiated with PMA and p115 depleted directly on culture slides. Cells then were stained with anti-p115 and anti-GM130 (Abcam) Abs and Golgi structure was observed by fluorescent microscopy. GM130 served as marker for the Golgi (28).

The Block-it RNAi designer (Invitrogen) was used to design a short hairpin RNA (shRNA) molecule specific to human p115. Two constructs were designed: p115-1 from position 2625 to 2645 (5′-GCAGCTGGATTCATCTAATAG-3′) and p115-2 from position 1601 to 1620 (5′-GCAGTTGGTCCAAGGCTTAT-3′). Neither construct shares sequence homology with other genes as analyzed by BLAST search. For negative controls, shRNA constructs against lacZ were generated. Oligonucleotides were cloned into the RNA expression vector pENTR/U6 and confirmed by sequencing. shRNAs against p115 and control were transfected into human THP-1 monocytes (1 × 10⁶/ml) using electroporation (Amaxa) or HeLa cells (1 × 10⁵/well) using Lipofectamine 2000 (Invitrogen). GM130 was targeted using a RNAi construct directed against the sequence ATGAGAACATGGAGATCACC (29).

The shRNA constructs p115-2 and lacZ in the pENTR/U6 vector were recombined with the adenoviral vector pAd/BLOCK-iT-DEST (Invitrogen) using Gateway LR Clonase according to the manufacturer’s protocol. The resulting p115/pAd plasmid was transfected into HEK293A and amplified following the standard procedure described in the manual. The viral titer was determined using an AdenoX Rapid Titer Kit (BD Clontech). PMA-differentiated THP-1 macrophages were transfected with a multiplicity of infection (MOI) of 50 and incubated overnight; cells were additionally cultivated for 96 h after infection. For infection of primary macrophages, cells were transfected with a MOI of 100 for 2 h and then cultured for an additional 48 h.

Human PBMCs were isolated by Ficoll-Hypaque gradient centrifugation. The cells were resuspended in RPMI 1640 medium supplemented with 20% human AB serum (Cambrex) and plated at 2.5 × 10⁶ cells/well in 24-well tissue culture plates. After 2 h of culture, the adherent cells were washed extensively with PBS and cultured for 1 wk to allow differentiation into monocyte-derived macrophages. THP-1 cells (1 × 10⁶/ml) were infected with Chlamydia trachomatis serovar L1 at a MOI of 10 for 1 h at room temperature by slow rotation in RPMI 1640 medium. Cells were washed with medium and resuspended in RPMI 1640 containing 10% FBS without antibiotics and cultivated at 1 × 10⁶ cells/ml in 12-well plates. The infection rate was analyzed by flow cytometry and staining with anti-Ct LPS FITC Ab (ViroStat) (30). Bone marrow-derived macrophages were isolated from C57BL/6 mice as reported previously (31). For experiments, the cells were replated at a concentration of 0.5 × 10⁶/ml.

HeLa cells were genetically modified to express FGF-2-GFP and FGF-4-GFP fusion proteins in a doxycycline-dependent manner (32). To quantify the secretion of FGF fusion proteins, cells were transfected with knockdown constructs targeting either p115, PIP-K (16), or a control sequence. After 96 h, fusion protein expression was induced by doxycycline for 24 h. Cells were prepared for flow cytometry to detect cell surface-associated FGF-2-GFP and FGF-4-GFP, respectively. Primary Abs were decorated with allophycocyanin-conjugated secondary Abs and GFP fluorescence (intracellular and cell surface) and allophycocyanin fluorescence (cell surface) was determined using a FACSCalibur flow cytometer (BD Biosciences). Untransduced HeLa cells were used to determine autofluorescence signals.

To identify intracellular proteins that mediate MIF secretion, we used the CytoTrap yeast two-hybrid screen, which relies on the cytoplasmic rescue of a temperature-sensitive mutation in the Ras signaling pathway (33). MIF cDNA served as “bait” and a human pituitary cDNA library was the “prey” (34). After cotransformation of bait and prey plasmids into the Saccharomyces cerevisiae strain cdc25H, 90 colonies were identified in a primary screen. The secondary and tertiary screenings ruled out 75 colonies as false positives that showed growth in the absence of the library protein. Prey proteins in the remaining 15 colonies bound specifically to MIF (Fig. 1,A). By DNA sequencing, one of the positive clones contained the carboxyl-terminal 780 bases of the cDNA for p115. The remaining 14 colonies all contained the coding sequence of MIF; these may be regarded as an internal positive control because MIF forms a homotrimer (35, 36). p115 was characterized initially as a vesicle-docking protein that is localized predominantly to the cytosolic side of vesicular tubular intermediate clusters and the cis-Golgi (22, 37, 38). The carboxyl-terminal region of p115 consists of four coiled-coiled domains and a short acidic tail (37) (Fig. 1 B).

To biochemically confirm the interaction between MIF and p115, we produced the carboxyl-terminal region of p115 (p115^702–962) in E. coli (Fig. 1,C) and evaluated its binding to MIF in an in vitro competition binding assay. As shown in Fig. 1 D, increasing concentrations of MIF, but not the equimolar addition of a control protein (lysozyme), inhibited the interaction of biotinylated MIF with p115. Protein-protein interaction was verified by performing the reverse assay, i.e., measuring the binding of biotinylated p115^702–962 to immobilized MIF.

The intracellular interaction between MIF and p115 was further investigated by coimmunoprecipitation. COS-7 cells that were transfected with a V5 epitope-tagged p115 served as a source for p115. After the addition of recombinant MIF, a MIF-p115 complex was coprecipitated using an anti-MIF Ab. p115 also coprecipitated after adding GST-tagged MIF (Fig. 1 E). An endogenous interaction between p115 and MIF also was observed by Western blot analysis for p115 after immunoprecipitation with anti-MIF in the prostate carcinoma cell line LNCap (27). These results, taken together, confirm an interaction between p115 and MIF in mammalian cells.

We studied the cellular distribution of MIF in macrophages in response to LPS stimulation. In unstimulated macrophages, MIF appears present throughout the cytoplasm and p115 is predominantly in the perinuclear area in a pattern that is consistent with the Golgi (Fig. 2,A), which is in agreement with previous reports (22, 38). In the time course study of LPS-activated primary macrophages, MIF gradually redistributes from the cytoplasm to a plasma membrane-proximal area and a portion of p115 now also appears dispersed in the cytoplasm. Furthermore, in the course of LPS stimulation, a colocalization of MIF and p115 at the area of the Golgi and in the cytoplasm can be observed (Fig. 2 B).

MIF is released from cells in both a regulated and a specific manner (5, 39). We stimulated human THP-1 monocytes with LPS and observed an increase in the concentration of MIF in supernatants that was detectable at 2 h and reached a plateau at 4 h (Fig. 3,A). The observed increase in the supernatant content of MIF was not the result of cell death as assessed by LDH analysis, and it was not accompanied by an increase in the level of MIF or p115 mRNA measured over 6 h (Fig. 3 B and data not shown). These data indicate that the initial release of MIF from THP-1 monocytes occurs from preformed pools and does not require the transcription of MIF mRNA. The results are in accord with previous reports that were based on the immunostaining and in situ hybridization of murine tissues after LPS administration in vivo (2, 3).

p115 plays a role in vesicle transfer from the endoplasmic reticulum to the Golgi complex (40). MIF does not enter the Golgi (5) and it was of interest to examine whether p115 is necessary for the release of MIF in response to activating stimuli. We generated RNAi constructs for the gene-specific silencing of p115 and tested their efficacy in human THP-1 cells. As shown in Fig. 4 A, two RNAi constructs specific for p115 (p115-1 and p115-2) reduced the concentration of p115 protein by 50–70%. Of importance, this level of RNAi-mediated reduction in p115 production was not associated with any detectable alteration in the morphology of the Golgi complex, as assessed by direct observation with immunofluorescent microscopy (data not shown).

We next examined whether p115 knockdown reduced the amount of MIF produced by LPS-stimulated THP-1 monocytes. Transfection of the p115-1 or p115-2 RNAi plasmids resulted in a 40–60% decrease in MIF release in response to LPS (Fig. 4 A). Of note, the LPS-stimulated secretion of two conventionally secreted cytokines, TNF-α and IL-6, which have hydrophobic leader sequences, was not affected by a reduction in cellular p115 protein. These data point to the potential specificity of p115 in mediating MIF secretion. TNF-α and IL-6 secretion also affirms that the depletion of p115 by the RNAi approach used herein does not appreciably affect the functioning of the ER/Golgi pathway. We further studied the effect of depleting GM130, which is a p115 binding partner that interacts with the carboxyl-terminal acidic tail of p115 (22), but there was no measurable effect of GM130 depletion on MIF secretion (data not shown).

To gain a broader understanding of the role of p115 in unconventional protein secretion, we studied the export of the two structurally related cytokines FGF-2 and FGF-4. Although both proteins are released from cells in response to tissue injury (41, 42, 43), FGF-4 is secreted conventionally, whereas FGF-2 translocates directly across the plasma membrane (15). Using a previously established doxycycline-inducible expression system (32), we measured both total and cell surface-associated FGF-2 or FGF-4 protein (Fig. 4 B). Cellular depletion of p115 did not influence the export of either FGF-2 or FGF-4, while the secretion of FGF-2 was inhibited by depletion of PIP-K as previously reported (16). The finding that the release of both conventionally secreted cytokines (TNF-α, IL-6, FGF-4) and the unconventionally secreted cytokine FGF-2 is not affected by p115 depletion suggests a specific role for p115 in the stimulated export of MIF.

Human THP-1 monocytes acquire many of the characteristics of mature, primary macrophages upon differentiation with PMA, such as an increase in the expression of the LPS coreceptor CD14 (26). Differentiated THP-1 macrophages thus respond to low doses of LPS and show a plateau in the MIF secretion response at a LPS concentration of 0.1 μg/ml (Fig. 5,A). Differentiated THP-1 macrophages also secrete ∼3-fold more MIF protein than undifferentiated THP-1 monocytes. Of note, ELISA and Western blot analyses of LPS-stimulated THP-1 macrophages showed a coordinated increase in secreted MIF and p115 and a corresponding decrease in the cellular protein levels over the observed time period of 24 h (Fig. 5 B). Moreover, the supernatants of unstimulated macrophages contained only a modest amount of MIF and p115, and no apparent reduction in the cellular MIF and p115 content was measurable. These results taken together suggest that MIF and p115 are cosecreted. Further evidence that MIF and p115 are cosecreted comes from the observation that glyburide, an pharmacologic inhibitor known to inhibit the secretion of MIF (5), also shows an inhibitory effect on the secretion of p115 (data not shown). Of note, glyburide is also known to inhibit the release of other unconventionally secreted proteins (44, 45).

We used an adenovirus construct to deliver p115 siRNA to differentiated THP-1 macrophages and observed that adenoviral transfection reduced significantly intracellular p115 protein levels (Fig. 5,C). The transfection of adenovirus encoding p115 siRNA, but not a control adenovirus, significantly reduced LPS-stimulated MIF release over the 24 h experimental period as demonstrated by Western blot analysis (Fig. 5,D) and ELISA (Fig. 5,E). Of note, the decrease in MIF secretion in p115- depleted cells is not due to cell death or permeabilization as assessed by LDH assay. LDH levels were measured over the course of the experiment; LDH concentration never exceeded 14 mU/ml (<5% total cellular LDH), and no statistical differences between mock-treated and p115-depleted cells were detectable. By contrast, the release of TNF-α, IL-6, or IL-1β was not affected by adenovirus-encoded p115 siRNA, confirming that partial p115 reduction did not interfere with the functional integrity of the Golgi apparatus (Fig. 5,E). The inhibition of MIF secretion also appeared to be dose dependent with respect to the level of p115 depletion (Fig. 5,F). When compared with controls, an ∼50% depletion of p115 was associated with an ∼50% inhibition of MIF release, while an ∼75% depletion of p115 was associated with an ∼75% inhibition of MIF release. Adenoviral-mediated delivery of p115 siRNA also reduced p115 protein production in human peripheral blood monocyte-derived macrophages (Fig. 6,A) and reduced by almost 80% MIF release after stimulation with LPS (Fig. 6,B). Finally, the LPS stimulation of bone marrow-derived macrophages leads to the export of both MIF and p115 protein (Fig. 6 C), thus affirming the results obtained from THP-1 macrophages.

Chlamydiae are Gram-negative, obligate intracellular bacteria that infect and replicate within monocytes/macrophages (46, 47). Infection of cultured THP-1 monocytes with C. trachomatis induces the release of proinflammatory cytokines (48), including MIF, which was most evident at 48 h (Fig. 7,A). We studied the effect of C. trachomatis infection on MIF production in differentiated THP-1 macrophages that had been pretreated with adenovirus encoding either a mock control or a p115 siRNA. p115 depletion had no effect on the infection or replication rate of organisms in macrophages, as measured by flow cytometry for the C. trachomatis Ag LPS (24 h after infection: mock siRNA: 38 ± 5% vs p115 siRNA: 39 ± 8%, p = NS; data not shown). Nevertheless, macrophages treated with p115 siRNA showed a significant decrease in MIF secretion after C. trachomatis infection. By contrast, the secretion of TNF-α, which was evident at 8 h after infection, was unaffected by p115 siRNA (Fig. 7 B). These data indicate that cellular p115 plays an important role in mediating the secretion of MIF by macrophages in response both to a defined microbial ligand such as LPS, as well as to live infection by an intracellular pathogen such as C. trachomatis.

Prototypic small molecules have been identified that bind to the MIF amino-terminal, tautomerase region (24, 49). Mitchell and colleagues recently reported on a suicide substrate, 4-IPP, which was identified by a computational virtual screening for MIF antagonists, and cocrystallization of MIF and 4-IPP demonstrated that it covalently modifies the MIF amino terminus and inhibits MIF-dependent cell motility and growth in vitro (23). It was of interest to investigate whether 4-IPP, which is cell permeable, inhibits the stimulated release of MIF. Because of previous studies showing that the covalent modification of MIF by a small molecule may affect its immunorecognition in a two-Ab, sandwich ELISA (24), we first determined whether 4-IPP influences the detection of MIF by ELISA. Indeed, MIF that had been incubated with 4-IPP showed increased immunorecognition by a MIF-specific mAb (clone MAB289) when compared with native MIF (Fig. 8 A). By contrast, when analyzing 4-IPP-modified MIF with a different anti-MIF mAb (clone 3H2F), a diminished recognition compared with native MIF was observed (>2-fold; data not shown). We concluded that 4-IPP significantly alters the native conformation of the MIF protein, and we therefore used Western blotting for analyzing the effect of 4-IPP on MIF secretion. 4-IPP modification did not affect MIF detection by Western blot (data not shown).

We found that increasing concentrations of 4-IPP dose-dependently block the release of MIF, with 1 μM 4-IPP (10-fold molar excess) resulting in ∼80% inhibition of MIF secretion (Fig. 8,B). 4-IPP also inhibits the basal secretion of MIF (Fig. 8,C). Of note, 4-IPP effectively inhibits the release of both MIF and p115 from LPS-stimulated THP-1 macrophages and, correspondingly, the cytoplasmic levels of MIF and p115 did not decrease when LPS-stimulated cells were treated with 4-IPP (Fig. 8,D). 4-IPP did not affect the release of TNF-α and IL-1β (Fig. 8,E), while reducing the secretion of MIF from monocytes infected with C. trachomatis (Fig. 8,F). We furthermore studied the influence of the reversible MIF-binding molecule (S,R)-3-(4-hydroxyphenyl)-4,5-dihydro-5-isoxazole acetic acid methyl ester (ISO-1) (49) on MIF secretion from stimulated THP-1 macrophages but observed no effect (data not shown). The observation that 4-IPP enhanced the recognition of MIF by an anti-MIF mAb (Fig. 8,A) suggests that 4-IPP imparts a significant conformational affect on MIF that may also influence its binding to p115. We tested this possibility by examining the effect of 4-IPP on the competitive binding of MIF to p115 in vitro. As shown in Fig. 8 G, 4-IPP, but not the small molecules ISO-1 or NAPQI (24, 49), increased MIF binding to p115. The enhanced binding of MIF to p115 in the presence of 4-IPP thus may play a role in the inhibitory action of 4-IPP on MIF release from cells, perhaps by interfering with additional protein interactions necessary for secretion.

MIF plays an important upstream role in the regulation of diverse cellular responses (39, 50, 51, 52) and its role in human pathology has been emphasized by the finding that high expression MIF alleles are associated with the incidence or the severity of inflammatory and oncologic diseases (25, 53, 54, 55, 56). The precise mechanism underlying the secretion of MIF, which lacks an N-terminal signal sequence, has remained enigmatic (34, 57). MIF has been reported to be associated with vesicles and exosomes (18, 19) and its release from stimulated monocytes is reduced by glyburide, which inhibits ATP-binding cassette transporter-dependent secretion (5).

We used a yeast two-hybrid approach to identify intracellular proteins that interact with MIF. The specificity of our screen was confirmed by the finding that 93% of the interacting clones were MIF itself, which associates into a noncovalent homotrimer (35, 36). However, we also identified a cDNA clone encoding the carboxyl-terminal 260 aa of the Golgi-associated protein p115. Identification of p115 as an interacting partner for MIF was confirmed by in vitro binding assays and endogenous coimmunoprecipitation. The competition binding assay using recombinant MIF and p115 protein suggests that no other proteins are necessary for the binding of the two proteins in vitro. Nevertheless, the competition curve levels at 50–60%, indicating that additional protein interactions may occur in vivo.

The immunofluorescent microscopy studies show that MIF is present in large cytoplasmic pools which disperse toward the periphery upon inflammatory stimulation. Similarly, p115 is also partially redistributed upon LPS stimulation from the Golgi toward the plasma membrane. It remains possible that LPS stimulation induces additional unknown proteins, which in turn might be necessary for the interaction between MIF and p115 and its export out of the cell.

Because the cellular depletion of p115 reduces stimulated MIF export, p115 may be essential for the transport of MIF from the perinuclear ring to the plasma membrane and then out of the cell. This pathway for release also is supported by the data demonstrating the accumulation of MIF and p115 in the supernatants of stimulated cells.

The partial depletion of p115 by the RNAi technique used in this study neither affected the structural morphology of the Golgi nor reduced the export of ER-dependent proteins such as TNF-α, IL-6, and FGF-4. The release of the unconventionally secreted protein, IL-1β, which occurs by vesicles (13), and FGF-2, which translocates across the plasma membrane (15), also were not affected by p115 depletion. That p115 is necessary for the release of MIF but not other cytokines, whether conventionally or unconventionally secreted, suggests that p115 has a specific role in MIF export. It also is possible that p115 mediates a generalized secretory response by activated monocytes/macrophages that results in the release of a p115 macromolecular complex containing numerous proteins, of which MIF is one. Further elucidation of both the signals and the protein-protein interactions underlying this export pathway may be informative and enhance our understanding of the acute secretory response of activated monocytes/macrophages.

That MIF is cosecreted with p115 was affirmed by the finding that the small molecule MIF inhibitor 4-IPP reduces the release of both proteins from monocytes/macrophages. These data identify 4-IPP as a potentially selective inhibitor of the p115/MIF secretory pathway. 4-IPP covalently modifies the MIF amino terminus (23) and it alters the protein’s conformational integrity so as to increase its binding interaction with p115. These results indicate that a “native” MIF-p115 interaction is essential for the efficient release of MIF from cells. A non-native MIF-p115 interaction may disrupt trafficking and downstream protein interactions that are necessary for efficient MIF release. 4-IPP also provides proof-of-concept for the possibility of reducing MIF-dependent responses in the earliest phase of cell activation by interfering with MIF’s cytoplasmic release. Inhibitors such as 4-IPP may be attractive in clinical application in those settings, such as severe inflammation, in which rapid intervention is desired to prevent the initiation of a tissue-damaging cytokine cascade (58). We note that our findings do not exclude additional actions for 4-IPP with respect to inhibiting MIF binding with its cell surface receptor or with other effector proteins. Whether p115 is involved exclusively in the stimulated export of MIF or also plays a role in the constitutive release of MIF under basal conditions remains to be answered. The observation that 4-IPP inhibits MIF secretion under basal and stimulated conditions indicates a role for p115 in the constitutive release, but we were unable to observe a significant influence of genetic p115 depletion on MIF release under nonstimulated conditions.

In summary, these data describe a novel function for the Golgi-associated protein p115 in the mediation of MIF unconventional secretion. Pharmacologic targeting of the MIF-p115 interaction may offer a powerful approach for inhibiting innate immune responses by interfering with the secretion of the upstream cytokine, MIF.

We gratefully acknowledge discussions with the late Dennis Shields (1948–2008), who visited our laboratory in 2005 to encourage our investigations of p115. We also thank Jason Griffith and Utako Kaneyuki (Yale University School of Medicine) for helpful discussions.

The University of Louisville has applied for a patent describing the potential therapeutic value of the MIF inhibitor 4-IPP.