|Home | About | Journals | Submit | Contact Us | Français|
Dendritic cells (DCs) are professional APCs that reside in peripheral tissues and survey the body for pathogens. Upon activation by inflammatory signals, DCs undergo a maturation process and migrate to lymphoid organs, where they present pathogen-derived Ags to T cells. DC migration depends on tight regulation of the actin cytoskeleton to permit rapid adaptation to environmental cues. We investigated the role of hematopoietic lineage cell-specific protein 1 (HS1), the hematopoietic homolog of cortactin, in regulating the actin cytoskeleton of murine DCs. HS1 localized to lamellipodial protrusions and podosomes, actin-rich structures associated with adhesion and migration. DCs from HS1−/− mice showed aberrant lamellipodial dynamics. Moreover, although these cells formed recognizable podosomes, their podosome arrays were loosely packed and improperly localized within the cell. HS1 interacts with Wiskott-Aldrich Syndrome protein (WASp), another key actin-regulatory protein, through mutual binding to WASp-interacting protein. Comparative analysis of DCs deficient for HS1, WASp or both proteins revealed unique roles for these proteins in regulating podosomes with WASp being essential for podosome formation and with HS1 ensuring efficient array organization. WASp recruitment to podosome cores was independent of HS1, whereas HS1 recruitment required Src homology 3 domain-dependent interactions with the WASp/WASp-interacting protein heterodimer. In migration assays, the phenotypes of HS1- and WASp-deficient DCs were related, but distinct. WASp−/y DCs migrating in a chemokine gradient showed a large decrease in velocity and diminished directional persistence. In contrast, HS1−/− DCs migrated faster than wild-type cells, but directional persistence was significantly reduced. These studies show that HS1 functions in concert with WASp to fine-tune DC cytoarchitecture and direct cell migration.
Dendritic cells (DCs) are professional APCs that play a unique role in bridging innate and adaptive immunity (reviewed in Refs. 1-3). DCs reside in peripheral tissues and continually sample the environment for pathogens. In response to pathogen-derived inflammatory molecules, these cells undergo a maturation program that induces their migration to lymphoid organs, where they present Ags obtained in the periphery to naive T cells to initiate an adaptive immune response. DC function is critically dependent on the ability to migrate long distances, traverse barriers, and navigate diverse tissues with variable surface characteristics (3). DCs achieve this by mechanical adaptation of cytoskeletal dynamics. Depending on the nature of the substrate with which they are interacting, DCs can move by integrin-independent amoeboid protrusion into an open space within a three-dimensional matrix, or by pushing against integrin-based adhesive contacts with extracellular substrates (4, 5). In this latter mode, movement is driven by the combined force of actin polymerization and myosin contractility. This mechanism is characterized by extension of an actin-rich lamellipodium at the front of the cell, often accompanied by the formation of adhesive contacts termed podosomes just behind the edge of this protrusion. Podosomes are short-lived structures composed of actin-rich cores surrounded by adhesion molecules, including vinculin, talin, and integrins (reviewed in Refs. 6-9). Although the exact function of podosomes is still unclear, these structures serve as sites of matrix metalloproteinase deposition (10, 11) and are thought to facilitate adhesion and migration through tissue barriers such as the lymphatic endothelium. In addition, podosomes may function as part of the mechanosensing mechanism that allows DCs and other hematopoietic cells to alter their cytoskeletal dynamics in response to changing substrates.
The plasticity of DC migration is mediated by tightly regulated changes in actin dynamics. Several individual actin regulatory proteins have been shown to be important for controlling specific aspects of DC migration. One key protein is WASp, the gene for which is mutated in the immunodeficiency disorder Wiskott–Aldrich Syndrome. DCs deficient for WASp show an almost complete lack of migratory capacity (12-15). WASp and its binding partner Wiskott–Aldrich Syndrome protein-interacting protein (WIP) colocalize with F-actin in podosome cores and are essential for the formation of podosomes (16-19). WASp functions by activating the Arp2/3 complex, a seven-subunit protein complex that promotes actin polymerization by generating new actin filaments on the sides of pre-existing filaments (20). Two other proteins that have been shown to be important for DC migration, CDC42 and Vav1, also function to activate Arp2/3-dependent actin polymerization (21, 22). Formation of branched actin filaments is important for generating lamellipodial protrusions as well as for generating podosome cores, which turn over rapidly and exchange actin continuously (19, 23, 24).
Another important actin regulatory protein in DCs is hematopoietic lineage cell-specific protein 1 (HS1, also called HCLS1 or LckBP1) (25). HS1 is the hematopoietic homolog of cortactin, a protein involved in adhesion, spreading, endocytosis, and migration in many cell types (26-29). Cortactin is upregulated or hyperphosphorylated in a number of metastatic cancers and plays an important role in the formation of invadopodia, structures that resemble, but are distinct from, podosomes (30-32). Cortactin stabilizes branched actin filaments in vitro (33, 34) and enhances the persistence of actin-rich lamellipodial protrusions in fibroblasts (26). Like cortactin, HS1 is involved in the stabilization of branched actin filaments (35). Both of these proteins have a modular structure, with an N-terminal region that binds Arp2/3 complex and actin filaments, and a C-terminal adaptor region that can connect multiple proteins within the actin network, including Lck, Itk, Vav1, WASp, WIP, and Nck (36-39). HS1 promotes lamellipodial protrusion in T cells (36, 37) and regulates adhesion and migration in NK cells and B cells (40, 41). Thus, HS1 and cortactin appear to carry out similar functions. Although HS1 has recently emerged as an important actin-regulatory protein in hematopoietic cells, its role in DC function has not been investigated.
In this study, we show that HS1 is the sole cortactin family member expressed in murine bone marrow-derived dendritic cells (BMDCs). HS1 localizes to actin-rich structures involved in cell migration, including lamellipodia and podosomes, and localization of HS1 to podosomes requires Src homology (SH)3 domain-dependent interaction with WASp through its binding partner WIP. Comparative analysis of DCs lacking HS1, WASp, or both proteins reveals that these proteins play distinct roles: WASp is essential for podosome formation, whereas HS1 is necessary to organize the podosome array within the cell. Similarly, WASp is required for overall migration of DCs as well as for directional persistence during chemotaxis, whereas HS1 is primarily required for directional persistence. These studies show that HS1 functions to fine-tune DC cytoarchitecture and direct cell migration.
Rabbit anti-human HS1 (36) and rabbit anti-mouse HS1 (37) were described previously. Mouse anti-HS1 (3A3) was purchased from StressGen Bioreagents. Anti-GAPDH was purchased from Calbiochem. Anti-cortactin (GK-18), anti-FLAG (M2), anti-vinculin, and LPS were obtained from Sigma-Aldrich. Anti-myc, anti-GFP, Alexa Fluor 594 phalloidin, anti-mouse IgG1 Alexa Fluor 488, anti-mouse IgG Alexa Fluor 594, anti-rat Alexa Fluor 488, anti-rabbit IgG Alexa Fluor 488, anti-goat IgG Alexa Fluor 488, and FITC-gelatin were obtained from Invitrogen. Anti-cortactin (4F11), anti-phosphotyrosine (4G10), and anti-WASp were purchased from Upstate Biotechnology. Anti-β2 integrin (CD18, C71/16) was obtained from BD Pharmingen. Anti-talin (C-20) and anti–GST-HRP were obtained from Santa Cruz Biotechnology. Anti-HA tag was purchased from Roche.
Recombinant HS1 was made as described previously (42). To generate recombinant cortactin, full-length human cortactin cDNA was subcloned into pGEX-4T-2 vector (GE Healthcare) and expressed in BL21-DE3 bacteria. The recombinant cortactin was purified using glutathione Sepharose 4B (GE Healthcare). FLAG-WIP and FLAG-WASp were described previously (43). Mutations in WIP (Δ460 and P463A) to abrogate WASp binding were generated based on Ref. 44, and a WASp mutant (Δ40–154) that does not bind WIP was made using standard site-directed mutagenesis (Stratagene).
HS1−/− mice on the C57BL/6J background have been previously described (45), and WASp knockout mice were purchased from The Jackson Laboratory. To generate HS1 and WASp double-knockout (DKO) mice, HS1+/− and WASp−/y mice were bred, and the F1 progeny were then interbred. All mice were housed under pathogen-free conditions in the Children’s Hospital of Philadelphia animal facility. All studies involving animals were reviewed and approved by the Children’s Hospital of Philadelphia Institutional Animal Care and Use Committee.
GM-CSF was produced from the B78Hi/GMCSF.1 cell line provided by T. Laufer (University of Pennsylvania, Philadelphia, PA). Bone marrow was isolated from leg bones, cleaned of muscle tissues, and sterilized in 70% EtOH using IMDM (Life Technologies) containing 1% FBS (Atlanta Biologicals). The cells were centrifuged at 1600 rpm and 4°C for 10 min and resuspended in DC culture media (IMDM, 10% FBS, penicillin/streptomycin, GlutaMax, 55 μM 2-ME, and 3% supernatant GM-CSF) at a concentration of 1 × 106 cells/ml. Cell suspension (1 ml) was added to wells of 24-well plates and supplemented with 1 ml of media on day 2. Starting on day 5, 1 ml media was replaced daily. Differentiation into DCs was verified on day 6 by flow cytometric analysis of surface CD11c levels (typically 70–80%). Cultures were used between days 6 and 9. To induce maturation, DCs between days 6 and 8 were cultured with 100 ng/ml LPS for 24 h.
A vector expressing the GFP-variant Venus (Venus/pCS2 (46)) was provided by A. Miyawaki (Brain Science Institute, RIKEN, Yokohama, Japan). Human HS1 cDNA, described in Ref. 36, and Venus were amplified and ligated into the pMSCV2.1 retroviral vector (provided by W. Pear, University of Pennsylvania). GFP-actin (BD Clontech) was amplified and ligated in place of existing GFP in the MiGR retroviral vector (provided by W. Pear). GFP-WASp (47) and GFP-Lifeact (provided by M. Sixt and R. Wedlich-Solner, Max Planck Institute of Biochemistry, Martinsried, Germany) (48) were amplified and ligated into MiGR and pMSCV2.1 retroviral vectors, respectively. Retrovirus was produced by calcium phosphate cotransfection of 293T cells with 30 μg of the DNA of interest, and constructs encoding the viral envelope protein for mouse ectopic virus and the gag and pol genes. Supernatant was harvested at 24 and 30 h posttransfection and titered using NIH-3T3 fibroblast cells.
BMDCs were transduced by spin infection with retrovirus expressing Venus-HS1, GFP-WASp, GFP-Lifeact, or GFP-actin on day 2 of culture. Retrovirus and 4 μg/ml Polybrene (Sigma-Aldrich) were added to the wells of a 24-well culture plate and centrifuged at 2000 rpm and 32°C for 2 h. Retrovirus-containing media were then replaced with DC culture media, and the cultures were cared for as described above. Transduction efficiency (typically ≥45%) was determined on day 6 by detection of Venus or GFP expression by flow cytometry.
RAW/LR5 cells were a gift from D. Cox (Albert Einstein College of Medicine, New York, NY) and were cultured and retrovirally transduced as described previously (19). Cells were either transduced with control virus or virus to knockdown HS1, and stable lines were selected with puromycin. HS1 suppression was verified by Western blotting.
For analysis of cortactin, HS1, and WASp expression, cells were lysed in lysis buffer (20 mM HEPES [pH 7.5], 1% Nonidet P-40, 0.5% deoxy-cholate, 0.1% SDS, 50 mM NaCl, 5 mM EDTA, 10 μg/ml leupeptin, 500 mM AEBSF, 1 mM Na3VO4, and 5 mM NaF) on ice, cleared by centrifugation, and protein concentration was determined by BCA assay (Pierce). Lysates were resolved on 4–12% NuPage gels (Invitrogen) or tris-glycine SDS-PAGE gels, transferred to nitrocellulose membranes, blocked in 5% milk in PBS, and probed with primary Abs in TBST, followed by secondary Abs (goat anti-mouse IgG-Alexa Fluor 680 [Invitrogen] or goat anti-rabbit IgG-IR Dye 800 [Rockland]). Proteins were visualized and analyzed ratiometrically using the Licor Odyssey infrared fluorescence system, taking care to remain within the linear range.
BMDCs were harvested and cultured on coverslips at 2 × 105 cells/well in 6-well plates overnight. The coverslips were washed in HBSS, followed by fixation in 3% paraformaldehyde/PBS. Cells were permeabilized with 0.3% Triton X-100 and blocked with 0.05% saponin/1.25% fish skin gelatin in TBS. Cells were labeled for F-actin with Alexa Fluor 594-phalloidin and with primary Abs followed by appropriate fluorescently tagged secondary Abs. For endogenous HS1 staining, cells were fixed and permeabilized simultaneously using a protocol from Ref. 49. For visualization of Venus or GFP-tagged proteins, anti-GFP was used, followed by anti-rabbit IgG Alexa Fluor 488. Cells were imaged using a PerkinElmer Ultraview ER6 spinning disk confocal system equipped with a Zeiss Axiovert 200 microscope and a ×63 1.4 NA objective. Images were collected using an Orca ER camera (Hamamatsu) and analyzed using Volocity v.5 software (PerkinElmer).
For array analysis, images were collected without bias using spinning disk confocal imaging. Cell profiles were determined using the “find object” function in Volocity, and the borders of each podosome array were drawn by hand. The areas were calculated in Volocity. The percentage of the total area of individual cells covered by podosome arrays was then calculated. To determine the number of podosomes per cell, the actin cores were identified and counted using the “find objects” function in Volocity, with verification and correction by eye.
For analysis of array localization and packing, slides were blinded to experimental conditions. Cell polarization was determined, based on the presence of an actin-rich, spread lamellipodium. The number of arrays per cell was counted and placed into one of the following groups: touching the leading edge (touch), behind the leading edge but not touching it (behind), in the middle of the cell not touching an edge (middle), opposite of the leading edge (back), lateral to the leading edge (side), or circular rosettes (rosette). Array packing was based on the tightness of podosome packing within individual arrays, with cells scored as tight if most of the podosomes contacted one another in a regular pattern and loose if gaps were evident between many of the podosomes. Approximately 200 cells were analyzed per experiment. For add-back experiments, slides were blinded to experimental conditions, and cells were scored for presence or absence of podosomes, whether the arrays were loosely packed, and whether HS1 or WASp was in the podosome cores. Graphs are averages from three to five independent experiments.
To measure podosome lifetime, wild-type (WT) and HS1−/− DCs were transduced with GFP-Lifeact retrovirus on day 2. On days 7 or 8, DCs were harvested and cultured overnight at 1 × 105 cells/well of a 4-well Lab-Tek chamber slide, prior to imaging. DCs were imaged every 5 s for 15 min by confocal microscopy. Fifteen WT and HS1−/− cells were analyzed, representing 4000–6000 individual podosomes per cell type. The “find objects” function in Volocity was used to identify podosomes in each timepoint, with visual verification. Objects were then tracked throughout the video using the “Track Object” function. Tracks that contained objects in the first two or last two time points were excluded from analysis, as were tracks that existed for less than four time points. Podosome lifetime was then calculated, based on the number of time points in any given track. Outliers (>2 SDs from the mean) were removed, and data were plotted as box and whiskers plots with the median of each population represented.
Reformation of podosomes was assayed, based on a modification of Ref. 19. Briefly, BMDCs were cultured on coverslips overnight. DC culture media containing 1 μM cytochalasin D (Calbiochem) were added for 30 min at 37°C. The cells were then washed twice with warm DC culture media and incubated for the indicated times at 37°C, fixed, and labeled for actin and vinculin. Slides were blinded, and ~200 cells were scored for the presence or absence of podosome arrays.
To measure actin turnover within podosomes, BMDCs transduced with GFP-actin were cultured in 4-well Lab-Tek II chambered coverglasses (Nalge Nunc) at 5 × 104 cells/chamber overnight. Fresh media were added and overlaid with mineral oil before imaging. Cells were imaged by spinning disk confocal microscopy using the Volocity v.5 fluorescence recovery after photobleaching (FRAP) plug-in. The cells were imaged every 3 s before bleaching. A 20-μm2 area within the podosome array of each cell was then bleached for 50 cycles, and images were captured at the fastest speed for 15 s, every second for 45 s, and every 3 s for 180 s. Analysis was conducted using a single constrained exponential algorithm. Results are shown as τ1/2, the time required for fluorescence to recover to half the original value.
WT and HS1−/− DCs were transduced with GFP-Lifeact retrovirus on day 2. On days 7 or 8, DCs were harvested and cultured overnight at 1 × 105 cells/well of a 4-well Lab-Tek chamber slide, prior to imaging. DCs were imaged every 5 s for 15 min by confocal microscopy. Fifteen WT and 15 HS1−/− cells were analyzed. Two lines per cell were drawn through lamellipodial projections perpendicular to the cell edge and kymographs were produced using Volocity software. Periods of protrusion, retraction or stationary behavior were identified and measured to determine the distance (x) and time (y). From this, distance and velocity were calculated. Outliers (>2 SDs from the mean) were removed, and data were represented as box and whiskers plots with the median of each population represented.
For zymography, 2.5 × 106 BMDCs were cultured in serum-free media in bacteriological 10-cm dishes for 24 h. Supernatant was concentrated using a <30-kDa cutoff centrifugal filter device (Millipore). Proteins were separated on SDS-PAGE gels containing 1 mg/ml gelatin (Sigma-Aldrich). The gel was then incubated in renaturation buffer (2.5% v/v Triton X-100, 50 mM Tris-HCl, and 0.05% NaN3) for 3 h at 37°C and in developing buffer (50 mM Tris-HCl, 150 mM NaCl, 10 mM CaCl2, and 0.05% NaN3) at 37°C overnight. Nondegraded gelatin was visualized by Coomassie brilliant blue staining and imaged on a Licor Odyssey fluorescence scanner.
FITC-gelatin degradation assays were performed as described in Ref. 50. Briefly, coverslips were acid-washed, coated with 2% FITC-gelatin (Invitrogen), and quenched in serum-free media for 1 h at 37°C. BMDCs were cultured on coverslips overnight, washed, and fixed. Cells were labeled for actin and imaged by spinning disk confocal microscopy. The degraded area for each cell was quantified using Volocity and expressed as a percentage of total cell area, as defined by phalloidin labeling.
Ninety-six-well Transwell plates, 5-μm pore size, were from NeuroProbe. Immature and mature WT and HS1−/− BMDCs were harvested, pelleted at 1500 rpm and room temperature for 5 min, and resuspended at 2 × 106 cells/ml. Migration media alone (IMDM, 1% serum) or containing chemokine (200 ng/ml CXCL12 [PeproTech], 500 ng/ml CCL21, or 2.5μg/ml CCL19 [R&D Systems]) was added to the bottom well. Cells (5 × 104) were placed on the filter, and the plate was incubated at 37°C for 3 h. The filter was removed, and the cells in the bottom well were counted using a hemocytometer. The percentage of migrated cells was calculated by dividing the number of cells in the bottom well by the number of input cells. Chemokine receptor expression and cell maturation were verified by flow cytometry of input cells.
For microfluidic migration assays, a microfluidic gradient generator was fabricated in polydimethylsiloxane (PDMS, Sylgard 184; Dow Corning) using soft lithography as previously described (51) with modifications. Briefly, soft lithography was used to create an SU-8-2050 photoresist (MicroChem) on a silicon master. Positive replicas with embedded channels were fabricated by molding PDMS against the master. The PDMS replica and a glass microscope slide were activated by oxygen plasma treatment then irreversibly contact bonded. The adhesion surface was functionalized by incubation with 10 μg/ml fibronectin (Sigma-Aldrich) for 1 h at 20°C and blocked with 1% BSA (Sigma-Aldrich) in PBS for 2 h at 20°C. The microfluidic chemotaxis assay was performed as previously described (52) with the following changes. BMDCs were cultured as described in Ref. 53 and matured for 24 h with 100 ng/ml LPS, harvested by pipetting and loaded into a syringe. The chemoattractant solution used was CCL19 (PeproTech).
The structural changes associated with DC migration are orchestrated by several actin-regulatory proteins, including Rho family GTPases, Vav1, and WASp (12, 14, 21, 22, 54). The cortactin homolog HS1 functions as part of this actin regulatory complex in lymphocytes and NK cells (36, 37, 40, 41), but its role in DCs has not been addressed. HS1 and cortactin usually exhibitmutually exclusive expression patterns, with HS1 expressed in hematopoietic lineage cells and cortactin expressed in other cell types. However, DCs have been reported to express both proteins (10, 17). Thus, we initiated our studies by carefully characterizing the expression patterns of these two proteins in BMDCs. In addition to testing BMDCs from WT mice, we tested BMDCs generated from HS1−/− mice, to ask whether cortactin expression is upregulated to compensate for loss of HS1. BMDCs lacking HS1 differentiate normally in culture and upregulate costimulatory molecules (CD80 and CD86), CD40, and MHC class I and II similarly to WT BMDCs upon maturation with LPS (data not shown). As shown in Fig. 1, a polyclonal anti-mouse HS1 Ab raised in our laboratory reacted with recombinant human HS1 but not cortactin and with a ~70-kDa band in lysates from hematopoietic cells (T cells and DCs) from WT mice but not HS1−/− mice. As expected, this Ab failed to interact with lysates from non-hematopoietic cell types (mouse 3T3 and human 293T). This reagent binds human HS1 weakly, as indicated by its ability to detect recombinant human HS1, but not HS1 in human Jurkat T cells. In addition, a polyclonal anti-human HS1 Ab reacts specifically with human HS1 as a recombinant protein or from Jurkat T cells, but not with mouse HS1 (DCs and T cells). Several commercially available Abs tested displayed different patterns of HS1 and cortactin recognition. One monoclonal anti-cortactin Ab, 4F11, reacted with mouse and human cortactin (~70-kDa band in 3T3 and 293T cells, respectively), but failed to detect recombinant HS1, or HS1 expressed in mouse T cells or BMDCs. However, another widely used Ab, GK-18, cross-reacted with HS1 and cortactin in all cell types tested and with both recombinant proteins. In keeping with these biochemical data, GK-18 labeled actin-rich structures in WT, but not HS1−/− DCs by immunofluorescence microscopy, whereas 4F11 showed only background labeling in both cell types (see Fig. 2C). Taken together, these findings demonstrate that murine DCs express HS1 but not cortactin. Furthermore, they verify that DCs from HS1−/− mice lack HS1 expression and show that these cells do not exhibit compensatory upregulation of cortactin.
We next investigated HS1 localization in murine BMDCs. The polyclonal Ab raised in our laboratory did not work well for immunofluorescence microscopy, but mAb 3A3 (StressGen Bioreagents) specifically labeled WT DCs but not DCs from HS1−/− mice (Fig. 2C). Labeling with this Ab revealed that HS1 colocalizes with F-actin in podosome cores (Fig. 2A, 2C). Similar results were obtained with cross-reacting anti-cortactin Ab GK-18 but not the more cortactin-specific Ab 4F11 (Fig. 2C). We also observed HS1 colocalization with F-actin at the edges of lamellipodia (Fig. 2B). This distribution is consistent with the localization of cortactin in nonhematopoietic cells (55) and with the idea that HS1 functions to regulate actin-rich structures associated with cell migration.
To ask whether HS1 is required to organize DC cytoarchitecture, WT and HS1−/− DCs were plated onto coverslips, and the actin cytoskeleton was analyzed by immunofluorescence microscopy. Adhesion and spreading of HS1−/− DCs on both fibronectin-coated and uncoated coverslips were grossly normal, although we noted that HS1−/− DCs were somewhat more likely to exhibit multiple lamellipodial protrusions. Labeling with phalloidin and anti-vinculin revealed that both WT and HS1−/− DCs were able to make podosomes with actin-rich cores surrounded by vinculin rings (Fig. 3A). In some HS1−/− DCs, the boundaries of actin rich cores and vinculin rings seemed diffuse in HS1−/− DCs (Supplemental Fig. 1), but this phenotype was also observed in the population of WT cells. No differences were observed in the number of cells with podosome arrays or the number of arrays percell (data not shown). Moreover, other markers of podosomes, including talin, β2 integrin, and phosphotyrosine (56), localized normally within the podosomes of HS1−/− DCs (Supplemental Fig. 2). In addition to exhibiting normal composition, the podosomes of HS1−/− DCs were functionally competent as sites of extracellular matrix degradation. Supernatants from WT and HS1−/− DCs contained similar levels of functional matrix metalloproteinases (MMPs). On the basis of their gelatinase activity and mobilities, the predominant bands for both cell types correspond to pro-MMP9 and pro-MMP2 (Supplemental Fig. 3A). Moreover, when WT or HS1−/− DCs were cultured on FITC-gelatin-coated coverslips, there was no difference in the size, placement or frequency of holes formed in the matrix, or in average area of matrix degradation per cell (Supplemental Fig. 3B, 3C; data not shown). We conclude that HS1 is not required for the formation of podosomes that contain many of the characteristic proteins and function as sites of metalloproteinase release.
Although the organization of podosome arrays varies widely, even among WT cells, we observed clear differences in the podosome arrays of WT and HS1−/− DCs. First, although HS1−/− DCs were as likely to contain podosomes as WT DCs (data not shown), they exhibited significantly fewer podosomes per cell (Fig. 3B). To determine whether HS1 promotes podosome formation, WT or HS1−/− DCs were treated with cytochalasin D to dissolve podosomes. After drug washout, the cells were allowed to recover for varying times, and the number of cells with podosome arrays was assessed. Prior to treatment, a similar number of WT and HS1−/− DCs exhibited podosomes (Fig. 3C, points on the y-axis). In both cell types, podosomes were lost upon drug treatment, and actin cores were recovered within 30–60 min of drug washout (Fig. 3C). However, recovery of podosome cores in HS1−/− DCs was consistently delayed. In both WT and HS1−/− cells, recovery of vinculin into rings was not observed until after actin cores were formed (data not shown), consistent with the idea that a viable actin core is needed for recruitment of ring proteins (15).
Because HS1 stabilizes branched actin filaments, we hypothesized that loss of HS1 would decrease the stability of podosome cores. To test this, DCs were transduced with GFP-Lifeact, which selectively labels F-actin in DCs without affecting lamellipodial dynamics (48), and the lifetime of individual podosomes was assessed by video microscopy. As shown in Fig. 3D, the median podosome lifetime was significantly reduced in HS1−/− DCs, but this represents a minor shift in the population given the wide distribution of values. Finally, the exchange of actin molecules within podosome cores was assessed by FRAP in DCs transduced with retrovirus expressing GFP-actin. No significant differences were observed between WT and HS1−/− DCs, indicating that loss of HS1 does not affect actin turnover within pre-existing podosomes (Supplemental Fig. 3D, Supplemental Videos 1, 2). Taken together, these data indicate that HS1 accelerates the early stages of podosome biogenesis, but is not essential for podosome formation or stability in DCs.
The most striking cytoarchitectural defects we observed in HS1−/− DCs involved podosome array organization. Podosomes in the HS1−/− DCs were not packed as tightly as those in WT DCs and the arrays were more randomly distributed throughout the cell (Fig. 3A, Supplemental Fig. 2). To assess differences in the positioning of podosome arrays, cells were categorized into one of several groups: touching the leading edge (touch), behind the leading edge (behind), centrally located within the cell (middle), opposite the leading edge (back), lateral to the leading edge (side) or forming rosettes within the center of the cell not adjacent to any edges (rosettes). Images exemplifying each group are shown in Fig. 4C. Whereas WT DCs more frequently showed arrays that touched the leading edge, arrays in HS1−/− DCs tended to be further behind the leading edge (Fig. 4A). In addition to array localization, array packing was affected (Fig. 4B, example images shown Fig. 4D). Whereas podosomes in WT DCs tended to be tightly packed within the array, HS1−/− DCs showed more cells with loosely packed podosome arrays (arrays with significant space between adjacent podosomes). This qualitative finding is consistent with our quantitative data showing that HS1−/− DCs have fewer podosomes than WT DCs, distributed in arrays that occupy a similar area (Fig. 3B, Supplemental Fig. 3E).
To verify that the phenotypes we observe do not reflect developmental changes in the HS1−/− mice and to ask whether these results extend to other myeloid cell types, HS1 function was tested in RAW/LR5 macrophages, a cell line that efficiently forms podosomes (19, 57). Like the BMDCs, these cells express HS1 but not cortactin (data not shown). Fig. 5A shows that HS1 could be efficiently silenced in these cells using shRNA. HS1-suppressed RAW/LR5 cells were able to form podosomes with normalfrequency (data not shown), but the podosome array in these cells became more loosely packed (Fig. 5B, 5C). We conclude that HS1 is not required for podosome formation or stability in myeloid cells, but is required for proper organization of podosome arrays.
Because podosome formation and dissolution are mechanistically linked to lamellipodial dynamics (15, 17, 58), and HS1 is present at the edges of lamellipodial protrusions (Fig. 2B), we asked whether lamellipodial dynamics are altered in HS1−/− DCs. DCs were transduced with GFP-Lifeact and monitored by video microscopy and kymographic analysis. As shown in Fig. 6A, lamellipodial protrusion and retraction events occurred over longer distances in HS1−/− DCs. The velocity of these events was also increased (Fig. 6B). These results are in good agreement with the effects of HS1 on lamellipodial dynamics in T cells (36) and the effects of cortactin on lamellipodial dynamics in fibroblasts (26).
Cortactin interacts via its SH3 domain with the WASp/WIP heterodimer (59, 60), but the ability of HS1 to interact with these proteins has not been directly addressed. To test this, lysates from cells expressing FLAG-tagged WASp or WIP were subjected to pulldown assays using the GST-tagged HS1 SH3 domain. As shown in Fig. 7A, both WASp and WIP interacted with the HS1 SH3 domain. To ask whether these proteins can interact in intact cells, we overexpressed FLAG-tagged HS1 with WASp and WIP (tagged with HA and myc, respectively) in 293T cells. As shown in Fig. 7B, immunoprecipitation of HS1 led to coimmunoprecipitation of both WIP and WASP, and mutation of the critical tryptophan residue in the HS1 SH3 domain abolished this interaction. Interestingly, deletion of the N-terminal half of HS1 enhanced binding to both WASp and WIP. Because this region contains the binding sites for Arp2/3 complex and F-actin, this demonstrates that binding of HS1 to WASp and WIP does not require mutual binding to F-actin. Moreover, the enhanced binding to this mutant suggests that the actin regulatory portion of HS1 may limit accessibility of the SH3 domain in the full-length molecule.
WASp and WIP exist as a heterodimer in cells. To ask whether HS1 can bind WASp independently of WIP, cells were transfected with FLAG-tagged WASp or with a WASp deletion mutant that lacks the WIP binding region (Δ40–154). Because of its inability to bind WIP, this WASp mutant is unstable (not shown), but it is expressed at reasonable levels using this overexpression system. As shown in Fig. 7C, WT WASp binds well to the HS1 SH3 domain, but the WIP binding mutant does not. To test WIP-HS1 binding, similar studies were performed in cells transfected with FLAG-tagged WIP or WIP mutants that fail to bind WASp (P436A or Δ460). The two WIP mutants failed to bind efficiently to WASp, but interacted strongly with the HS1 SH3 domain, indicating that WIP binding to HS1 is independent of WASp (Fig. 7D). Finally, cell lysates from FLAG-WIP expressing cells were probed with recombinant WT HS1 SH3 domain or the inactive tryptophan mutant using a gel overlay approach. As shown in Fig. 7E, the WT SH3 domain, but not the mutant, bound specifically to WIP. This confirms direct interaction between HS1 and WIP. Taken together, these studies indicate that WIP binds directly to the HS1 SH3 domain and mediates indirect interactions between HS1 and WASp.
Like HS1, WASp is involved in podosome formation and lamellipodial protrusion (12, 16, 18, 19, 58), and HS1 is thought to stabilize branched actin filaments generated by WASp (35, 36). To investigate the functional relationship between HS1 and WASp in DCs, we compared the phenotypes of DCs cultured from mice lacking HS1 alone, WASp alone, or both proteins. DCs culturedfrom these mice exhibited loss of the appropriate proteins, and knockout of one had no effect on the expression levels of the other (Fig. 8A). As shown in the black bars in Fig. 8B, HS1−/− and WASp−/y DCs differed with respect to the proportion of cells exhibiting podosomes. Although significantly fewer WASp-deficient cells displayed podosomes, HS1-deficiency had no effect on this parameter (the slight increase relative to WT cells in this experiment was not reproducible). The defect in podosome formation in WASp−/y DCs is consistent with previous reports (61), although the magnitude of the defect is somewhat less severe in our hands. It has been proposed that HS1 may contribute to residual podosome formation in WASP−/y DCs (62); however, we found that DCs deficient for both HS1 and WASp were indistinguishable from cells deficient for WASp alone. We next compared the effects of loss of HS1 or WASp with respect to podosome organization. As shown in the black bars in Fig. 8C, HS1−/− DCs showed defective packing of the podosome array, a phenotype that was not observed in WASp-deficient cells. Podosome packing in DCs deficient in both HS1 and WASp was indistinguishable from packing in cells lacking HS1 alone. Taken together, these results show that WASp is required for efficient formation of podosomes, whereas HS1 is important for organizing the podosome array.
To confirm these findings and to ask whether HS1 and WASp show interdependent function, DCs lacking these proteins individually or together were transduced with WASp, HS1, or with the HS1 SH3 domain mutant that abrogates interaction of HS1 with the WIP/WASp heterodimer. As shown in Fig. 8B, hatched bars, ectopic expression of WASp restores the number of WASP−/y DCs cells displaying podosomes to WT frequency. Transfection with WASp also significantly increases the number of double-deficient DCs displaying podosomes. In contrast, expression of HS1 (gray bars) in double-deficient DCs does not rescue this defect, supporting the idea that WASp, but not HS1, is essential for efficient podosome formation.
When transduced DCs were analyzed with respect to podosome packing, reciprocal results were obtained (Fig. 8C). The abnormally loose packing observed in HS1−/− DCs and DKO cells was rescued by ectopic expression of HS1 (gray bars). Expression of GFP-WASp (hatched bars) in DKO DCs did not rescue this defect, supporting the idea that HS1, but not WASp, is needed to organize a closely packed podosome array. Interestingly, the HS1 SH3 domain mutant (open bar) was unable to rescue podosome organization in HS1−/− DCs, suggesting that interactions mediated by the SH3 domain are required for HS1 function.
To complement our functional analysis of HS1 and WASp interactions, we asked whether these proteins depend on one another for recruitment to podosomes. As shown in Fig. 9A and 9C, Venus-HS1 localized efficiently to podosome cores when expressed in HS1−/− DCs. Costaining with anti-phosphotyrosine or anti-vinculin showed that Venus-HS1 exhibits podosome core localization similar to that observed with endogenous HS1 (Supplemental Fig. 4). In DKO DCs, however, Venus-HS1 localization to podosome cores was faint or nonexistent, and the protein instead exhibited a diffuse cytoplasmic distribution (Fig. 9A, 9C). Interestingly, HS1 was sometimes apparent at lamellipodial protrusions in these cells (compare Fig. 9C, HS1 in DKO), suggestingthat the requirements for podosome localization and lamellipodial localization may differ. Similar results were obtained when the SH3 domain mutant of HS1 was expressed in HS1−/− DCs. As shown in Fig. 9B and 9C, WASp localized efficiently to the podosome core when expressed in either WASp−/y or DKO cells. Taken together, these results indicate that WASp localization to podosome cores is independent of HS1, but HS1 localization to podosome cores depends on SH3 domain-mediated interactions with the WIP/WASp heterodimer.
Podosomes are thought to promote cell migration in some settings, and WASp−/y DCs are defective in migration (13, 14, 63). We therefore asked whether HS1 is also required for DC migration. Initial studies were performed using transwell assays. As reported previously (64), immature and mature BMDCs from WT mice preferentially migrated toward CXCL12 (SDF1α) and CCL19 (MIP3β), respectively (Fig. 10A, 10B). HS1−/− DCs migrated with the same efficiency as DCs from WT mice and showed the same switch in chemokine sensitivity with maturation. Although immature DCs migrated slightly less well to CXCL12 in the assay shown Fig. 10A, this difference was not statistically significant and was not reproducible over multiple assays. Migration of WT and HS1−/− DCs toward CCL21 (SLC) was very similar to migration toward CCL19 (data not shown). Furthermore, WT and HS1−/− DCs exhibited similar sensitivity in chemokine dose-response studies, and flow cytometry analysis showed that these two populations express similar surface levels of chemokine receptors (data not shown).
Transwell assays measure only end-point effects and can fail to detect defects in directional migration. Thus, we used video microscopy to compare the ability of mature WT, HS1−/−, WASp−/y and DKO DCs to undergo chemotaxis in a gradient of CCL19. Only mature DCs were studied using this assay, because immature cells were tightly adherent to the microfluidic chamber and largely immotile. As shown in Fig. 10C and 10D and Supplemental Videos 3–6, HS1−/− DCs moved significantly faster than WT DCs, although WASp−/y DCs and DKO cells moved significantly slower. Analysis of directionality revealed HS1−/− DCs, like WASp-/y DCs, exhibit diminished directional persistence (chemotactic index) (Fig. 10C, 10E). The magnitude of this defect was greater in WASp−/y and DKO DCs, but all three mutants weresignificantly less persistent than WT DCs. Taken together, these data show that although WASp is required for overall DC migration as well as directional persistence, the primary role of HS1 is to promote persistent directional migration.
This study addresses for the first time, to our knowledge, the function of the cortactin family member HS1 in DCs. In general, hematopoietic lineage cells express HS1, whereas nonhematopoietic cells express cortactin; however, megakaryocytes, platelets, and osteoclasts express both proteins (65, 66). Thus, an important starting question was whether these cells express HS1, cortactin, or both. Using carefully controlled non-cross-reactive Abs, we find that HS1 is the only cortactin family member present in murine BMDCs. Because cortactin expression has been reported in murine splenic DCs (10, 17), this raises the possibility that these related, but distinct, cortactin family members are differentially expressed in DC subsets.
Our analysis of BMDCs from HS1−/− mice revealed three related defects: disorganization of the podosome array, altered lamellipodial dynamics, and diminished directional migration. WASp, which interacts indirectly with HS1, also affects these processes, but our analysis shows that the roles of these two proteins are distinct. It is well established that WASp and its obligate binding partner, WIP, are essential for formation of the actin-rich cores that nucleate podosome biogenesis (10, 12, 15-17, 19, 24, 61, 67-71). Depending on the experimental system, WASp-deficient DCs and macrophages either lack podosomes altogether or show severe reductions in podosome numbers, and our data are consistent with this. In contrast, we find that HS1-deficient DCs and macrophages can form podosomes containing many, if not all, of the characteristic components. However, HS1-deficient cells show disordered podosome array packing and mislocalization of the arrays with respect to the leading edge of the cell. Similar effects were noted in a recent study in which WIP−/− DCs were reconstituted with a WIP mutant lacking the HS1/cortactin binding site (10). Thus, we conclude that HS1 is not required for podosome formation, but rather for organization of the podosome array.
The mechanisms through which HS1 controls podosome array organization are unclear. Because HS1 stabilizes branched-actin filaments generated by WASp and other Arp2/3 complex activators, it seems likely that it stabilizes actin filaments within podosomes. Although our FRAP data show that loss of HS1 does not affect actin exchange in mature podosomes, it does delay podosome reformation. Thus, HS1 may stabilize newly formed actin cores, such that the diminished numbers and loose packing of podosomes in HS1-deficient cells could result from stochastic disassembly of some newly formed podosome cores within the array. This interpretation is consistent with the modest decrease in average podosome lifetime in HS1−/− DCs; this may represent increased instability of a small number of podosomes. Another appealing possibility is that HS1 aids in stabilizing the long actin filaments that form connections between adjacent podosomes (72). Interestingly, these interconnecting filaments are frequently decorated with clathrin coated endocytic pits (72), and the HS1 homolog cortactin associates with coated vesicle components (27). Finally, it should be noted that in addition to directly regulating actin dynamics, HS1 functions as an adaptor molecule and can recruit other signaling molecules to sites of actin polymerization (36, 37). We show in this study that HS1 is not needed for recruitment of WASp to podosomes, but HS1 could recruit Vav1 or phospholipase Cγ, proteins that are important regulators of podosome dynamics and directional persistence in DCs (22, 73).
In addition to podosomes, HS1 is enriched at lamellipodial edges in DCs. HS1 was sometimes enriched in lamellipodial protrusions in the absence of WASp (see Fig. 9C, second row), but we were unable to quantify the frequency of this localization pattern. The mechanism through which HS1 is localized to lamellipodial protrustions is not known. In addition to binding to the WIP/WASp heterodimer, the SH3 domain of HS1 can interact with other proteins, including Src kinases and Dynamin 2 (Ref. 25 and D.A. Klos Dehring, unpublished observations). In other systems, interaction of phosphotyrosines with Src kinases and other SH2 domain-containing molecules has been shown to mediate plasma membrane targeting (36, 41, 74, 75). Finally, actin binding could be involved. HS1−/− DCs show increased protrusion and retraction distance and velocity, indicating that HS1 functions to stabilize lamellipodial dynamics in DCs. This finding is consistent with our previous work showing that HS1 stabilizes lamellipodial protrusions in T cells (36) and with studies showing that cortactin stabilizes lamellipodial protrusion in nonhematopoietic cells (26, 76). Function of HS1 in podosomes and at the lamellipodial edge need not be mutually exclusive processes. On the contrary, HS1 is likely to function at both sites, with outcomes that are functionally intertwined via a feedback process. Because forward movement of the DC lamellipodium is closely linked to the cycle of podosome formation and dissolution (12, 15, 24, 68, 71), erratic leading edge dynamics in HS1-deficient DCs could result in disorganization of the podosome array. Indeed, this seems the likeliest mechanism to create the observed mislocalization of the array with respect to the leading edge of the cell. Conversely, it is thought that the podosome array stabilizes the dominant leading edge of the cell, such that a disordered or misplaced podosome array may lead to lamellipodial instability, and diminished directional persistence during migration.
The importance of HS1 function in podosomes and at the leading edge of the cell is demonstrated by the diminished ability of HS1−/− DCs to undergo directional chemotaxis. In this study, too, the phenotypes of HS1−/− and WASp−/y DCs are related but distinct. As reported previously (13, 14, 63), we found that WASp expression is essential for DC migration. WASp−/y DCs migrating in a chemokine gradient showed a large decrease in velocity, and those cells that did migrate exhibited diminished directional persistence. In contrast, HS1-deficient DCs actually migrated faster than WT cells, but directional persistence was significantly reduced. The defects in directional persistence in HS1−/− cells may reflect the increased lamellipodial instability in these cells. Alternatively, the defects in directional persistence may reflect the role of podosomes in stabilizing a dominant leading edge (15, 17, 58). In this scenario, HS1 would function to fine-tune the packing and localization of podosomes formed by WASp to aid the stabilization of the leading edge, promoting efficient directional cell migration. These two possibilities are not mutually exclusive and, in fact, are likely to represent intertwined aspects of HS1 function.
Our data point to a hierarchical relationship between WASp and HS1 in controlling DC actin dynamics. HS1 interaction with WASp is indirect and involves binding of the HS1 SH3 domain to WIP. Using add-back experiments, we found that WASp can localize to podosomes independently of HS1, but HS1 is not recruited efficiently to podosomes in the absence of WASp or if its WIP-binding SH3 domain is mutated. This indicates that HS1 is recruited to podosomes through SH3-domain-dependent interactions with the WIP/WASp complex. These results are consistent with a recent study showing that cortactin fails to localize to podosomes in WIP−/− DCs and that the proline rich region of WIP responsible for binding to cortactin’s SH3 domain is important for podosome architechture (10). Indeed, WIP seems to play a key role in podosome assembly, because WASp also fails to localize to podosomes in WIP−/− DCs (17). Given our data showing that HS1 binds directly to WIP rather than to WASp, it would be interesting to ask whether WASP−/Y DCs fail to correctly localize WIP.
Our functional studies also support the view that the WASp/WIP heterodimer serves a central role in podosome biogenesis, whereas HS1 fine-tunes podosome array organization. In cells lacking both WASp and HS1, the defects in both cell morphology and chemotaxis were indistinguishable from cells deficient for WASp alone. One finding, however, suggests that HS1 function may be at least partially independent of WASp. In analyzing the 30–40% of DCs that generated podosomes in the absence of WASp, we observed loose packing of podosome arrays if HS1 was also absent, but tight packing of podosome arrays if HS1 was expressed (either WASp single knockout cells or DKO cells transduced with HS1). This result is particularly surprising given our finding that HS1 fails to localize to podosomes efficiently in the absence of WASp. This apparent discrepancy may reflect WASp-independent HS1 function at the leading edge. Alternatively, it may reflect the ability of HS1 to interact weakly with podosomes by binding to F-actin. Evidence that such binding occurs is shown by the higher frequency of podosome localization of the HS1 SH3 domain mutant in WASp-sufficient cells as compared with WT HS1 in WASp−/y DCs (see Fig. 9A).
The mild podosome phenotype we observe in HS1−/− DCs is somewhat surprising given that cortactin is essential for formation of invadopodia in metastatic tumor cells (77, 78). Although we cannot exclude the possibility that HS1−/− mice undergo compensatory developmental changes that blunt the DC phenotype, we deem this unlikely because we found no upregulation of WASp or cortactin, and because similar defects were observed with HS1 shRNA in a macrophage cell line that also lacks cortactin. A more likely possibility is that HS1 and cortactin are functionally distinct. Because HS1 is expressed in hematopoietic cells that generate podosomes, whereas cortactin is typically expressed in nonhematopoietic cells that generate invadopodia, it will be interesting to explore the differential role of these proteins in generating podosomes versus invadopodia. Interestingly, cortactin is required for formation of podosomes in osteoclasts, hematopoietic lineage cells that express both HS1 and cortactin (66). Although this may represent an exception to the podosome/invadopodium distinction, osteoclast podosomes resemble invadopodia in that they are key sites of matrix degradation. This points to another interesting difference between HS1 and cortactin. Although cortactin is required for matrix metalloproteinase release at invadopodia in nonhematopoietic cells (79) and at podosomes in splenic DCs (10), we found no requirement for HS1 in metalloproteinase release in BMDCs. It will be interesting to explore these distinctions by asking whether it is possible to rescue HS1−/− cells with cortactin and vice versa.
The relationship between HS1 and WASp defined in this study is also somewhat different from the relationship between cortactin and N-WASp in other cell types. Although we find that WASp recruits HS1 to podosomes, cortactin has been shown to recruit and promote N-WASp activity at sites of actin polymerization (80, 81). Moreover, we find that the WIP-binding SH3 domain of HS1 is needed for recruitment to podosomes in DCs, but this domain of cortactin has been shown to be dispensible for podosome targeting in osteoclasts (66). Phosphorylation of cortactin has been shown to play an important role in its ability to regulate N-WASP (38, 39, 77, 82). Phosphorylation of HS1 is important for its actin-regulatory function in T cells and NK cells (36, 41), but its role in DCs remains to be explored.
An important open question in this field is the extent to which podosomes are important for DC function in vivo. WASP−/y DCs have significant migration defects in vivo, but it is unclear to what extent this reflects a requirement for podosome formation. It has long been assumed that podosomes are sites for integrin-dependent adhesion to the extracellular matrix, but the importance of integrins in regulating DC migration is complex and highly dependent on environmental cues (4, 5, 83, 84). In this context, an appealing possibility is that these structures are important as mechanosensors, to allow DCs to adapt to movement along variable surfaces (9, 21, 85). Finally, because podosomes are most prominent in immature DCs, these structures may a play an important role in maintaining cell anchorage and/or dynamics of dendritic processes in peripheral tissues. By identifying and characterizing individual proteins that control distinct aspects of podosome function, we will have a better understanding of whether and how these structures contribute to the regulation of DC movements during an in vivo immune response.
We thank Dr. D. Cox for providing materials and guidance for studies in RAW/LR5 macrophages. We also thank Drs. T. Laufer, S. Gallucci, and P. Oliver and members of the Oliver and Burkhardt laboratories for helpful discussions and critical reading of the manuscript.
The authors have no financial conflicts of interest.