Recombinant Human Vimentin Binds to P-Selectin and Blocks Neutrophil Capture and Rolling on Platelets and Endothelium

Leukocyte (WBC) interaction and adhesion to platelets and vascular endothelium plays an integral role in the initiation of inflammation. These early steps are mediated through cell surface adhesion molecules on the leukocytes (e.g., P-selectin glycoprotein ligand-1 [PSGL-1] and CD11b/18 [Mac-1]) to counterreceptors on platelets and endothelium (e.g., P-selectin, GP1bα, E-selectin, and ICAM-1) (1–6). In most cases, the presence of inflammation plays an important role in the maintenance of health, such as in infection (7) and wound healing (8). However, in pathologic inflammation, these processes may lead to more tissue injury and organ dysfunction, such as in acute lung injury (ALI) (9, 10), ischemia/reperfusion injury (11, 12), and inflammation from extracorporeal life support devices (13, 14). Therefore, blocking these interactions may have a beneficial role in decreasing pathologic inflammation in specific disease states.

One way to block leukocyte interaction with platelets and endothelium is through inhibiting the binding of leukocyte surface adhesion molecules to those on platelets and endothelium. We and other investigators have reported an important role for PSGL-1–P-selectin interactions on leukocyte adhesion to (11) and transmigration across (4) vascular endothelium. In a murine model of experimental ALI, blocking P-selectin using a mAb improved oxygenation and decreased leukocytic infiltration into the lungs (9).

Vimentin is an intracellular type 3 intermediate filament protein that is important for maintaining the cytoskeleton, as well as intracellular transport. It is highly conserved, with ∼96% homology between humans and mice and, in its native state, it is mainly found as helical dimers and tetramers (15). Vimentin is primarily located within mesenchymal cells; however, recently, it has also been reported on the surface of cells (16), as well as secreted into the plasma (17). Previous reports on the role of vimentin in inflammation have been conflicting, with reports of vimentin-knockout mice having increased phagocytic capabilities (18), as well as decreased lymphocyte adhesion and transmigration across endothelial cells (19). This discrepancy may be due to the differential functions of intracellular and extracellular vimentin (20).

Although the exact role of vimentin in inflammation is complex, vimentin has recently been reported to bind to N-acetylglucosamine (21), a moiety that is present on PSGL-1 on leukocytes (22, 23). In addition, soluble CD44, the soluble form of E-selectin ligand normally found on neutrophils (24), was recently reported to bind to vimentin (25). In addition to E-selectin, CD44 variants have been reported to interact with P-selectin and L-selectin (26). Based on these data, we tested the hypothesis that our recombinant human vimentin (rhVim) may inhibit leukocyte adhesion to platelets and endothelium through blocking interactions between the selectins and their ligands.

Human and animal subject research were approved by the Institutional Review Boards and Institutional Animal Care and Use Committees at Baylor College of Medicine and the Michael E. DeBakey Veterans Affairs Medical Center.

Escherichia coli (strain M15) was purchased from QIAGEN. HRP-conjugated anti-polyhistidine Ab, 1× protease inhibitor, BSA, tetramethylbenzidine, and mepacrine were purchased from Sigma-Aldrich. Dulbecco’s PBS (DPBS) was purchased from Life Technologies. Recombinant human IL-1β (rhIL-1β), recombinant human IL-4 (rhIL-4), P-selectin/Fc, E-selectin/Fc, and PSGL-1/Fc were purchased from R&D Systems. Human IgG was purchased from Pierce. Sheep anti-vimentin Ab was purchased from Affinity Biologicals. Rabbit anti-vimentin (H-84; N terminus), rabbit anti-vimentin (C-20; C terminus), and mouse anti-vimentin (E-5; C terminus) Abs were purchased from Santa Cruz Biotechnology. Mouse anti-vimentin (V9) Ab was purchased from Invitrogen. The following functional blocking Abs were used: anti–P-selectin (AK4; BioLegend), anti–PSGL-1 (KPL-1; BD), anti–E-selectin (BBIG-E1; R&D Systems), anti–VCAM-1 (BBIG-V1; R&D Systems), anti–ICAM-1 (11C81; R&D Systems), and mouse IgG isotype control (BD). HUVECs were purchased from Lonza (Clonetics). Fibronectin was purchased from Advanced BioMatrix. BioFlux microfluidic plates were purchased from Fluxion Biosciences. CM5 sensor chips were purchased from GE Healthcare. Amine-reactive second-generation sensors were purchased from Pall ForteBio. Endotoxin (LPS; E. coli O111:B4) was purchased from Sigma-Aldrich.

The human vimentin sequence was synthesized using sequence NM_003380.3 (National Center for Biotechnology Information) from nt 418–1814 (the mRNA sequence) and placed into pQE-30 plasmid (GenScript). The resulting pQE-30–human vimentin plasmid was transformed into E. coli (strain M15) for protein expression, following IPTG induction. E. coli bacteria were pelleted down and lysed to obtain inclusion bodies. Then the inclusion body pellets were lysed with lysis buffer (6 M guanidine hydrochloride, 25 mM Tris, [pH 7.4]) supplemented with 1× protease inhibitor at room temperature for 2 h with constant stirring. Bacterial lysates were centrifuged at 19,000 rpm, and the supernatants were filtered using a 0.45-μm filter before loading onto the Ni²⁺ affinity column, which had been equilibrated with binding buffer (7 M urea, 25 mM Tris [pH 8]). Bound proteins were eluted with elution buffer (200 mM imidazole, 8 M urea, 25 mM Tris [pH 8]). The purified sample was serially dialyzed against decreasing concentrations of urea in Tris buffer at 4°C. Samples were ultimately dialyzed against 20 mM sodium phosphate buffer [pH 8] overnight at 4°C and filtered using a 0.22-μm filter under sterile conditions. Absorbance was measured at 280 nm to determine protein concentration using an extinction coefficient of 21,425 l/M/cm and molecular mass ∼ 57 kDa. Purity was confirmed using Coomassie blue stain and Western blot using an HRP-conjugated anti-polyhistidine Ab under reduced and nonreduced conditions (Fig. 1). rhVim was stored at 4°C until use, with new protein purified monthly. The ultimate dialysis buffer was used as the negative (vehicle) control. In some experiments, rhVim was boiled for 20 min to heat denature the protein for use as a secondary protein control.

After informed consent was obtained, blood was collected from healthy unmedicated adult human subjects. To determine the effect of soluble vimentin in whole blood, 3.2% citrated whole blood (1:9) was labeled with 10 μM mepacrine for 20 min at 37°C to label platelets and leukocytes (11). To assess the effect of soluble vimentin on isolated platelets and neutrophils, whole blood was processed, as previously described, to collect washed platelets (27) and neutrophils (4). In some experiments, isolated neutrophils were labeled with mepacrine, as described above, for better visualization. The purity of isolated PMNs was evaluated by flow cytometry using an ImageStream Mark II imaging flow cytometer (Amnis; Millipore Sigma). Isolated PMNs were labeled with anti-CD45/allophycocyanin and anti-CD62P/PE Abs (BD Pharmingen) to determine the presence, if any, of P-selectin on isolated PMNs. As positive control, isolated platelets were added to isolated PMNs and similarly labeled for flow cytometry. In our study, <3% of isolated PMNs were CD62P⁺ compared with 34% when isolated platelets were added to isolated PMNs (Supplemental Fig. 1).

Parallel-plate flow-adhesion studies were performed using BioFlux microfluidic plates, as we described previously (28). All assays were imaged using the FITC channel (200-ms exposure) and a 10× objective lens (Zeiss); this allowed for image capture of two channels within a single field of view. Static and time-series images (100 frames) were captured. Images were analyzed using ImageJ software (National Institutes of Health). The PMNs imaged were counted as captured or adhered within identical areas of study. Rolling analyses were performed on 20 random PMNs per channel, and the results were averaged based on distance traveled within 100 frames (200 milliseconds per frame) or until the PMN was outside the field of view.

To study the effect of rhVim on leukocyte–platelet interactions in whole blood, channels were coated with fibrin(ogen) (100 μg/ml) for 1 h at room temperature to capture platelets (28). The channels were washed with DPBS without calcium or magnesium (pH 7.2–7.4) to remove excess fibrin(ogen). Mepacrine-labeled whole blood, in the presence of rhVim or vehicle, was perfused over fibrin(ogen)-coated chambers at 4 dyn/cm² for 10 min to allow for initial platelet activation and adhesion to fibrin(ogen). The shear stress was then reduced to 2 dyn/cm², and images were recorded for 10 min to study leukocyte–platelet adhesion in the presence of rhVim (Supplemental Video 1).

To study the effect of rhVim on the adhesion of isolated neutrophils, channels were coated with isolated platelets (400 × 10⁶ per milliliter) for 1 h at room temperature or with fibrin(ogen) (100 μg/ml) or were left uncoated. Channels were washed with DPBS without calcium or magnesium. Mepacrine-labeled neutrophils (4 × 10⁶ per milliliter) were perfused through the channels in the presence of various concentrations of rhVim; blocking Abs to P-selectin and PSGL-1 were used as positive controls. Additional experiments were performed using channels coated with P-selectin/Fc chimeric protein (10 μg/ml) to determine the effect of rhVim on neutrophil adhesion to P-selectin.

To study the effect of rhVim on leukocyte adhesion to endothelium, microfluidic channels were coated with fibronectin (100 μg/ml) for 1 h at room temperature. HUVECs (passage 2–4) were seeded in the channels and incubated at 37°C/5% CO₂. Once the HUVECs reached confluence, they were stimulated with rhIL-1β (10 U/ml) for 4 h at 37°C to activate them (4). Then, mepacrine-labeled whole blood or isolated PMNs were perfused (2 dyn/cm²) over IL-1β–stimulated HUVECs in the presence of rhVim or boiled rhVim. To determine whether the effect of rhVim on blocking PMN adhesion to HUVECs was indeed through P-selectin, we stimulated HUVECs with a combination of rhIL-4 (20 ng/ml) and rhIL-1β (100 U/ml) for 24 h to induce P-selectin upregulation, as described previously (29). Initial experiments were performed with isolated PMNs in the presence of anti–P-selectin Abs or isotype control to test whether P-selectin was involved in PMN capture and rolling adhesion in this system. Isolated PMNs were perfused over activated HUVECs at 2 dyn/cm² in the presence of 50 μg/ml rhVim and blocking Abs to E-selectin (25 μg/ml; BBIG-E1), P-selectin (10 μg/ml; AK4), VCAM-1 (30 μg/ml; BBIG-V1), ICAM-1 (10 μg/ml; 11C81), or mouse IgG (30 μg/ml) isotype control, using the manufacturers’ recommendations. PMNs with vehicle only were used as a negative control. Images were recorded as described above (Supplemental Video 2).

To determine the proteins to which rhVim binds, we performed ELISAs, as we described previously (16). We coated immunoplate wells overnight at 4°C with P-selectin/Fc or PSGL-1/Fc (10 μg/ml in DPBS). Wells were washed with DPBS and then blocked with 1% BSA for 1 h. After blocking, rhVim was added in increasing concentrations and detected using an HRP-conjugated anti-polyhistidine Ab. Tetramethylbenzidine solution was used as the substrate, and the reactions were stopped using 1 N sulfuric acid. All ELISA measurements were performed in duplicate, and the data reported were averaged from two separate experiments performed.

To determine the binding affinity of P-selectin and E-selectin to rhVim, we measured K_D of P-selectin binding to rhVim using surface plasmon resonance (SPR; Biacore; GE Healthcare) (30). rhVim (15 mg/ml diluted in 10 mM sodium acetate [pH 4.5]) was immobilized on CM5 sensor chips using 0.1 M N-hydroxysuccinimide and 0.4 M 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide. The amine-coupling reaction was quenched with 1 M ethanolamine-HCl (pH 8.5). P-selectin/Fc, E-selectin/Fc, or human IgG (control) was perfused over the chip using a concentration range of 10–1000 nM. Regeneration was performed using 20 mM sodium hydroxide and 10 mM glycine (pH 2). DPBS was used as the running buffer. Data were analyzed using Biacore analysis software to determine K_D.

In addition to SPR, we used bio-layer interferometry (BLI; Octet Red 384; Pall ForteBio) to determine the K_D of P-selectin, E-selectin, and PSGL-1 to rhVim. We immobilized rhVim (50 μg/ml) onto amine-reactive second-generation sensors (Pall ForteBio) using N-hydroxysuccinimide and 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide. The amine-coupling reaction was quenched with 1 M ethanolamine-HCl (pH 8.5) and washed with DPBS. rhVim-immobilized sensors were used to assess binding kinetics to the following analytes at concentrations ranging from 0 to 1000 nM (diluted in DPBS): P-selectin/Fc, E-selectin/Fc, PSGL-1/Fc, human IgG, and sheep anti-vimentin Ab. We also performed BLI to determine whether rhVim (30 μg/ml) bound to the following immobilized (onto amine-reactive second-generation sensors, as described above) commercial anti-vimentin Abs (10 μg/ml): H-84, C-20, E-5, V9, and sheep anti-vimentin (Fig. 2). DPBS was used as buffer for all association and dissociation kinetics. Regeneration was performed using 10 mM glycine (pH 1.75). Data were analyzed using Octet System Data Analysis software (Pall ForteBio).

Male and female C57BL/6J wild-type mice, ages 10–11 wk, were used for this study. To assess whether rhVim would decrease leukocyte accumulation into inflamed lungs, mice were injected i.p. with 3 mg/kg rhVim, followed 1 h later by a sublethal dose of LPS (15 mg/kg i.p.). Blinded investigators evaluated the mice, using the mouse clinical activity score for sepsis (31). Ninety-six hours after LPS injection, mice were euthanized under a surgical plane of anesthesia. Lung fixation was accomplished by intratracheal instillation of 10% formalin in situ fixation. The trachea was tied with surgical suture to maintain inflated lung tissue for proper fixation. Tissues were submerged in 10% formalin for 72 h prior to tissue processing and paraffin embedding. Tissues were sectioned (5 μm thick) and slides were prepared with H&E. Images were captured on an Olympus BX-61 microscope with a JVC KY-F75U camera with ImageJ Micro Manager software. Two blinded investigators scored the H&E sections for histologic ALI using the method defined by the American Thoracic Society (32). Briefly, 20 random 400× fields were imaged and independently scored using a weighted average evaluating the number of alveolar neutrophils, the number of interstitial neutrophils, the presence of hyaline membranes, the proteinaceous material filling alveoli, and alveolar wall thickness. The weighted average provides a score between 0 and 1, with 1 being the most severe. The scores were averaged to determine the score for that animal.

Data analysis was performed using Prism 6 (GraphPad Software). In vitro data were analyzed using a Student paired t test, repeated-measures ANOVA with the Dunnett or Tukey multiple-comparison test, or two-way repeated measures ANOVA with the Bonferroni multiple-comparison test, where appropriate. In vivo data were analyzed using a Student t test. A p value < 0.05 was considered significant. K_D values were considered indeterminate if R² < +0.8.

Based on previous reports of vimentin being secreted and its ability to bind to N-acetylglucosamine, we studied the effect of rhVim on WBC capture and adhesion. We perfused (2 dyn/cm²) mepacrine-labeled whole blood over surfaces coated with fibrin(ogen), which exposes fibrin-specific sequences upon surface adsorption (Fig. 3A). In this assay, platelets adhere to fibrin(ogen) and become activated, consequently capturing the flowing WBCs. The addition of rhVim had no effect on platelet adhesion to the fibrin(ogen) surface based on fluorescence imaging (data not shown). However, 50 μg/ml rhVim significantly decreased the number of adhered WBCs flowing over fibrin(ogen) monolayers by ∼55% (Fig. 3B).

To determine whether our observation that rhVim decreases WBC adhesion to fibrin(ogen) is due to disruption of WBC–fibrin(ogen) interactions (33) or WBC–platelet interactions (1, 4), we perfused isolated human PMNs over microfluidic channels coated with fibrin(ogen), platelets, or no coating (glass; Fig. 4A–C). Isolated PMNs did not adhere to fibrin(ogen), with or without rhVim. Additionally, PMN adhesion to glass was not affected by the presence of rhVim; however, the presence of rhVim significantly decreased PMN adhesion to platelet monolayers (Fig. 4D). Furthermore, rhVim decreased PMN adhesion to platelet monolayers in a dose-dependent manner, with maximal inhibition at an rhVim concentration of 40 μg/ml (Fig. 5A). Blocking Abs to P-selectin (AK4; 10 μg/ml) and PSGL-1 (KPL-1; 2.5 μg/ml) similarly decreased PMN adhesion to platelet monolayers; however, the addition of 5 μg/ml rhVim did not further decrease PMN adhesion (Fig. 5B).

In addition to its effect on platelet–neutrophil interactions, we tested whether rhVim impairs the capture and adhesion of neutrophils to endothelial cells. We cultured HUVECs on fibronectin-coated channels (Fig. 6A); once they reached 100% confluence, HUVECs were stimulated with rhIL-1β for 4 h (4). We then perfused isolated PMNs in the presence of rhVim (40 μg/ml) or vehicle control over the HUVEC monolayer. The presence of rhVim significantly decreased the number of adherent PMNs to HUVEC monolayers by ∼67% (Fig. 6B).

We then compared the effect of rhVim (50 μg/ml) with vehicle control and boiled rhVim (50 μg/ml) to determine whether the effects on PMN adhesion to HUVECs was specifically due to rhVim. Like our previous observations, rhVim significantly decreased PMN adhesion to HUVEC monolayers compared with vehicle control and boiled rhVim. Furthermore, we found no difference in PMN adhesion between vehicle control and boiled rhVim (Fig. 6C). Similarly, average PMN rolling velocity was significantly increased in the presence of rhVim compared with vehicle control and boiled rhVim (Fig. 6D).

To better assess the role of P-selectin on PMN adhesion to inflamed HUVECs, we stimulated HUVECs with a combination of rhIL-4 and rhIL-1β to increase their surface expression of P-selectin, in addition to VCAM-1 and ICAM-1 (29). When isolated PMNs were perfused (2 dyn/cm²), the addition of anti–P-selectin blocking Abs significantly reduced PMN adhesion to rhIL-4/rhIL-1β–stimulated HUVECs compared with isotype control (43.9 ± 4.2% of the isotype-control group, p < 0.01, n = 3 paired subjects). Next, we perfused isolated PMNs in the presence of rhVim (50 μg/ml) and blocking Abs to E-selectin, P-selectin, VCAM-1, or ICAM-1 or isotype control to determine whether the addition of blocking Abs to rhVim would further decrease PMN adhesion to inflamed endothelium. PMNs perfused with only vehicle were used as control. The combination of rhVim + IgG decreased PMN adhesion by ∼50% compared with control, similar to our previous observations with rhVim alone. There was no additional decrease in PMN adhesion when E-selectin or P-selectin was added to rhVim. However, the addition of VCAM-1 or ICAM-1 blocking Abs to rhVim modestly decreased PMN adhesion to HUVEC monolayers (Fig. 6E).

Based on these observations, we further tested the effect of rhVim on leukocyte capture onto HUVECs when mixed in whole blood. We perfused mepacrine-labeled whole blood over IL-1β–stimulated HUVECs in the presence of increasing concentrations of rhVim. Similar to its effect on isolated PMNs, rhVim blocked leukocyte capture and adhesion to HUVEC monolayers by ∼60% (Fig. 6F).

Because of the multiple adhesive interactions between PMNs and platelets, we turned our attention to the cell adhesion molecules. We perfused isolated PMNs over channels coated with P-selectin/Fc chimeric protein (10 μg/ml) in the presence of increasing concentrations of rhVim. Similar to our observations with platelet monolayers, the presence of rhVim decreased PMN adhesion to P-selectin/Fc monolayers in a dose-dependent fashion (Fig. 7A). Furthermore, the addition of rhVim increased the mean rolling velocity of PMNs over the P-selectin/Fc monolayers (Fig. 7B).

The effect of rhVim blocking PMN adhesion to P-selectin could be attributed to rhVim affecting P-selectin, PSGL-1, or both. Therefore, we performed ELISAs with wells coated with PSGL-1/Fc protein (10 μg/ml) or P-selectin/Fc protein (10 μg/ml) and incubated with increasing concentrations of rhVim. Increasing amounts of rhVim bound to P-selectin/Fc protein and reached saturation at 20 μg/ml. There was no binding of rhVim to PSGL-1/Fc at any concentration (Fig. 7C).

Based on our ELISA data that rhVim binds to P-selectin but not to PSGL-1, we performed binding kinetics analysis of P-selectin/Fc to rhVim using SPR, with human IgG as a control for the Fc fragment. In addition, we tested the binding kinetics of E-selectin to rhVim, because E-selectin is expressed on endothelium and participates in leukocyte capture and rolling adhesion. We found that P-selectin/Fc exhibited strong binding to immobilized rhVim (K_D = 67 ± 1.2 nM), whereas there was no binding to E-selectin/Fc or human IgG (Table I).

We also performed binding kinetic analyses of P-selectin/Fc, E-selectin/Fc, and PSGL-1/Fc to rhVim using BLI. Human IgG and sheep anti-vimentin Ab served as negative and positive controls, respectively. We found that, similar to our ELISA and SPR data, P-selectin/Fc bound to rhVim with great affinity (K_D = 53 ± 16 nM). However, neither E-selectin/Fc nor PSGL-1/Fc bound to rhVim (Table I). There was no binding to human IgG and expected binding to sheep anti-vimentin Ab.

It has been established that P-selectin–PSGL-1 interactions have been implicated in animal models of ALI (9). Therefore, based on published and our in vitro data, we tested the hypothesis that rhVim has a beneficial role in attenuating inflammation in vivo. There were no differences in mortality when using the sublethal dose of LPS; however, mice receiving rhVim had a significantly lower histologic ALI score than mice receiving vehicle control, with decreased neutrophilic infiltration into the lungs (Fig. 8).

Previously published data suggested that cell surface vimentin binds to N-acetylglucosamine (21), a moiety present in PSGL-1. That report led us to examine the possibility that vimentin binds to PSGL-1 and could inhibit the interplay between platelet P-selectin and PSGL-1 on leukocytes; however, the results from using several methods to study protein–protein interactions demonstrated that our rhVim, in fact, binds to P-selectin. Our data suggest that the interaction between rhVim and P-selectin is calcium independent, because all three methods to determine K_D (ELISA, SPR, and BLI) were conducted in the absence of calcium. These data are congruent with our dose-dependent flow adhesion studies in which we perfused isolated neutrophils over platelets, HUVECs, and P-selectin/Fc–coated surfaces. Moreover, our observations in endotoxin-treated mice suggest that rhVim attenuates histologic signs of ALI, including neutrophilic infiltration. Taken together, rhVim binds specifically to P-selectin to block neutrophil adhesion to platelets and endothelium under venous shear stress. This may imply how rhVim attenuates endotoxin-induced lung injury.

Our data show that rhVim effectively blocked the interaction of neutrophils with platelets and endothelial cells in vitro and decreased neutrophilic infiltration into the lungs in endotoxemia in vivo. These observations are congruent with our and other investigators’ published data regarding the role of P-selectin–PSGL-1 interactions on inflammation (i.e., platelet–leukocyte–endothelial interactions) (4, 9, 34). However, one limitation of this study is that the effects seen in vivo may be due to effects of rhVim, in addition to inhibition of P-selectin–PSGL-1 interactions. We also noticed that, unlike its effect on neutrophil–platelet or neutrophil–P-selectin interactions that were performed in a more isolated system, a higher concentration of rhVim was required to block neutrophil rolling adhesion to IL-1β–stimulated HUVECs.

We used IL-1β to stimulate HUVECs because of its role in the development of sepsis (35) and ALI (36, 37), as well as its effect in promoting P-selectin–mediated rolling in vivo (38). IL-1β is known to stimulate ICAM-1 expression on HUVECs but not that of P-selectin, whose expression is induced in endothelial cells (39). Thus, the ability of rhVim to block leukocyte and isolated PMN adhesion to IL-1β–stimulated HUVECs may be due to the presence of other receptors mediating cell interactions with the endothelium, such as Mac-1 (CD11b/18) to ICAM-1 (6), and/or binding of rhVim to other endothelial surface receptors. Because of the limitation of IL-1β stimulation alone, we also stimulated HUVECs with a combination of IL-4 and IL-1β to maximally increase P-selectin surface expression, in addition to VCAM-1, ICAM-1, and E-selectin, as previously described (29). In our assay, adhesion of isolated PMNs to IL-4/IL-1β–stimulated HUVECs was partially blocked by anti–P-selectin blocking Abs. We observed a similar reduction in PMN adhesion to IL-4/IL-1β HUVECs in the presence of rhVim + isotype control Ab, without a further reduction in PMN adhesion when anti–P-selectin Abs were added to rhVim. However, the combination of anti–VCAM-1 or anti–ICAM-1 blocking Ab with rhVim modestly decreased PMN adhesion to IL-4/IL-1β–stimulated HUVECs more than rhVim with isotype control. Although the addition of anti–VCAM-1 or anti–ICAM-1 Ab with rhVim did not fully abrogate PMN adhesion at the concentrations used, their synergistic effects suggest that they block PMN adhesion via distinct mechanisms. Finally, the region of rhVim that binds to P-selectin is unknown. In our experiments with the anti–P-selectin Ab AK4, we found minimal synergy with the addition of rhVim to block adhesion of neutrophils to platelets and endothelium. This suggests that rhVim may bind to part of the epitope recognized by AK4. Further studies are needed to elucidate the exact molecular mechanism by which rhVim binds to P-selectin and whether there are other mechanisms by which rhVim blocks leukocyte adhesion to platelets and endothelium.

Although our bacteria-derived rhVim protein retained some structural features found in native vimentin, as demonstrated by its interaction with anti-human vimentin Abs (Fig. 2), we do not know whether this recombinant protein mimics endogenous extracellular vimentin. This is because vimentin, which is glycosylated and phosphorylated, is secreted into plasma by activated macrophages in response to proinflammatory stimuli to promote bacterial killing (17). To our knowledge, there is no report describing the purification of soluble vimentin from plasma. This may be due, in part, to the fact that vimentin has historically been considered a cytoskeletal protein. It has not been until recently that some studies reported the presence of circulating vimentin in the peripheral blood in certain clinical situations (40, 41). Whether circulating soluble vimentin, especially with different posttranslational modifications, behaves similarly to our rhVim remains unclear. Therefore, further research is required to determine whether naturally occurring circulating vimentin is also involved in immune function by regulating leukocyte–platelet interactions via P-selectin.

To our knowledge, our work describes the first direct evidence that rhVim blocks platelet–leukocyte–endothelial interactions. However, other studies have proposed rhVim as a potential therapeutic agent for other disease states (i.e., cancer). One group has studied the effect of rhVim on Wnt signaling and cancer cell invasion, observing that rhVim binds to the surface of cancer cells in vitro and induces increased expression and nuclear accumulation of β-catenin (42). Another area of interest is the role of the immune system’s response to modified vimentin, such as after citrullination, in which immunization with peptide fragments of citrullinated vimentin leads to Ab production and improved survival in mice with various cancer types (43).

Based on our data, rhVim blocks leukocyte adhesion to platelets and endothelium by blocking P-selectin–PSGL-1 interactions and decreases histologic signs of endotoxin-induced ALI. Therefore, rhVim may be a useful agent to attenuate inflammation in a variety of diseases, such as ALI (9, 44) and ischemia-reperfusion injury (11). Further studies are warranted to validate our findings and to further evaluate the efficacy of rhVim in attenuating inflammation in those diseases.

The authors have no financial conflicts of interest.