Suzanne Van Helden

General discussion and conclusions. MUTZ-3 cells are poorly adhesive and therefore less suited as a model for DC adhesion. For these studies, THP-1 cells or HL-60 cells provide a good alternative. Although there is not one cell line that can fulfill all aspects of DC-biology, the use of combination of cell lines, each for a certain application, will contribute to our understanding of DC-biology.

DC maturation and regulation of adhesion and migration
In the tissues iDCs interact with the surrounding extracellular matrix (ECM) and it turns out that iDCs are able to bind to several components of the ECM. DCs bind the ECM component fibronectin through the α5β1-integrin [5]. During maturation strongly adherent iDCs change into highly migratory mDCs (chapter 3). This is reflected both by changes in the cytoskeleton of these cells as well as the activity of α5β1-integrin. While iDCs show prominent cell-substrate adhesions, in mDCs those structures are absent. Instead mDCs display characteristic dendritic extensions that give these cells their name. Although in iDCs focal adhesions (FAs) can be found, the main adhesion structures in these cells are podosomes. In chapter 3 we describe that the active form of the α5β1-integrin is enriched in podosomes and that the activity of this integrin is downregulated during maturation. Moreover, keeping the integrin in the active form prevents the loss of podosomes and the induction of migration, showing that a tight regulation of the activity of integrins is essential for the proper DC function.

We observed that upon addition of a maturation-inducing combination of monocyte-conditioned medium (MCM), TNFα and prostaglandin E2 (PGE2) DCs show a rapid loss of podosomes (chapter 3). PGE2 can substitute for this combination and is sufficient to induce podosome loss and migration. By using live-cell imaging in combination with interference reflection microscopy (IRM), we have investigated the dynamics of podosome loss in DCs (chapter 5). IRM allows one to image adhesion structures without a need for fluorescently-tagged molecules [6-8]. Structures close to the glass substrate appear dark with IRM, hence podosomes appear as dark spots and FAs as dark stripes. Prior to dissolution of a podosome, a bright ring surrounding the podosome-core was observed, suggesting a local change in the proximity of the cell to its substrate. In addition to a loss of podosomes, PGE2 stimulation led to contraction of the DCs accompanied by an increase in the number of FAs (chapter 5). In other cell types, the reversed process, a loss of FAs accompanied by podosome formation, was shown [9, 10], suggesting that FAs and podosomes are oppositely regulated. After prolonged stimulation the number of DCs with FAs decreased again, with mDCs being completely devoid of FAs. Loss of FAs is associated with increased Rac1 activity [9] and may involve the microtubule network [11]. However, much remains unknown about the function and regulation of FA loss, especially during DC maturation. The dendritic extensions displayed by mDCs can be induced by CCL19, a chemokine that regulates migration to the lymph nodes [12]. Although the regulation of this process is largely unknown, the actin-bundling protein fascin and the Rho GTPases Rac and Cdc42 appear to be involved [13, 14]. However, detailed information on the mechanisms responsible for loss of FAs and subsequent induction of dendritic extensions during DC maturation is still lacking.

Reactions to bacteria and TLR4 signaling in DCs
The bacterial cell wall component lipopolysaccharide (LPS), a potent DC maturation stimulus, also induced podosome dissolution in DCs (chapter 3). However, this effect occurred much more slowly than the PGE2-induced podosome loss. LPS is specifically found on Gram-negative bacteria and binds to Toll-like receptor 4 (TLR4) [15]. Components of Gram-positive bacteria activate TLR2, which dimerizes either with TLR1 or TLR6 [16]. In chapter 4 we have compared the effects of Gram-negative and Gram-positive bacteria on DC-maturation. We observed that expression of costimulatory molecules as well as secretion of inflammatory cytokines is higher in response to Gram-negative bacteria than to Gram-positive bacteria. Moreover, Gram-negative bacteria were found to be more effective inducers of migration and expression of CCR7, the chemokine receptor required for CCL19/21-induced migration to the lymph nodes. Furthermore, only Gram-negative bacteria are able to induce podosome dissolution. These findings show that Gram-negative bacteria are superior to Gram-positive bacteria in inducing DC-maturation. An explanation for these difference could be that evolutionarily the number of Gram-negative bacteria infections led to the development of an APC dedicated to respond to this class of pathogens; the DC. Other APCs, such as monocytes and macrophages, may have evolved to coordinate immune responses against Gram-positive as well as Gram-negative bacteria. Indeed, macrophages effectively secrete cytokines, such as IL-6 and TNFα, in response to both classes of bacteria. At least our findings demonstrate that DCs discriminate between Gram-positive and Gram-negative bacteria. It will be interesting to see whether DCs also differentially respond to distinct classes of other pathogens, such as viruses.

Stimulation with purified ligands for TLR2 did not induce podosome loss. In addition, the amount of cytokines produced in response to TLR2 ligands was significantly lower than in response to LPS. This suggests that activation of TLR4 induces better DC activation than TLR2. Responses induced by LPS or Gram-negative bacteria could be efficiently blocked by TLR4-receptor antagonists, suggesting that these responses are fully dependent on TLR4 (chapter 4). By using mouse bone marrow derived DCs (BMDCs), we showed that these responses are conserved between mouse and men. Responses induced either by LPS or Gram-negative bacteria were absent in TLR4-/- BMDCs, which further confirms the TLR4 dependency of these responses (chapter 4). TLR4 can activate signaling pathways that act either downstream of the myeloid differentiation primary response gene 88 (MyD88) or the TIR domain-containing adaptor-inducing IFNβ (TRIF) adaptor [17, 18]. We used MyD88- [19] and TRIF-deficient [20] BMDCs to investigate the role of these signaling pathways (chapter 4). The effects of LPS or Gram-negative bacteria are mainly TRIF dependent, which is in line with the inability of TLR2 activation to induce potent DC maturation. However, MyD88 appears to also be involved, suggesting that activation of both signaling pathways is needed to induce full DC maturation.

The importance of PGE2
The combination of MCM, TNFα and PGE2, to induce DC maturation, is currently used in DC vaccination trials in melanoma patients. The presence of PGE2 was shown to be essential for the induction of migration and consequently efficient anti-tumor immune responses [5, 21-24]. This effect is partly mediated by the effects on expression and activation of CCR7, the chemokine receptor needed for migration to the lymph nodes [22, 23, 25]. Interestingly, we found that PGE2 alone is sufficient to induce rapid podosome dissolution and migration in DCs (chapter 3). Therefore, we tested whether the effects of LPS or Gram-negative bacteria on DCs were dependent on the production of prostaglandins. LPS-induced podosome dissolution and the induction of migration could be inhibited by indomethacin, a pharmacological inhibitor that blocks the production of prostaglandins by inhibition of cyclooxygenase (COX) enzymes (chapter 3). Similarly, also the Gram-negative bacteria induced effect could be inhibited by using indomethacin (chapter 4). These findings reinforce the notion that the production of prostaglandins by DCs is important for the induction of an efficient immune response. At the moment it is unclear which prostaglandins are involved in these processes. PGE2 is very potent in inducing podosome loss and migration, both in vitro and in vivo, and is produced by DCs in response to LPS. However, the involvement of other prostaglandins cannot be ruled out at this point.

Given the importance of prostaglandins, in particular PGE2, in the regulation of adhesion (i.e. podosome dissolution) and migration during DC-maturation, we investigated the signaling pathways involved in PGE2-mediated podosome dissolution (chapter 5). From the four receptors known for PGE2, EP1-4 [26], we detected EP2 and EP4 on DCs. PGE2-induced responses were mediated through both of these receptors. Activation of the EP2 and EP4 receptor induces signaling through Gs to elevated cAMP levels [26] and these PGE2-induced responses can be mimicked by elevating cAMP levels. Cell shape and the organization of the cytoskeleton are influenced by numerous processes regulated by the family of small Rho GTPases [27-29]. Therefore, the role of RhoA, Rac1 and Cdc42 in PGE2-induced podosome loss was investigated (chapter 5). The active (GTP bound) form of RhoA increases when cells were stimulated with PGE2, EP agonists or cAMP raising drugs. At the same time, levels of active Rac1 and Cdc42 decreased. This suggests that PGE2 induces podosome loss by activation of RhoA. In most cell types cAMP-mediated activation of PKA antagonizes the effects of RhoA on actomyosin contractility, leading to cytoskeletal relaxation [30-32]. In DCs however, PGE2 stimulation appears to induce a contractile response while activating RhoA. At the moment, it is not clear how the elevation of cAMP levels in DCs leads to activation of RhoA. Probably, PKA activation is involved, however activation of the EPAC-Rap1 pathway may also play a role [33]. We showed that the function of Rho kinase, a downstream effector of RhoA, is needed for PGE2, EP agonist or cAMP elevation-induced podosome loss. Together our findings show that PGE2 induced podosome loss occurs by activation of the RhoA-Rho kinase axis.

Podosome regulation: a balance of forces
Activation of the RhoA-Rho kinase axis is known to promote the formation of FAs and induce cell contraction [27, 34, 35]. In addition, Rho kinase activity enhances stress fiber formation by inactivation of the actin severing protein cofilin [35], whereas direct activation of mDia by RhoA stimulates actin polymerization [27, 36]. Contraction of the cytoskeleton is generated through the function of the motor protein myosin II [37] and Rho kinase promotes phosphorylation of the myosin II regulatory light chain [38, 39]. Of the three myosin II isoforms (A, B and C) [40-42], myosin IIA is the predominant isoform expressed in DCs. The involvement of the RhoA-Rho kinase axis and the contraction observed in DCs upon PGE2 stimulation, suggest a role for myosin II-mediated contraction in these responses. Indeed, by blocking myosin II with blebbistatin, a selective inhibitor of myosin IIA ATPase activity, we were able to show the importance of myosin II-mediated contraction in PGE2-induced podosome loss (chapter 5). In addition, we found that the localization of myosin II changes dramatically upon PGE2 stimulation from a diffusely cytosolic to a more fibrillar distribution. These fibrillar structures colocalize with the bright rings observed by IRM, most likely reflecting an increased association with F-actin and/or a change in solubility of myosin filaments. In some cells, myosin II function was needed for podosome formation and maintenance [43, 44], whereas in other cells inhibition of myosin II function was important for podosome formation [9, 10]. We propose that basal myosin IIA activity is required for the formation and maintenance of podosomes, whereas a sudden (stimulus induced) increase in myosin II activity may trigger podosome dissolution.

The regulation of basic cellular functions, such as differentiation, apoptosis, adhesion and migration, is dependent on the balance between forces generated by myosin II-mediated contraction inside the cell and forces generated by the ECM outside the cells [45, 46]. Integrins organized into adhesion structures form the link between the surrounding of the cell and the cytoskeleton. Physical properties of the substrate can influence the generation of forces outside the cell. Therefore, we investigated the effects of some of these physical aspects on DC-adhesion and podosome formation (chapter 6). The ability of DCs to adhere and form podosomes and FAs appears to be independent of the type or hydrophobicity of the substrate, although DCs more frequently form FAs on hydrophobic substrates. It will be interesting to determine, by using blocking antibodies, whether integrins are important for the adhesion and formation of podosomes and FAs on substrates other than ECM components. It has been suggested that the dynamic behavior of podosomes, rather than their size is affected by the substrate [44]. Therefore it will be important to determine whether the life-time of podosomes varies on those different substrates or whether the kinetics of podosome dissolution in response to PGE2 is affected by substrate composition. To investigate effects of spacing and size of substrate on DC adhesion characteristics, we searched for a substrate that is non-adhesive for DCs but that can be treated to locally allow DC-adhesion. We found that hydrogels form such a non-permissive substrate for DCs. Moreover, we found that (ECM) proteins or even large particles can be spotted onto these hydrogels leading to local DC-substrate interactions and podosome formation. Hydrogels can be used in combination with microcontact printing, which allows better controlled size and spacing. We conclude therefore that hydrogels, combined with microcontact printing, provide good tools to explore the role of substrate size and spacing on DC adhesion (chapter 6).

In addition, we investigated the effects of topography on DC-adhesion and podosome formation by using substrates with ridges or other forms of height structures. Surprisingly, DCs aligned podosomes on the topological boundaries of these substrates (chapter 6). This suggests that the angle of the substrate induces a curvature of the cell membrane resulting in mechanical stress, giving rise to local podosome formation. By using pharmacological inhibitors, varying the angle of the ridges or by using a stretchable substrate, regulation of podosome formation by mechanical stretching could be further investigated. The profound effects of substrate topology on podosome formation show that podosomes are regulated by and can respond to mechanical stress, which suggests that podosomes, similar to focal adhesions [47], are mechanosensing structures. In line with this, we find zyxin, which has been suggested to be a mechanosensor [48, 49], in the podosome rings. The mechanical effects on formation and regulation of podosomes could have implications for DC-behavior in different tissues and be important when DC cells encounter tissue boundaries.

Together, our findings suggest that podosomes, in addition to FAs, are regulated by the contractile state of the cell as well as the forces applied to the cell by the extracellular matrix. Interestingly, podosomes appear to be oppositely regulated by force when compared to FAs. While contraction induces FA formation and relaxation leads to a loss of these adhesions [9, 10, 34], (myosin II-based) cytoskeletal contraction leads to podosome loss while cytoskeletal relaxation promotes the formation of podosomes. Hence, such a balance of forces may be key to the regulation of cell adhesion and migration in DCs and consequently for the induction of adaptive immune responses.

The function of podosomes in DCs
Although there is a time gap between the loss of podosomes (minutes) and the induction of migration (16 hours), we see a strong correlation between the loss of podosomes and migration (chapter 3 and 4). This suggests that podosomes prevent the induction of this high-speed, T cell-like migration needed for proper functioning of mDCs.

Dissolution of podosomes is a tightly regulated process. By fixing DCs at the moment of bright ring formation and subsequent staining for various podosome components, we found that vinculin could no longer be detected at the time that the actin core was still largely intact (Chapter 5). At this time, zyxin was still present in the podosome-ring, while myosin II redistributes to fibrillar-like structures. We conclude that loss of podosomes involves the sequential loss of podosome components. Podosome formation probably involves a similar sequential acquisition of podosome components and detailed information on the acquisition or loss of components can shed more light on the function of podosomes.

Although the role of podosomes in iDCs is still somewhat elusive, some suggestions can be made. Similar to FAs, podosomes link integrins to the actin cytoskeleton. Podosomes have a very distinct morphology with a core that contains actin and actin binding proteins surrounded by a ring that contains adaptor molecules and integrins (Figure 1). In addition, they are highly dynamic adhesion structures with a life-time of 2 to 12 minutes that often localize to the leading edge of the iDCs. Therefore, it is likely that podosomes are important for adhesion, the formation of polarity and directional crawling, resembling slow, fibroblast-like migration. This could facilitate the scavenging for antigens by iDCs in the tissues. In addition, local protein degradation takes place underneath podosomes as the result of focalized release of matrix metalloproteases [50, 51]. Although the adhesion-mediating integrins are found in the podosome ring, the dark spots observed by IRM correspond in size and localization to the actin-rich cores of podosomes, suggesting that the actin-rich core is closer to the substrate. Actin polymerization in the core could drive protrusion towards or into the substrate [52-54]. Moreover podosomes have been suggested to palpate the substrate [54] and they have been implicated in leukocyte transmigration [6]. Together, these observations suggest that podosomes allow DCs to either enter or leave the peripheral tissues and help the DCs to move from the tissue into a lymph vessel.

Figure 1. Model of the morphology of podosomes
In the upper left an IRM image of a DC with podosomes (dark spots) is depicted. In the upper right the same DC is stained for vinculin (green) and actin (red). An enlargement of a podosome is depicted in the lower right corner of this image. In the lower part of the figure a schematic model of a podosome is shown. A podosome consists of an actin-rich core (red) connected to a surrounding ring structure (green). In the core actin polymerization takes place. Here, the actin nucleation complex Arp2/3 and its regulator WASp, as well as the actin crosslinking protein α-actinin and the actin-bundling protein cortactin are localized. Integrins binding their extracellular ligands, such as ECM molecules, are found in the ring structure. Also the integrin binding proteins talin and paxillin, the talin- and paxillin-binding protein vinculin, zyxin and myosin II are found there. In podosomes integrins are connected to the actin cytoskeleton, mediating adhesion and migration.

Concluding remarks
In this thesis we show the importance of adhesion regulation during DC-maturation and more specifically the need for these cells to dissolve podosomes in order to acquire migratory properties. Gram-negative bacteria are more capable of inducing DC maturation relative to Gram-positive bacteria. This difference is due to differential activation of TLR4. Prostaglandins such as PGE2 are key mediators of DC maturation in response to either pathogen-derived molecules such as LPS or Gram-negative bacteria. A model showing TLR4 activation and subsequent PGE2 signaling is depicted in Figure 2. PGE2-induced podosome dissolution, which either involves the EP2 or the EP4 receptor, raises cAMP levels. This increase in cAMP leads to activation of the RhoA-Rho kinase pathway and consequently an increase in myosin II-mediated contraction. This contractile response leads to a rapid loss of podosomes and the (transient) formation of FA. In conclusion, our results show that regulation of the actomyosin cytoskeleton, mediated by Rho GTPases in response to extracellular signals, plays an important role during the transition from a (tissue resident) iDC to a highly motile mDC.

Figure 2. Regulation of podosomes loss during DC maturation
Stimulation with LPS or Gram-negative bacteria activates TLR4, MyD88 and TRIF dependent signaling, resulting in COX-mediated production of prostaglandins, such as PGE2. PGE2 activates the EP2 and EP4 receptors, resulting in cAMP elevation and PKA activation. This leads to activation of RhoA and inactivation of Rac1 and Cdc42. Active RhoA induces Rho kinase activation. Activation of the RhoA-Rho kinase axis induces myosin II-mediated contraction, which results in podosome loss and temporary FA formation.