Extracellular nicotinamide phosphoribosyltransferase boosts IFNγ-induced macrophage polarization independently of TLR4

Summary Nicotinamide phosphoribosyltransferase (NAMPT), alongside being a crucial enzyme in NAD synthesis, has been shown to be a secreted protein (eNAMPT), whose levels are increased in patients affected by immune-mediated disorders. Accordingly, preclinical studies have highlighted that eNAMPT participates in the pathogenesis of several inflammatory diseases. Herein, we analyzed the effects of eNAMPT on macrophage-driven inflammation. RNAseq analysis of peritoneal macrophages (PECs) demonstrates that eNAMPT triggers an M1-skewed transcriptional program, and this effect is not dependent on the enzymatic activity. Noteworthy, both in PECs and in human monocyte-derived macrophages, eNAMPT selectively boosts IFNγ-driven transcriptional activation via STAT1/3 phosphorylation. Importantly, the secretion of eNAMPT promotes the chemotactic recruitment of myeloid cells, therefore providing a potential positive feedback loop to foster inflammation. Last, we report that these events are independent of the activation of TLR4, the only eNAMPT receptor that has hitherto been recognized, prompting the knowledge that other receptors are involved.


INTRODUCTION
Intracellular nicotinamide phosphoribosyltransferase (iNAMPT) has received significant attention over the years, as it represents the cytosolic rate-limiting enzyme of the NAD salvage-pathway in mammals and catalyzes the synthesis of nicotinamide mononucleotide (NMN) from nicotinamide (NAM,vitamin B3,or PP) and 5-phosphoribosylpyrophosphate (PRPP) (Garten et al., 2015). NAMPT has also been shown to be a secreted protein. Indeed, extracellular NAMPT (eNAMPT) is the same protein that was described as pre-B-cell enhancing factor (PBEF) for its ability to synergize with interleukin-7 (IL-7) and stem cell factor, increasing the number of pre-B-cell colonies, and as visfatin, a cytokine first described as released from adipose tissue Revollo et al., 2007;Samal et al., 1994). A number of groups, including ours, have shown that eNAMPT can be secreted by immune cells (Audrito et al., 2015;Curat et al., 2006;Halvorsen et al., 2015;Laudes et al., 2010) as well as by other cell types in a classic manner (Grolla et al., 2015;Tanaka et al., 2007), and recently it has been shown that eNAMPT can also be present in secreted microvesicles (Grolla et al., 2015;Yoshida et al., 2019). How eNAMPT exerts its extracellular functions has not been fully elucidated (Camp et al., 2015). Van der Bergh et al. proposed a direct binding to CCR5 in macrophages and PBMCs in vitro ( Van den Bergh et al., 2012), and we have indeed confirmed that eNAMPT may have an antagonistic role on this receptor, although it does not appear to be the principal pathway by which it exerts most of its actions (Torretta et al., 2020). Controversially, a different group brought evidence that eNAMPT might instead also have agonist properties, acting on muscle stem cells and promoting muscle regeneration (Ratnayake et al., 2021). On the other hand, it has also been shown that eNAMPT leads to TLR4 activation. Evidence for this comes from surface plasmon resonance (Camp et al., 2015;Managò et al., 2019), from an effect on human macrophages, and from an antagonistic effect of a TLR4 antibody (Camp et al., 2015;Managò et al., 2019). myelopoiesis and functional skewing of monocytes, eNAMPT further enhances the expression of immunosuppressive M2 genes such as IL-10, IDO, CD206, and CD163 (Audrito et al., 2015). Moreover, eNAMPT appears to foster macrophage phagocytic activity (Yun et al., 2014) and to favor macrophage migration by inducing the expression of matrix metalloproteinases (Dahl et al., 2007). Despite this encouraging evidence, a thorough characterization of the actions of eNAMPT on macrophages is lacking.
Although the mechanisms underpinning eNAMPT activity remain largely unclear, it is well established that eNAMPT participates in the pathogenesis of several inflammatory conditions, as demonstrated by the beneficial effects of its neutralization in experimental models of colitis  and inflammatory lung injury Quijada et al., 2021).
Given that macrophages are pivotal orchestrators of both initiation and resolution of inflammation, we undertook a full investigation of the effects of eNAMPT on primary murine peritoneal macrophages (PECs), an approach that has the advantage of giving insights on the physiological role of this protein using primary cells. Our data show that eNAMPT promotes macrophages-driven inflammation mainly in a Toll-like receptor 4 (TLR4)independent manner. Specifically, we found out that eNAMPT (1) promotes chemotactic recruitment of inflammatory cells, (2) activate macrophages to express an M1-skewed transcription program, (3) boosts IFNg-driven macrophage activation by enhancing STAT1/3 activation, and (4) is strongly released in response to IFNg treatment, thereby providing a potential positive feedback loop supporting exacerbation of inflammation.

RESULTS eNAMPT is an M1-skewing stimulus
To unravel the effect of eNAMPT on macrophages, according to the guidelines (Murray et al., 2014), we stimulated PECs with murine recombinant eNAMPT (500 ng/mL, endotoxin levels less than 0.1 EU/mL). To ascertain the specific role of eNAMPT on gene expression, cells were treated in the presence or absence of C269 (10 mg/mL), an eNAMPT-neutralizing monoclonal antibody that we have recently generated and validated . Given the different kinetics of M1 and M2 gene induction, we analyzed transcript levels after 4 and 18 h, respectively ( Figure 1A). In comparison with untreated PECs, qPCR results showed that eNAMPT induced all inflammatory M1-related genes tested, including Il6, Il1b, Cxcl10, Cxcl9, Nos2, Cox2, Tnf, and Il12b, whereas neither the anti-inflammatory cytokine Il10 nor the M2(IL-4)-associated genes were modulated ( Figure 1B). As shown in Figure 1B, C269 (blue bars) abrogated the effect of eN-AMPT demonstrating the specificity of the effect. To evaluate whether the effect of eNAMPT could be attributed to its enzymatic activity, we next stimulated PECs with eNAMPT H247E , a mutant that has been shown to lose the catalytic activity (Wang et al., 2006). eNAMPT H247E was able to induce M1-associated genes ( Figure 1C) to the same extent as wild-type eNAMPT, conclusively proving that the extracellular enzymatic activity is dispensable for macrophage skewing.
Given that in several inflammatory conditions NAMPT has been shown to increase and act as an exacerbator of inflammation (Travelli et al., 2018), we explored the crosstalk between eNAMPT and other inflammatory stimuli. We treated PECs with interferon gamma (IFNg), lipopolysaccharide (LPS), IL-6, IL-1b, granulocytemacrophage colony-stimulating factor (GM-CSF), and IL-4 either as single stimuli ( Figure 1D) or in combination with eNAMPT (Figures 1E and S1A-S1E). The results indicated that eNAMPT strongly enhanced the expression of IFNg and LPS-induced genes ( Figures 1E and S1A). The pattern of potentiation was not identical between IFNg and LPS, although in both settings Il6 was the most upregulated gene over the respective stimulus. On the contrary, IL-6, IL-1b, and GM-CSF responsive genes were not further induced by eNAMPT (Figures S1B-S1D). We confirmed the specific boosting effect of eNAMPT by using C269, which completely prevented the increased expression of IFNg-induced genes ( Figure 1E). Moreover, we verified that the synergism is maintained also with catalytically inactive eNAMPT H247E mutant ( Figure 1F; see Figure S1F for residual enzymatic activity of the mutant), confirming that also this phenomenon is not dependent on the enzymatic activity of the protein. We next investigated whether these observations could have relevance to humans by evaluating the effect of eNAMPT on human macrophages differentiated in vitro from monocytes of healthy donors. Of the selected gene panel, we confirmed that human recombinant eNAMPT alone significantly increased Il6, Il1b, and Il12b and in combination with IFNg further enhanced the expression of IFNg-induced genes Cxcl9 and Cxcl10 ( Figure 1G). The combination of eN-AMPT and IFNg also potentiated the induction of the inflammatory genes Il6, Il1b, Il12b, and Tnf, whereas, as expected, no effect by eNAMPT, IFNg, or the combination was observed on Il10 and Arg1 expression ( Figures 1G and 1H). Overall, these data indicate that eNAMPT is a cytokine endowed with selective M1-skewing activity and with a potent boosting activity on IFNg-induced activation in both murine and human macrophages.

Characterization of the M1 signature elicited by eNAMPT
To fully characterize the effect of eNAMPT on macrophage-polarized activation, we carried out a comprehensive analysis of the transcriptional profile of PECs by RNA sequencing. Cells were stimulated for 4 h with eNAMPT (500 ng/mL) or with IFNg (200 U/mL), as a reference stimulus inducing classic M1-polarized activation (Adams and Hamilton, 1984). When using a log2 fold-change of at least 1 with an FDR below 0.05, eNAMPT upregulated 407 genes over control ( Figure 2A). When validating a selected 20-gene set by qPCR, we found a strong correlation between the two techniques, thereby validating our findings ( Figure S2A). The IFNg-induced gene expression pattern was coherent with the literature (Das et al., 2018;Piccolo et al., 2017) and resulted in the induction (with the same cut-offs as above) of 947 genes ( Figure 2B). The concordance between the two stimuli was low, with only 134 out of 1,219 genes (11%) that were significantly upregulated by both ( Figure 2C). These results suggest that eNAMPT and IFNg activate two different pathways that ultimately regulate distinct transcriptional programs. Notably, a poor superimposition was confirmed also in terms of extent of gene expression, indeed only a few genes (12 out of the 42 genes) that were upregulated by eNAMPT with at least a log2 fold-change above 2 were also upregulated by IFNg (red bars) ( Figure 2D). Although some of these differences may be attributable to the cut-offs chosen (for example, Icosl has a fold-change of 1.5 with IFNg), most genes were selectively upregulated by eN-AMPT (light blue bars), thereby representing an inflammatory signature that is distinct from the IFNg ones. For example, Il1b, CxCl1, and CxCl3 are strongly induced by eNAMPT and repressed by IFNg (Il1b log2 fold-change À0.19, CxCl1 log2 fold-change À2.44, and CxCl3 log2 fold-change À1.0). We also analyzed the genes downregulated by eNAMPT or IFNg. Using the same cut-offs as above, 241 and 489 genes were repressed by eNAMPT and IFNg, respectively ( Figure S2B). Again, concordance between the two stimuli was low, and the 20 most downregulated genes by eNAMPT are displayed in Figure S2C. Figure 2E shows the top 10 most enriched pathways by eNAMPT using gene ontology (GO) analysis. As expected, there is an enrichment in inflammatory response genes, including those involved in LPS and cytokine (TNFa, IL-6, and IL-1b) responses as well as genes associated with the activation of ERK and NF-kB cascades. Interestingly, the same analysis also highlighted an enrichment in genes involved in chemotaxis. Moreover, the GO molecular function confirmed that the binding of eNAMPT to a receptor (not shown) is the most plausible mechanism whereby eNAMPT modulates gene expression. We also performed KEGG pathway enrichment analyses of the genes upregulated by eNAMPT ( Figure 2F), and we corroborated the involvement of several inflammatory signaling pathways including TNF, NF-kB, JAK-STAT, MAPK, PI3K-AKT, and TLRs. We also analyzed the 241 downregulated genes, but no enrichment was found using either the GO or KEGG databases. Last, we performed predictive analysis of transcription factors driving the upregulated DEGs via Pscan and JASPAR. The results highlighted NF-kB, KLF, and TBX family members as the most enriched transcription factors ( Figure 2G).

eNAMPT promotes chemotaxis in a TLR4-independent manner
The above results pointed out three functional observations: (1) NF-kB appears to be an important mediator of eNAMPT responses; (2) eNAMPT responses seem to be not dissimilar to LPS responses, posing the question on whether eNAMPT acts via TLR4, as previously proposed (Camp et al., 2015); and (3) eNAMPT could play a role in chemotaxis. We also proceeded in analyzing the DEGs with STRING, that predicts interaction between gene products. As shown in Figure S2D, eNAMPT-responsive genes could be clustered in five networks, the three highlighted in the above points as well as IL-6 and TNF networks.
Given that it has been firmly demonstrated that NF-kB acts down-stream of eNAMPT (Camp et al., 2015; Managò et al., 2019), we did not pursue this further. We instead decided to investigate the responses of Figure 2. eNAMPT triggers a unique M1 signature (A and B) Volcano plot of the differentially expressed genes by eNAMPT (500 ng/mL) or IFNg (200 U/mL), respectively, using RNAseq analysis on PECs (n = 5 replicates/condition); FDR % 0.05; (C) Venn diagram of the relationship between eNAMPT-and IFNg-regulated genes (FDR % 0.05 and log2 foldchange > 1); (D) heatmap and histogram representation depicting the most upregulated genes by eNAMPT. Light blue bars represent those genes that are not regulated by IFNg (log2-fold change < 0.5), orange represents those genes that are moderately regulated by IFNg (0.5 < log2-fold change < 1.5), and red represents those genes that are highly regulated by IFNg (log2-fold change > 1.5); (E) gene ontology analysis of eNAMPT-upregulated genes; (F) top 12 pathways highlighted by KEGG analysis emerging from eNAMPT-upregulated genes; number of genes annotated in each pathway (purple) and fold enrichment (green) are shown; (G) patterns of transcription factor motif enrichment within the promoters of the eNAMPT-upregulated genes. iScience Article TLR4-KO PECs to eNAMPT. As shown in Figure 3A, eNAMPT triggered a similar response in wild-type and TLR4-KO PECs, whereas, as expected, LPS did not elicit any response on PECs derived from TLR4-KO mice (data not shown). Among the 22 genes evaluated, CxCl10 only was statistically reduced in TLR4-KO PECs treated with eNAMPT. On a descriptive front, the expression of a few genes was blunted (e.g., Nos2, Il23a, CxCl9, or Il12b), and the others were virtually unchanged. These data suggest a minor contribution of TLR4 in eNAMPT-induced gene expression along with the existence of an alternative receptor for eNAMPTdriven M1 macrophage activation.
Next, to unravel the potential impact of eNAMPT on macrophage migratory behavior, we performed functional in vitro and in vivo assays. First, we carried out a wound healing assay. PECs were seeded at the concentration required to cover cell culture area, scratched and treated with 500 ng/mL eNAMPT or 1 mM of the chemotactic peptide N-formyl-methionyl-leucyl-phenylalanine (fMLP) as a positive control (Ortiz-Masiá et al., 2010). We monitored wound closure at different time points, and we found that eNAMPT and fMLP similarly accelerated wound closure compared with control ( Figures 3B and 3C). Next, using a Transwell migration assay, we evaluated the chemotactic response of PECs toward eNAMPT or fMLP. The results confirmed a remarkable increase of PECs recruited in response to either eNAMPT or fMLP ( Figures 3D and E).
To corroborate this chemotactic activity in vivo, we performed the subcutaneous air pouch model (Figures 3F-3H and S3), enabling the analysis of inflammatory cell response to local chemoattractants (Lu et al., 2020). eNAMPT (50 mg), LPS (1 mg; as a positive control), or an equal volume of PBS were injected subcutaneously, in the air pouch, and, after 6 h, cells recruited were harvested and analyzed by flow cytometry. The results showed a significant accumulation of leukocytes (CD45 + cells) including neutrophils (CD11b + Ly6G high Ly6C low/cells), monocytes (CD11b + Ly6G À Ly6C high cells), and macrophages (CD11b + F4/80 + Ly6C low/À cells) in the air pouches injected with eNAMPT-or LPS as compared with PBS. Moreover, macrophages showed a CD86 high CD206 low phenotype that implies an M1-skewed polarized activation ( Figure 3H).
To determine the potential contribution of TLR4 in eNAMPT chemotactic activity, we carried out in vitro migration assays with TLR4-KO PECs, and we found that eNAMPT still promoted PEC migration in both wound healing model ( Figure 3C) and in Transwell migration assay ( Figure 3E). In keeping with WT PECs, the effect of eNAMPT was comparable to fMLP on TLR4-KO PEC, strengthening that eNAMPT induces PEC migration in a TLR4-independent manner. Consistently, TLR4-KO mice showed an impaired recruitment of inflammatory cells in the air pouch upon LPS treatment but maintained responsiveness to eNAMPT ( Figure 3H).

eNAMPT boosts IFNg responses in a TLR4-independent manner
To get insight into the inflammatory activities of eNAMPT, we decided to explore the effect that eNAMPT exerts toward IFNg responses ( Figure 1D). As shown in Figure 4A, the co-stimulation of PECs with eNAMPT and IFNg regulated a significantly higher number of genes (1715 genes upregulated versus 895 downregulated) than control and single treatments ( Figure 2B). Also in this case, we confirmed the correlation between RNAseq and qPCR ( Figure S4A). As shown in Figure 4B, most genes upregulated by eNAMPT or IFNg as single stimuli ( Figure 2C) are also upregulated by the combination. Indeed, most of the genes that are induced by IFNg (86.1%) are still significantly induced in presence of eNAMPT, whereas approximately two-thirds of the genes (66.1%) upregulated by eNAMPT alone emerged also upon the combination. Of note, a considerable number of genes induced by the combination (44.2%) were not significantly modulated by the single stimuli, indicating that the co-presence of eNAMPT and IFNg might activate new transcriptional programs or enhance the expression of weakly induced genes, leading to an increase of those that overcome the threshold (log2 fold-change>1; FDR <0.05). We next focused on the 50 top-ranked genes that were upregulated by the combination (Figure 4C). Among these genes, we could find a few iScience Article genes that were part of the NAMPT signature (i.e., CxCl9, Il1b, Il12b, Il6, and Marcksl1 and Cd38, Figure 4C), whereas most genes belonged to the IFNg signature. Overall, the combination significantly induced a more pronounced upregulation of genes than the single agents ( Figures 2D and 2E); indeed, all the genes have a log2 fold-change higher than 3 rather than 2. We also evaluated whether the effects of the combination could be additive or synergic (see ''Combinatory evaluation'', Table S3), and we observed that most genes were induced in an additive manner along with a small group of genes, including Cxcl9, Cxcl10, Cxcl11, Gbp4, Il1b, and Il12b, that were synergistically upregulated. The effect of the combination IFNg and eNAMPT is less than additive for only a few of the genes induced. Next, we evaluated the genes that were selectively upregulated by the combination of eNAMPT and IFNg, and we found that only 10 genes ( Figure 4D) showed a log2 fold induction above 2. These results suggest that the main biological functions modulated by the combination are likely associated with the genes belonging to either eNAMPT or IFNg signatures. Therefore, we performed gene ontology (GO) analysis on the top-ranked genes, and we found out that the majority of the pathways enriched by the combination (e.g. immune system processes, cellular response to IFNg, Figure 4F) are also typically associated with IFNg response ( Figure 4E). These results strengthened the concept that eNAMPT has a powerful boosting effect of IFNg response. We also analyzed the 895 downregulated genes ( Figures S4B and S4C), but no obvious trend was observable.
Last, we evaluated the contribution of TLR4 in eNAMPT-dependent promotion of IFNg-induced gene expression. We observed that the response of TLR4-KO PECs to the combination of eNAMPT and IFNg was similar to WT PECs ( Figure 4G), demonstrating that eNAMPT boosts IFNg-driven inflammatory gene expression in a TLR4-independent manner.

The boosting effect of eNAMPT on IFNg is mediated by STAT1/3
We next performed the KEGG analysis of the combination dataset. As shown in Figures 5A and 5B, we observed an enrichment of the genes associated with the activation of the JAK-STAT pathway in the combination compared with IFNg alone. A modest increase of this pathway also emerged in the eNAMPT dataset ( Figure 2G). Moreover, we analyzed the putative transcription factors regulating gene expression programs via Pscan and JASPAR. As expected, IFNg showed an enrichment of IRFs, NF-kB family members, and STATs ( Figure S5A). The combination did not highlight any new transcriptional signatures but showed an enrichment of the transcription factors that were associated with either eNAMPT or IFNg ( Figure 5C). We also focused the analysis on the 757 genes that were upregulated by the combination only ( Figure 4B). Strikingly, we observed only transcription factors (e.g., KLF and SP families) that are associated with eN-AMPT signature. In contrast, STAT emerged only in association with IFNg, either alone or in combination ( Figure S5B). We therefore investigated the effects of eNAMPT on IFNg-induced STAT activation. We stimulated PECs with eNAMPT, IFNg, or their combination, and we analyzed the phosphorylation of STAT1, which is the main transcription factor regulating IFNg-induced gene expression, and STAT3, which is already known to be activated by eNAMPT (Li et al., 2008) and to be a modulator of IFNg biological activity (Qing and Stark, 2004). The results showed that the combination of eNAMPT and IFNg induced a higher level of phosphorylated STAT1 and STAT3 than IFNg alone at 30 0 , followed by a reduced level of both phospho-STATs at 60 0 of stimulation ( Figures 5D and 5E). These results suggest that eNAMPT boosts STAT1/3 signaling and accelerates the kinetics of IFNg-induced STAT1 and 3 phosphorylation. We confirmed this by using a specific STAT3 inhibitor, Stattic (3 mM), and observing the loss of the eNAMPT-mediated boosting effect on IFNg response is reset ( Figure S5C).
To rule out an effect of TLR4 in eNAMPT-induced STATs activation, we analyzed phosphorylation of STAT1 and STAT3 in TLR4-KO PECs. Although we found a faster decay of STAT phosphorylation, the results showed a consistent increase of both phospho-STAT1 and phospho-STAT3 levels after 30 0 of treatment ( Figures 5F-5G), thus confirming that eNAMPT boosts IFNg signaling in a TLR4-independent manner. . eNAMPT acts as a boosting-IFNg response (A) Volcano plot of the differentially expressed genes by eNAMPT (500 ng/mL) and IFNg (200 U/mL) using RNAseq analysis on PECs (n = 5 replicates/ condition); FDR % 0.05; (B) Venn diagram of the relationship between eNAMPT-, IFNg-, and combination-regulated genes (FDR % 0.05 and log2 foldchange > 1); (C) heatmap and histogram representation depicting the most upregulated genes by the combination; red bars represent genes with fold changes higher than expected, blue bars represent genes with fold changes as expected, and yellow bars less than expected (see ''Combinatory evaluation'', Table S3). Genes are indicated with (*) or ( Our RNAseq analysis shows that the Nampt is one of the most IFNg-upregulated genes ( Figure 2D; log2 fold-change of 2.1 over control) and is strikingly further potentiated by the co-stimulation of PECs with eN-AMPT (log2 fold-change of 2.75 over control). Accordingly, it has been recently reported that IFNg upregulates iNAMPT expression in a STAT-dependent manner (Huffaker et al., 2021). We confirmed RNAseq results by qPCR analysis. Despite in THP-1 cells iNAMPT is selectively induced by LPS (Halvorsen et al., 2015), for PECs we observed that Nampt transcription is mostly induced by IFNg ( Figure 6B). These results suggest that Nampt selectively belongs to the IFNg signature and prompted us to explore the relationship between IFNg stimulation and eNAMPT production. First, we evaluated iNAMPT (whole lysates) and eN-AMPT (supernatants) levels after 48-h stimulation with IFNg via western blot. As shown in Figures 6C and  6D, densitometric analysis confirmed the upregulation of both intracellular and extracellular forms of NAMPT, upon IFNg treatment. Importantly, we investigated the mechanism whereby IFNg induced a consistent and robust release of eNAMPT by treating. IFNg-activated PECs with brefeldin A (1 mg/mL) or monensin (1 mM). Both inhibitors of the protein transport from ER to Golgi apparatus significantly reduced eNAMPT release ( Figures 6C and 6D). To corroborate these findings, we analyzed cell-free supernatants by ELISA, and we obtained superimposable results ( Figure 6E). Overall, these results demonstrate that IFNg triggers PECs to increase eNAMPT production by inducing Nampt gene expression and by favoring protein release through the canonical pathway.

DISCUSSION
eNAMPT is increasingly explored as a drug target in a variety of inflammatory diseases Garcia et al., 2021;Quijada et al., 2021). Being the orchestrators of both initiation and resolution of inflammation, macrophages are pivotal players in many disorders and consequently promising targets for new therapeutic strategies (Sica et al., 2015).
Macrophages are distributed through the body where they act as crucial gatekeepers of tissue homeostasis and key players of innate and adaptive immune response (Amit et al., 2016). Plasticity is the hallmark of monocytes/macrophages that carry out different responses to the plethora of physiologic and pathologic microenvironmental signals (e.g., microbial products, endogenous alarmins, metabolites, ROS) they are exposed to (Gordon and Mantovani, 2011). Different studies have investigated the effect of eNAMPT on macrophage polarization; however, this issue has remained debated (Travelli et al., 2018). Indeed, looking at the expression of only a few genes, some studies indicate that eNAMPT has an M1 skewing ability (Bermudez et al., 2017;Halvorsen et al., 2015), whereas others show an enhancement of M2-skewed activation. A potential explanation might be found on the complexity of macrophage activation and the source of macrophages used to perform such studies (Murray et al., 2014). For the first time, we have provided a comprehensive transcriptional analysis by RNAseq of primary murine macrophages (PECs) activated by eNAMPT. Therefore, the results allow us to conclusively determine the effect of this protein on iScience Article macrophage-polarized activation. The description of circulating eNAMPT and its biological activity dates back in time Samal et al., 1994), nevertheless its receptor has remained elusive. We have recently shown that this protein may bind and antagonize CCR5 on murine melanoma cells (Torretta et al., 2020); however, in a muscle injury model in zebrafish, the binding of eNAMPT to the CCR5 expressed by muscle stem cells triggers a signaling cascade that supports muscle regeneration (Ratnayake et al., 2021). These studies suggest that eNAMPT might modulate CCR5 activity in a cell-type-dependent manner. Nonetheless, we have observed that eNAMPT-induced M1 PEC activation is not affected by maraviroc, ruling out a contribution of CCR5 for eNAMPT activities in macrophages (not shown).
TLR4 is an alternative receptor of eNAMPT that has emerged by SPR studies and then confirmed by additional evidence (Camp et al., 2015). For example, in human monocytes a TLR4-neutralizing antibody was able to reduce eNAMPT-mediated NF-kB activation (Managò et al., 2019). Here, we investigated the contribution of TLR4 on the effects of eNAMPT by using PECs from TLR4 KO mice. Our results demonstrate that eNAMPT exerts its effects through a TLR4-independent pathway. Given that KEGG analysis points out a consistent enrichment of TLR signaling, it is reasonable to assume that eNAMPT activities are receptor specific and that the receptor could yet belong to the TLR family.
Alongside a receptor interaction, a second line of thought hypothesizes that the enzymatic activity of eNAMPT could be important (Revollo et al., 2007). In the present contribution, we demonstrate that the effect of eNAMPT is superimposable to that of the catalytically inactive NAMPT H247E mutant, thereby ruling Strikingly, we also observed that IFNg induces the expression and release of eNAMPT, thereby providing a positive feedback loop for macrophage-driven inflammation. It is worth noting that eNAMPT has been found to be increased in numerous pathological conditions that are also associated with elevated levels of IFNg, including autoimmune disorders and sepsis (Chung et al., 2009;Managò et al., 2019;Starr et al., 2017). Our results confirmed the boosting effect of eNAMPT on IFNg-induced gene expression in human monocyte-derived macrophages, therefore strengthening the potential relevance of eNAMPT neutralization in IFNg-dependent inflammatory disorders. Accordingly, both we and another group have generated eNAMPT neutralizing antibodies that are able to mitigate inflammation in preclinical models of inflammatory bowel disease, acute lung injury, and ventilatory-induced lung injury (Camp et al., 2015;Colombo et al., 2020;Quijada et al., 2021).  (Audrito et al., 2015). Last, Li et al. showed an effect of eNAMPT on STAT3 phosphorylation (Li et al., 2008). Our manuscript complements these observations and shows for the first time that, in unskewed macrophages, eNAMPT induces a M1 phenotype and strongly synergizes with IFNg.
In conclusion, we have demonstrated that eNAMPT promotes inflammation by favoring both the recruitment of myeloid cells and the induction of an inflammatory transcriptional program. Moreover, IFNg triggers macrophages to upregulate and release eNAMPT that boosts IFNg-driven transcriptional activation, thereby suggesting eNAMPT as a new amplifier of the cytokine storm.

Limitation of the study
The main limitation of the study is given by the fact that although it presents solid data excluding the involvement of TLR4 and of the enzymatic activity, the responsible receptor remains unknown. Other limitations may be as follows: (1) the high concentration of eNAMPT that does not reflect the amount of the cytokine in the inflammatory milieu, but we used amounts that are coherent with the literature; (2) we did not investigate all the possible pathways that may be activated by eNAMPT, but they will be one of our interests in the future.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Armando A. Genazzani (armando.genazzani@uniupo.it).

Materials availability
This study did not generate new materials or reagents.

Data and code availability
All data generated or analysed during this study are available upon request. The RNA-seq data have been deposited in the Gene Expression Omnibus (GEO) database under the accession GSE189104. This paper does not report original code. Any additional required to reanalyse the data reported in this paper is available from the lead contact upon request.

Isolation of murine peritoneal macrophages
Animal care was in compliance with Italian regulations on protection of animals used for experimental purposes and were authorized by the Ministry of Health (120/2018 DB064.27 of 04/10/2017 and 983/2020-PR DB064.62 of 14/10/2020). C57BL/6 (WT or TLR4-KO, Jackson Laboratory) male 8-weeks-old mice were injected in the peritoneal cavity with 1 mL of 3% Brewer thioglycollate medium (BD Bioscience). After 5 days, the mice were euthanized. After retracting the abdominal skin, exposing the peritoneal wall, 5 mL of sterile PBS were injected closed to abdominal adipose tissue. The liquid in the peritoneal cavity was shacked, aspirated with the syringe closed to sternum and collected for macrophage purification. 2 or 3 3 10^6 cells peritoneal exudate cells (PECs) were seeded in RPMI-FBS free Medium (RPMI, with 10 U/ mL Penicillin, 100 mg/mL streptomycin and 1% L-glutamine, Merck Life Science) and left 1 h in incubator at 37 C 5% CO 2 . Next, the non-macrophage cells were vigorously washed away with PBS and culture in complete RPMI-medium (RPMI with addition of 10% of FBS, Gibco, Thermo Fisher Scientific) at 37 C 5%CO 2 for at least 1 h. Macrophages were treated as following described.

Air pouch
Ten-week-old C57Bl/6 (WT and TLR4-KO) mice were used, and all experiments were performed under isoflurane anaesthesia. Mice were subcutaneously injected with 3 mL of sterile air on the dorsal region, at days 0 and 3. At day 6, 500 ng/mL of eNAMPT or 100 ng/mL of LPS were injected in the pouches. Control mice were administered with PBS. After 6 h, the cells recruited in the pouches were harvested with PBS, stained and analysed by flow cytometry.

Measurement of eNAMPT levels in cell medium
For eNAMPT measurement, 3 3 10^6 cells were seeded in 6-well plates and cultured in serum-free conditions, with or without treatments, for 48 h. Then, the conditioned medium was collected and 50mL were analysed by Western blotting. Experiments were performed in serum-free conditions to avoid aspecific immunoglobulin signals and because of the possible presence of eNAMPT in FBS. In parallel, some nostarved samples were analysed for eNAMPT concentrations using a commercially available sandwich enzyme-linked immunosorbent assay for human or murine NAMPT (ELISA kit from AdipoGen Inc, Seoul Korea).

Gene expression analysis
Cells were lysed with Trizol reagent (Life-technologies) and RNA was extracted with chloroform. 1 mg RNA was reverse transcripted with SENSIFAST kit as manufacturer's protocol (Aurogene) and 20 ng of cDNA were used to perform qPCR with SYBR-green (Bio-Rad) and CFX96 Real-Time System (Bio-Rad). Gene expression results were normalized to actin as housekeeping gene. The sequences of gene-specific primers are reported in Table S1.