Structural determinants of MIF functions in CXCR2-mediated inflammatory and atherogenic leukocyte recruitment

Abstract

We have recently identified the archaic cytokine macrophage migration inhibitory factor (MIF) as a non-canonical ligand of the CXC chemokine receptors CXCR2 and CXCR4 in inflammatory and atherogenic cell recruitment. Because its affinity for CXCR2 was particularly high, we hypothesized that MIF may feature structural motives shared by canonical CXCR2 ligands, namely the conserved N-terminal Glu-Leu-Arg (ELR) motif. Sequence alignment and structural modeling indeed revealed a pseudo-(E)LR motif (Asp-44-X-Arg-11) constituted by non-adjacent residues in neighboring loops but with identical parallel spacing as in the authentic ELR motif. Structure–function analysis demonstrated that mutation of residues R11, D44, or both preserve proper folding and the intrinsic catalytic property of MIF but severely compromises its binding to CXCR2 and abrogates MIF/CXCR2-mediated functions in chemotaxis and arrest of monocytes on endothelium under flow conditions. R11A-MIF and the R11A/D44A-MIF double-mutant exhibited a pronounced defect in triggering leukocyte recruitment to early atherosclerotic endothelium in carotid arteries perfused ex vivo and upon application in a peritonitis model. The function of D44A-MIF in peritoneal leukocyte recruitment was preserved as a result of compensatory use of CXCR4. In conjunction, our data identify a pseudo-(E)LR motif as the structural determinant for MIF's activity as a non-canonical CXCR2 ligand, epitomizing the structural resemblance of chemokine-like ligands with chemokines and enabling selective targeting of pro-inflammatory MIF/CXCR2 interactions.

Chemokines govern leukocyte trafficking and deployment during immune responses and inflammatory reactions by signaling through corresponding G_i protein-coupled chemokine receptors of the CXC chemokine receptor (CXCR) and CC chemokine receptor (CCR) family (1–3). Molecular mimicry of the chemokine system is exploited by viruses, e.g., HIV-1, which invades host cells following interaction of its capsid protein gp120 with host CXCR4, or to evade the host defense, e.g., by herpes viruses expressing macrophage inflammatory proteins, which operate as CCR agonists/antagonists (4–6). It is increasingly appreciated that various host proteins, which cannot be classified according to the consensus nomenclature of chemokines, also rely on direct interactions with chemokine receptors to regulate inflammatory and immune processes, thus acting as “non-cognate” ligands for CXCRs and CCRs. For instance, antimicrobial human β-defensins display chemotactic activity for T and dendritic cell subsets by binding to CCR6. In turn, the cognate ligand of CCR6, CCL20, shares anti-microbial properties with β-defensins as a result of a similar surface topology of positively charged residues (7, 8). Autoantigenic aminoacyl-tRNA synthetases (AaaRS) and their fragments released from damaged cells induce chemotactic cell migration through binding to CCR5 (i.e., HisRS) and CCR3 (i.e., AsnRS) (9, 10), and fragments of TyrRS mediate pro-angiogenic activity by direct binding to CXCR1 through a CXCL8 (also known as interleukin-8; IL-8)-like N-terminal motif consisting of residues Glu, Leu, and Arg (ELR) (11).

Macrophage migration inhibitory factor (MIF) is a long-known T cell cytokine discovered more than four decades ago that more recently has been recognized to be a key mediator of innate immunity and pleiotropic inflammatory cytokine. MIF plays a pivotal role in the pathogenesis of acute and chronic inflammatory diseases such as septic shock, rheumatoid arthritis, inflammatory lung disease, and atherosclerosis by promoting and amplifying involved inflammatory reactions such as monocyte/macrophage survival, MAPK signaling, or inflammatory cytokine release (12–14). We have recently demonstrated that MIF, contrary to its historic and eponymous name, is a non-cognate ligand of the CXC chemokine receptors CXCR2 and CXCR4. Importantly, through interaction with these receptors, MIF is instrumental in inflammatory leukocyte recruitment in atherosclerosis, targeting monocytes and neutrophils through CXCR2 and T cells through CXCR4 (15, 16). MIF is strongly over-expressed in the arterial wall of human atherosclerotic tissue, and blockade or genetic deletion of MIF in animal models of both native and injury-induced atherogenesis leads to a marked reduction in arterial inflammation and lesion size, including regression of established plaques (14, 17). MIF binds to CXCR2 with low nanomolar affinity and induces CXCR2-mediated leukocyte arrest and chemotaxis (15).

CXCR2 signaling induced by the known cognate ligands, such as CXCL8, requires an ELR motif (18). For example, mutagenesis experiments or alanine scanning indicated that substitution of the ELR residues in ELR⁺ CXC chemokines such as CXCL8 or CXCL7 resulted in a dramatic loss of CXCR2 binding affinity or elastase release from neutrophils, whereas its introduction into related chemokines, namely CXCL4, but not CXCL10 or CCL2, conferred neutrophil-activating properties, indicating further structural determinants required for binding to CXCR2 (18, 19). Structural studies have revealed that the ELR residues are critically involved in CXCR2 ligand recognition as a constituent of the protein–protein interaction area (20–22). Although it is tempting to speculate that MIF harbors structural elements, which resemble the ELR motif and thereby convey receptor binding, functional activity, and specificity, as observed for the “classical” ELR+ CXC chemokines, no structural or functional data are available to date.

Here we have used sequence and structural analysis and have identified a pseudo-(E)LR motif in MIF. We have mutated this motif to probe its relevance for CXCR2 binding and CXCR2-mediated functions of MIF in atherogenic and inflammatory cell recruitment.

Results

Generation and Characterization of Pseudo-(E)LR MIF Mutants.

We were intrigued by an apparent architectural homology between the MIF monomer and the CXCL8 dimer (Fig. 1 A and B). However, despite this similarity in 3D structure, MIF does not share any significant sequence homology with CXCL8. MIF contains neither N-terminal cysteine residues nor an ELR motif in the N-terminal sequence (Fig. 1C). However, inspection of the structure of the MIF monomer (23) revealed that residues Asp-44 and Arg-11 are located in neighboring loops in a parallel and adjacent position in 3D space to form an ELR-like motif (Fig. 1 A and B) that forms upon folding and which we termed a pseudo-(E)LR motif. Because it has not been unequivocally resolved whether bioactive MIF occurs in a monomeric or trimeric state, we examined the position of the Asp-44 and Arg-11 residues in the 3D structure of the MIF trimer. Surface topology analysis indicates that both residues would be solvent-exposed and would form a ring-like structure under conditions favoring a trimeric state of MIF (Fig. 1D).

Fig. 1.

Structural homology between CXCL8 and MIF and their ELR and pseudo-(E)LR motives, respectively. The 3D structures of the CXCL8 dimer (A) and the MIF monomer (B) share an architectural homology. (A) The three N-terminal amino acids Glu-3 (E3), Leu-4 (L4), and Arg-5 (R5) of each CXCL8 monomer form an ELR motif known to be essential for signaling through CXCR2. (B) Conformational pseudo-(E)LR motif of MIF formed by the two non-adjacent residues Arg-11 (R11) and Asp-44 (D44), which reside in an ELR-like spacing in exposed neighboring loops. Note that the N-terminal methionine of MIF is processed; thus, numbering of residues starts with Pro-1. (C) Schematic illustration indicating the positions of the ELR and pseudo-(E)LR residues in CXCL8 and MIF, respectively. (D) Surface model of the trimeric structure of MIF. Following the color code of the scheme in C, the location of R11 in each subunit is depicted in red; D44 is highlighted in green.

To test the functional relevance of the pseudo-(E)LR motif, we constructed the site-specific mutants R11A-MIF, D44A-MIF, and R11A/D44A-MIF in which Arg-11, Asp-44, or both residues were exchanged for Ala. The MIF pseudo-(E)LR mutants were over-expressed in bacteria using the pET11b/BL21-DE3 expression system. Despite substituting charged residues, mutants could be purified to homogeneity (>98%; negligible endotoxin content of <10 pg LPS/μg MIF mutant) by an almost identical procedure as that established for WT MIF (24) [Fig. 2A and supporting information (SI) Fig. S1]. Circular dichroism (CD) spectropolarimetry showed that the pseudo-(E)LR mutants folded in a native-like manner (Fig. 2B). Overall structural integrity of the mutants was further verified by comparing the intrinsic catalytic tautomerase property of WT MIF (24) with that of the pseudo-(E)LR mutants. Fig. 2C shows that the mutants exhibited an identical tautomerase activity toward D-dopachrome methyl ester. We further explored whether mutations in the pseudo-(E)LR motif may induce more subtle conformational changes. Indeed, guanidine hydrochloride (GdnHCl)-induced unfolding studies in combination with CD spectropolarimetry showed that the mutants had slightly, yet distinguishably, different conformational stabilities as reflected by their midpoints of unfolding. The unfolding behavior of the double mutant was closest to that of WT MIF (K_U50[WT MIF], 1.71 M; K_U50[R11A/D44A/MIF], 1.77 M), whereas the R11 mutant was more stable (K_U50[R11A/MIF], 1.94; D44A/MIF displayed a lower conformational stability than WT MIF (K_U50[D44A/MIF], 1.34 M; Fig. 2B and Fig. S2). This indicates that the mutants may behave distinctly in protein–protein interactions, which, in conjunction with charge effects, could be relevant for interactions with the MIF receptor CXCR2.

Fig. 2.

Characterization of the pseudo-(E)LR MIF mutants. The pseudo-(E)LR mutants were prepared by an almost identical procedure as that established for WT MIF. (A) Purification of the pseudo-(E)LR mutants. Representative, silver-stained SDS gel of the purification of R11A-MIF. Cell lysates were prepared by high-spin centrifugation (L and P). Upon concentration (C, FT), mutants were purified by anion exchange chromatography (MQ) and by C8 reverse-phase chromatography (washing with 20% acetonitrile [W], elution with 60% acetonitrile [E1, E2]). The molecular weight (M_r) is indicated on the Left. (B) Overall structural integrity of the pseudo-(E)LR mutants as evidenced by CD spectroscopy. Renatured mutants were compared with WT MIF. Mean residue ellipticities per residue (θ) are plotted against the wavelength. Spectra are representative of three independent recordings. (C) Pseudo-(E)LR mutants show an identical tautomerase activity as WT MIF. Monitoring of tautomerase activity by measuring the decrease in absorbance at 475 nm over a 4-min time period. Data represent means ± SEM of three independent experiments. (D) Distinct conformational stabilities of the mutants. GdnHCl-induced unfolding of WT MIF and pseudo-(E)LR mutants as followed by CD spectroscopy. Unfolding curves are presented as the percentage of unfolded relative to native protein calculated from the change in ellipticity at 222 nm over the concentration of GdnHCl. Data represent means ± SEM of three independent experiments.

CXCR2 Binding Activity of the Pseudo-(E)LR Mutants.

Binding activity of the pseudo-(E)LR mutants to CXCR2 was tested in a receptor competition assay using the radiolabeled cognate ligand CXCL8 as established for WT MIF (15). Contrary to WT MIF, which strongly and specifically competed with [I¹²⁵]CXCL8 for CXCR2 binding, none of the mutants showed a significant CXCR2 receptor binding activity, but only exhibited a minor residual activity (Fig. 3 A and B). This indicated that the pseudo-(E)LR motif is a critical determinant of the MIF/CXCR2 protein–protein interaction site.

Fig. 3.

Abrogated CXCR2 binding activity of the pseudo-(E)LR mutants. HEK293 cells ectopically expressing surface CXCR2 were incubated with radioiodinated [I¹²⁵]CXCL8 tracer together with WT MIF or pseudo-(E)LR mutants as competitor as indicated. Plots represent percent of specific [I¹²⁵]CXCL8 binding at indicated competitor concentrations. (A) Curve diagram of the receptor competition assay over a range of competitor concentrations. (B) Comparison of the competitor effects at a competitor concentration of 1 × 10⁻⁶ M MIF or pseudo-(E)LR mutants and 1 × 10⁻⁷ M CXCL8. Data represent means ± SEM of 7 to 10 independent experiments. Asterisks indicate that the impaired binding activity of the pseudo-(E)LR mutants differs significantly from the effect of WT MIF at the indicated concentration (*, P < 0.01; **, P < 0.005; ***, P < 0.0005).

Effects on CXCR2-Mediated Monocyte Arrest and Chemotaxis.

The strongly reduced CXCR2 binding activity of the pseudo-(E)LR mutants could imply that the pseudo-(E)LR motif could be crucial for the leukocyte recruitment activity of MIF. In fact, we have previously reported that immobilization of MIF on human aortic endothelial cells triggers CXCR2-mediated monocyte arrest under flow conditions; MIF thus exhibits CXCR2-dependent chemokine-like functions (15). Here we confirmed that a preincubation of activated aortic endothelial cells for 2 h indeed results in a 2-fold increase of firm shear-resistant monocyte adhesion under flow conditions. Strikingly, exposure of the endothelial layer to the R11A, D44A, or R11A/D44A mutants of MIF failed to trigger the arrest-inducing activity (Fig. 4A). Thus, the pseudo-(E)LR motif appears to be a crucial determinant for the CXCR2-mediated arrest function of MIF.

Fig. 4.

The pseudo-(E)LR motif is crucial for the monocyte recruitment activity of MIF in vitro. (A) Effect on CXCR2-mediated monocyte arrest. HAoECs were preincubated with WT MIF, pseudo-(E)LR mutants, or buffer before the HAoEC monolayer was perfused with calcein AM-stained MonoMac6 cells. Adherent cells were quantified by counting multiple high-power fields; data (means ± SEM of 7 to 9 independent experiments) are represented as relative increase of adherent monocytes over control. (B) Effect on CXCR2-mediated monocyte chemotaxis. THP-1 monocytes were added to the upper compartment of a Transwell chamber and allowed to migrate in response to WT MIF or pseudo-(E)LR mutants in the lower chamber. The plot represents the chemotactic index at a protein concentration of 100 ng/ml. Data represent means ± SEM of 5 to 6 independent experiments.

It is of note that replacing the native Asp at position 44 with Glu to yield the D44E mutant of MIF resulted in a slightly increased binding activity to CXCR2, and did not alter the MIF-triggered monocyte arrest function and in vivo leukocyte recruitment compared with WT MIF (Fig. S3, data not shown). The preserved function of the D44E mutant of MIF thus justifies the use of the term pseudo-(E)LR.

In addition, MIF exhibits chemotactic activity for monocytes and T cells via CXCR2 and CXCR4, respectively. In chemotaxis assays with THP-1 monocytes, R11A-MIF and R11A/D44A-MIF showed a significantly impaired chemotactic activity compared with the WT protein, whereas the D44A mutant largely retained its chemotactic activity (Fig. 4B and Fig. S4).

Role of the Pseudo-(E)LR Motif in Atherogenic Leukocyte Recruitment.

To corroborate the relevance of the pseudo-(E)LR motif for CXCR2-mediated functions of MIF and to assess its role in atherogenic leukocyte recruitment, we analyzed the effect of MIF loading on monocyte recruitment in ex vivo perfused carotid arteries of mif^−/−ldlr^−/− mice with early atherosclerotic endothelium, which has been previously shown to be CXCR2-dependent (15). Exposure of arteries to WT MIF but not to R11A-MIF or R11A/D44A-MIF for 90 min triggered monocyte recruitment in early atherosclerotic arteries (Fig. 5). Again, the D44A mutant retained its activity in mediating monocyte arrest in the arteries (Fig. 5), which was also observed with T cells and was thus likely mediated by CXCR4 (data not shown). The difference seen for the D44A mutation, when comparing its effect on monocyte arrest on human aortic endothelial cells in vitro (Fig. 4A) versus monocyte arrest in ex vivo perfused mouse carotid arteries (Fig. 5), may be related to the atherogenic stimulation or to differences in the requirements for surface immobilization and presentation of MIF between different vascular beds and different species, and may also account for the somewhat less marked functional impairment of the double mutant in Fig. 5. Thus, CXCR2 functions of MIF clearly require the pseudo-(E)LR motif, in particular the Arg-11 residue, which parallels the structure–function requirements of other CXCR2 ligands.

Fig. 5.

The pseudo-(E)LR motif is critical for MIF-mediated monocyte arrest in atherosclerotic carotid arteries. Effect of the pseudo-(E)LR mutants on the arrest of MonoMac6 monocytes in carotid arteries from mif^−/−ldlr^−/− mice fed a Western diet for 6 weeks and comparison with WT MIF. Upon perfusion with calcein AM-labeled monocytes for 10 min, adherent cells per carotid artery were counted. The number of adherent cells after treatment with pseudo-(E)LR mutants as indicated or buffer (control) was normalized in comparison to treatment with WT MIF. Data represent means ± SEM of three independent experiments.

Function of Pseudo-(E)LR Mutants in a Model of Peritoneal Inflammation.

The activity of MIF in leukocyte recruitment, in particular that directed at neutrophils, has also been evident in a model of peritoneal inflammation in vivo (15). Indeed, we confirmed that injection of MIF into the peritoneum of C57BL/6 mice resulted in a potent recruitment of neutrophils. Again, the recruitment function of MIF was fully abrogated by the R11A and R11A/D44A mutations of MIF, whereas neutrophil infiltration induced by D44A/MIF was not impaired and was even more pronounced (Fig. 6). The preserved recruitment function triggered by the D44A mutant was mediated by CXCR4, as indicated by a blockade of this effect with AMD3465. Thus, CXCR4 can partly compensate for a defect in CXCR2 binding activity in the D44A mutant to mediate MIF-induced activities. Overall, these data point at the importance of the pseudo-(E)LR motif for CXCR2-mediated functions of MIF in the inflammatory recruitment of leukocytes.

Fig. 6.

Role for the pseudo-(E)LR motif in MIF-triggered neutrophil recruitment in an acute in vivo murine peritonitis model. C57BL/6 mice (n = 3 per group) were i.p. injected with 200 ng WT MIF or pseudo-(E)LR mutants as shown, either alone or together with the CXCR4 antagonist AMD3465 as indicated. Peritoneal neutrophil infiltration, expressed as percent of CD45⁺ cells, was determined by FACS analysis. Data represent means ± SEM of three independent experiments.

Discussion

The evolutionary conserved cytokine MIF has more recently been recognized as a potent inflammatory mediator in various diseases (12–14). We have recently identified the CXC chemokine receptors CXCR2 and CXCR4 as the previously elusive signaling receptors for MIF in inflammatory and atherogenic cell recruitment (15). Although of remarkably high affinity, in particular to CXCR2, the structural basis for this non-cognate interaction has remained unknown.

Here we demonstrate that MIF shares a functional ELR-like motif with the canonical CXCR2 ligands (25), which we term the pseudo-(E)LR motif. This motif comprises the non-adjacent residues Asp-44-X-Arg-11 (i.e., DXR) localized in neighboring loops of a MIF monomer (23) and displays a spacing identical to that in the canonical ELR motif. Structure–function analysis revealed that substitution of these residues, in particular R11, severely compromises the binding of MIF to CXCR2 and abrogates MIF/CXCR2-mediated functions in leukocyte recruitment in various experimental settings in vitro, ex vivo, and in vivo.

Whereas substitutions of R11 as studied through the mutants R11A/MIF and R11A/D44A/MIF more or less result in a complete loss of function in all assays tested, the D44A single mutant shows preserved activity depending on the model used, for instance in peritoneal neutrophil recruitment. These data reveal the importance of the pseudo-(E)LR motif, in particular the Arg-11 residue, in CXCR2-mediated functions of MIF. The Asp-44 residue is also critical for predominantly CXCR2-mediated functions, but appears to be dispensable for CXCR4-dependent activities of MIF, as indicated by an observed specific inhibition with the CXCR4 antagonist AMD3465. Alternatively, the preserved functions of the D44A mutant in some but not all assay systems tested may be related to a different extent in the contribution of CXCR2 to the assay-specific signal. For instance, the more stringent (regarding MIF dependence) arrest assay under flow conditions in vitro may heavily rely on CXCR2; or the contact area with both residues of the ELR motif is a prerequisite for effective presentation and receptor function of MIF in a reconstituted local environment, whereas monocyte recruitment in ex vivo perfused carotid arteries may be more profoundly affected by a complex set of mediators including other interaction partners in the (patho-)physiological and proatherogenic context of feeding a high-fat diet over 8 weeks (3).

Interestingly, substitution of ELR residues in ELR⁺ CXC chemokines (e.g., CXCL8) results in a dramatic loss of CXCR2 binding affinity and neutrophil-activating properties, and introduction of ELR residues into CXC chemokines such as CXCL4, but not CXCL10, confers such properties (18–22). This demonstrates that, although possible and effective in principle, an embedding of the ELR motif is critically dependent on the specific structural context to achieve functional relevance in receptor binding and activity. In fact, MIF not only features an “embedded” ELR-like motif, but the appropriate structural environment appears to have co-evolved in this ancient mediator to allow for the pseudo-(E)LR motif in MIF to serve as a potent CXCR2 agonist structural moiety.

In conjunction, our data offer a structural explanation for the surprising identification of MIF as a non-cognate yet functional CXCR ligand (15) and provide an illustrative example for an intriguing “molecular mimicry” within the chemokine system by epitomizing the structural resemblance of a chemokine-like ligand with classical chemokines. The discovery that the cytokine MIF, which is secreted by a variety of host cells following inflammatory stimulation, indicates that not only “alarmins” such as the defensins (8, 26) and host autoantigens such as the AaaRS that are liberated from dying cells or after pathogen challenge to recruit antigen-presenting cells (26), but also MIF can serve as a recruitment signal during host defense and inflammation by targeting chemokine receptors. Thus, host exploitation of the chemokine receptor system expands much further to encompass mononuclear and neutrophil inflammatory cell recruitment (15).

Recent studies have provided convincing evidence for an important role of CXCR2 in atherosclerotic lesion formation and macrophage infiltration by demonstrating that deficiency of CXCR2 in bone marrow cells more markedly inhibited lesion formation than deletion of single canonical ligands, e.g., CXCL1, in atherosclerosis-prone mice fed a high-fat diet (27). This had led to speculations about the involvement of additional hitherto-unidentified CXCR2 ligands in this process. Indeed, it has recently been substantiated that interference with MIF results in a reduction of lesion formation upon its genetic deletion (15, 28) and even in a regression of advanced lesions upon antibody inhibition (15). Together with this emerging importance of MIF as a CXCR2 ligand in the inflammatory pathogenesis of atherosclerosis, this study paves the way for the rational design of strategies selectively targeting proinflammatory and atherogenic MIF/CXCR2 interactions by pinpointing the pseudo-(E)LR motif as being integral for the structure–function relationships. Beyond their potential therapeutic relevance, our data have unraveled critical determinants of the MIF/CXCR receptor interface. This should enable future studies aiming at a prediction of MIF-specific CXCR2-mediated signaling responses as they may occur in inflammatory or pro-atherogenic environments, as opposed to those triggered by the cognate CXCR2 ligands, and at dissecting the specific characteristics of CXCR2/ligand interactions among the various CXCR2 ligands, including MIF.

Materials and Methods

Cell Culture.

Human aortic endothelial cells (HAoECs) were from PromoCell. Cells were used at passages 3 to 5. MonoMac6 cells were a gift of Prof. Ziegler-Heitbrock (University of Leicester, U.K.) and were cultured as described (29). HEK293-CXCR2 transfectants were a gift of Dr. Ben-Baruch (Tel Aviv University, Israel). THP-1 cells (ATCC) were grown in RPMI 1640 containing 10% FCS. Miscellaneous cell culture reagents were from Invitrogen.

Proteins and Reagents.

Biologically active human MIF was expressed, purified, and renatured as described (30, 31). Human MCP-1 was from PeproTech. Calcein AM was purchased from Calbiochem. All other reagents were from Sigma, Merck, Roth, or Calbiochem.

Site-Directed Mutagenesis and Cloning of the Pseudo-(E)LR Mutants of MIF.

The pseudo-(E)LR mutants were cloned using the QuikChange Site-Directed Mutagenesis Kit (Stratagene). Briefly, the mutants were cloned from the pET11b expression vector pET11b-huMIF (30) according to the manufacturer's protocol using the following primers: MIF-R11A forward, 5′-CAC CAA CGT GCC CGC CGC CTC CGT GCC GGA CG-3′; MIF-R11A reverse, 5′-CGT CCG GCA CGG AGG CGG CGG GCA CGT TGG TG-3′ for mutant R11A-MIF; MIF-D44A forward, 5′-GTG CAC GTG GTC CCG GCC CAG CTC ATG GCC TCC-3′; and MIF-D44A reverse, 5′-GAA GGC CAT GAG CTG GGC CGG GAC CAC GTG CAC-3′ for D44A-MIF. PCR reactions were performed using the following conditions: 30 s at 95°C followed by 16 cycles of 30 s at 95°C, 1 min at 55°C, and 7 min at 68°C using PfuTurbo DNA polymerase. The double mutant R11A/D44A-MIF was cloned from pET11b-R11A-MIF using primers D44A-MIF-forward and D44A-MIF-reverse. For D44E-MIF, the following primers were used: MIF-D44E-forward, 5′-GTG CAC GTG GTC CCG GAA CAG CTC ATG GCC TTC-3′; and MIF-D44E reverse, 5′- GAA GGC CAT GAG CTG TTC CGG GAC CAC GTG CAC -3′. DNA sequences were confirmed by bidirectional sequencing (MWG). Expression, purification, and renaturation of the pseudo-(E)LR mutants were performed essentially using the protocol for WT MIF (30) except that proteins were expressed at 20°C overnight following induction with 0.1 mM IPTG and that a French press (1,240 psi) was used for cell disruption.

Biochemical Characterization of the Pseudo-(E)LR Mutants.

To verify expression levels and identity of the mutants, SDS/PAGE electrophoresis in combination with silver staining was performed as described (30) with slight modifications. Gels were silver-stained by a routine protocol (30). Mutations were confirmed as well by quantitative amino acid analysis as described (32), which also served for quantification of the mutant proteins preparations

D-Dopachrome Tautomerase Assay.

The D-dopachrome methyl ester substrate was prepared in situ through oxidation of L-dopa methyl ester with sodium periodate, and catalytic activities were analyzed as previously described (32) and as outlined in detail in the SI Methods.

Far-UV CD and Analysis of Conformational Stability.

Far-UV CD spectra were recorded in a 202SF spectropolarimeter (AVIV Instruments). Scans were recorded at 25°C in the range between 195 and 250 nm and collected at 1.0-nm intervals with a bandwidth of 1 nm. Quartz cells with a path length of 10 mm were used. Spectra are presented as a plot of the mean molar ellipticity per residue ([θ], deg cm² dmol⁻¹) versus the wavelength. Renatured MIF or mutant protein was diluted from the respective stock solution to a final concentration of 1 μM in 20 mM sodium phosphate buffer, pH 7.2.

GdnHCl-induced unfolding of mutants in comparison to WT MIF was followed by CD spectropolarimetry as described previously and as outlined in detail in the SI Methods (30).

Flow Chamber Adhesion Assay.

Laminar flow assays were performed as described (15), using HAoEC monolayers, which were preincubated with MIF (50 ng/ml) or control buffer and MonoMac6 cells labeled with calcein AM. (See also SI Methods.)

Competitive Receptor Binding Assays.

For MIF/CXCL8 receptor competition assays, stable HEK293-CXCR2 transfectants and radiolabeled [I¹²⁵]CXCL8 tracer were used and assays performed as described (15). The competitive effect of the pseudo-(E)LR mutant proteins was compared with that of MIF. Plots represent the percentage of specific [I¹²⁵]CXCL8 binding at the indicated ligand/competitor concentrations. Margins for displacement curves refer to tracer binding in the absence (100%) versus the presence (0%) of 100 nM cold CXCL8, with non-specific binding rates ranging from 10% to 20% (15). For control, mock-transfected HEK cells were used, which exhibited negligible binding rates of ≈500 cpm. This value was subtracted.

Chemotaxis Assay.

Chemotaxis assays were performed in 96-well plates (MultiScreen; Millipore) with a pore size of 5 μm. THP-1 cells were labeled with calcein AM and resuspended in RPMI 1640 containing 0.5% BSA at 1 × 10⁶ cells/ml. Fifty microliters of the cell suspension was transferred to the upper chamber of the plates. The lower chamber contained different concentrations of MIF or MIF mutants varying from 0.1 to 250 ng/ml in 150 μl RPMI 1640/0.5% BSA as chemoattractant. Monocytes were allowed to transmigrate for 1.5 h. Filters were removed and 100 μl of the cell suspension in the lower well transferred to a black 96-well plate (Nunc). Transmigrated cells were quantified using a Victor² fluorescent plate reader (Perkin-Elmer).

Ex vivo Perfusion of Atherogenic Murine Carotid Arteries.

Mif-deficient mice (mif^−/−) (33) were crossbred with LRL receptor (Ldlr)-deficient mice (ldlr^−/−; Charles River Laboratories) to establish mif^+/−ldlr^−/− mice (all C57BL/6). mif^+/−ldlr^−/− animals of the F9 generation were used to generate mif^−/−ldlr^−/− mice and ldlr^−/− littermates. Transgenic mice were fertile and showed no phenotypical abnormalities. Preparation for ex vivo perfusion of murine carotid arteries was performed as described (34) and as outlined in the SI Methods. Arteries were perfused with 50 ng/ml of WT MIF or pseudo-(E)LR mutants over 90 min at 4 μl/min and afterward at 4 μl/min with MonoMac6 cells labeled with calcein AM (10⁶/ml). Total monocyte arrest after treatment with the pseudo-(E)LR mutants or untreated control (i.e., Mops-buffer) was normalized to arrest obtained with WT MIF. All studies were approved by local authorities and complied with German animal protection law.

Model of Acute Peritonitis.

WT C57BL/6 mice were injected i.p. with 200 ng MIF or mutant protein in sterile PBS solution. Additionally, AMD3465 (200 ng) was injected i.p. together with WT MIF or mutant D44A-MIF (200 ng each) to see whether neutrophil recruitment was CXCR4-dependent. The number of infiltrated neutrophils was analyzed as described previously (15).

Statistical Analysis.

Data are expressed as means ± SEM. Analysis of variance and Student t tests were performed to compare experimental groups. Differences with a P value <0.05 were considered significant.