Regulated Osteopontin Expression by Dendritic Cells Decisively Affects their Migratory Capacity

TO THE EDITOR

Osteopontin is a secreted arginine-glycine-aspartate (RGD)-containing phosphoglycoprotein that interacts with different cell-surface receptors, including αvβ3, αvβ5, α4β1, α9β1, and certain CD44 isoforms (Denhardt et al., 2001; Sodek et al., 2006; Scatena et al., 2007). Osteopontin (OPN) has been implicated to have important cytokine and chemokine functions in Th1/Tc1-mediated immunity against viral and bacterial pathogens and in type-1-mediated diseases such as rheumatoid arthritis, autoimmune encephalitis, and granuloma formation (Ashkar et al., 2000; Diao et al., 2004; Tanaka et al., 2004; Shinohara et al., 2005; Hur et al., 2007). We demonstrated that OPN-null mice are impaired in their capacity to mount allergic contact hypersensitivity against trinitro-chlorobenzene (TNCB) (Weiss et al., 2001). The initiation of Th1-mediated immunity depends upon the function of antigen-presenting myeoloid dendritic cells (DCs) (Banchereau et al., 2000; Kapsenberg 2003). Contact hypersensitivity is a Th1-mediated immune response that is elicited once a hapten has been recognized during the initial sensitization phase. DCs are central to the sensitization process, as they recognize and transport the allergen into skin draining lymph nodes for presentation to naive T cells (Romani et al., 2006). OPN is likely to influence contact hypersensitivity response because it activates DCs and induces their maturation toward an IL-12 secreting, Th1-polarizing phenotype (Renkl et al., 2005). We previously described that OPN is chemotactic for DCs and OPN-null mice are impaired in their capacity to attract DCs to skin draining lymph nodes (Weiss et al., 2001). As we and others found that myeloid DCs produce OPN when differentiating from monocytes, we went on to investigate the role of autocrine effects of OPN produced by DCs themselves (Kawamura et al., 2005; Renkl et al., 2005). As convincing evidence exists that the migratory phenotype of macrophages, endothelial cells, and various types of tumour cells is associated with a high OPN-expressing phenotype, we determined which factors regulate OPN expression during murine DC maturation and investigated the role of DC-expressed OPN for their migratory abilities (Bruemmer et al., 2003; Rangaswami et al., 2006; Sodek et al., 2006; Nystrom et al., 2007).

In murine bone marrow cultures, we found that under the addition of GM-CSF and IL-4 OPN mRNA and protein secretion increased constantly from day 3 to 7 of DC maturation (Figure 1a and b). When separating CD11c+DC from bone marrow cultures, these were the major OPN-secreting cells as they expressed high levels of OPN mRNA in contrast to the CD11c-depleted fraction (Figure 1c). Our findings demonstrate that OPN expression is strongly upregulated early during DC differentiation from bone marrow precursors. The fact that immature DCs highly express OPN is interesting especially as other immune cells, such as T cells, NK-cells, and monocytes, express OPN only following their activation in an inflammatory context (Shinohara et al., 2005; Sodek et al., 2006; Hur et al., 2007).

Figure 1. Upregulation of constitutive OPN secretion by TNF-α and IL-1α correlates with the migratory capacity of DCs.

DCs were generated from bone marrow of mice (C57BL/6) in the presence of GM-CSF (40 ng ml^-1; PromoCell, Heidelberg, Germany) and IL-4 (10 ng ml^-1; Cell Concepts, Umkirch, Germany) for 5 or 6 days as described previously (Weiss et al., 2001). During DC differentiation, supernatants and cells were obtained daily. In all the figures, data representative of at least four independent experiments is shown. (a) OPN concentration in supernatants obtained on the indicated days of culture was measured by OPN specific ELISA (IBL-Japan, Gunma, Japan) according to the manufacturer’s instructions. (b) OPN quantitative real-time PCR of cells harvested on the indicated days was performed with a LightCycler (Roche Diagnostics, Mannheim, Germany) using the following primers: OPN (108-bp product): sense, 5′-GGTGATAGCTTGGCTTATGGACTG-3′; antisense, 5′-GCTCTTCATGTGAGAGGTGAGGTC-3′; glyceraldehyde-3-phosphate dehydrogenase (147-bp product): sense, 5′-TGGCCTTCCGTGTTCCTACC-3′; antisense, 5′-GGTCCTCAGTGTAGCCCAAGATG-3′. Thermocycling conditions were as follows: hold at 95 °C for 15 minutes, 40 cycles of 95 °C for 15 seconds, 60 °C for 20 seconds, and 72 °C for 10 seconds followed by melting-curve analysis. The relative expression of target gene in different samples was normalized to the endogenous glyceraldehyde-3-phosphate dehydrogenase and was calculated with the $2_{\mathrm{t}}^{-\Delta \Delta C}$ method (Livak et al., 2001). (c) For further DC enrichment, CD11c+ cells were positively selected by magnetic cell sorting (Miltenyi, Bergisch Gladbach, Germany) and quantitative real-time PCR was performed for OPN as described above. Unseparated bone marrow DC culture cells were used as control. (d) DCs were harvested from bone marrow cultures on day 5, washed, replated (1 Mio. cells per well) and stimulated with LPS (serotype 0111:B4; Sigma-Aldrich, Hamburg, Germany) at the indicated concentrations. Viability was determined by propidium iodide staining by fluorescence-activated cell sorting and did not significantly differ for unstimulated controls and the indicated LPS concentrations (93-95% viability, after 48 hours of culture). (e) DCs were harvested from bone marrow cultures on day 5, washed and cultured with or without GM-CSF (20 ng ml^-1) for 48 hours without additional cytokines, GM-CSF or as indicated additional IL-4 (5 ng ml^-1), TNF-α (100 U ml^-1; Cell Concepts), IL-1α (500 pg ml^-1; Cell Concepts) or LPS (1 μg ml^-1, serotype 0111:B4; Sigma-Aldrich) at 0.5 Mio. Cells per well. Supernatants were obtained after 48 hours of stimulation to detect OPN concentration by ELISA. To investigate the effect of OPN production on their migratory capacity, DCs were compared in their migratory function towards TNF-α in modified Boyden chamber assays. Cells that had migrated to the filter bottom were quantified counting four ocular grids (×40). Results are expressed as mean number of migrated cells per mm^-2 ±SD of six chambers. *Statistically significant compared with GM-CSF control (t-test P<0.001).

Both tumor-necrosis factor-α (TNF-α) and IL-1α induce a highly migratory phenotype in DCs (Cumberbatch et al., 2002). To establish a possible correlation between the migratory phenotype of DCs and their OPN-secreting potential, DCs were stimulated with TNF-α, IL-1α, or lipopolysaccharide (LPS). TNF-α and IL1-α induced OPN secretion (Figure 1d and e). Whereas IL-4 only moderately downmodulated OPN expression, LPS that induces terminally activated, non-migratory DCs strongly inhibited their OPN secretion, which was not due to significantly reduced viability (Figure 1d and e). In Boyden chamber assays, we investigated how the level of OPN secretion correlates with the migratory potential of DCs (Figure 1e). Indeed, the highly OPN-secreting, TNF-α- or IL-1α-treated DCs efficiently migrated toward TNF-α (Figure 1e). In contrast, LPS-matured, low OPN secreting DCs were non-migratory (Figure 1e). Similar to our findings with DCs, it was described that LPS suppresses OPN in macrophages (Shinohara et al., 2006), however, not correlating this OPN-low state with migration. We can only speculate that LPS in both cell types may induce a low OPN-expressing less migratory phenotype.

To demonstrate that OPN deficiency is sufficient to influence DC migratory function, DCs generated from wild type and OPN-null mice were compared in their potential to migrate toward TNF-α, OPN, and CCL19. Whereas OPN wild-type DCs showed the previously described migration toward TNF-α, OPN, and CCL19, OPN-deficient DCs were unable to sufficiently migrate toward any of these stimuli (Figure 2a and b). Interestingly, in both monocytes and DCs, the high OPN-expressing state is correlated with high mobility. In vivo, macrophages migrating toward dermally injected N-formyl-met-leu-phe highly express OPN and Nystrom et al. (2007) recently reported that silencing of OPN expression by RNA interference impaired macrophage migration in vitro (Giachelli et al., 1998).

Figure 2. OPN-deficient DCs are impaired in their capacity to migrate in vitro and to enter skin draining lymph nodes in vivo.

DCs were generated from bone marrow of OPN-deficient C57BL/6 mice (-/-) (tenth generation of backcrossing) and wild-type littermates (+/+) for 6 days and used in Boyden chamber migration assays to migrate toward (a) OPN (2.5 μg ml^-1, recombinant OPN), (a) TNF-α (50 and 250 U ml^-1), or (b) CCL19, R&D Systems, Wiesbaden, Germany (100 or 300 ng ml^-1). Hank’s balanced salt solution medium was used as control. Cells that had migrated to the filter bottom were quantified as described above. *Statistically significant compared with corresponding +/+ DCs (t-test P<0.001). Representative data of four independent experiments is shown. (c) In vivo DC migration assays were performed as described previously (Lappin et al., 1999). All animal protocols were approved by the local Committee of Animal Research. DCs were generated from OPN-null (-/-) C57BL/6 mice or wild-type littermates (+/+) and labeled with PKH2-GL (Sigma Immunochemicals, Hamburg, Germany) and injected at 250,000 DCs in 30 μl of phosphate-buffered saline into each quadrant of the abdominal skin of wild-type mice. Twelve wild-type mice were injected with OPN wild-type DCs and 14 mice with OPN-null DCs. Inguinal and axillary lymph nodes were obtained after 48 hours, mechanically disaggregated, pooled, stained with phycoerythrin-labeled antibody against mouse CD11c (clone HL3, hamster IgG; Pharmingen, Heidelberg, Germany). CD11c, PKH2-GL double positive cells were quantified by a FACS. Using Mann-Whitney rank-sum test the difference between the two groups is statistically significant (P=0.042).

To demonstrate that OPN expression also correlates with DC migratory potential in vivo, fluorescently labeled wild-type DCs or OPN-deficient DCs were injected into the abdominal skin of wild-type mice (Lappin et al., 1999). Quantitative fluorescence-activated cell sorting analysis of pooled inguinal and axillary lymph nodes revealed that OPN-deficient DCs are significantly limited in their migratory efficiency into skin draining lymph nodes (Figure 2c). These data provide evidence that autocrine OPN production is centrally involved in DC migratory abilities.

The mechanism by which OPN mediates DC migration remains to be elucidated. Chemokine receptors CCR5 and CCR7 are both crucially involved in migration of DCs. We speculated that these receptors are modulated by OPN. However, DC treatment by recombinant OPN revealed that their expression is not influenced by OPN (data not shown). In addition to the secreted form, an intracellular form of OPN may be important for DC migration by modulating CD44 activity, as described for peritoneal macrophages (Zohar et al., 2000; Zhu et al., 2004). Because we have previously shown that CD44 isoforms are involved in DC migration, further investigations on the role of the intracellular form of OPN in DC migration are under way (Weiss et al., 1997).

In conclusion, our work demonstrates that OPN is highly expressed by myeloid DCs and is differentially modulated by DC-stimulating factors. In vivo and in vitro optimal DC migration depends on the expression of OPN.

Abbreviations

DC

dendritic cell

LPS

lipopolysacharide

OPN

osteopontin

TNF-α

tumor-necrosis factor-α