Unique Toll-Like Receptor 4 Activation by NAMPT/PBEF Induces NFκB Signaling and Inflammatory Lung Injury

Ventilator-induced inflammatory lung injury (VILI) is mechanistically linked to increased NAMPT transcription and circulating levels of nicotinamide phosphoribosyl-transferase (NAMPT/PBEF). Although VILI severity is attenuated by reduced NAMPT/PBEF bioavailability, the precise contribution of NAMPT/PBEF and excessive mechanical stress to VILI pathobiology is unknown. We now report that NAMPT/PBEF induces lung NFκB transcriptional activities and inflammatory injury via direct ligation of Toll–like receptor 4 (TLR4). Computational analysis demonstrated that NAMPT/PBEF and MD-2, a TLR4-binding protein essential for LPS-induced TLR4 activation, share ~30% sequence identity and exhibit striking structural similarity in loop regions critical for MD-2-TLR4 binding. Unlike MD-2, whose TLR4 binding alone is insufficient to initiate TLR4 signaling, NAMPT/PBEF alone produces robust TLR4 activation, likely via a protruding region of NAMPT/PBEF (S402-N412) with structural similarity to LPS. The identification of this unique mode of TLR4 activation by NAMPT/PBEF advances the understanding of innate immunity responses as well as the untoward events associated with mechanical stress-induced lung inflammation.

mortality to ARDS 10,11 . In preclinical models of VILI, NAMPT/PBEF expression was spatially localized to lung epithelium, tissue leukocytes and the lung vascular endothelium 10 with direct participation in ARDS/VILI pathobiology. Furthermore, intra-tracheally-instilled NAMPT/PBEF induces a neutrophilic alveolitis 12 and heterozygous PBEF +/− mice are dramatically protected from severe murine VILI 12 . As reductions in extracellular NAMPT/PBEF availability, via neutralizing antibodies 12 or liposomes encargoed with NAMPT siRNAs, provide significant protection from LPS-and VILI-induced murine lung inflammation 12 , together these findings indicate that NAMPT/PBEF is an attractive therapeutic target in ARDS and VILI.
NAMPT regulates intracellular nicotinamide adenine dinucleotide (NAD) biosynthesis and apoptosis pathways 11,[13][14][15] . However, it is the increased NAMPT/PBEF expression and extracellular secretion into blood and bronchoalveolar lavage fluid that produce the profound inflammatory effects of NAMPT/ PBEF in response to inflammatory stimuli such as excessive mechanical stress 10,12 . In the absence of an inflammatory stimulus such as LPS, recombinant PBEF alone directly exerts inflammatory responses that are similar to the ARDS condition 12 . Contributing to potential mechanisms of NAMPT/PBEF-mediated lung pathobiology, we demonstrated that exogenous NAMPT/PBEF elicits robust inflammatory gene transcription in murine lungs 12 , including dysregulated genes in the transcriptome related to leukocyte extravasation, the transcription factor NFκ B 12 , and expression of Toll-like receptors (TLR) 12,16 .
These data, supporting NAMPT/PBEF as a regulator of lung innate immunity pathways, led us to systematically explore the biochemical and molecular basis for NAMPT/PBEF involvement in the inflammatory pathophysiology associated with mechanical ventilation and acute lung injury. Utilizing complementary system biology in vivo and in vitro approaches, including genetically-engineered mice and computational modeling, we now define novel and rapid NAMPT/PBEF-mediated NFκ B transcriptional activities via the ligation of TLR4. Computational analysis revealed substantial sequence identity between NAMPT/PBEF and MD-2, a TLR4-binding protein essential for LPS-induced TLR4 activation. Importantly, NAMPT/PBEF and MD-2 share similar α helix and β sheet structures and strong similarity in regions containing the majority of MD-2-TLR4 binding residues. We further speculate that a protruding region of NAMPT/PBEF (S402-N412), with structural similarity to LPS, serves as the site of direct TLR4 binding. Wheras MD-2 binding of TLR4 in the absence of LPS fails to induce NFκ B activation, our identification of a novel mechanism of direct TLR4 activation by NAMPT/PBEF, occurring in the absence of bacterial infection and cofactor requirements, increases the understanding of lung innate immunity responses and the untoward inflammatory effects of mechanical stress-induced lung injury.

Results
Exogenous NAMPT/PBEF induces robust in vitro and in vivo NFκB activation in human and murine tissues. Leveraging our prior reports of NFκ B transcriptome induction by recombinant NAMPT/PBEF (rPBEF) 12,17,18 , complementary in vitro approaches were utilized to functionally examine the direct role of extracellular NAMPT/PBEF in NFκ B pathway activation and innate immunity gene expression. Initial experiments assessed exogenous NAMPT/PBEF-mediated phosphorylation of NFκ B (p-NFκ B at Ser 536 ) in human lung endothelial cells (EC) as an indication of NFκ B activation. rPBEF, similar to LPS and TNF-α , increases phosphorylation of p-NFκ B within 30 min with persistent elevation at 1 hr (Fig. 1A). Heat denaturing of rPBEF (HD-rPBEF) resulted in the elimination of NAMPT/PBEF's capacity to induce NFκ B activation, demonstrating that NAMPT/PBEF-mediated NFκ B phosphorylation/activation does not reflect endotoxin contamination (Fig. 1A,B). rPBEF-induced NFκ B activation was unaffected by inhibition of NAMPT enzymatic activity (FK-866), indicating that phosphoribosyl-transferase activity is not required (Fig. 1C).
Time-dependent NFκ B translocation from the cytoplasm to the nucleus was next assessed as a reflection of activation of the canonical NFκ B pathway. Consistent with biochemical indices of NFκ B activation, rPBEF challenge resulted in NFκ B translocation to the nucleus in human lung EC (Fig. 1D) temporally similar to TNF-α (positive control) (Fig. 1D). Finally, EC transfected with a NFκ B promoter luciferase reporter and challenged with rPBEF exhibited increased luciferase activity (up to three hours) (Fig. 1E). These data strongly suggest that extracellular NAMPT/PBEF is a rapid and potent NFκ B activator in human lung endothelium.
Extracellular NAMPT/PBEF-mediated NFκ B activation was next evaluated in preclinical models of ARDS/VILI. Phospho-NFκ B immunostaining of murine lung tissues from VILI-exposed mice revealed prominent VILI-induced NFκ B phosphorylation/activation in vascular endothelium and alveolar epithelium ( Fig. 2A). Genome-wide gene expression analysis in wild type mice revealed marked similarities in NFκ B Toll receptor pathway signaling gene upregulation evoked by rPBEF, LPS, and VILI, with heterozygous PBEF +/− mice exhibiting dramatic attenuation of VILI-mediated NFκ B pathway gene expression (Fig. 2B). These studies provide compelling evidence for extracellular NAMPT/PBEF involvement in induction of the NFκ B transcriptome and in lung innate immunity directly contributing to ARDS/VILI pathobiology.
These in vitro results were extended to in vivo studies utilizing mice pretreated with the TLR4 inhibitor, RS-LPS, as well as TLR4 −/− mice (Fig. 4). Consistent with our prior report 12 , rPBEF instillation produced marked increases in lung inflammatory indices in wild type mice (BAL protein levels, BAL PMNs, and BAL cell counts (Fig. 4A)) that were significantly attenuated by RS-LPS pretreatment (100 μ g/ mouse) (Fig. 4A) in wild type mice. Similar to NAMPT/PBEF challenge, RS-LPS pretreatment produced attenuation of VILI-induced pulmonary inflammation in wild-type mice (Fig. 4B). rPBEF-induced lung inflammation was also reduced in TLR4 −/− mice compared to wild type mice (Fig. 4C). TLR4 −/− mice demonstrated abolishment of the prominent rPBEF-induced NFκ B phosphorylation in murine pulmonary EC (Fig. 5A) and reduced basal levels of NFκ B signaling (Fig. 5B). Both rPBEF and LPS triggered similar, robust increases in expression of NFκ B signaling genes in wild type animals that were markedly reduced in TLR4 −/− mice (Fig. 5B). More importantly, the NFκ B pathways gene ontology was significantly dysregulated by either rPBEF or LPS (as the top regulated pathways) with significant suppression  Effects of TLR4 inhibitory strategies on rPBEF-and LPS-mediated NFκ B phosphorylation in ECs. Cell lysates were immunoblotted for phospho-specific or total NFκ B with experiments independently performed in triplicate (representative blots shown). TNF-α (100 ng/ml, 15 min) serves as a positive control for NFκ B signaling activation. (Panel A) Inhibition of both rPBEF (1 μ g/ml, 1 hr)-and LPS (5 μ g/ml, 1 hr)-mediated NFκ B phosphorylation (Ser 536 ) in EC pretreated (1 hr) with neutralizing TLR4 polyclonal (pAb, 20 μ g/ml) or TLR4 monoclonal antibodies (mAb, 10 μ g/ml). Premixing of rPBEF, but not LPS, with neutralizing NAMPT/PBEF pAb (100 μ g/ml, 30 min) similarly reduced NFκ B phosphorylation. (Panel B) Inhibitory effects of 1 hr pretreatment with TLR4 pharmacologic inhibitors (RS-LPS [10 μ g/ml], CLI-095 [5 μ M] and OxPAPC [30 μ g/ml]) on rPBEFand LPS-mediated NFκ B phosphorylation in ECs. (Panel C) Densitometric summary of the attenuation of rPBEF-and LPS-induced NFκ B phosphorylation by TLR4 neutralizing antibodies and inhibitors. rPBEF-induced NFκ B phosphorylation was significantly reduced to levels similar to those observed with pretreatment with anti-NAMPT/PBEF specific antibody. LPS-induced NFκ B phosphorylation was unaffected by pretreatment with anti-NAMPT/PBEF specific antibody. Bar graphs represent data as integrated density normalized to rPBEF-or LPSstimulated control. n = 3 independent experiments per condition; *p < 0.05 versus rPBEF-stimulated control, **p < 0.05 versus LPS-stimulated control. Pretreatment with inhibitors or neutralizing antibodies alone (without rPBEF or LPS stimulation) did not significantly differ from unstimulated controls (data not shown). (Panel D) Surface plasmon resonance (SPR) analysis (using Bio-Rad ProteOn XPR36 instrument and GLC sensor chips) demonstrates TLR4-NAMPT/PBEF binding interaction. Bethyl PBEF antibody was covalently bound to the chip surface, using standard direct immobilization, at a final immobilization level of 5000 RUs. rPBEF only (100 nM), rTLR4 only (1 μ M), and rPBEF-rTLR4 pre-mixed (100 nM or 1 μ M, respectively) analytes were then injected over the PBEF antibody coated surface. While rTLR4 did not bind to the PBEF antibody coated surface, premixed rPBEF-rTLR4 resulted in increased binding response over rPBEF alone.

Figure 4. TLR4 is the novel receptor for extracellular NAMPT/PBEF-and VILI-induced NFκB activation and inflammatory lung injury in vivo. Panel A.
Compared to saline-challenged controls (VEH), intratracheal instillation of rPBEF (40 μ g/mouse) in wild type mice produces robust increases in BAL protein levels, in the percentage of BAL PMNs, and in BAL total cell counts, findings consistent with prior reports 12 . These rPBEF-mediated inflammatory lung indices were significantly reduced both in mice pretreated with the TLR4 inhibitor RS-LPS (100 μ g/mouse, i.p.) (Panel A) and in TLR4 −/− mice (Panel C) indicating that TLR4 is required for NAMPT/PBEF-induced pro-inflammatory activities and lung injury. Results are expressed as mean ± SEM; n = 3-6 per condition; *p < 0.05 for VEH/WT vs VEH/WT + rPBEF and **p < 0.05 VEH/WT + rPBEF vs RS/TLR4 −/− + rPBEF, using Anova non-parametric Newman-Keuls Multiple Comparison Test. (Panel B) Exposure to VILI (40 ml/kg, 4 hr) in wild type mice produces significant increases in BAL protein levels, in the percentage of BAL PMNs, and in BAL cell counts. These rPBEF-mediated inflammatory indices were significantly reduced in mice pretreated with the TLR4 inhibitor RS-LPS (100 μ g/mouse), indicating that TLR4 is required for NAMPT/PBEF-induced pro-inflammatory activities and lung injury. Results are expressed as mean ± SEM; n = 3-6 per condition; *p < 0.05 for VEH vs VEH + VILI and **p < 0.05 VEH + VILI vs RS + VILI.

Figure 5. Extracellular NAMPT/PBEF-and LPS-induced NFκB pathway gene dysregulation is mediated by TLR4. (Panel A)
Paraffin-embedded lung tissue was sectioned (10 μ m) and immunostained with a p-NFκ B antibody. Shown is a representative image demonstrating prominent NFκ B activation and expression in capillary endothelium and alveolar epithelium from rPBEF-challenged wild type mice, whereas p-NFκ B immunoreactivity was significantly reduced in rPBEF-challenged TLR4 −/− mice. Scale bar = 100 μ m. See Supplemental Figure 1 for IHC staining isotype controls. (Panel B) Heat maps reflecting the critical involvement of TLR4 in rPBEF-and LPS-mediated upregulation of NFκ B pathway gene expression. Both rPBEF (40 μ g/mouse, 4.5 hr) and LPS (2.5 mg/kg, 4 hr) mediate robust NFκ B pathway increases in wild type mice whereas this expression was markedly reduced in TLR4 −/− mice. Blue color indicates reduced gene expression, red color reflects increased gene expression. Bar graphs represent enriched pathways in mice challenged with rPBEF (40 μ g/mouse, 4.5 hr) (Panel C) or LPS (2.5 mg/kg, 4 hr) (Panel D). The top ranking BIOCARTA pathways are listed for the genes differentially expressed between controls and wild-type mice challenged with rPBEF and LPS, respectively. The corresponding pathway patterns for the genes differentially expressed between the wild-type controls and TLR4 −/− mice treated with rPBEF or LPS are also indicated. The genes dysregulated by LPS were identified using criteria of a false discovery rate (FDR) of < 5% and a minimum of a 2-fold change. The genes dysregulated by rPBEF were identified by a cutoff of < 10% FDR and > 1.5-fold change. The gray dash line indicates the cutoff of significance (P-value < 0.05).
In silico modeling reveals NAMPT/PBEF binding surface similarities with MD-2, an essential LPS cofactor in TLR4 activation. The crystal structure of NAMPT/PBEF 19 , the TLR4 receptor, as well as the LPS-and TLR4-binding protein, MD-2, have been resolved 20,21 . In silico protein structure analysis revealed that despite only ~30% total sequence identity between murine NAMPT/PBEF and MD-2, striking structural similarities exist between a loop region on NAMPT/PBEF and a loop region on MD-2 known to be involved in LPS binding to TLR4 and critical to subsequent TLR4 activation (Fig. 6A,B). Using a sequence order independent structural alignment method 22 , a loop (99D-111E, purple Fig. 6C) in the TLR4-binding region on MD-2 was well aligned in the N-to-C order with a loop residing in NAMPT/ PBEF in the reverse C-to-N order (457L-445E, red, Fig. 6D, 30.3% identity). Six of the 7 MD-2 residues known to be important for TLR4 binding reside within this loop 20 and this loop also contains a prominent motif consisting of a consecutive triplet of residues K109, G110, and E111, a D residue and an S/Y residue 23 (Fig. 6A) fully conserved among different species. This motif is also present in the NAMPT/PBEF loop (thus indicating a potentially conserved biological function). We speculate that this loop represents the TLR4-binding region for NAMPT/PBEF is supported by protein analysis of the TLR4-binding protein, Der-p2, a house mite allergen with MD-2 homology (sequence identity ~26%) [24][25][26] . The TLR4-binding regions of Der-p2 are highly conserved and similarly align to the loop in NAMPT/PBEF. Finally, residue R434, in spatial proximity of the predicted NAMPT/PBEF loop binding residues, is structurally aligned to R90 on MD-2 implicated to participate in TLR4-MD-2 interactions 27 . Thus, the observed similarity in the TLR4 binding environment between NAMPT/PBEF and MD-2, in conjunction with the biochemical and SPR studies, support our hypothesis that NAMPT/PBEF is a TLR4-binding protein.
We next constructed a surface model of signature binding site pockets 22,28 characterizing LPS-binding regions based on three structures from species-diverse LPS-binding proteins sharing less than 64% sequence identities with MD-2 ( Fig. 6E) but that exhibited identical LPS binding modes as MD-2. According to the degree of residue preservation in their geometric locations in the calculated signature LPS binding surface, the most important signature residues on MD-2 for LPS binding were identified as F119, L74, L94, and I52 (Fig. 6E), residues within the three proteins used to construct the signature pocket. These signature LPS-binding residues, while tightly clustered in MD-2, in NAMPT/PBEF are spatially separated from each other suggesting that NAMPT/PBEF is unlikely to directly bind LPS. For example, the distance between LPS-binding residues F119 (red in Fig. 6F) and I52 (green in Fig. 6F) in MD-2 is 3.8 Å, whereas the distance between the corresponding residues F399 (red in Fig. 6G) and I114 (green in Fig. 6G) in NAMPT/PBEF is 9.3 Å. Furthermore, although a well-defined surface pocket exists in MD-2 for LPS binding, no such surface pocket exists on NAMPT/PBEF to contain the LPS molecule. Thus, it is unlikely that NAMPT/PBEF directly interacts with either LPS or MD-2 physically at this region. Interestingly, within the structure of the MD-2-LPS complex, the LPS molecule maps to the protruding region of NAMPT/PBEF (S402-N412) (Fig. 6G), suggesting that NAMPT/PBEF may have intrinsically adopted a conformation capable of directly binding and activating TLR-4. Future mutation studies of these residues will provide further insight into structure-function relationship of NAMPT/ PBEF-induced TLR4 signaling.

Discussion
Originally named for facilitation of B cell maturation, NAMPT/PBEF is a "cytozyme" dually functioning as an intracellular dimeric type II nicotinamide phosphoribosyltransferase enzyme (NAMPT) involved in NAD biosynthesis 29,30 and as an extracellular pro-inflammatory cytokine 12 . NAMPT/PBEF expression is markedly elevated in acutely inflamed lungs in ARDS and VILI 10,12 , in cardiac tissues during cardiac arrest and resuscitation 31 , in amniotic membranes during gestation 32 and is released from visceral fat during the development of obesity, an observation resulting in its naming as visfatin 9 . We recently demonstrated that NAMPT expression in the lung is transcriptionally regulated by excessive mechanical stress via STAT5-dependent increases in NAMPT promoter activity 10,33 and via post-transcriptional epigenetic mechanisms involving 5′ UTR promoter demethylation and 3′ UTR miRNA binding 33,34 . Furthermore, NAMPT expression is influenced by promoter SNPs that function to increase NAMPT/PBEF expression and confer enhanced susceptibility to ARDS as well as increased ARDS mortality 10,11,33 .
We previously explored mechanisms of NAMPT/PBEF-mediated inflammatory lung injury relevant to both ARDS and VILI utilizing multiple complementary in vitro and in vivo approaches and demonstrated that extracellular NAMPT/PBEF is a direct neutrophil chemoattractant augmenting lung injury induced by excessive mechanical stress (VILI) and inducing robust inflammatory cytokine expression 10,12,35 . Furthermore, reductions in NAMPT/PBEF bioavailability (neutralizing antibodies or siRNAs) attenuated VILI-induced lung inflammation 12 . The current study expands support for NAMPT/PBEF as a novel ARDS/VILI candidate gene and biomarker directly involved in ARDS/VILI pathobiology 10,12 .
We have now detailed robust NAMPT/PBEF-mediated NFκ B activation that is independent of NAMPT enzymatic activity (Fig. 1C). Lung gene ontology signatures evoked by extracellular NAMPT/PBEF exposure demonstrated prominent overlap with Toll-like receptor signaling 12 and led to interrogation of TLR4 as a putative NAMPT/PBEF receptor. This hypothesis was confirmed by studies utilizing TLR4 pharmacologic inhibitors, TLR4 neutralizing antibodies, TLR4 siRNAs, NAMPT/PBEF-TLR4 SPR analysis, and  Signature pocket for the LPS-binding surface constructed from three diverse LPS binding proteins, each less than 64% of sequence identities with MD-2 but all exhibiting the same LPS binding mode as that of MD-2. Binding pockets of the three LPS binding proteins (red = oxygen; blue = nitrogen; green = carbon atoms) are shown along the hierarchical tree based on sequence order independent structural alignment of the binding surfaces 28 . The large yellow pocket at the root of the tree is the overall signature pocket constructed from the 7 LPS binding pockets. Spatially preserved signature LPS binding residues are in red. (Panel F) Signature residues in the LPS binding pocket of the MD-2-LPS complex including F119 (red); I52 (green); L74, L94 (blue); overall LPS binding pocket (cyan); TLR4binding loop (purple). (Panel G) NAMPT/PBEF structure with residues corresponding to signature residues for LPS binding mapped in MD-2: F399 corresponding to F119 on MD-2 (red); L432 and L458 corresponding to L74 and L94, respectively (blue); I114 corresponding to I52 (green); other residues of LPS binding pocket (cyan); loop residues for TLR4-binding loop (purple). Signature residues for LPS binding exhibit greater spatial separation in NAMPT/PBEF compared to MD-2 with the prominent surface pocket for LPS binding, present in MD-2, absent in NAMPT/PBEF. Potential TLR4 activation by the LPS molecule (transparent-wheat) can be mapped to the protruding region of S402-N412 residues on NAMPT/PBEF (transparent-wheat).
Toll-like receptors (TLRs) are essential to innate immunity responses 39,40 and have been implicated in the pathobiology of acute inflammatory lung injury 41 . Similar to other TLRs, TLR4 recognizes conserved microbial-specific patterns 39,42 , shares NFκ B as a common downstream effector and transcription factor, and exhibits TLR structural similarity 20 . TLR4 is uniquely critical to the regulation of innate immunity responses to gram-negative bacteria infection via binding of LPS, an outer membrane glycolipid of gram-negative bacteria. LPS is recruited by LBP (LPS-binding protein) and by CD14 to bind the TLR4-MD-2 complex 43 . Crystal structure analysis has revealed that TLR4 and its binding partner, MD-2, form a heterodimer that recognizes LPS present in gram-negative bacteria 20 and structural analysis of TLR4-MD-2 interactions in the presence or the absence of the LPS antagonist, eritoran, indicates that MD-2 is essential for LPS binding to TLR4 20 .
Utilizing in vitro and in vivo approaches and in silico modeling analyses, we now demonstrate that NAMPT/PBEF directly induces TLR4-mediated NFκ B activation without the requirement for MD-2-TLR4 binding and in the absence of additional LPS chaperones or cofactors. Multi-pronged in silico protein analysis revealed striking unconventional NAMPT/PBEF structural similarity with loop regions on MD-2 known to be involved in TLR4 binding and LPS-induced TLR4 activation. Both NAMPT/PBEF and MD-2 contain fully conserved K109, G110, E111 residues residing in these loops 20 critical for TLR4 binding 20,23 . Furthermore, the residue R434 on NAMPT/PBEF exists in spatial proximity of the predicted loop binding residues and structurally aligns to R90 on MD-2, an amino acid that directly participates in TLR4-MD-2 interactions 27 . Together, these structural similarities in the TLR4 binding environment support the hypothesis that this loop functions as the TLR4-binding region for NAMPT/PBEF.
In contrast to the strong loop alignment between NAMPT/PBEF and MD-2 implicated in TLR4 binding, our model of signature binding surface of LPS-binding pockets 22,28 determined that NAMPT/PBEF, unlike MD-2, fails to contain a well defined surface pocket for LPS binding. Furthermore, whereas MD-2 contains several highly clustered signature residues critical for LPS binding (F119, L74, L94, I52), these residues on NAMPT/PBEF are spatially separated and unlike MD-2, do not reside in close proximity to one another. Thus, NAMPT/PBEF is unlikely to directly and physically interact with either MD-2 or LPS within this region. Our analysis of the structure of the MD-2-LPS complex, however, indicates that LPS binding to TLR4 maps to the protruding region of NAMPT/PBEF (S402-N412), suggesting that NAMPT/PBEF may intrinsically adopt a conformation capable of direct TLR4 binding and activation.
In addition to NAMPT/PBEF and LPS/MD-2, Der-p2 and high-mobility group box 1 (HMGB1) proteins also bind TLR4. Similar to NAMPT/PBEF, HMGB1 is a ubiquitous dually functioning protein with intracellular activities as a DNA-binding protein regulating transcription 13,14 as well as extracellular cytokine-like activities 12,15 via TLR4 binding 11,12 . HMGB1 binds at least eight distinct receptors, including TLR4 4,8,13 and is a key mediator of severe sepsis 44 reproducing extracellular NAMPT/PBEF-mediated pathologic features (lung vascular permeability, interstitial edema, neutrophil infiltration) in rodent ARDS models 24 . HMGB1 levels in patients with sepsis and ARDS are highly correlated with morbidity and severity 24,32,35 and HMGB1-TLR4 signaling has been implicated in the pathogenesis of sterile injury 26 . Unlike NAMPT/PBEF, however, HMGB1 does not directly bind TLR4 45 utilizing a signaling cascade similar to LPS with high affinity binding to the MD-2/TLR4 complex 46 . In contrast to LPS and HMGB1, NAMPT/PBEF-mediated TLR4 binding and activation appears similar to Der-p2, a house mite allergen and TLR4-binding protein homologous to MD-2 (sequence identity 26.3%) 25 . The TLR4-binding region of Der-p2 is highly conserved and aligns to the NAMPT/PBEF loop 47,48 and like NAMPT/PBEF, Der-p2 reconstitutes LPS-driven TLR4 signaling without the requirement for MD-2 20,25 .
Taken together with the recognition that NAMPT promoter polymorphisms confer ARDS/VILI susceptibility and influence ARDS severity and mortality 10,25 , these studies significantly increase the mechanistic understanding of the untoward inflammatory effects of mechanical ventilation-induced mechanical stress. Although NAMPT/PBEF is not entirely unique in its capacity to bind and activate TLR4, MD-2, HMGB1 and LPS fail to directly induce TLR4-mediated NFκ B activation. Furthermore, unlike the house mite allergen, Der-p2, NAMPT/PBEF is endogenously expressed in man with critical participation in normal cellular homeostasis as well as in pathologic stress responses to excessive mechanical stress such as observed in the critical care setting. Thus, NAMPT/PBEF is unique in serving as an endogenous innate immunity molecule capable of directly binding and activating TLR4 in the absence of bacterial infection and cofactor requirements, thereby delineating a novel dimension to the induction of lung inflammatory and innate immunity responses by non-infectious mechanisms.

Materials and Methods
Reagents. Commercially  Cell Culture. Human pulmonary artery endothelial cells (EC) and human lung microvessel EC (Lonza, Walkersville, MD) were cultured as described previously 49 in endothelial growth medium-2 (EGM-2 or EGM-2-MV). Cells were grown at 37 °C in a 5% CO 2 incubator, and passages 6 to 9 were used for experiments. Media were changed one day before experimentation.
Western Blotting. After treatment as outlined for individual experiments, EC were subsequently washed with cold (4 °C) Ca 2+ /Mg-free PBS and lysed with 0.3% SDS lysis buffer containing protease inhibitors (1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate, 1 mM sodium fluoride, 0.2 trypsin inhibitor unit/ml aprotinin, 10 μ M leupeptin, and 5 μ M pepstatin A). Sample proteins were separated with 4 to 15% SDS-PAGE gels (Bio-Rad, Hercules, CA) and transferred onto Immobilon-P polyvinylidene difluoride membranes (Millipore Corporation, Bedford, MA). Membranes were then immunoblotted with primary antibodies (1:1000, 4 °C, overnight) followed by secondary antibodies conjugated to horseradish peroxidase (1:5000, room temperature, 30 min) and detected with enhanced chemiluminescence (Pierce ECL or SuperSignal West Dura; Pierce Biotechnology, Rockford, IL) on Biomax MR film (Carestream Health, Rochester, NY). Western blot densitometry analysis was performed on inverted images with Adobe Photoshop (San Jose, CA) software, using the same selection box size for all band histogram means readings, where all means were then compared to the appropriate control on the same gel. The number of replicates analyzed is a minimum of three per experiment, with specific number of replicates and comparison control listed in each corresponding figure legend.
Immunostaining. HLMVEC were cultured in EGM-2-MV culture medium in a 12-well plate format for immunostaining. The next day, the culture medium was changed and EC were incubated for up to six hours with either vehicle, TNF-α (100 ng/ml concentration), or rPBEF (10 μ g/ml). Coverslips were dipped in D-PBS, immersed in 3.7% formaldehyde/PBS, pH 7.4, for 20 min at room temperature, washed and quenched in 50 mM NH 4 Cl/PBS 3 × 5 min, permeabilized in 0.1% TritonX100/PBS for 2-3 min, and blocked in 5% BSA/PBS for 60 min at room temperature. EC were incubated 60 min with NFκ B monoclonal antibody (BD Transduction Labs) diluted 1:100 in blocking buffer, washed in PBS, incubated with goat anti-mouse IgG/AlexaFluor488 (Life Technologies, Carlsbad, CA) and subjected to autoradiography measurements. For the immunofluorescence experiments, EC were washed in PBS and mounted onto a drop of ProLong Gold with DAPI (Life Technologies). Images were acquired on a Leica TCS SP5 AOTF laser-scanning confocal microscope system scanning at 400 Hz with an Ar488 nm laser and a multiphoton 740 nm laser, a Leica DMI 6000 microscope, and an HCX PL APO CS 63X NA1.4 oil objective lens. Twelve-bit 512 × 512 images were acquired sequentially scan line-by-scan line and with a line average setting of 16 with Leica LAS AF software and detected with a photomultiplier tube. Images were analyzed using ImageJ v1.39 software (Wayne Rasband, National Institutes of Health, Bethesda MD).
Transgenic and Control Mice. All in vivo mouse methods/experiments were approved by and performed in accordance with University of Illinois at Chicago IACUC Committee guidelines and regulations. Animals were housed under standard conditions. NAMPT/PBEF +/mice were generated as previously described 12 with C57BL/6 and TLR4-deficient (TLR4 −/− ) (B6.B10ScN-Tlr4 lps-del /JthJ) mice (8-12 weeks old) purchased from Jackson Laboratory (Bar Harbor, ME).
Bronchoalveolar Lavage (BAL) Analysis. Mice underwent lavage with 1 ml of HBSS buffer into the intratracheal catheter for BAL protein, total BAL cell counts and differential counts assessments as previously described 12 . Briefly, BAL fluid recovered was centrifuged, and the supernatant assessed for total protein content, using a kit assay (Bio-Rad) and expressed in mg/ml. In addition, the BAL pellet was utilized for counting the total number of cells with a TC20 cell counter (Bio-Rad), and for cytospin analysis on stained slides using Diff-Quik dye for differential counts (PMNs neutrophil percentage) from each mouse sample. BAL fluid and pellet were frozen for further analysis.
Scientific RepoRts | 5:13135 | DOi: 10.1038/srep13135 Lung Tissue Histology. Excised mice left lungs were placed immediately in formalin overnight, followed by embedding in paraffin for histological evaluation by hematoxylin-eosin (H&E) staining. These sections were examined under microscope and representative images were recorded by camera 50 . Immunohistochemistry. Paraffin blocks of lung tissues were prepared and 10 μ m microscope slides were obtained. A serial section of each specimen was de-paraffinized and rehydrated in serial graded ethanol. Antigens retrieval was achieved with 100 mM Tris base buffer (pH 9) and heating slides in a 98 °C water bath for 15 min. Endogenous peroxidase activity was blocked in methanol containing 3% hydrogen peroxide. The section was incubated with the p-NFκ B (anti-phospho-RELA (p65/pSer 536 ), Sigma) antibody produced in rabbit in 1:100 dilution for 40 min, followed by 30 min incubation with Dako labeled polymer-HPR anti-rabbit secondary antibody (K4011, Dako Inc, Carpinteria, CA). The DAB/DAB+ Chromogen solutions were used serially, and the slides were counterstained with hematoxylin.
Microarray Analysis. RNA extraction was performed using RNeasy kits (Qiagen); Affymetrix Mouse Genome 430 2.0 and Mouse Gene 2.0 ST arrays used to detect genome-wide expression levels summarized by the gcrma package in Bioconductor with GC robust multichip average (GCRMA) normalization 51 . The expression level of each transcript in Mouse Gene 2.0 ST arrays was summarized by the RMA method in "oligo" package in Bioconductor 52 . SAM (Significance Analysis of Microarrays) 53 , implemented in the samr library of the R Statistical Package, for comparing log2-transformed gene expression levels between groups. Enriched BIOCARTA pathways were searched among differentially-expressed genes using NIH/DAVID 54 . In Silico Computational Modeling. In silico analysis are as previously described 22,28 . The structures of LPS binding pockets analyzed include: 1N12 (E. coli), 4GGM (caulobacter), 3MU3 (jungle fowl), 3RGY (cattle), 3VQ1 (mouse), 4G8A and 2E59 (human).
Statistical Analysis. For all in vitro (n of 3 or more) or in vivo (n of [3][4][5][6] experiments, values are shown as the mean ± SEM and data were analyzed using standard student's t test or two-way ANOVA. Significance in all cases was defined at p < 0.05.