Cell Surface Biliverdin Reductase Mediates Biliverdin-induced Anti-inflammatory Effects via Phosphatidylinositol 3-Kinase and Akt*

Biliverdin reductase A (BVR) catalyzes the reduction of biliverdin (BV) to bilirubin (BR) in all cells. Others and we have shown that biliverdin is a potent anti-inflammatory molecule, however, the mechanism by which BV exerts its protective effects is unclear. We describe and elucidate a novel finding demonstrating that BVR is expressed on the external plasma membrane of macrophages (and other cells) where it quickly converts BV to BR. The enzymatic conversion of BV to BR on the surface by BVR initiates a signaling cascade through tyrosine phosphorylation of BVR on the cytoplasmic tail. Phosphorylated BVR in turn binds to the p85α subunit of phosphatidylinositol 3-kinase and activates downstream signaling to Akt. Using bacterial endotoxin (lipopolysaccharide) to initiate an inflammatory response in macrophages, we find a rapid increase in BVR surface expression. One of the mechanisms by which BV mediates its protective effects in response to lipopolysaccharide is through enhanced production of interleukin-10 (IL-10) the prototypical anti-inflammatory cytokine. IL-10 regulation is dependent in part on the activation of Akt. The effects of BV on IL-10 expression are lost with blockade of Akt. Inhibition of surface BVR with RNA interference attenuates BV-induced Akt signaling and IL-10 expression and in vivo negates the cytoprotective effects of BV in models of shock and acute hepatitis. Collectively, our findings elucidate a potentially important new molecular mechanism by which BV, through the enzymatic activity and phosphorylation of surface BVR (BVR)surf modulates the inflammatory response.

Biliverdin reductase (BVR) 2 mediates the rapid conversion of biliverdin to bilirubin (1,2). BVR also functions as a dual tyro-sine and serine/threonine kinase (3,4) and as a transcription factor that binds promoters within Ap-1 sites (5). Biliverdin and bilirubin both possess potent cytoprotective properties in a variety of animal models (6,7) including those for ischemia/ reperfusion injury following small bowel or liver transplantation (8 -10), vascular injury (11), and endotoxic shock (6,12). The mechanisms underlying these effects are still poorly understood and to date have not been linked to BVR activity per se, but attributed rather to the antioxidant power of biliverdin and bilirubin.
The concept that BVR may be an ideal docking protein for SH-2 domains (4) combined with the known anti-inflammatory effects of the bile pigments and rapid conversion of BV to BR in vivo, led us to hypothesize that BVR is expressed on the membrane and through its kinase activity regulates Akt signaling via recruitment of phosphatidylinositol 3-kinase (PI3K). We present novel findings that the ability of biliverdin to prevent LPSinduced morbidity and mortality most likely requires BVR surface expression for specific activation of PI3K-Akt and IL-10. We suggest that this is the major pathway by which biliverdin inhibits the inflammatory response and that these molecular events are natural protective mechanisms involved in the innate immune response to bacterial endotoxin that enables homeostasis.

Cell Culture and Treatment
The mouse macrophage cell line, RAW 264.7 (RAW), and HEK cells were purchased from ATCC and maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and 50 g/ml gentamicin (Invitrogen). For treatment, cells were seeded 24 h before experiment. Biliverdin (Frontier Scientific) was freshly prepared in DMSO (Sigma) and kept in the dark before and during treatment. Final concentration of DMSO in medium was Ͻ0.01%. LPS (Escherichia coli serotype 0127:B08, Sigma) was dissolved in PBS and used for treatment at concentrations ranging from 1 to 1000 ng/ml. GGTI-287, a selective inhibitor of geranylgeranyl transferase I (Calbiochem), geranylgeranyl pyrophosphate (Sigma), and DL-␤-hydroxymyristic acid (Sigma) were dissolved in DMSO and used at concentrations of 1, 3, and 100 M, respectively. LY294002 (Sigma; 10 M) was used as a selective inhibitor of PI3K.
Bone marrow-derived macrophages were isolated and cultured as previously described (13). Adenovirus containing Cre recombinase was used at 50 multiplicity of infection on the third day of culture. Macrophages were treated and harvested on the fifth day of culture.

Animal Treatment
C57BL6/J mice were purchased from The Jackson Laboratories. PI3K p85␤ Ϫ/Ϫ /p85␣ loxp mice were kindly provided by Prof. Lewis Cantley (BIDMC, Harvard Medical School). All animals were held under pathogen-free conditions and the experiments were approved by the BIDMC Animal Care and Use Committee. Lung, liver, and spleen as well as blood samples were harvested for immunohistochemical and immunostaining analyses from control and LPS-injected mice (5 mg/kg, intraperitoneal) for 6 h. Biliverdin/bilirubin was freshly dissolved in 0.2 N NaOH, adjusted to a final pH of 7.4 with HCl, and kept in the dark. Mice were administered biliverdin or bilirubin (35 mg/kg, intraperitoneal) 16 h and again 2 h prior to LPS/D-galactosamine (250 g/kg, intraperitoneal/750 mg/kg, intraperitoneal; E. coli serotype 0127:08, Sigma). Serum bilirubin levels were evaluated spectrophotometrically (Sigma Kit), according to the manufacturer's protocol in a separate group of non-LPStreated mice. For adenovirus experiments, mice were administered either Ad-BVR-siRNA or Ad-Y5 (2 ϫ 10 9 plaque-forming units intraperitoneally) 5 days prior to LPS/D-Gal.

Immunohistochemistry
Liver, lung, and spleen tissue samples were embedded in freezing medium and stored at Ϫ80°C. Five-m sections were fixed in cold acetone and embedded in paraffin followed by immunohistochemistry using fluorescently tagged or horseradish peroxidase-tagged secondary antibody as previously described (10).

Immunofluorescence and Flow Cytometry and Total Internal Reflective Fluorescence (TIRF)
RAW cells were seeded on glass (Fisher) and treated with 10 -100 ng/ml LPS for various time points as described. For detection of surface antigens, cells were rinsed in PBS and blocked with 0.5% BSA (bovine serum albumin, RIA grade, Sigma) in PBS for 1 h followed by overnight incubation with primary antibodies at 4°C. Staining of intracellular antigens was performed by membrane permeabilizers with Triton X-100 or methanol, followed by rinsing with PBS and blocking with 0.5% BSA in PBS. Fluorescent-labeled secondary antibodies were applied for 1 h at room temperature. Finally, cells were washed, fixed with 0.25% paraformaldehyde, and covered with 4Ј,6-diamidino-2-phenylindole (ϩ/Ϫ) mounting medium. Fluorescence was viewed with a Zeiss Apotome Axiovert microscope at ϫ40 or ϫ60 magnification.
For flow cytometric analysis of surface antigens, cells were gently harvested, centrifuged (300 ϫ g for 5 min), and washed in PBS. Cells were blocked in 0.5% BSA in PBS for 30 min followed by 30 min incubation with primary antibodies. Afterward, cells were washed with blocking buffer and resuspended in fluorescently labeled antibodies and incubated in the dark at room temperature for 30 min followed by extensive washing with PBS. Cells were immediately analyzed by flow cytometry (BD Biosciences FACS Sorter).
TIRF measurements were performed on cells culture and stained as for immunofluorescence analysis with few modifications. Briefly, cells were seeded on the glass and after treatment, washed with 1ϫ PBS. Cells were fixed with 2% paraformaldehyde for 15 min followed by permeabilization with 0.1% Triton X-100 for 15 min. Cells were rehydrated with 5 washes of PBS and blocked for 45 min with 2% BSA. Following staining cells were washed, stained with Hoechst for 30 s, and coverslipped with gelvatol and refrigerated until analysis with a TIRF microscope.

Cell Fractionations
Cytosolic and membrane cell fractions were isolated following the manufacturer's protocol of the Plasma Membrane Protein Extraction Kit (BioVision, CA). Briefly, cells were rinsed in ice-cold PBS, resuspended in homogenization buffer, and incubated on ice for 1 h. To isolate total cellular membrane proteins, cell or liver homogenate was centrifuged at 700 ϫ g for 10 min. The supernatant was collected and further centrifuged at 10,000 ϫ g for 30 min. To purify plasma membrane protein, the pellet was resuspended in a 1:1 upper phase solution:lower phase solution, followed by 5 min incubation on ice. The lower phase was extracted twice with the upper phase solution by centrifugation at 3,000 ϫ g for 5 min. The upper phase was collected and diluted in 5 volumes of water for 5 min on ice. The pellet of the plasma membrane protein was obtained by centrifugation for 10 min (14,000 ϫ g). Protein extracts were subjected to further analyses by immunoblotting, immunoprecipitation, and for biliverdin reductase activity assay as previously described (14).

Plasmids and Transfections
BVR198 (Tyr 3 Phe) Construction-Overlap extension PCR was applied for construction of point mutation in hBVR cDNA and followed by ligation to pcDNA3.1 plasmid between KpnI and XhoI restriction sites. The following primers were used with the underlined mutation and marked restriction sites: 1) 5Ј-CGACGGTACCGAAGGAAGAGACCAAGATGAA-3Ј; 2) 5Ј-GAAAGGAAGATCAGTTCATGAAAAT-3Ј; 3) 5Ј-GAC-ACTCGAGTGGAAGTGCTACATCACCT-3Ј; 4) 5Ј-CTTTC-CTTCTAGTCAAGTACTTTTACTGTCA-3Ј. In PCR-1, primers 1 and 4 were used to amplify a fragment of 609 bp and in PCR-2, primers 2 and 3 gave a fragment of 335 bp. Both fragments gave the whole BVR cDNA with point mutation in the PCR-3 with primers 1 and 3. Proper cloning and mutagenesis were verified by restriction site analysis and sequencing.
Mutant Vectors-The Glu 97 and Glu 47 mutants were generated as previously described and characterized (27). BVR97 (Glu 3 Ala) was cloned as a single fragment between the KpnI and NotI sites of pcDNA3.1 and the mutation was introduced using the following primer: 5Ј-TCC CGA GGG CCC TTC TCG ACA CGA AGC C. BVR47 (Glu 3 Ala) was cloned between the KpnI and XhoI sites of pcDNA3.1 using the following primers: 1) 5Ј-CTG AGC GGC CGC CAA TGA CAG TGT CAT GGG GTA GGC CAC AAG G; and 2) 5Ј-CGA GAA GGG CCC TCG GGA GCA TTG ATG GAG.
Dominant-negative Akt and control constructs were a kind gift from Dr. Alex Toker (BIDMC). RAW cells were transfected using Lipofectamine 2000 (Invitrogen) or Amaxa nucleofection reagents (Amaxa) according to the vendor's protocol. HEK cells were transfected with Lipofectamine 200 according to the manufacturer's protocol. Transfection efficiency estimated by FACS analysis (BD Biosciences FACSort) was 30 -40%.
RNAi Transfections-Pre-designed and verified annealed RNAi against BVRa, GFP RNAi, and negative control RNAi were purchased from Ambion (Ambion Inc., Austin, TX). Fifty nM RNAi was used for each transfection together with 5 l of Lipofectamine 2000 reagent in Opti-MEM medium (Invitrogen) according to the manufacturer's instructions. In some experiments re-transfection with RNAi after 2 days of incubation was performed. BVR protein knock-down was confirmed 100 h after transfection as previously described (15).

Immunoblotting and Immunoprecipitation
Snap-frozen tissue samples were homogenized in ice-cold tissue lysis buffer (250 mM NaCl, 5 mM EDTA, 1% Triton X-100, 10 mM Tris-HCl, pH 7.5, containing the protease inhibitor mixture Complete Mini) (Roche, Indianapolis, IN). Cells were lysed by a freeze-thaw cycle in ice-cold lysis buffer (0.5% Nonidet P-40, 0.1% sodium deoxycholate, 0.1% SDS, 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, pH 8.0, 1 mM sodium fluoride, and the protease inhibitor mixture). Samples were centrifuged for 30 min at 14,000 ϫ g at 4°C and supernatants were collected. Protein concentration was measured using the BCA Protein Assay Kit (Pierce). 10 -20 g of each protein sample was electrophoresed on NuPAGE 4 -12% BisTris gel (Invitrogen) in NuPAGE MES SDS Running Buffer (Invitrogen). For co-immunoprecipitation, 100 -500 g of protein lysates in RIPA buffer (50 mM Tris-HCl, pH 7.5, 50 mM sodium fluoride, 150 mM NaCl, 1% Nonidet P-40, 2 mM EDTA, pH 8.0, 10 mM phenylmethylsulfonyl fluoride, and the protease inhibitor mixture) were mixed with appropriate antibodies and 30 l of Protein A/G Plus-agarose beads (Santa Cruz) and then rocked for 2.5 h at 4°C. The proteins were washed with RIPA buffer and eluted (95°C, 5 min in SDS loading buffer). The proteins were then subjected to SDS-PAGE electrophoresis followed by transfer to polyvi- nylidene difluoride membrane (Ready Gel Blotting Sandwiches; Bio-Rad). The membranes were blocked with 5% nonfat dry milk, probed with appropriate primary antibodies followed by horseradish peroxidase-conjugated secondary antibodies at a dilution of 1:5000, and visualized using Super Signal West Pico chemiluminescent substrate (Pierce) or Femto Maximum Sensitivity substrate (Pierce), followed by exposure to the autoradiography film (ISC BioExpress, Kaysville, UT).

ELISA for IL-10
IL-10 was measured in culture medium using Quantikine Immunoassay (R&D Systems) according to the manufacturer's protocol.

Statistical Analysis
Results are presented as the mean Ϯ S.D. Statistical comparison was performed by use of Student's t test (SPSS Inc) (p Ͻ 0.05).

Membrane-associated Biliverdin Reductase Is a Functional
Surface Antigen in Macrophages-We employed several techniques, including TIRF microscopy, immunoblotting, immunofluorescence, and flow cytometry (FACS) to demonstrate that BVR is expressed on the external plasma membrane. Live cell imaging of BVR activity showed rapid conversion (seconds) of non-fluorescent biliverdin to fluorescent bilirubin (emission at 512 nm), which was lost in the presence of trypsin ( Fig. 1A and supplemental Fig. S1). These observations lead us to hypothesize that BVR may be a membrane protein with an active site present on the external leaflet of the membrane. Coimmunostaining with BVR and the macrophage-specific F4.80 antibodies show the co-localization of surface BVR with the well characterized surface antigen in unstimulated RAW 264.7 (RAW) macrophages (Fig. 1B). BVR and F4.80 co-stained cells were then imaged by Z-stack analysis with similar results (Fig.  1C). The induced expression of BVR by LPS on the cell surface was observed via TIRF microscopy (Fig. 1D), which permits specific detection of surface-only expressed proteins.
We next tested the effects of LPS on BVR expression in macrophages to potentially understand the anti-inflammatory actions of biliverdin that others and we have reported (6). There was a clear induction of BVR on the cell surface as measured by flow cytometry of non-permeabilized RAW cells ( Fig. 2A). LPS administration resulted in a time-and dose-dependent increase in BVR surface expression as early as 30 min after treatment with LPS, suggesting post-transcriptional activation of BVR ( Fig. 2, B and C). LPS had no effect on the level of total BVR (data not shown), whereas it up-regulated membrane-bound BVR. We confirmed the time-dependent increase in membrane BVR after LPS (100 ng/ml) treatment by immunoblotting the lysates of membrane fractions in macrophages (Fig. 2D). This was specific for membrane expression as no change in BVR expression in total lysates was observed Fig. 2D). To evaluate whether our in vitro data were reflected in vivo, we performed immunohistochemical staining for BVR in the livers of mice treated with LPS. Staining results revealed induction of BVR in  hepatic macrophages 6 h after LPS injection (Fig. 2E) with surface expression detected by double staining with antibodies to BVR and the macrophage marker F4.80 (Fig. 2E).
Finally, to substantiate BVR membrane localization and activity, we performed quantitative experiments with purified membrane fragments. Addition of biliverdin to cells resulted in rapid generation of bilirubin with an increase in absorbance at 450 nm at 1 h from 0.003 Ϯ 0.001 to 0.007 Ϯ 0.001 relative activity change in the cytosolic fraction versus an increase from 0.005 Ϯ 0.002 to 0.015 Ϯ 0.002 (p Ͻ 0.02) in purified membrane fractions. Collectively, the data presented in Figs. 1 and 2 support membrane-bound, enzymatically active BVR on the external leaflet of the plasma membrane.
Sequence Analysis of BVR Identifies a Transmembrane Spanning Region-A structure analysis of BVR using the transmembrane prediction program (16) identified a 24-amino acid transmembrane segment among three ␣-helixes (163-186 amino acids) (Fig. 3A). The C terminus of BVR comprises specific prenylation sites for geranylgeranyl transferase (CLGL), which in addition to five myristoylation sites may be crucial for recruitment and association of BVR to the membrane (Fig. 3, A and B). Lipid modifications of many proteins enable them to localize to the plasma membrane (17). Inhibition of prenylation or myristoylation prevented LPS-induced cell surface BVR expression (p Ͻ 0.005), whereas addition of GGPP resulted in augmentation of LPS-induced BVR surface expression ( Fig. 3C; p Ͻ 0.01 versus LPS). These data demonstrate that BVR is translocated from the cytosol to the cell membrane by post-translational protein modification, i.e. prenylation and myristoylation. The tyrosine phosphorylation motif (YMKM) is also present in the BVR sequence as previously described (4) (Fig. 3A). Although the data in Fig. 3 do not distinguish between internal and external BVR leaflet localization, the data in Figs. 1 and 2 support the concept that BVR is localized to the external plasma surface.
Phosphorylation of BVR in Response to BV Binding Leads to Interaction of BVR with the p85␣ Subunit of PI3K-To test our hypothesis that surface BVR interacts with a downstream kinase proximal to the membrane via its tyrosine phosphorylation motif, we explored a possible complex formation between BVR and the p85␣ subunit of PI3K in RAW macrophages. The p85␣ subunit of PI3K co-immunoprecipitates with BVR in BV-treated RAW cells without changing the total BVR or p85␣ expression in the lysates (Fig. 4A). After biliverdin treatment, BVR is phosphorylated on tyrosine residues (Fig. 4B) and there is greater binding of p85␣ to BVR (Fig. 4A). Mutation of the tyrosine that is responsible for the kinase activity (Tyr 3 Phe 198 ) or catalytically inactive mutant (Glu 3 Ala 97 ) resulted in a reduced interaction of BVR with p85␣ in response to BV (Fig. 4C). The Tyr 198 -Met 199 -Lys 200 -Met 201 (YMKM) motif within BVR appears to be the binding site for the p85␣ subunit of PI3K leading to BV-BVR-mediated induction of Akt. Overexpression of BVR or either mutant had any effect on BVR or p85 expression (Fig. 4D).
Given that Akt is downstream of PI3K, we next tested the effects of biliverdin administration on Akt activation. BV induced a time-and dose-dependent phosphorylation of Akt in RAW cells (Fig. 5, A and B) and in murine bone marrow-derived macrophages (Fig. 5D). To test whether the effect of BV-induced activation of Akt is dependent on a PI3K-BVR interaction we used pharmacological and genetic approaches. The effect of BV on Akt phosphorylation is blocked by LY290024, a selective inhibitor of PI3K (Fig. 5C). We show that the effect of biliverdin is specifically mediated by the p85␣ subunit of PI3K but not the PI3K␤ subunit by employing bone marrow-derived macrophages from p85␤ PI3K Ϫ/Ϫ and conditional loxp p85␣ PI3K mice. The absence of p85␣ resulted in an inability of BV to activate Akt, whereas BV was able to activate Akt in bone marrow macrophages in wild type cells (Fig. 5D). Although there was increased phosphorylation of Akt after expression of WT BVR, introduction of a mutated (Tyr 3 Phe 198 ) BVR blocked Akt phosphorylation (data not shown).
To confirm and validate the activation of Akt by BVR, we employed BVR mutants on BV-induced downstream targets. We observed strong phosphorylation of S6 protein (Fig. 5E) as supportive evidence of Akt activation, which was inhibited by pharmacologic inhibition of PI3K (data not shown). Introduction of a catalytically inactive mutant Glu 3 Ala 97 resulted in a loss in the ability of BVR to convert biliverdin to bilirubin (supplemental Fig. S2). Importantly, inhibition in the binding of p85␣ to PI3K to BVR (Fig. 5B), and subsequent loss of S6 phosphorylation (Fig. 5F) in cells treated with BV suggests that the activity of BVR is critical for the receptor function of BVR surf . We also generated an additional BVR mutant Glu 3 Ala 47 , which is catalytically active (supplemental Fig. S3) but the edge of the binding pocket for BV and NADH/NADPH is disrupted. With expression of this particular mutant, we observed inhibition of biliverdin-mediated Akt signaling, whereas WT BVRmediated BV signaling remained normal (Fig. 5F).
BVR-mediated Signaling through Akt Increases IL-10 Production in Macrophages-Previously we have shown that BV enhances IL-10 expression in macrophages and in vivo in rats. Therefore, we evaluated whether a functional link exists between BVR-mediated Akt signaling and production of antiinflammatory cytokines. Overexpression of BVR (70 Ϯ 10% efficiency) led to increased IL-10 production in BV-stimulated RAW macrophages (Fig. 6A), whereas the Tyr 3 Pro 198 mutant BVR blocked BV-induced signaling through PI3K-Akt to increase IL-10 (Fig. 6A). Moreover, introduction of a dominant-negative Akt abrogated BV-mediated induction of IL-10 (Fig. 6C).
LPS stimulates rapid induction of HO-1 and biliverdin thereby becomes available as a substrate for BVR as heme is degraded. To test the function of BVR in LPS-mediated IL-10 production, we overexpressed or silenced BVR (Fig. 6,  B-D). Overexpression of BVR enhanced LPS-induced production of IL-10 (Fig. 6B), whereas inhibition of BVR or Akt with siRNA or dominant-negative mutants led to significant inhibition of IL-10 production (Fig. 6, C and D). These data suggest that BV acts in part via a BVR-PI3K-Akt pathway to mediate its anti-inflammatory effects in response to LPS through augmented expression of IL-10 and offers one explanation for BV-elicited protection, in addition to its antioxidant power. The regulation of IL-10 expression by BV/BVR, whereas significant is not striking and we speculate that IL-10 regulation is only part of the mechanism by which BV modulates the inflammatory response to impart protection in the animal models described below.
BV Protects against Lethal Endotoxemia and Acute Hepatitis via BVR-We next evaluated the in vivo conversion of biliverdin to bilirubin and the efficacy in a model of lethal endotoxic shock. Biliverdin given as a bolus to mice was rapidly converted to bilirubin, reaching a maximum concentration of bilirubin of 50 Ϯ 7.4 M at 5 min (the earliest time point measured) from a baseline of 6 Ϯ 2.8 M at 5 min (p Ͻ 0.001), and returning to baseline by 6 h (data not shown). We also tested the ability of biliverdin to prevent acute inflammation following a lethal injection of LPS alone or LPS/D-Gal (50 mg/kg, intraperitoneal) to induce acute liver failure. One group of mice received biliverdin (35 mg/kg, intraperitoneal) 16 and 2 h prior to LPStreated mice or LPS/D-Gal-treated mice, whereas controls received saline in both models. In the experiment with LPS alone, all saline-treated mice were dead by 24 h, whereas 100% of biliverdin-treated mice survived long-term (Fig. 7A). We further confirmed our observation with LPS/D-Gal treatment and observed 80% survival rate in the BV-treated group (Fig. 7B). To test whether the protection conferred by biliverdin required the action of BVR, we blocked BVR with an adenovirus expressing BVR-siRNA based on the published sequence (7). Separate groups of mice were administered 2 ϫ 10 9 plaqueforming units/mouse intravenously of either Ad-BVR-siRNA or Ad-Y5 control virus, diluted in normal saline. After 5 days, the time when viral expression is maximal and results in high and specific expression of the vector primarily in the liver, we administered LPS/D-Gal in the presence and absence of biliverdin. Six to eight hours later, serum alanine aminotransferase was measured and the livers were harvested and snap frozen for subsequent Western blot analyses. Animals treated with BV or BVϩY5 control virus showed significant inhibition in LPS/D-Gal-induced hepatitis as assessed by serum levels of alanine aminotransferase. BV, but not bilirubin (BR) at the doses tested, blocked LPS/ D-Gal-induced increases in alanine aminotransferase, whereas those animals in which BVR was blocked with BVR-siRNA expressed serum alanine aminotransferase levels comparable with LPS/D-Gal controls (Fig. 7C). BVR knockdown was confirmed by Western blot in liver lysates (Fig. 7C,  inset). These data validate our in vitro data showing that the protection afforded by BV requires the activity of BVR.

DISCUSSION
We describe here a novel paradigm showing that BVR, the enzyme that converts biliverdin to bilirubin, is a surface protein (BVR surf ) that initiates a signaling cascade upon the binding of biliverdin involving PI3K, activation of Akt, and downstream production of IL-10. We suggest that the remarkable anti-inflammatory effects of biliverdin, previously attributed solely to the antioxidant properties of biliverdin and bilirubin, are in very significant part mediated via this novel mechanism (6,12). We submit that BVR functions in part as a ligand-activated signaling protein that resembles tyrosine kinase receptors and as such is directly involved in regulation of the innate immune response. We define this localized membrane form as BVR surf .
BVR has been studied with regard to functions other than converting biliverdin to bilirubin. It has been shown by others that BVR can function as a kinase (3,4) and as a transcription factor involved in the regulation of heme oxygenase-1 (5). BVR is present in various compartments of the cell including the inner cell membrane in endothelial cells in association with HO-1 (18,19).
There are to date many reports in the literature demonstrating the ability of biliverdin to modulate the inflammatory response. The protective potential of biliverdin/bilirubin stems from a history of studies of the enzyme HO-1 (20,21). HO-1 has emerged over the last decade as a powerful cytoprotective molecule, yet the mechanism(s) by which it functions remains unclear. HO-1 acts on heme and generates three products: biliverdin, carbon monoxide (CO), and Fe 2ϩ . Recent studies of these products have begun to shed light on the mechanisms of HO-1 action as a homeostatic gene (8,(22)(23)(24).
The sequence of amino acids in BVR is consistent with our findings regarding the signaling cascade initiated by biliverdin-BVR. The tyrosine motif of BVR present on the cytosolic side resembles the binding motif of platelet-derived growth factor receptor for the p85␣ subunit of PI3K. We suggest that biliverdin stimulates BVR phosphorylation of tyrosine 198, which is the critical residue within the membrane that enables BVR to bind the p85␣ subunit of PI3K and signal to activate Akt. The regulation of this interaction is complex and perhaps in part FIGURE 6. Surface BVR mediates biliverdin Akt-IL-10 signaling. A, IL-10 expression was measured by ELISA in the media of RAW cells transfected with WT BVR or Y198P mutant BVR in the presence or absence of BV for 10 h. Results represent percent of control Ϯ S.D. of triplicate measurements from three independent experiments. Vector transfection efficiency was determined to be 78 Ϯ 10% by immunofluorescence. *, p Ͻ 0.05 BV treated and BVR ϩ BV-treated RAW cells versus corresponding controls; #, p Ͻ 0.05 Tyr 198 ϩ BV versus BVR ϩ BV. Inset shows BVR levels in the RAW cells transfected with BVR and control vector. B, IL-10 expression by ELISA in the media of RAW cells transfected with WT BVR and then treated with 100 ng/ml LPS for 24 h. *, p Ͻ 0.01 LPS versus control; #, p Ͻ 0.05 BVR ϩ LPS versus Control ϩ LPS; data represent percent of control Ϯ S.D. from three independent experiments in duplicate. C, IL-10 ELISA results in cells transfected with dominant-negative Akt or pcDNA3 controls Ϯ BV (10 h), demonstrating that BV is unable to increase IL-10 in cells unable to signal through Akt. Data represent mean Ϯ S.D. from at least three independent experiments in triplicate. D, IL-10 ELISA was performed in supernatants from RAW cells transfected with siRNA control or BVR-siRNA and then treated with 100 ng/ml LPS for 24 h. Inset shows immunoblot of BVR in RAW macrophages transfected with BVR-siRNA or control RNAi. Blot is representative of two independent experiments to validate knockdown. Data represent percent of control Ϯ S.D. and are representative of three independent experiments. *, p Ͻ 0.01 LPS versus control; #, p Ͻ 0.05 siRNA BVR versus siRNA control. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.