Dual regulation of osteopontin production by TLR stimulation in dendritic cells

ABSTRACT

OPN, a cytokine produced, among others, by DCs, is involved in inflammation and defense against patho- gens. Here, we report that the activation of the MyD88 pathway by TLR2, TLR5, and TLR7/8 agonists or IL-1β induces high levels of OPN in human DCs. Conversely, LPS and Poly I:C, two TLR3 and TLR4 agonists that en- gage the TRIF pathway, were ineffective. TLR2 agonists were the strongest OPN inducers, and OPN production was highly stimulated by TLR2-triggering bacteria (Staphylococcus aureus) but not by TLR4-triggering Escherichia coli. Costimulation experiments revealed that TLR3 and TLR4 agonists, beyond being inactive by themselves, sharply limited TLR2-dependent OPN pro- duction by activating a TRIF-dependent inhibition of the MyD88-dependent OPN production. MyD88 silencing impaired TLR2-dependent OPN induction, whereas TRIF pathway blockage by chloroquine, dynasore, or TRIF knockdown prevented the TLR3/4 agonist-medi- ated inhibition, which was independent from the endog- enous production of type I IFN, IL-29, IL-10, or TGF-β.

LPS and Poly I:C inhibitory activity was associated with the release of a >10-kDa protein factor(s). We also demonstrated that the higher OPN levels produced by S. aureus-treated DCs compared with E. coli-treated DCs were responsible for a markedly increased pro- duction of IL-17 by CD4+ T cells. These results high- light the biological relevance of the differential OPN induction by TLR2 and TLR4 agonists and emphasize the importance of TLR cross-talk in OPN induction. This implies that OPN regulation by TLR signaling is critical in shaping inflammatory responses and may modulate IL-17 production in response to pathogens. J. Leukoc. Biol. 94: 000 – 000; 2013.

Introduction

OPN is a glycosylated phosphoprotein, originally isolated from bone matrix and later identified as a factor secreted by acti- vated T cells (from which its original definition is early T lym- phocyte activation-1) [1, 2]. During inflammation, OPN regu- lates adhesion and recruitment of neutrophils, macrophages, and T cells through the binding to β1 integrins and to the variant of the hyaluronic acid receptor CD44v6-v9 [3–5]. It is also involved in angiogenesis, functioning as a chemoattractant for vascular endothelial cells and as a VEGF and IL-1 inducer [6, 7]. The role of OPN in the regulation of adaptive immu- nity is complex and still debated. OPN was associated with Th1 immune responses through the induction of IL-12 and the suppression of IL-10 production [2, 8, 9]. Accordingly, data from OPN-deficient mice and human patients underline the importance of OPN in several Th1-associated diseases, such as sarcoidosis, rheumatoid arthritis, multiple sclerosis, and Crohn’s disease [10 –13]. In view of its role on Th1 polariza- tion, OPN was also implicated in the host immune response against pathogens, such as Herpes simplex virus type 1, rotavi- rus, Listeria monocytogenes, Mycobacterium bovis, and Klebsiella pneumoniae [2, 14 –16]. However, OPN was also implicated in allergic diseases of the airways mediated by Th2 cells [17, 18]. Recent works highlight the role of OPN in the Th17-related immune responses involved in asthmatic inflammation [19]. In addition, OPN promotes the differentiation of Th17 cells in rheumatoid synovium [20] and regulates IL-17 production during the pathogenesis of hepatitis [21] and acute coronary syndrome [22].OPN regulates DC biology by inducing DC migration, sur- vival, and maturation [23–25]. DCs, as a link between innate and adaptive immunity, are key players in determining the in- duction of T cell responses and in orchestrating the immune response against invading microbes [26]. The analysis of the contribution of OPN to the immunological activities of DCs suggested an involvement of this protein in the DC-dependent regulation of Th1 and Th17 cell responses [19].

To accomplish the function of orchestrating the immune response against invading microbes, DCs express an array of PRRs, including TLRs [27, 28]. The human TLR family, so far, includes 11 members. TLR1, -2, -4, and -6 are expressed on the cell surface, where they mainly recognize bacterial compo- nents, such as lipopeptides and peptidoglycan (TLR1, -2, and -6), LPS (TLR4), or flagellin (TLR5). TLR3, -7, -8, and -9 are instead located into cellular vacuoles and respond to nucleic acids, such as dsRNA (TLR3), ssRNA (TLR7 and -8), and CpG DNA (TLR9) [29 –31].

Upon microbe recognition, TLRs recruit one or more adap- tor molecule(s), among MyD88, TIRAP (Mal), TRIF, and TRAM, to activate overlapping but distinct signal transduction pathways. MyD88 is used by all TLR except TLR3 and activates the transcription factor NF-nB and MAPKs to induce inflam- matory cytokines. TRIF is used by TLR3 and TLR4 and in- duces an alternative pathway that leads to the activation of the transcription factors IRF3 and NF-nB and to the consequent induction of type I IFN and inflammatory cytokines. TRAM and TIRAP recruit TRIF to TLR4 and MyD88 to TLR2 and TLR4, respectively. TLR4 is the only TLR that uses all four adaptors and activates the MyD88- and TRIF-dependent path- ways [32–35]. As pathogens can engage multiple TLRs through the complex expression of PAMPs, pathogen recogni- tion by DCs involves the integration of multiple TLR-derived signaling pathways. Thus, the simultaneous engagement of dif- ferent TLRs will result in complementary, synergistic, or antag- onistic effects that ultimately modulate the immune response [36 –38].

DCs produce conspicuous levels of OPN, but so far, no information is available on the ability of TLR agonists to regulate OPN production. This study was designed to inves- tigate the regulation of OPN production by human DCs fol- lowing the engagement of different TLRs and the OPN in- fluence on the proinflammatory response following patho- gen recognition by DCs.

MATERIALS AND METHODS

Mo-DC preparation and stimulation

Monocytes were isolated from PBMCs, obtained from healthy donor buffy coats (through the courtesy of the Centro Trasfusionale, Spedali Civili, Brescia, Italy) by immunomagnetic selection with CD14 microbeads (MACS monocyte isolation kit; Miltenyi Biotec, Bergisch Gladbach, Germany). This procedure yields a ≥98% pure monocyte population, as assessed by FACS analysis (FACSCalibur; BD Biosciences, Franklin Lakes, NJ, USA). To ob- tain Mo-DCs, monocytes were cultured for 6 days at 106 cells/ml in RPMI- 1640 medium (Gibco, Invitrogen, Life Technologies, Carlsbad, CA, USA) containing 10% heat-inactivated FCS (Lonza Group, Switzerland) in the presence of GM-CSF (50 ng/ml) and IL-4 (20 ng/ml; both from Pepro- Tech, Rocky Hill, NJ, USA).
DCs were stimulated with the following agonists: FSL-1, Pam3CSK4, Poly I:C, flagellin, R848, and Porphiromonas gingivalis LPS (all from InvivoGen,
San Diego, CA, USA) and LPS (E. coli 055:B5; Sigma-Aldrich, St. Louis, MO, USA). Heat-killed E. coli, Legionella pneumophila, L. monocytogenes, S. aureus, and Acholeplasma laidlawii were purchased from InvivoGen. Bartonella henselae Houston I strain (ATCC 49882; American Type Culture Collection, Manassas, VA, USA) was grown in Schneider’s medium containing 10% FCS and 5% sucrose for 6 days [36] and killed by heating to 56°C for 30 min. In some experiments, cells were treated with IL-1β (PeproTech).

B18R, a vaccinia virus-encoded, neutralizing type I IFNR (eBioscience, San Diego, CA, USA), and anti-IFNAR2 mAb (MMHAR-264G12; PBL Interferon Source, Piscataway, NJ, USA) were used to inhibit the type I IFNR signal- ing. IFN-β-1a was from R&D Systems (Minneapolis, MN, USA). Where indi- cated, cells were pretreated with the TBK-1 inhibitor BX795 (InvivoGen), the endosome-acidifying maturation inhibitor chloroquine (InvivoGen), and the dynamin inhibitor dynasore (Sigma-Aldrich) or vehicle (DMSO). For TLR blocking experiments, DCs were treated with 10 µg/ml anti-TLR2 (anti-human TLR2-IgA; InvivoGen), anti-TLR4 (HT52; eBioscience), or ap- propriate isotype control (InvivoGen and eBioscience) for 60 min at 37°C before addition of the appropriate stimuli.

In some experiments, cells were preincubated with anti-TGF-β, anti-IL- 10R, anti-IL-29, and mouse IgG1 control antibody, all purchased from R&D Systems.

Real-time PCR

Total RNA, isolated with the Qiagen RNeasy Mini Kit, was treated with DNase I (Qiagen, Hilden, Germany) and retrotranscribed into cDNA by using the iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA, USA). Gene-specific primers were: HPRT (sense, 5=-TGACCTTGATT- TATTTTGCATACC-3=; antisense, 5=-CGAGCAAGACGTTCAGTCCT-3=); OPN (sense, 5=-GCCGAGGTGATAGTGTGGTTTATG-3=; antisense, 5=- CGCTTTCCATGTGTGAGGTGATG-3=); TNF-α (sense, 5=-CAAGCCTG- TAGCCCATGTTGTAG-3=; antisense, 5=-CCTGGGAGTAGATGAGGTACAGG-3=). The iQ SYBR Green Supermix (Bio-Rad Laboratories) for quan- titative real-time
PCR was used according to manufacturer instructions.

Reactions were run in duplicate on an iCycler Chromo4 (Bio-Rad Labora- tories) and the generated products analyzed by Opticon Monitor 3.0 soft- ware (Bio-Rad Laboratories). Gene expression was normalized based on HPRT mRNA content.

Silencing of MyD88 and TRIF

Monocytes were cultured at 106 cells/ml in RPMI-1640 medium containing 10% heat-inactivated FCS in the presence of GM-CSF and IL-4, as described above. After 48 h, differentiating cells were transfected with a MyD88 Si- lencer Select Validated siRNA, with a TRIF Silencer Select Predesigned siRNA, or with a control siRNA (all at 50 nM final concentration; Ambion, Invitrogen, Life Technologies) using Opti-MEM I reduced serum medium and Lipofectamine RNAiMAX transfection reagent (Invitrogen, Life Tech- nologies), according to the manufacturer’s protocol. Transfected cells were incubated for 72 h and then stimulated for 24 h with different TLR ago- nists as indicated.

Western blot analysis

For Western blot analysis, transfected cells were washed twice with PBS and lysed in ice-cold lysis buffer containing 50 mM Tris-HCl (pH 8.0), 250 mM NaCl, 1 mM EDTA, 0.1% Nonidet P-40, 10% glycerol, 1 mM sodium or- thovanadate, 2 mM DTT, and protease inhibitor cocktail (all from Sigma- Aldrich). Total cell lysates (20 µg) were analyzed by 10% SDS-PAGE, fol- lowed by Western blotting with antibodies against MyD88, TRIF (Cell Sig- naling Technology, Beverly, MA, USA), or β-actin (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Protein bands were detected with SuperSignal West Pico chemiluminescent substrate (Pierce, Rockford, IL, USA).

Preparation of CM

To prepare Ctrl-CM or LPS-CM and Poly I:C-CM, Mo-DCs were left un- treated or incubated for 4 h with 100 ng/ml LPS or 10 µg/ml Poly I:C.After six washes, fresh medium was added to the cells. Following incuba- tion at 37°C for 18 h, the CM was separated from the cells by centrifuga- tion and applied to fresh cells to examine its effects on OPN release. Some aliquots were heated at 95°C for 20 min. Other aliquots were filtered by using Microcone-YM-10 and -YM-50 (Millipore, Billerica, MA, USA).

CD4+ T cell isolation and stimulation

CD4+ T cells were enriched (purity >95%) by immunomagnetic negative depletion using a CD4+ T cell isolation kit and MACS columns (Miltenyi Biotec), according to the manufacturer’s instructions. CD4+ T cells were cultured with plate-bound anti-CD3 (eBioscience) and anti-CD28 (BioLeg- end, San Diego, CA, USA) mAb (0.3 µg/ml) in the presence or absence of the indicated stimuli. Human rOPN, anti-CD44, and mouse IgG1 (control IgG) were purchased from R&D Systems, and anti-CD29 (integrin β1) was obtained from LifeSpan BioSciences, Seattle, WA, USA. The mAb MPIIIB10(1), developed by Michael Solursh and Ahnders Franzen, was ob- tained from the Developmental Studies Hybridoma Bank (Iowa City, IA, USA), developed under the auspices of National Institute of Child Health and Human Development and maintained by the University of Iowa, De- partment of Biology (Iowa City, IA, USA). Supernatants were harvested af- ter 60 h and assayed for IL-17 content by ELISA.

Cytokine evaluation by ELISA

Cytokine levels in culture supernatants were measured by ELISA kits pur- chased from R&D Systems for OPN, TNF-α, IL-6, and CXCL10 and from PBL Interferon Source for IFN-β. IL-17A ELISA MAX Deluxe Set was from BioLegend.

Statistical analysis

Statistical significance between the experimental groups was determined by using one-way ANOVA with Dunnet’s post hoc test (GraphPad Prism ver- sion 4.00 for Windows; GraphPad Software, La Jolla, CA, USA).

RESULTS

Induction of OPN by selected TLR agonists and Gram-positive bacteria

OPN production was evaluated in the supernatants of human DCs exposed to increasing concentrations of different TLR ligands. As shown in Fig. 1A, unstimulated DCs released con- siderable levels of OPN (331±173 ng/ml; n=20). Interestingly, TLR2 stimulation with FSL-1 (TLR2/6) or Pam3CSK4 (TLR1/2) induced the highest levels of OPN even at the low- est concentration used of 10 ng/ml, with maximum induction observed at 100 ng/ml (Fig. 1A; 3.7- and 3-fold increase over basal production for FSL-1 and Pam3CSK4, respectively).

Higher concentrations of the TLR5 (flagellin) and TLR7/8 (R848) agonists were required to increase OPN levels with peak induction attained at 1 µg/ml (Fig. 1A; 4.3-fold over basal levels for flagellin and 2.6-fold for R848). No OPN pro- duction increase was found when the TLR3 and TLR4 agonists Poly I:C and E. coli LPS were used (Fig. 1B). On the contrary, OPN production was increased by stimulation with P. gingivalis LPS, which is a TLR2 agonist [37]. These results indicate that in DCs, OPN is induced by a selected pattern of TLR agonists, among which the most active are those that stimulate TLR2.

Figure 1. TLR ligands differentially affect OPN production by Mo-DCs. (A and B) Mo-DCs were cultured in the presence of increasing concentrations of FSL-1, Pam3CSK4, flagellin, or R848 (A) and Poly I:C, E. coli LPS (LPS), or P. gingivalis LPS (LPS PG; B). After 24 h, superna- tants were collected, and the OPN con- tent was measured by ELISA. Data repre- sent means ± seM (n=5). *P < 0.05 ver- sus unstimulated Mo-DCs. (C) Mo-DCs were stimulated for 24 h with heat-killed E. coli, A. laidlawii, L. pneumophila, L. monocytogenes, S. aureus, or B. henselae at 1:10 Mo-DC/bacteria ratio, and the OPN content in the supernatants was assessed by ELISA. OPN in the supernatant of un- treated cells was set as 100%. Data repre- sent means ± seM (n=6). *P < 0.05 ver- sus unstimulated Mo-DCs. (D) Mo-DCs were stimulated for 24 h with heat-killed E. coli or S. aureus (Mo-DC/bacteria 1:10) in the presence of blocking TLR4 or TLR2 mAb (10 µg/ml), and the OPN content was measured in the supernatants by ELISA. OPN in the supernatant of un- treated cells was set as 100%. Data repre- sent means ± seM (n=3). *P < 0.05 ver- sus S. aureus-stimulated Mo-DCs in the presence of the isotype antibody.

TLR2 is mostly responsible for the detection of Gram-posi- tive bacteria, whereas TLR4 is mainly involved in the recogni- tion of Gram-negative bacteria [38]. However, certain Gram- negative bacteria, characterized by a LPS with weak intrinsic activity, activate cells via TLR2 instead of TLR4. To investigate the ability of different bacteria to stimulate OPN production, DCs were exposed to whole heat-killed, TLR4-stimulating, Gram-negative bacteria (E. coli) [38]; TLR2-stimulating, Gram- negative bacteria (L. pneumophila, B. henselae) [39, 40]; and TLR2-stimulating, Gram-positive bacteria (L. monocytogenes, S. aureus) [41] and with the TLR2-stimulating, wall-less bacterium A. laidlawii [42]. As with the synthetic TLR2 agonists, whole bacteria that mainly signal through TLR2 caused a significant increase in OPN release. In contrast, heat-killed E. coli, used at similar concentrations, did not induce OPN production (Fig. 1C). Instead, we observed a reduction of OPN levels (Fig. 1C; 100% OPN in the supernatant of untreated cells, 46% ± 13% in the presence of E. coli), although not statistically significant. To further confirm the contribution made by the different TLRs to the DC response to whole bacteria, we exposed the DCs to S. aureus and E. coli following 1 h preincubation with blocking anti-TLR2 and anti-TLR4 mAb, respectively. As ex- pected, OPN production by DCs treated with S. aureus was in- hibited by anti-TLR2, whereas a slightly higher OPN produc- tion was observed in DCs stimulated with E. coli following TLR4 blockade (Fig. 1D).

TLR4 and TLR3 agonists inhibit TLR2-induced OPN production

It has been reported that the simultaneous stimulation of DCs with multiple TLR agonists results in an integrated effect on the production of soluble factors compared with the engage- ment of single TLRs [43, 44]. We observed that the simultane- ous stimulation of TLR4 with LPS and of TLR2 with FSL-1 or Pam3CSK4 resulted in a dramatic inhibition of OPN produc- tion (Fig. 2A, left). A similar inhibitory effect was observed when Poly I:C (TLR3) was used instead of LPS, whereas the treatment of DCs with the combination of the two TLR ago- nists, FSL-1 and Pam3CSK4, had no additive effect. The simul- taneous stimulation of DCs with TLR2 agonists and LPS also led to a reduction of OPN mRNA levels, as evaluated by real- time PCR (Fig. 2B). The negative cross-talk between TLR3/4 and TLR2 was apparently specific for OPN, as no negative in- teraction was observed when the effect of the combination of the same agonists was evaluated on the induction of TNF-α mRNA expression or production (Fig. 2A and B, right). We then investigated whether the timing and order of agonist ad- dition would affect the extent of inhibition of OPN produc- tion. The ability of LPS to down-regulate OPN was more pro- nounced when TLR4 stimulation preceded TLR2 stimulation of 4 h. In addition, LPS stimulation that followed FSL-1 stimu- lation was still able to reduce OPN induction (Supplemental Fig. 1). Finally, Fig. 2C shows that the inhibition of OPN pro- duction by TLR4 and TLR3 agonists was concentration-depen- dent, already detectable at 1 ng/ml LPS and 1 µg/ml Poly I:C and reaching maximal values at 100 ng/ml and 10 µg/ml, re- spectively. These results indicate that TLR3 and TLR4 inter- fere with the TLR2-dependent production of OPN mRNA and protein.

Negative regulation of OPN production by the TRIF-dependent pathway

TLR signaling involves a MyD88-dependent pathway or a MyD88-independent, TRIF-dependent pathway. All the agonists reported previously as OPN inducers activate TLRs that selec- tively signal through the MyD88 pathway, which is also shared by certain cytokine receptors, such as IL-1R [45]. Accordingly, in Fig. 3A, we show that the DC stimulation with 50 ng/ml IL-1β induced the production of OPN, and as expected, this effect was abrogated completely in the presence of LPS or Poly I:C. The two negative regulators, LPS and Poly I:C, engage two receptors that activate MyD88 and TRIF (TLR4) or that exclu- sively use the TRIF-dependent pathway (TLR3). TLR4 triggers MyD88 signaling when still located at the plasma membrane and activates the TRIF pathway after moving to the early endo- somes [46]. TLR3 activates TRIF in the endosomes where it is located. The two TRIF activation pathways can be discrimi- nated by the use of chemical inhibitors. Chloroquine inhibits endosome maturation and/or acidification, a process crucial for TLR3 signaling [47], whereas TRIF signaling by TLR4 is blocked by dynasore, a dynamine GTPase inhibitor [46]. Fig- ure 3B shows that DC pretreatment with 10 µM chloroquine prevented the inhibitory effect of Poly I:C on TLR2-induced OPN production. As expected, the action of LPS was not influ- enced by chloroquine but was inhibited significantly by the pretreatment with 40 µM dynasore. In this set of experiments DCs were stimulated with 10 ng/ml LPS, as dynasore does not inhibit the expression of TRIF-dependent genes when the stimulation is performed with higher LPS concentrations [46, 48]. Cell viability was not affected by chloroquine and dyna- sore, as monitored by trypan blue staining (data not shown). Finally, to further confirm the role of TLR pathways, we deter- mined OPN production under MyD88 or TRIF knockdown conditions obtained by siRNA treatment. MyD88 knockdown greatly reduced S. aureus-induced OPN production, whereas the inhibitory effect of LPS on TLR2-induced OPN was par- tially prevented by TRIF knockdown (Fig. 3C). Collectively,
our data indicate that OPN is up-regulated in a TLR/IL-1R- MyD88-dependent manner and suggest that LPS and Poly I:C may exert their negative regulation through the activation of a TRIF-dependent pathway.

The TLR3/4-negative regulation of OPN production is independent of IFN-β autocrine induction

The engagement of the TRIF-dependent pathway results in the activation of the downstream kinases TBK-1 and IKKϵ and eventually, in the IRF3 phosphorylation and type I IFN pro- duction. Type I IFNs, such as IFN-β, regulate gene expression after TLR signaling in an autocrine–paracrine manner [49]. As shown in Fig. 4A, the addition of recombinant IFN-β, strongly inhibited OPN production induced by FSL-1 and Pam3CSK4 stimulation. The role of the endogenously produced type I IFN on OPN production was investigated by the use of a blocking anti-type I IFNR mAb. As shown in Fig. 4B, the anti- IFNR did not affect the OPN release as a result of the con- comitant stimulation of TLR3/4 and TLR2. Conversely, the release of CXCL10, an IFN-dependent chemokine, was strongly inhibited. To substantiate this observation further, DCs were stimulated in the presence of the soluble vaccinia virus-encoded protein B18R, an inhibitor of type I IFN antivi- ral activity [50]. B18R, used at a concentration (0.2 µg/ml) able to neutralize the effect of exogenous IFN-β, failed to re- verse the inhibitory effects of LPS and Poly I:C on OPN pro- duction (Fig. 4C). Finally, the effect of BX795, a TBK-1 inhibi- tor [51], on TLR2-induced OPN production was investigated. In these experiments, DCs were preincubated with 0.5 µM BX795 and then stimulated with Pam3CSK4 in the presence of LPS or Poly I:C. BX795 blocked the secretion of IFN-β in re- sponse to TLR3 and TLR4 agonists but did not affect OPN inhibition mediated by these agonists (Fig. 4D). These results indicate that the inhibitory action of TLR3 and TLR4 on TLR2-dependent OPN secretion is independent of IFN-β pro- duction. Moreover, the addition of antibodies neutralizing TGF-β, IL-10R, or the IL-10-related cytokine IL-29 (IFNh1), did not reverse the inhibition of OPN mediated by LPS, thus excluding a role for these cytokines in OPN-negative regula- tion (Supplemental Fig. 2).

A released soluble factor mediates the inhibitory effects of TLR4 and TLR3 agonists on OPN production

To evaluate the possibility that TLR3/4 agonists induce the release of a soluble inhibitor, DCs were activated for 4 h with 100 ng/ml LPS or 10 µg/ml Poly I:C and then washed exhaustively and incubated further with fresh medium for 24 h. CM collected from these cultures (LPS-CM and Poly I:C-CM) were added to fresh DC cultures just before stimu- lation with TLR2 agonists. Figure 5A shows that LPS-CM and Poly I:C-CM inhibited FSL-1-induced OPN production in a dose-dependent manner. The possibility that the inhibi- tory activity was a result of a residual amount of LPS in the CM was ruled out by the use of a blocking anti-TLR4 mAb. In fact, the addition of anti-TLR4 mAb inhibited the effect of exogenously added LPS on FSL-1-induced OPN produc- tion but not that of LPS-CM (Fig. 5B). Moreover LPS-CM significantly inhibited FSL-1-induced OPN release, whereas it did not alter the production of other FSL-1-stimulated cytokines, such as TNF-α or IL-6, suggesting that the effect of the inhibitor is OPN-selective (Fig. 5C). To gain insight into the nature of the putative, soluble OPN inhibitor, ali- quots of LPS-CM and Poly I:C-CM were heated at 95°C for 20 min before testing. As shown in Fig. 5D, this treatment abrogated the CM inhibitory activity almost completely, sug- gesting the proteinaceous nature of the factor(s). Molecular weight fractionation of both CM indicated that the inhibi- tory activity was present in the fraction comprised of be- tween 10 and 50 kDa (Fig. 5D, and data not shown). These results provide evidence that DCs stimulated with LPS or Poly I:C release a protein factor(s) that inhibits OPN re- lease.

OPN present in the supernatants of S. aureus-stimulated DCs triggers IL-17 production by CD4+ T cells

APCs, in response to microbial stimuli, release cytokines, which in turn, act on T cells to induce the production of IL-17. As OPN regulates IL-17 production in autoimmune and inflammatory conditions [21, 22, 52], we hypothesized a link between the different levels of OPN released by Mo-DC stimu- lated with S. aureus or E. coli and IL-17 production by T cells. To test this hypothesis, we stimulated CD4+ T cells with anti- CD3 and anti-CD28 in the presence of 50% vol/vol superna- tants from Mo-DCs stimulated with S. aureus or E. coli (amount of OPN present in the supernatants: following stimulation with S. aureus, 500 –700 ng/ml; with E. coli, 70 –90 ng/ml). Human rOPN was used as control. As shown in Fig. 6, CD4+ T cells incubated with supernatants from Mo-DCs stimulated with S. aureus produced a significantly larger amount of IL-17 com- pared with T cells incubated with supernatants from Mo-DCs stimulated with E. coli. Interestingly, no such effect was ob- served with the addition of an anti-OPN-neutralizing antibody (Fig. 6A). Furthermore, the blockade of the OPNRs CD29 and CD44 also significantly reduced the levels of IL-17 released in the presence of the S. aureus supernatant (Fig. 6B). These re- sults indicate that high levels of OPN, released by Mo-DCs through TLR2 signaling, amplify IL-17 production by T cells.

DISCUSSION

OPN is expressed in chronic inflammatory diseases, which are often caused by Gram-positive bacteria, such as staphylococci, streptococci, mycobacteria, or by Gram-negative bacteria con- taining LPS variants that do not stimulate TLR4 (Bartonella spp., Chlamydia spp., Helicobacter) [53, 54]. DCs were previously reported to produce conspicuous levels of OPN and to be present in mycobacterial and B. henselae granulomas, as well as in Helicobacter pylori-infected gastric epithelium [41, 55, 56].

Thus, it is reasonable to infer that DCs represent a source of OPN contributing to the inflammatory response. Here, we show that heat-killed, Gram-positive bacteria, mainly interact- ing with TLR2, boost OPN release by DC, whereas Gram-negative enterobacteria and other TLR3/TLR4 agonists do not. In line with these findings, the up-regulation of OPN production was also observed in the presence of nonenteric, Gram-negative bacteria, such as Bartonella spp. and Legionella, which con- tain LPS variants unable to activate TLR4 but mainly recog- nized by TLR2.

The host-pathogen interaction involves the simultaneous recognition of several microbial products, each one with spe- cific agonist activities, and it is now clear that the combined activation of different receptors may result in complementary, synergistic, or antagonistic effects modulating innate and adap- tive immunity. Numerous reports emphasize the synergy be- tween different TLR signaling to enhance the expression of costimulatory molecules and the production of inflammatory cytokines [43, 44, 57, 58]. Nevertheless, growing evidence also demonstrates that simultaneous triggering of selected TLR may result in antagonism and cross-tolerance [59 – 63]. Selec- tive down-modulation of specific cytokines and chemokines by TLR cross-talk may be a mechanism evolved to regulate the DC response to pathogens and to prevent potential detrimen- tal effects. Moreover, in certain situations, the activation of particular TLR responses might represent an escape mecha- nism implemented by some microorganisms.

In the present work, we show that in the presence of simul- taneous triggering of TLR2 and TLR3/4, the induction of OPN is abrogated completely. The observation that the same effect is obtained when TLR2 is stimulated in the presence of CM from TLR3/4-stimulated DCs points to the existence of a soluble mediator(s) that may account for this regulation. Pre- liminary characterization demonstrated that the molecule is heat-labile and has a molecular mass >10 kDa. These results are in line with many reports that envisage autocrine cyto- kines, such as IFN-β, IL-10, and TGF [60, 61] as regulators of the TLR cross-talk. In our experimental setting, IFN-β ap- peared a particularly interesting candidate, as it is induced by TLR3 and TLR4 and is known to suppress OPN in hu- man CD4 T cells [64]. Consistent with this report, we show that exogenous IFN-β inhibits TLR2-mediated OPN produc- tion in a dose-dependent manner. However, when DCs were treated with combinations of TLR2 and TLR3/TLR4 ago- nists, neither the neutralization of endogenous type I IFNs nor the blockade of IFNAR2 could rescue OPN secretion, demonstrating that type I IFN is not the response element accounting for TLR interference in our system. In addition, we also excluded a role for TGF-β, IL-10, the IL-10-related cytokine IL-29 (IFNh1), and factors preferentially induced by TLR3 and/or TLR4, such as IFN-inducible protein 10, (IP-10), a proliferation-inducing ligand (APRIL), and B cell activating factor (BAFF) (data not shown). Further efforts are being dedicated to the purification and identification of this putative inhibitory molecule(s).

Figure 6. The OPN present in the supernatants of S. aureus-stimulated DCs contributes to IL-17A production by CD4+ T cells. CD4+ T cells, activated with anti-CD3 and anti-CD28 antibodies, were treated with 50% vol/vol supernatants (sup) collected from Mo-DCs, which were stimulated previously with S. aureus or E. coli for 18 h. Antibody against OPN (20 µg/ml; A) and OPN receptor-blocking antibodies anti-CD29 and anti-CD44 (10 µg/ml each; B) or specific isotype control (control IgG; 20 µg/ml) antibody was added in the culture. T cell supernatants were collected after 60 h, and IL-17 was determined by ELISA (means ± seM; n=3 in duplicate). *P < 0.05 versus isotype control-pretreated cells.

To gain insight into the mechanism leading to the release of the inhibitory factor(s), we investigated the importance of some steps in the TLR3/4 pathways. Our results indicate that OPN is preferentially induced by those TLR that signal exclu- sively through the MyD88 pathway. In keeping with this, OPN is up-regulated by IL-1β, which also signals via MyD88. Impor- tantly, MyD88 silencing led to a significant decrease in OPN up-regulation in Mo-DCs following S. aureus stimulation, imply- ing that MyD88 is a crucial component of this response.

On the contrary, TLR4 or TLR3 stimulation sharply limits the MyD88-dependent OPN production. Among TLR, TLR4 is unique for its ability to make use of two adaptor mole- cules, i.e., MyD88 and TRIF, whereas TLR3 uses TRIF exclu- sively. As dynasore and chloroquine, respectively, prevented the inhibitory effect of LPS and Poly I:C on TLR2-induced OPN, we propose that the signaling through the TRIF path- way blocks the MyD88-dependent OPN up-regulation. In line with these evidences, the inhibitory effect of LPS on TLR2-induced OPN was partially prevented by TRIF knock- down.
However, this is independent from TBK activation, as the TBK inhibitor BX795 did not rescue OPN production. These results suggest that the TLR3/4- and TRIF-dependent pathways leading to IFN-β induction and to the inhibition of MyD88- mediated OPN production bifurcate upstream of TBK-1, fur- ther excluding a role of an IFN-mediated, negative feedback loop. In general, there are very little data in the literature de- scribing the suppressive role of TRIF on TLR/PRR signaling. Recently, Seregin et al. [63] showed that in vivo TRIF acts as a negative regulator of the TLR/MyD88 signal in multiple cell types (macrophages; NK, B, and T cells; and DCs), but the molecular mechanism remains unexplored, as well as the key question of whether TRIF-negative regulation is dependent on ligand binding. Another group showed that TRIF can suppress TLR5-mediated proinflammatory responses in intestinal epi- thelial cells by inducing the degradation of this receptor as well as that of TLR3, -6, -7, -8, -9, and -10 [5]. However, this is unlikely in our experimental system, as TLR2 and TLR1 cir- cumvent the TRIF-mediated proteolytic degradation. Further- more, it is important to stress here that the observed inhibi- tory effect is gene-specific, if not restricted to OPN, as other proinflammatory mediators, such as TNF-α, are induced in our experimental setting. In addition to TRAF3/TBK, TRIF associ- ates with TRAF6 and receptor-interacting protein 1 (RIP-1) and with the help of TNFR1-associated death domain (TRADD) and TAK1, activates late-phase NF-nB and MAPKs [65]. This pathway, together with other yet-unidentified path- way(s), may thus be responsible for the induction of some in- hibitory mediator(s), which in our view, must be OPN-specific. In addition to the release of soluble mediators, the engage- ment of selected TLR pairs may result in synergy or inhibition, also because of intracellular interference among signaling pathways or induction of molecular effectors. In keeping with this hypothesis, we are currently working at the identification and validation of OPN-binding microRNA that are induced upon simultaneous triggering of TLR2 and TLR3/4.

OPN is mainly studied as a secreted protein, but an iOPN was also identified [19, 66]. The characterization of iOPN is fairly advanced in the mouse system. Recent evidences indicate that iOPN is induced following TLR4 stimulation and nega- tively regulates TLR4-induced IFN-β production in murine macrophages [67]. Moreover, iOPN is involved in TLR9-de- pendent expression of IFN-α in plasmacytoid DCs [19] and in the downstream signaling of TLR2/dectin-1 pathways in the antifungal response [68]. Because of the role of iOPN in the regulation of TLR signaling pathways, it would be of great in- terest to evaluate the iOPN involvement in our experimental system. Unfortunately, the current understanding of iOPN ex- pression and functions in human cells is still limited. Based on histology data from human tissues, cytoplasmic OPN is present in human cells, but it is not clear whether the alternative translation initiation site is the same in human and mice iOPN, and it is difficult to discriminate between iOPN and se- creted OPN in human cells [66]. For this reason, we could not explore a possible involvement of iOPN in our experimental system.

In this study, we also provide evidence that the observed OPN modulation by different TLR agonists is biologically relevant and may play a critical role in the polarization of the immune response. In fact, we showed that the higher levels of OPN produced by Mo-DCs following stimulation with S. aureus, compared with E. coli, lead to a more robust response of CD4+ T cells in terms of IL-17 production. IL- 17, a cytokine critical for the recruitment of phagocytes op- erating bacterial clearance, is mainly produced by Th17 cells, which have been shown recently to mediate many in- flammatory and autoimmune diseases [69]. Moreover, OPN plays a critical role in the differentiation of Th17 cells in rheumatoid synovium [20]. A high expression of OPN regu- lating IL-17 production in the pathogenesis of hepatitis and acute coronary syndrome was reported [21, 22]. Nevertheless, the exact role of OPN in pathogen-induced human Th17 cells remains poorly defined. Our data indicate that a selective regulation of OPN by pathogens may influence a Th17 response. In line with our observation that TLR2 sig- naling can be an important mediator of Th17 cell re- sponses, Kim et al. [70] reported that S. aureus can induce Th1 and Th17 inflammation, mainly in a TLR2-dependent manner, and TLR2 signaling has been shown to be an im- portant molecular mediator of effective Th17 cell responses during Mycobacterium tuberculosis infections [71]. In addition, recent data show that OPN represents an important regula- tory factor of the protective Th17 immunity in Trypanosoma cruzi infection [72], in which TLR2 has a predominant, im- munoregulatory role [73].

Figure 7. Model for the contribution of the MyD88 and TRIF signaling pathways in the TLR-dependent OPN regulation in Mo-DCs. SF, Solu- ble factor.

All in all, this study provides novel insights into the regula- tion of OPN production in Mo-DCs (represented in Fig. 7) and indicates that the differential production of OPN in re- sponse to TLR agonists may be relevant in shaping the patho- gen-induced inflammatory response through the regulation of IL-17 secretion by CD4+ T cells.