Survival and Migration of Human Dendritic Cells Are Regulated by an IFN-α-Inducible Axl/Gas6 Pathway¹

Axl is an important member of the receptor tyrosine kinase family constituted by Tyro3, Axl, and Mer (TAM⁴ family) (1, 2, 3). Each member shares a similar extracellular domain structure, and a conserved catalytic kinase domain in the cytoplasmic portion. Both Axl and Mer undergo proteolytic processing to yield a soluble form (soluble Axl (sAxl) and soluble MER) (4, 5, 6). sAxl, generated by ADAM10-mediated cleavage, is present in cell-conditioned medium of primary and transformed cell lines and in human and mouse serum (7). The TAM receptor ligands are two closely related vitamin K-dependent proteins: Gas6, the product of the growth arrest-specific gene, and protein S, a negative regulator of blood coagulation (8, 9, 10, 11, 12). Gas6 binds the three receptors with different affinity (Axl > Tyro3 > Mer), whereas protein S seems to be a specific agonist for Tyro3 and Mer only (8).

TAM receptors are broadly expressed by cells of the immune, nervous, reproductive, and vascular systems, and by different tumor cell lines (2, 3). Several studies indicate that the Gas6/Axl system plays an important role in cell adhesion and migration (13, 14). Furthermore, Gas6/Axl signaling modulates cell growth and inhibits apoptosis. For example, Axl promotes survival of endothelial and neuronal cells (15, 16), and protects murine fibroblasts and human endothelial cells from apoptosis induced by TNF or other stimuli (17, 18, 19). Both Gas6 and protein S mediate the recognition of apoptotic cells and their subsequent phagocytosis by macrophages through the recognition of phosphatidylserine exposed on apoptotic cell membranes (1, 3, 20). Mutant mice lacking the three TAM receptors show defective clearance of apoptotic cells and develop severe lymphoproliferative disorders accompanied by broad-spectrum autoimmunity (21).

Recently, Rothlin et al. (22) reported that, in murine macrophages and dendritic cells (DC), TAM receptor signaling limits the TLR-induced production of proinflammatory cytokines through the induction of the inhibitory proteins suppressor of cytokine signaling (SOCS) 1 and SOCS3. Intriguingly, Axl was found to associate with the type I IFN receptor (IFNAR1). Also, IFNAR1 as well as its transcription factor STAT1 were shown to be essential for SOCS induction (22). In this murine model, type I IFNs, which are induced downstream of TLR activation, also up-regulate Axl expression via IFNAR-STAT1 signaling (22).

Type I IFNs are pleiotropic cytokines that play an important role in direct antiviral defense, and link the innate and adaptive immune response (23). Although many different cell types of both hematopoietic and nonhematopoietic origin produce type I IFNs in response to infectious agents or inflammatory stimuli, it is well established that plasmacytoid DC are the most potent IFN-α-producing cells after microbial challenge (24, 25). Mounting evidence shows that type I IFNs modulate DC biology at different levels. For example, IFN-α (in combination with GM-CSF), rapidly induces the differentiation of monocytes into potent, functional DC (IFN/DC) (26, 27, 28, 29, 30, 31, 32, 33, 34). These IFN/DC show increased expression of costimulatory molecules and exhibit more potent Ag-presenting activities compared with DC generated in the presence of IL-4 (IL-4/DC) (26, 27, 28, 29, 30, 31, 32, 33, 34). IFN/DC undergo complete maturation upon LPS stimulation and migrate in response to β-chemokines, as a consequence of CCR7 up-regulation (30). Functionally, IFN/DC exhibit potent allostimulatory activity in an MLR assay and induce humoral and Th1-polarized cellular responses in SCID mice reconstituted with human PBMCs (26). IFN/DC also exhibit a cytotoxic activity that is mainly mediated by a subset expressing CD56 (33, 34).

Because IFN-α is a crucial regulator of DC functions, and because Axl expression is regulated by IFN-α, we addressed the question as to whether there is direct evidence of a regulation of Axl expression by IFN-α in human DC. We focused on human IFN/DC, because they are generated in the presence of IFN-α, and there is accumulating evidence that IFN/DC differentiate from monocytes in systemic lupus erythematosus, an autoimmune condition characterized by relatively large amounts of circulating IFN-α (35). In addition, IFN/DC are considered promising candidates for immunotherapeutic studies (36). For comparison, we investigated conventional IL-4/DC (37). IFN/DC and IL-4/DC were generated from human monocytes, and their expression of Axl, Tyro3, Mer, and the ligand Gas6 was investigated in immature and mature DC. In addition, the activity of Gas6 on DC survival and chemotaxis was assessed. These studies revealed that IFN/DC, but not IL-4/DC, acquire cell surface Axl during their differentiation. TLR-dependent maturation stimuli (LPS, poly(I:C), TLR7/8 ligand) significantly down-regulate the Axl expression on IFN/DC through increased proteolytic cleavage. Gas6 protects DC from serum deprivation-induced apoptosis, and promotes chemotaxis of DC in an Axl-dependent manner. These results suggest that IFN-α can regulate DC survival and migration through the up-regulation of Gas6/Axl-mediated signaling.

Human PBMC were obtained from the venous blood of voluntary healthy donors by Histopaque density gradient centrifugation (Sigma-Aldrich) and enriched in monocytes with an isolation kit (Miltenyi Biotec). The resulting preparations were consistently >95% CD14⁺ as determined by FACSCalibur (BD Biosciences). Monocytes were incubated in six-well culture plates (1 × 10⁶ cells/ml) for the indicated times in RPMI 1640 medium with 10% heat-inactivated FBS (Life Technologies) supplemented with 100 ng/ml GM-CSF and 50 ng/ml IL-4 (both PeproTech) to generate IL-4/DC, or with 100 ng/ml GM-CSF and 1000 IU/ml IFN-α or IFN-β (PeproTech) to generate IFN/DC. In some experiments, after 3 days of culture, IFN/DC, previously labeled with anti-Axl primary mAb (R&D Systems) and with a PE anti-mouse Ig (DakoCytomation), were purified with anti-PE microbeads (Miltenyi Biotec) to a purity of ≥95% Axl⁺ cells. Axl⁺ and Axl⁻ IFN/DC were then used for further analyses.

On day 3, IFN/DC were stimulated for 24 h with different stimuli: LPS derived from Escherichia coli O55:B5 (100 ng/ml) and poly(I:C) (15 μg/ml) (both from Sigma-Aldrich); TLR7/8 ligand (5 μg/ml; InvivoGen); and CD40L-transfected cells J558 (1:4 ratio; provided by S. Sozzani, University of Brescia, Brescia, Italy). In the same manner, IL-4/DC were treated with IFN-α (1000 IU/ml) and LPS (100 ng/ml) alone or in combination. Where indicated, IFN/DC were cultured with a broad-spectrum hydroxamic acid-based metalloproteinase inhibitor GM6001 or the TNF-α protease inhibitor TAPI-1 (Calbiochem).

RNA was extracted from cells with TRIzol reagent, and cDNA was synthesized from 2 μg of total RNA by using random oligonucleotides as primers and a SuperScriptII kit (all reagents were from Invitrogen). The cDNA was then amplified for the housekeeping gene hypoxanthine phosphoribosyltransferase (HPRT), Axl, and Gas6 using the following primers: HPRT (sense, 5′-TGA CCT TGA TTT ATT TTG CAT ACC-3′; antisense, 5′-CGA GCA AGA CGT TCA GTC CT-3′); Axl (sense, 5′-CGT AAC CTC CAC CTG GTC TC-3′; antisense, 5′-TCC CAT CGT CTG ACA GCA-3′); and Gas6 (sense, 5′-AAC TCC CCA GGG AGC TAC A-3′; antisense, 5′-GCA CGG CAA GAT GTC CTC-3′). The Universal Probe Library system (Roche) was used to select specific probes. Real-time RT-PCR analysis was performed on Light Cycler (Roche) using TaqMan assay (Invitrogen) with the following thermal steps: initial denaturation at 95°C for 10 min, followed by 45 cycles of denaturation at 95°C for 10 s, annealing at 56°C for 30 s, and extension at 72°C for 60 s.

The cDNA was also amplified for the genes ADAM10 and ADAM17/TNF-α converting enzyme (TACE) using the following primers: ADAM10 (sense, 5′-ACC TGG GAA ACA GTG CAG TCC-3′; antisense, 5′-GGT CAG ATG CTG GGC AGA GAG-3′) and ADAM17 (sense, 3′-ATG AGGACC AGG GAG GGA AAT ATG-5′; antisense, 5′-CAC TCC TGG GCC TTA CTT TCA ATG-3′).

The iQ SYBR Green Supermix (Bio-Rad) was used to run relative quantitative real-time PCR of the samples, according to the manufacturer’s instructions. Reactions were run in triplicate on an iCycler (Bio-Rad), and generated products were analyzed with the iCycler iQ Optical System Software (version 3.0a; Bio-Rad). Gene expression was normalized based on 18S rRNA contents with overlapping results, and the data are evaluated as 2^−ΔCt (where Ct represents cycle threshold) values.

Cells were washed and resuspended in PBS (Sigma-Aldrich) supplemented with 0.2% BSA and 0.01% sodium azide, and incubated with fluorochrome-conjugated mAbs and isotype-matched negative controls (DakoCytomation) after blocking nonspecific sites with rabbit IgG (Sigma-Aldrich) for 30 min at 4°C. The following mAbs were used: anti-Axl, anti-Mer, and anti-Tyro3 (R&D Systems); anti-CD80, anti-CD83, and anti-CD86 (BD Biosciences); anti-MHCII (Ancell); and anti-CCR5, anti-CD11c, and anti-CD123 (BD Biosciences). To evaluate the NK phenotype, we also used Abs against CD56 (Southern Biotechnology Associates), CD49b, NKG2D, TRAIL, IFN-γ (BD Biosciences), IL-15 (R&D Systems), and granzyme B (Hölzel Diagnostica). Type I IFN receptor expression was monitored by staining cells with anti-IFNAR1 or anti-IFNAR2 (PBL InterferonSource). For intracellular staining, cells were fixed with cold 2% formaldehyde solution and permeabilized with 5% saponin solution before staining. For granzyme B, IFN-γ, and IL-15 analysis, the cells were also previously treated for 4 h with brefeldin A. Samples were then collected and analyzed using a FACSCalibur CellQuest (BD Biosciences). Cells were electronically gated according to their light-scatter properties to exclude cell debris.

Concentrations of sAxl, TNF-α, and IL-6 in cell supernatants were evaluated by DuoSet ELISA kits (R&D Systems), according to the manufacturer’s recommendations. Values are given as the mean concentration ± SEM of three independent experiments.

IFN/DC were cultured in RPMI 1640 medium/1% FBS in presence or absence of Gas6 to evaluate its capacity of protection. Human or mouse rGas6 (R&D Systems) was used in this study with identical results. The cells were treated for 15 and 30 h and then evaluated for apoptosis with the annexin V-FITC apoptosis detection kit (Calbiochem).

Migration of IL-4/DC and IL-4/DC treated with IFN-α for 24 h was measured in duplicate with a Transwell system (24-well plates; 8.0 μm pore size; Corning Costar). RPMI 1640 medium alone or 5, 25, 50 nM Gas6 (R&D Systems) or 250 ng/ml rCCL4 (PeproTech) was added to the lower chamber. Wells with medium only were used as a control for spontaneous migration. A total of 2.5 × 10⁵ cells in 100 μl was added to the upper chamber and incubated at 37°C for 4 h. In some migration experiments, IL-4/DC treated with IFN-α were pretreated with a polyclonal Ab (pAb) against Axl (10 μg/ml) or isotype control pAb (10 μg/ml) (both from R&D Systems) for 30 min at 4°C. For receptor desensitization, IL-4/DC treated with IFN-α were first stimulated with 250 nM Gas6 for 30 min at 4°C, then allowed to migrate toward 25 nM Gas6. Cells migrated into the lower chamber were harvested, concentrated to a volume of 200 μl, and counted by flow cytometry. Events were acquired for a fixed time of 60 s. The counts fell within a linear range of the control titration curves obtained by testing increasing cell concentrations. Values are given as the mean number of migrated cells ± SEM.

Statistical analysis was performed by Student’s t test (GraphPad Prism 4; GraphPad). Values of p ≤ 0.05 were considered significant. Values are expressed as the mean ± SEM.

Axl has been reported to be modulated by IFN-α treatment (22, 38). Therefore, we investigated whether the expression of Axl and/or Gas6 is regulated during the IFN-α-mediated differentiation of DC from human monocytes. IFN/DC were generated by culturing CD14⁺ monocytes with GM-CSF and IFN-α, and Axl expression was analyzed by flow cytometry. Conventional IL-4/DC were used for comparison.

As shown in Fig. 1,A, cell surface Axl was expressed by ∼50% of IFN/DC, whereas it was undetectable in IL-4/DC. A clear-cut induction of Axl expression (33 ± 5% of positive cells) was detected as early as 1 day after GM-CSF and IFN-α treatment. At day 3, 50% of the cells were Axl positive, and Axl expression remained stable until 5 days posttreatment (Fig. 1,A). Both IFN/DC and IL-4/DC showed typical phenotypes, as previously reported (28, 29, 30, 31). IFN/DC expressed higher levels of the costimulatory molecules CD80, CD86, and MHCII, compared with IL-4/DC (data not shown). In accordance with the Axl cell surface expression, Axl mRNA levels, measured by real-time RT-PCR, increased from day 1 to day 3 upon IFN-α treatment; Gas6 was also up-regulated, but in a delayed manner compared with its receptor (Fig. 1 B).

To confirm that IFN-α is the factor responsible for Axl up-regulation during DC differentiation, monocytes were treated with GM-CSF, IFN-α, or GM-CSF and IFN-α, and Axl expression was assessed after 48 h. GM-CSF-treated monocytes did not express Axl. IFN-α induced Axl expression in 15% ± 4 of monocytes; however, the combination of both cytokines increased Axl expression in a higher proportion of monocytes (33 ± 7%) (Fig. 1 C).

Axl expression induced by IFN-α was found to be dose dependent, with a maximal induction in the dose range of 1,000–10,000 IU/ml, and some induction with concentrations as low as 50 IU/ml (Fig. 1,D). Comparable results were obtained using IFN-β (data not shown). IFN-α specifically induced Axl, but did not affect the expression of Tyro3 and Mer, the other two members of the TAM family, whose level was comparable in both IFN/DC and IL-4/DC at day 3 of differentiation (Fig. 2 A). These results indicate that IFN-α specifically induces Axl expression during IFN/DC differentiation.

Because only about half of the IFN/DC population was positive for Axl (Figs. 1,A and 2,A), we next investigated whether Axl expression correlated with the expression of selected lineage and activation markers. Axl⁺ and Axl⁻ IFN/DC were sorted by microbeads and phenotypically characterized by FACS analysis. Both populations showed an almost identical pattern of the DC lineage markers CD11c, CD123, and CCR5, and of the cytotoxicity markers that are highly expressed by and shared between IFN/DC and NK cells (CD56, CD49B, granzyme B, TRAIL, NKG2D) (33, 34) (Fig. 2,B). Intracellular immunoreactivity for IFN-γ and IL-15, two cytokines shown to be present in IFN/DC (32, 34), was also expressed at comparable levels by both Axl⁺ and Axl⁻ IFN/DC. Interestingly, however, these two populations differed in their expression of IFNAR1 (mean fluorescence intensity (MFI) = 190 for Alx⁺; MFI = 36 for Axl⁻) (Fig. 2 C), a cytokine receptor that is physically associated with Axl (22). Therefore, we can postulate that, because they share DC lineage markers, Axl⁺ and Axl⁻ IFN/DC most likely originate from the same precursor cells, and later differentiate into two functionally distinct populations.

Given that Axl is regulated during IFN/DC differentiation, next we asked whether DC maturation influences Axl expression. To this end, IFN/DC were induced to mature with LPS, poly(I:C), TLR7/8 ligand, and CD40L, and Axl expression was measured by FACS analysis. Mer and CD83 were used for comparison (Fig. 3,A). As shown in Fig. 3 A, unexpectedly, Axl expression was significantly down-regulated by maturation stimuli, whereas the DC maturity indicator CD83 was highly up-regulated and Mer was not affected.

Because IFN-α regulates IL-4/DC maturation (39, 40, 41, 42), we analyzed Axl expression in IL-4/DC stimulated for 24 h with either IFN-α or LPS. As shown in Fig. 3,B, IFN-α-mediated activation of IL-4/DC resulted in Axl up-regulation comparable to the levels observed in IFN/DC. LPS did not affect basal Axl and Mer surface levels on IL-4/DC; however, it strongly inhibited Axl expression induced by IFN-α (Fig. 3 B). IFN-α, therefore, can induce Axl expression in IFN/DC as well as in IL-4/DC, and its effect is inhibited in both models by maturation stimuli. Therefore, only IFN-α, independently of whether it is used as a maturation or differentiation stimulus, specifically induces Axl expression in both DC cell types. In contrast, in both IFN/DC and IL-4/DC, maturation-inducing stimuli have the opposite effect, down-regulating Axl expression.

Rothlin et al. (22) reported an increase of Axl expression at the mRNA and protein level in murine bone marrow-derived DC (BMDC) treated with different TLR agonists. Because of this discrepancy with our data on human DC, we next studied whether the decreased amount of cell surface Axl seen in our system resulted from lower production or increased shedding. To elucidate this issue, we tested Axl mRNA expression in IFN/DC cultured in the presence or in the absence of LPS (100 ng/ml) or poly(I:C) (15 μg/ml) for 6 h. As shown in Fig. 4 A, Axl mRNA was markedly elevated upon treatment with both TLR agonists, in line with Rothlin’s results.

Cell surface Axl expression was measured by FACS analysis in IFN/DC after treatment with LPS, poly(I:C), and TLR7/8 ligand for different times (1, 3, and 6 h). A decrease in the percentage of Axl-positive cells was already evident after 3 h of treatment, and a further decrease was observed after 6 h (Fig. 4 B). No significant differences in Axl expression were observed at 1 h (data not shown).

The discrepancy between mRNA and cell surface Axl expression can be explained by our additional ELISA findings, which reveal that sAxl was significantly increased in the supernatants of IFN/DC treated with LPS, poly(I:C), and TLR7/8 ligand (Fig. 4 B). Notably, the augmented amount of sAxl in the culture medium was concurrent with a reduced expression of the transmembrane protein at the various times.

Members of the metalloproteinase superfamily have been shown to be responsible for the cleavage of the majority of shed proteins (43, 44). To test whether the release of sAxl is mediated by metalloproteinases, cells were first preincubated for 30 min with GM6001, a general inhibitor of matrix metalloproteinase (MMP) and a disintegrin and metalloproteinase (ADAM) family proteases (7), followed by direct addition of maturation stimuli. These experiments demonstrated the ability of GM6001 to restore the presence of Axl on the cell surface and to diminish sAxl release in maturation stimuli-treated cells (Fig. 4,B). Furthermore, we observed a capacity of GM6001 to suppress constitutive Axl shedding in IFN-DC, as demonstrated by a significant reduction in the release of sAxl (p = 0.0002) and by a higher amount of Axl on the cell membrane (p = 0.0462) at 6 h (Fig. 4 B). These data indicate that the TLR-induced reduction of surface Axl corresponds to an increased proteolytic cleavage.

Taking into account that ADAM10 and ADAM17/TACE have been implicated in the constitutive and inducible ectodomain shedding of a number of cell surface-expressed molecules and murine Axl (7, 43), we investigated the involvement of ADAM proteases in sAxl generation in human DC. First, we examined the expression of ADAM10 and ADAM17 in IFN/DC unstimulated or treated with LPS for 6 and 24 h by real-time RT-PCR. ADAM10 and ADAM17 mRNA is expressed by unstimulated IFN/DC. The level of ADAM10 expression was not altered by treatment with LPS, whereas a modest increase in the ADAM17 mRNA level was observed 24 h after stimulation (data not shown).

Next, cells were pretreated with the ADAM inhibitor TAPI-1 (45) for 30 min, followed by addition of LPS to the culture medium for 6 h. As shown in Fig. 4 C, TAPI-1 significantly inhibited LPS-induced Axl shedding, as demonstrated by a dramatic drop in the amount of sAxl and an increase in the percentage of Axl-positive cells.

Our data indicate that the activation of several TLR pathways, despite Axl induction at mRNA level, results in a dramatic decrease in the expression of cell surface Axl most likely mediated by ADAM family protease activity in human IFN/DC. The TLR-mediated regulation of cell surface Axl expression in human DC seems to be distinct from that in murine DC. To better understand these differences, mouse BMDC were stimulated with LPS for 24 h and analyzed for Axl cell surface expression (supplemental Fig. A).⁵ Axl was expressed on mouse DC in both unstimulated and stimulated conditions; only a decrease in MFI was observed in LPS-treated compared with untreated cells (MFI of unstimulated BMDC was 347 ± 23 vs 208 ± 4 (n = 4) in LPS-stimulated BMDC). Similarly to the results obtained in human IFN/DC, mouse DC constitutively released considerable amounts of sAxl, and treatment with LPS further elevated sAxl production. The inhibitor GM6001 significantly reduced both the constitutive and LPS-inducible shedding of Axl (supplemental Fig. B).

Because LPS activation induced the cleavage of Axl ectodomain on IFN/DC (Figs. 3,A and 4, B and C), we investigated whether, vice versa, the Axl ligand, Gas6, interfered with TLR-mediated signaling. This was assessed by measuring TNF-α and IL-6 levels in the supernatants of IFN/DC that had been concomitantly stimulated with LPS and Gas6, or that had been pretreated with Gas6 for 8 h and then stimulated with LPS (Fig. 5). Coincubation of Gas6 with LPS did not affect TNF-α and IL-6 release, as compared with LPS alone. However, Gas6 pretreatment significantly inhibited LPS-induced TNF-α and IL-6 production. Thus, the engagement of Axl by Gas6 renders IFN/DC unresponsive to subsequent LPS-mediated activation.

We investigated whether Gas6 stimulation affected apoptosis induced by growth factor deprivation in IFN/DC. Apoptosis was induced in IFN/DC by lowering the serum concentration from 10 to 1%, and the proportion of apoptotic cells was assessed by flow cytometry by annexin V-FITC and propidium iodide (PI) staining.

After 15 h of culture in low-serum conditions, 11% of IFN/DC was annexin V⁺ and PI⁻, reflecting the percentage of cells undergoing early stages of apoptosis, and 23% was already double positive (annexin V⁺-PI⁺), indicating a late stage of apoptosis. A higher percentage of early apoptotic cells was observed after 30 h of culture (18%). However, when IFN/DC were incubated in the presence of 50 nM Gas6, we observed a reduction in the percentage of early apoptotic cells: thus, only 6 and 8% of IFN/DC were annexin V⁺/PI⁻ at 15 and 30 h, respectively (Fig. 6). These results were consistently observed with three different donors.

Therefore, we can conclude that Gas6/Axl-mediated signaling promotes survival of human DC.

Because migration is pivotal for DC function as immune sentinels, we analyzed whether Gas6 exerted chemotactic effects on these cells, comparing the capacity of Axl⁺ (IFN-α-activated IL-4/DC and IFN/DC) and Axl⁻ (IL-4/DC) human DC to migrate in response to exogenous Gas6 in a Transwell system. As shown in Fig. 7, Gas6 induced the chemotaxis of IFN-α-activated IL-4/DC in a dose-dependent manner, generating a marked chemotactic response to increasing concentrations (5, 25, and 50 nM) of Gas6 (Fig. 7,A). Similar migratory response to Gas6 was obtained with IFN/DC (data not shown). In contrast, IL-4/DC did not show any chemotactic response to Gas6, suggesting that Axl expression is essential for Gas6 chemotactic effects. Both DC populations migrated in response to CCL4 used as positive control. As previously shown for IFN/DC, IFN-α-activated IL-4/DC displayed a higher spontaneous migration and chemotactic response to CCL4 compared with IL-4/DC (30). The role of Axl is supported by the observation that cell pretreatment with an excess of rGas6 abrogated the chemotactic response to Gas6 (Fig. 7,B). Final confirmation that these effects are Axl dependent was obtained by the demonstration that neutralizing anti-Axl Ab blocked Axl⁺ DC Gas6-driven migration (Fig. 7 B). Migration toward Gas6 was not affected by pretreatment with isotype-matched control Ab. This is the first evidence that Axl stimulation can mediate DC chemotaxis.

In this study, we demonstrate that Axl is up-regulated on human DC during differentiation and activation in an IFN-α-dependent manner. Furthermore, we show that Gas6 inhibits apoptosis and stimulates migration of human DC via Axl activation, providing the first evidence that Gas6/Axl-mediated signaling regulates human DC activities and can be considered a novel chemotactic pathway for DC. Previous detection of Axl on murine macrophages (20, 21, 38) and on BMDC (20) fostered the idea that Axl was of importance in murine APC biology (20). Moreover, it has recently been found that Gas6 regulates TLR-induced cytokine release by murine DC in vitro (22). Our new findings relative to human DC further corroborate the importance of Gas6/Axl-mediated signaling in DC biology.

IFN/DC and IL-4/DC share typical DC characteristics, but, probably due to a distinct transcriptional signature of IFN-α in comparison with other cytokines, IFN/DC express higher levels of several maturation markers involved in T cell activation (CD80 and CD86) (26, 27, 28, 29, 30, 31, 32, 33, 34), in the migratory capacity to the lymph node (CCR7) (30), in cell survival (32), and show a plasmacytoid-like phenotype associated with NK cell characteristics (31). We observed that Axl was absent on IL-4/DC, but within the IFN/DC we were able to identify an Axl-expressing subpopulation that displayed high migratory activity toward Gas6.

IL-15 up-regulates Axl expression in mouse fibroblasts (46), and IL-15-mediated NK cell development from hematopoietic progenitor cells is impaired upon blockade of the Axl/Gas6 pathway (47). In this study, we report that IFN/DC express high levels of IL-15, with no differences between Axl⁺ and Axl⁻ cells. Thus, it seems unlikely that IL-15 is involved in Axl up-regulation in IFN/DC. Interestingly, Axl⁺ and Axl⁻ differ in IFNAR expression. Current investigations are underway to elucidate whether this difference reflects some peculiar response.

Rothlin et al. (22) reported that Axl mRNA and protein expression was increased in murine BMDC treated with different TLR agonists. Our study demonstrates that, even though maturation stimuli such as LPS induce Axl gene transcription, they lower the expression of cell surface Axl by inducing its shedding. Human monocytes and DC express several metalloproteinases, some of which are increased following LPS-induced maturation (48). ADAM10-mediated proteolysis has been reported to constitute a major mechanism in sAxl generation by Axl shedding in mice (7). The inhibition of both basal and TLR agonist-inducible shedding by the general MMP and ADAM family inhibitor, GM6001, demonstrated that Axl shedding is metalloproteinase dependent. Moreover, also a broad ADAM inhibitor TAPI-1, which is extensively used to block ADAM17/TACE-mediated TNF-α shedding (45), inhibited the release of sAxl, suggesting a role of ADAM proteases in the Axl cleavage in human DC. However, because of the not complete selectivity of TAPI-1, additional experiments with a targeted deletion of ADAMs or the use of compound that preferentially blocks their activity are required to understand the individual contribution(s) of ADAM members to Axl release in IFN/DC.

The decreased amount of Axl on the cell membrane could result in a reduced sensitivity to Gas6. Thus, the sequence in which human DC encounter TAM ligands and LPS might influence DC activation. Our findings that Gas6 pretreatment inhibits LPS-induced production of TNF and IL-6 support this hypothesis. The inhibitory effect of Gas6 could be mediated by the induction of Twist transcriptional repressor that has been implicated in the suppression of TNF by type I IFN and Axl (38). Coincubation of Gas6 and TLR ligands does not inhibit IL-6 and TNF production in human DC, but does inhibit the production of these cytokines in murine DC, as reported by Rothlin et al. (22). We show that differently from the human DC in murine BMDC, Axl is expressed on the surface of all CD11c⁺ cells. Therefore, a more efficient Gas6-mediated Axl stimulation, compared with human DC, could result in an inhibitory signal for TNF-α and IL-6 production.

It has been shown that Gas6-Axl exerts antiapoptotic effects on a variety of different cell types, including endothelial cells and vascular smooth muscle cells (15, 16, 17, 18). In this study, for the first time, we report that the Gas6/Axl pathway regulates IFN/DC survival by inhibiting apoptosis induced by growth factor deprivation. Significant lifespan differences have been reported for various DC subsets and mature vs immature DC, but the underlying mechanisms remain unclear (49). Although it remains to be investigated whether Gas6 can prevent apoptosis induced by different conditions, our results strongly suggest that Gas6/Axl can selectively influence the DC lifespan.

Compared with IL-4/DC, IFN/DC exhibit enhanced chemotaxis to the inflammatory chemokines CCL3 and CCL5 (30, 50). This suggests a selective recruitment of IFN/DC with a highly efficient Ag-presenting function at the inflammatory site. Because of their homology to cell adhesion molecules, the Axl family receptors have been implicated in cell adhesion and motility (see Refs. 7, 8, 9 for reviews). Axl promotes fibroblast intercellular adhesion via homophilic and heterophilic mechanisms (13), whereas Gas6 induces vascular smooth muscle cell and neuronal cell migration (14, 15). Gas6 inhibition of vascular endothelium growth factor-A-dependent endothelial cell chemotaxis has also been reported (51). Interestingly, Gas6^−/− mice show impaired leukocyte recruitment (52). TAM^−/− mice with homozygous defects in each one of the three TAM receptors develop autoimmune diseases (21). However, the extent to which functional defects of individual TAM family members contribute to the development of human autoimmune diseases remains to be determined. It has recently been reported that high levels of Gas6 are present in the cerebrospinal fluid of patients with inflammatory autoimmune demyelinating diseases (53). This leads to the suggestion that in certain autoimmune conditions characterized by both high IFN-α (54) and Gas6 levels, Gas6 might affect IFN/DC functions. In this study, we report that Axl⁺ DC migrate in response to Gas6 and that this effect is Axl mediated. Therefore, DC responsiveness to Gas6 may also modulate DC trafficking during autoimmune conditions. In addition, preliminary data from our laboratory indicate that sAxl is 2-fold increased in a small cohort of patients with systemic lupus erythematosus (our unpublished observations). This encourages one to systematically assess the Axl expression patterns and Gas6-induced migratory properties of circulating DC in patients with autoimmune diseases so as to further explore the hypothesis that excessive inappropriate Gas6/Axl-mediated signaling may be involved in the pathogenesis of some autoimmune diseases.

In summary, our study reveals that functional, type I IFN-dependent Gas6/Axl signaling is required for human DC survival and migration in vitro. This novel pathway in the regulation of human DC biology deserves to be explored as a potential therapeutic target in autoimmune diseases in which type I IFN plays an important role.

The authors have no financial conflict of interest.