Semaphorin 7A Promotes Chemokine-Driven Dendritic Cell Migration Free

Dendritic cells (DCs) have major potential in new vaccination strategies for cancer and infectious disease because of their potent ability to initiate adaptive immune responses. DCs are dependent on their capacity to migrate to T cell areas in secondary lymphoid organs to induce effective immune responses (reviewed in Refs. 1, 2). Upon activation by pathogens or other danger signals, DCs mature, resulting in altered adhesive and migration capacity, as well as in the upregulation of chemokine receptor CCR7 that recognizes CCL21 and CCL19 present in lymphatic vessels and T cell zones of lymph nodes (LNs) (3–6). DCs can switch between two modes of migration, adhesion-independent and adhesion-dependent migration; actin polymerization is the driving force in both cases (7). In the absence of adhesion, DCs can increase retrograde actin flow to maintain migration speed, and they may use confinement-based pushing forces for locomotion in three-dimensional structures (7–9). When integrin ligands are available or on a two-dimensional substrate, DCs can make their migration more energy efficient by using adhesion receptors as an adhesive clutch (7). The molecular pathways modulating the adhesion and motility of activated DCs, allowing their effective chemokine-driven migration toward the LNs in vivo, have not yet been fully resolved.

Semaphorins represent a family of membrane-associated and secreted proteins that is characterized by an evolutionarily conserved “Sema” domain. More than 25 semaphorins have been identified that are broadly expressed in different organ systems, including the nervous, cardiovascular, and immune systems (reviewed in Refs. 10–12). The pleiotropic functions of semaphorin proteins range from axon outgrowth, angiogenesis, bone differentiation, and immune regulation to tumor metastasis. Semaphorin7A (Sema7A; also known as CD108 or SemaK1) is the only GPI-anchored semaphoring; thus, it lacks a cytoplasmic domain (13, 14). The extracellular domain of Sema7A contains a conserved Sema domain bearing an RGD motif, a “plexins semaphorins integrins” domain, and an Ig domain. The two known receptors for Sema7A are plexin C1 and β₁ integrin (α₁ and α_v heterodimers). Binding of Sema7A to plexin C1 in melanocytes leads to inhibition of the actin-binding protein cofilin and reduced cell spreading (15). In contrast, Sema7A binding to integrins in neurons or melanocytes activates FAK and MAPK pathways, leading to remodeling of the actin cytoskeleton and increased cell spreading (16, 17). Thus, Sema7A regulates two signaling pathways that have counteracting effects on the actin cytoskeleton and cell adhesion.

In the immune system, Sema7A expression was reported on activated T cells, B cells, macrophages, and DCs, where it is implicated in the regulation of immune cell activation (18–21). Our recent proteome study of human monocyte-derived DCs (moDCs) uncovered that Sema7A is one of the most highly upregulated proteins upon DC maturation (22). However, the biological function of Sema7A on DCs remains elusive. Based on the reported actions of Sema7A on melanocyte adhesion and axonal guidance, we hypothesized that Sema7A may be important for activation-induced DC migration. This study identifies Sema7A as a novel regulator of chemokine-induced migration of activated DCs that is central for initiating adaptive immune responses.

Human monocytes, myeloid DCs (myDCs), and plasmacytoid DCs (pDCs) were isolated from buffy coats of healthy donors, obtained after informed consent, using Ficoll density gradients. Monocytes were subsequently obtained by plate adhesion and differentiated into moDCs in 6 d in RPMI 1640 (Life Technologies) containing 10% FCS, glutamine, and antimycotics/antibiotics supplemented with IL-4 (300 U/ml) and GM-CSF (450 U/ml). For maturation, moDCs were cultured for an additional 2 d in the above medium supplemented with IL-1β plus IL-4 (300 U/ml) and GM-CSF (450 U/ml) or, alternatively, with IL-1β (5 ng/ml), IL-6 (15 ng/ml), TNF-α (10 ng/ml), and PGE₂ (10 μg/ml) (cytokine mixture), or LPS (1 μg/ml), or a combination of R848 (4 μg/ml; Alexis Biochemicals) and polyinosinic-polycytidylic acid (poly I:C; 20 μg/ml; Sigma-Aldrich), where indicated. Human myDCs were isolated from PBMCs using a MACS CD1c (BDCA-1) Dendritic Cell Isolation Kit and human pDCs were isolated from PBMCs using a MACS BDCA-4 Plasmacytoid Dendritic Cell Isolation Kit (both from Miltenyi Biotec). For maturation, myDCs and pDCs were cultured overnight in X-Vivo medium (Lonza) supplemented with 2% human serum and a combination of poly I:C (20 μg/ml) and R848 (4 μg/ml; myDCs) or with R848 alone (4 μg/ml; pDCs). Purity of isolated myDCs and pDCs was determined by staining for a combination of CD11c and CD1c (myDC) or for CD123 (pDCs; all Abs used were from BD Biosciences).

To generate bone marrow–derived DCs (BMDCs), bone marrow cells from mouse femurs were cultured in RPMI 1640 medium supplemented with 10% FCS, glutamine, antimycotics/antibiotics, and 2-ME in the presence of 20 ng/ml GM-CSF (PeproTech) for 8 d to generate GM-CSF BMDCs or in the presence of 5 ng/ml GM-CSF and 200 ng/ml Flt3L (CellGenix) for 15 d to generate Flt3L BMDCs (23).

Primary CD11c⁺ DCs were obtained from mouse spleens. Spleens were excised and injected with collagenase type II (Worthington) and DNase (Roche) for 45 min at 37°C. After addition of 10 mM EDTA and resuspending, cells were applied to nylon mesh to remove debris, and DCs were enriched by a 15/11.5/0% gradient of iodixanol in HBSS (prepared from 60% iodixanol OptiPrep stock; Sigma-Aldrich). For maturation, GM-CSF BMDCs and CD11c⁺ splenocytes were cultured overnight with a mixture of TLR ligands containing LPS (0.5 μg/ml; Sigma-Aldrich), poly I:C (10 μg/ml), and CpG 1668 (6.4 ng/ml; Sigma-Aldrich). Flt3L BMDCs were matured overnight with LPS alone (0.1 μg/ml).

Day-6 moDCs were electroporated (5–10 × 10⁶ cells) with 15 μM nontargeting (NT) small interfering RNA (siRNA; Dharmacon) or 15 μM SMARTpool of three Sema7A-targeting siRNAs (Stealth RNAi siRNA; Life Technologies; H55112366, H55112367, and H55112368), using the exponential program at 300 V, resistance at ∞, and capacitance at 150 μF in a 4-mm cuvette in the Bio-Rad Gene Pulser Xcell electroporator, and subsequently cultured for 2 d in the presence or absence of maturation stimuli, as indicated.

Human cells were stained with PBS containing 0.5% BSA with the following Abs: CCR7 (clone 150503, R&D Systems), CD83 (Immunotech), and Sema7A (clone MEM-150; Exbio). Murine cells were stained with PBS containing 0.5% BSA with the following Abs: CD86 (clone Gl-1), CD11c (clone N418), CD4 (clone RM4-5), and F4/80 (clone A3-1; all from BioLegend); CD3 (clone 145-2C11) and B220 (clone RA3-6B2; both from BD Biosciences); and CD8 (clone 53-6.7; Exbio). The presence of Sema7A protein on DCs of wild-type (WT) littermates and its absence on DCs from Sema7A^−/− mice were assessed using anti-murine Semaphorin7A (AF1835; R&D Systems). Cells were analyzed on a FACSCalibur (Becton Dickinson) using FlowJo software.

Proteins were separated by SDS-PAGE, transferred to Immobilon-P membranes (Millipore, Bedford, MA), and immunolabeled according to standard Western blotting procedures. Western blots were scanned using an Odyssey imager (LI-COR Biosciences). Goat anti-human Sema7A and rabbit anti-actin (clone 20-33; both from R&D Systems) Abs were used for Western blotting.

A total of 1 × 10⁵ allogeneic PBLs isolated from buffy coats were cultured overnight in the presence of DCs at a 1:5 or 1:25 ratio in 200 μl RPMI + 10% FCS. PBLs were subsequently stained for CD3 and CD69 to identify activated T cells and analyzed by flow cytometry.

Bovine type I collagen (55.5% [v/v]; stock concentration 3.1 mg/ml; final concentration 1.72 mg/ml [for moDCs] or 1.95 mg/ml [for BMDCs]; PureCol, Advanced BioMatrix) was turned into fibrillar collagen matrices by raising the pH to 7.4 using 3.7% (v/v) 0.75% Na-bicarbonate solution (Life Technologies), together with 7.4% (v/v) minimum essential Eagle’s medium (Sigma-Aldrich) and 33.3% (v/v) RPMI 1640 containing 10% FCS, and polymerizing at 37°C.

For Transwell-migration assays, a detached cytokine mixture of 5 × 10⁴–1 × 10⁵ matured moDCs was applied in 100 μl RPMI 1640 containing 10% FCS to 6.5-mm diameter Transwell inserts that were separated from the lower chamber by polycarbonate membranes containing 5-μm pores (Costar). The lower compartments were filled with RPMI 1640 containing 10% FCS and, when indicated, 1 μg/ml CCL21 (R&D Systems). Inserts were coated with 75 μl collagen matrix, as indicated, prior to applying the cells. Cells were allowed to migrate through the bottom of the chamber for 1–2 h (without collagen), for 4–5 h (BMDCs with collagen), or overnight (moDCs with collagen). The amount of transmigrated cells relative to input was measured by flow cytometry for a set time period, while keeping volumes equal between conditions.

For plate-adhesion assays, flat-bottom 96-well tissue culture–treated plates were coated with fibronectin (FN; Roche) or GRGDS peptides (AnaSpec) at 20 μg/ml and, when indicated, subsequently with 1 μg/ml CCL21 (R&D Systems). Cells were allowed to attach to the culture plates in RPMI 1640 with 10% FCS for 5 min, and nonadherent cells were removed by extensive washing. Adherent cells were subsequently detached by PBS containing 2 mM EDTA at room temperature, and the number of adherent cells relative to input was measured by flow cytometry for a set time-window while keeping volumes equal between conditions.

Force measurements on living mature DCs were performed in force-distance mode using a BioScope Catalyst AFM (Bruker, Santa Barbara, CA) mounted on an inverted microscope (TCS SP5 II; Leica, Mannheim, Germany). The temperature was kept at 37°C throughout the experiments by a petri dish heater (Bruker). Polystyrene microspheres (10 μm diameter; Polysciences) were glued onto atomic force microscopy (AFM) cantilevers (NP-S type D; Bruker) (24). Bead-functionalized cantilevers were cleaned by immersion in 2% Hellmanex III (Hellma, Müllheim, Germany) overnight and thoroughly rinsed with Milli-Q water and ethanol, and then (after a final rinse in Milli-Q water) were allowed to dry. Before experiments, cantilevers were incubated at 37°C in FN (20 μg/ml) in PBS. The spring constant of each cantilever was calibrated before adhesion measurements using thermal noise analysis (25). To measure the adhesion of a single DC to the FN-coated bead on the cantilever, siRNA-treated mDCs were immobilized on ConA-coated WillCo dishes (WillCo, Amsterdam, The Netherlands) and submerged in culture medium supplemented with 10 mM HEPES containing 10% FBS and 0.5% antibiotics/antimycotics (pH 7.4). ConA-coated WillCo dishes were prepared by overnight incubation at 4°C in ConA (Sigma-Aldrich; 0.2 mg/ml) in PBS; before seeding mDCs, dishes were rinsed with PBS. Adhesion of the FN-coated bead to different cells was measured by pushing the bead into contact with flat parts of the DCs and applying a 2-nN contact force for 10 s. Subsequently, the cell was retracted at 10 μm/s and allowed to recover for a time period equal to that during which the cell was in contact with the substrate before repeating adhesion measurements (26). Five measurements were performed per cell and used for subsequent calculation. Detachment forces were determined after baseline correction of F-D curves with a macro in Matlab (27). Force-step analysis and cell elasticity were determined from F-D curves using in-house Igor Pro 6 (WaveMetrics) algorithms (28). The cell cortex elasticity was determined using retraction F-D curves by linear fitting the region when adhesion built up to the maximum force (force-range 0 nN to F_max, Fig. 7C).

To monitor the appearance of moDCs after electroporation with siRNA, mature moDC cell cultures were directly imaged on a Leica DM IL inverted microscope using a 20× air objective (NA 0.4) and a Leica DC300 camera.

For confocal and time-lapse microscopy analysis, DCs (1 × 10⁶ cells/ml RPMI 1640) were embedded into collagen matrices and monitored by confocal microscopy or time-lapse microscopy experiments 1–2 h thereafter.

For time-lapse microcopy, a suspension of DCs in collagen was loaded in wax chambers (generated from cover glasses hermetically sealed with paraffin), which, after collagen polymerization, was overlaid with medium containing 1 μg/ml CCL21 (R&D Systems). Wax chambers were sealed to prevent a decrease in pH due to air exposure and were kept at 37°C while migration was monitored with a step interval of 2 min by time-lapse bright-field video microscopy using a 20× air objective (NA 0.3) on a bright-field inverted microscope (Leica) and a CCD camera (Sentech). A 16-channel frame grabber software (Vistek) was used for image acquisition. Cell tracks of randomly selected cells were followed for 3 h, beginning 30 min after the start of imaging, using the manual tracking and chemotaxis plug-ins that are incorporated into FIJI software. The speed was calculated as the length of each cell path/time, and xy trajectories were converted into mean square displacement (MSD) using the following equation:

where δt is the time interval and x(t) and y(t) are the position in x and y at time t. The MSD was calculated as a summation over time, for different time intervals.

For confocal analyses, DCs were overlaid with RPMI 1640 medium containing 10% FCS, with or without 1 μg/ml CCL21, and incubated for the indicated times, after which cells were fixed in 4% paraformaldehyde for 1 h, washed/permeabilized (PBS 1% BSA and 0.1% saponin), and labeled with Alexa Fluor 488–conjugated phalloidin (Invitrogen). Cells within the matrices were imaged using an Olympus FV1000 Confocal Laser Scanning Microscope with a 40× water objective (NA 0.8) and an Olympus U-CMAD3 camera. Images were acquired using Olympus imaging software and subsequently analyzed with FIJI software. Cell dendricity (1 − circularity) was measured for individual cells after applying a mask, as shown in Fig. 6A. For circularity measurement, FIJI software applies the following formula: circularity = 4π(area/perimeter²).

Sema7A^−/− mice were generated and genotyped previously (16). Sema7A^+/+ littermates (referred to as WT mice) or Sema7A^+/− littermates (referred to as heterozygous [HZ]) animals) were matched for age and gender and used between 9 and 14 wk of age. Mice were housed in top-filter cages and fed a standard diet with freely available water and food. All in vivo studies complied with national legislation and were approved by local authorities for the care and use of animals with related codes of practice. Animal studies were approved by the Animal Ethics Committee of the Nijmegen Animal Experiments Committee (DEC 2013-158).

In vivo DC migration was studied by FITC-painting experiments, as described previously (29). Briefly, abdominal skin was shaved prior to applying a FITC solution. FITC solution was prepared by diluting a 10% (w/v) solution of fluorescein-5-isothiocyanate in DMSO to a 1% final concentration with a 1:1 mixture of acetone and dibutyl phthalate. Draining (inguinal) and nondraining (brachial) LNs were taken after 24 h, and LN cells were isolated by DNase and collagenase digestion. LN cells were stained for FITC, CD11c, and CD86 to identify migrated DCs and analyzed by flow cytometry.

Ear emigration assays were performed as described previously (29). Briefly, dorsal and ventral skin sheets were placed in medium containing 1 μg/ml LPS alone or LPS and 1 μg/ml CCL21 and cultured for 48 h. Emigrated DCs were quantified using flow cytometry.

Mouse lymphoid organs were taken from Sema7A^−/− mice and WT Sema7A^+/+ littermates and embedded in OCT. Frozen sections (5 μm) were fixed in acetone for 10 min at −20°C. Sections were blocked with 5% goat serum and stained with anti-CD3 (clone CD3-12; Serotec), anti-B220 (clone RA3-6B2; BD Biosciences), anti-CD11c (clone N418; eBioscience), or CD68 (clone MCA1957; Serotec), followed by biotin-conjugated secondary Abs (Invitrogen). Staining was revealed using an SA-alkaline phosphatase labeling kit (Vector Laboratories) with Fast Red substrate (Sigma-Aldrich), and slides were counterstained with Mayer’s hematoxylin and embedded in Kaiser’s glycerol gelatin (Merck). Slides were imaged on a Leica DM LB microscope using a 10× air objective (NA 0.4) and a Leica DC300 camera.

Statistical analyses were performed in GraphPad Prism 5 software using the appropriate testing methods, as indicated in the figure legends.

Consistent with our previous proteome analysis (22), we found that Sema7A was strongly expressed on mature, but not resting, moDCs using both flow cytometry and Western blotting (Fig. 1A–C). Expression could be triggered by a mixture of IL-1β, IL-6, TNF-α, and PGE₂, as well as by various TLR ligands (R848, poly I:C, LPS), indicating that Sema7A is a general maturation marker of human moDCs (Fig. 1D). Sema7A expression was induced relatively late during maturation and similar in timing to the chemokine receptor CCR7 (Fig. 1E) (30).

Next, we studied whether Sema7A was also upregulated during the maturation of primary DCs from humans and mice. Indeed, all DC subsets tested increased Sema7A expression upon maturation, albeit to a variable extent (Fig. 2). Both human myDCs and pDCs isolated from the blood of healthy donors expressed Sema7A when matured with TLR ligands (Fig. 2A, 2B). In mice, most mature primary CD11c⁺ DCs derived from spleen, as well as the majority of Flt3L BMDCs and some (10–20%) GM-CSF BMDCs, expressed Sema7A (Fig. 2C, 2D; for gating see Supplemental Fig. 1A). Taken together, we show that, similar to CCR7, Sema7A expression is a general feature of maturation in murine and human DCs.

Because Sema7A was demonstrated to regulate cell adhesion/migration in the neuronal system and in skin tumors (16, 17), we hypothesized that Sema7A on DCs may be involved in the regulation of adhesion and/or migration during DC maturation.

To this end, we investigated the functional consequences for DC migration in vivo using Sema7A-knockout (Sema7A^−/−) mice DCs. We first analyzed the composition and architecture of secondary lymphoid organs in these mice. The architecture of the LNs (Fig. 3A) and spleen (data not shown) appeared unaltered with respect to B cell follicles, T cell zones, and localization of CD11c⁺ cells (DCs) in the T cell areas, as well as the presence of CD68⁺ cells (macrophages) throughout the LN. Importantly, there were no major differences between the total number (data not shown) and relative amounts of lymphoid and myeloid cells between Sema7A^−/− mice and their WT (Sema7A^+/+) littermates (Supplemental Fig. 2).

To test whether migration of DCs to the LN in vivo was affected by the absence of Sema7A, we performed FITC skin-painting experiments in Sema7A^−/− mice, as well as in WT and HZ (Sema7A^+/−) controls. Skin DCs in WT and HZ littermates readily migrated to the LNs and represented ∼50% of CD11c⁺ DCs in the inguinal LNs after 24 h (Fig. 3B; for gating see Supplemental Fig. 1B). In Sema7A^−/− mice, DC migration to the LNs was impaired, because only 40% of CD11c⁺ DCs in the LNs originated from the skin after 24 h. This reduction was not significant because of the high experimental variation, especially in Sema7A^−/− mice. This high variation was likely caused by the complexity of the migration process in vivo, which is composed of several sequential steps (e.g., peripheral detachment, migration through the lymphatics, LN invasion), each of which may be affected differentially by the absence of Sema7A. To further elucidate how Sema7A affects DC migration, we exploited the ear emigration model in which separated skin sheets of WT or Sema7A^−/− mice were exposed to LPS, with or without CCL21, to induce skin DC migration. DCs from both WT and Sema7A^−/− mice readily migrated out of the skin when exposed to LPS alone, demonstrating that maturation, basal migration, and peripheral detachment were not impaired by Sema7A deficiency (Fig. 3C; for gating see Supplemental Fig. 1C). Emigrating DCs also expressed high levels of Sema7A (Supplemental Fig. 1D). In the presence of CCL21, DC emigration from WT skin was clearly enhanced (Fig. 3C). In contrast, DC emigration from Sema7A^−/− mice was severely hampered upon CCL21 addition, demonstrating that Sema7A acts specifically on chemokine-induced DC migration ex vivo.

To investigate Sema7A function on human DCs, we used a mixture of three siRNAs to knockdown (KD) Sema7A expression in moDCs. Expression of Sema7A was almost completely inhibited by this approach (Fig. 4). Importantly, KD DCs were normal with respect to upregulation of maturation markers (CCR7, CD83), their morphology in culture, and their ability to stimulate allogeneic T cells (Fig. 4). Although the siRNA electroporation procedure itself reduced the number of matured DCs from 90% to ∼60%, this effect was not different between the NT siRNA–treated DCs and the Sema7A siRNA–treated DCs (Supplemental Fig. 3).

Having established efficient Sema7A KD, we investigated the effect of Sema7A KD on CCL21-induced migration of mature DCs using the Transwell system in the absence or presence of a collagen matrix. CCL21 triggered DC migration through membranes in control and KD DCs (Fig. 5A); however, migration through collagen was significantly impaired in DCs lacking Sema7A (Fig. 5B). Likewise, Flt3L BMDCs generated from Sema7A^−/− mice migrated less well through a collagen matrix (Supplemental Fig. 4A, 4B). Thus, in both human and murine DCs, the presence of Sema7A appeared most important for DC migration in a more complex environment where DCs encounter an abundance of integrin ligands, as well as in confined spaces. The migration defect could not be attributed to aberrant CCR7 expression (Fig. 4C) or a decreased ability of KD DCs to enter the collagen matrix because microscopic inspection demonstrated that KD DCs readily entered the matrix (data not shown). The migration defect was cell intrinsic (rather than mediated via soluble Sema7A or cell–cell contact), because the addition of an excess of Sema7A-expressing WT cells to the experiment could not rescue the migration defect of Sema7A-KD cells (data not shown).

To further investigate the mechanism underlying the migration defect of Sema7A-KD cells, we first studied cell morphology and the actin cytoskeleton (F-actin labeling) during CCL21-induced migration in a three-dimensional environment. Within the collagen matrix, Sema7A-KD DCs appeared less dendritic (Fig. 6A, 6B). Upon addition of CCL21, NT DCs greatly increased the number of dendrites (Fig. 6A, 6B). In contrast, Sema7A-KD cells remained impaired with regard to dendrite formation compared with NT DCs, indicating that DCs require Sema7A to efficiently assemble actin-based protrusions. Experiments with Sema7A^−/− murine Flt3L BMDCs showed a similar defect in dendrite formation and response to CCL21, although the responses were rather heterogeneous (Supplemental Fig. 4C, 4D).

Because actin polymerization regulates three-dimensional migration speed in DCs (7), we next used live cell microscopy to characterize the movement of individual DCs through collagen (Supplemental Videos 1, 2). DCs were incorporated into the collagen prior to matrix formation to abrogate any two-dimensional–dependent DC migration effects. In the absence of CCL21, migration speed and MSD of Sema7A-KD DCs were similar to those of NT DCs (Fig. 6C–E). However, in the presence of CCL21, KD DCs clearly migrated less far, and the MSD was 2.4-fold lower (p < 0.05) over the course of 3 h (Fig. 6C, 6D). The migration speed of NT DCs clearly increased upon addition of CCL21 in contrast to Sema7A-KD cells, which did not migrate faster in response to CCL21 (Fig. 6E). Sema7A-KD DCs did not appear to stop more frequently or to remain completely immobile. Instead they moved at a lower overall speed throughout the tracking time (data not shown). Taken together, these data demonstrate that, in the absence of Sema7A, DCs are less protrusive and migrate more slowly in a three-dimensional environment in response to CCL21.

Although DCs do not require adhesion for migration, especially in three-dimensional matrices (7), alterations in their adhesion capacity may still affect their migration speed when integrin ligands are present (e.g., by impairing cell detachment). Therefore, we investigated whether Sema7A KD affected the adhesive properties of DCs.

Initial plate-adhesion experiments suggested that Sema7A-KD DCs were slightly more adherent to extracellular matrix components than were NT DCs (Fig. 7A). However, experimental variation was high; therefore, we exploited AFM to decipher the effect of Sema7A on the individual factors governing DC adhesion by applying a mechanical force on extracellular matrix–binding adhesion receptors using an FN-coated bead. The bead was attached to the AFM cantilever that was subsequently brought into contact with a flat area of the DC surface and then retracted from the DC surface to break the adhesive bonds one by one (Fig. 7B). Analysis of obtained AFM-derived force-distance curves allowed us to distinguish effects on avidity (i.e., the binding strength exerted by the FN-binding receptors together), affinity (i.e., the binding strength/adhesion molecule), and cellular membrane elasticity (i.e., the amount of cytoskeletal support) (Fig. 7C–F) (28). Using AFM, we observed that DCs lacking Sema7A were indeed more adhesive, in line with the plate-adhesion experiments (Fig. 7F: adhesion strength). This was not caused by an increased affinity of single receptors for their ligands, because the distribution of rupture forces required to break single molecule bonds was not altered (Fig. 7E). Interestingly, the linear elasticity (i.e., the force required to pull out part of the membrane adhered to the FN-coated bead) of Sema7A-KD DCs was significantly lower compared with control DCs, indicating a reduced connection of membrane tethers to the cortical actin cytoskeleton in Sema7A-KD DCs (Fig. 7G). These results indicate that Sema7A-KD DCs are more adhesive and show an impairment of the actin cytoskeleton, similar to what we observed for intracollagen protrusion formation. Together, our data demonstrate that Sema7A promotes DC migration by increasing the connectivity of adhesion receptors to the actin cytoskeleton to support actin-based protrusion assembly.

Activated DCs are equipped with a complex molecular machinery to govern migration from peripheral tissues to the LNs. We identify Sema7A as an essential membrane receptor for chemokine-induced DC migration in vitro and ex vivo. The function of semaphorin proteins is mostly known from the neuronal system, where they regulate growth and branching of neuronal processes (11). In the immune system, Sema7A is implicated in the activation of T cells, monocytes, and macrophages, as well as in neutrophil migration (18, 20, 21, 31), although only in the case of T cells was the effect immune cell intrinsic (i.e., Sema7A is functional on the T cells themselves). We now report that Sema7A expressed on activated human and murine DCs controls adhesion and migration in response to CCL21. Sema7A was not required for DC maturation or induction of allogeneic T cell responses. Importantly, in mice lacking Sema7A, the migration of skin DCs to the LN was reduced, and CCL21-induced DC emigration from ear explants was severely impaired. The finding that Sema7A deficiency specifically affected CCL21-driven migration implies that Sema7A has an important role in guiding DCs from peripheral tissues into the lymphatic vessels and in intranodal DC mobility, processes that are known to be controlled by CCL21 (6, 32).

Sema3A produced by the lymphatics was described to support DC migration by acting on the cytoskeleton upon binding to plexin A1 expressed by DCs (33). Furthermore, the secreted viral semaphorins from pox and herpes viruses, which harbor some degree of homology to Sema7A, were demonstrated to affect DC adhesion, migration, and phagocytosis by acting on the cytoskeleton through plexin C1 (34, 35).

Our experiments revealed that Sema7A acts cell autonomously on DC adhesion and protrusion formation. Because Sema7A lacks a cytoplasmic tail, we anticipate that it alters the intracellular signaling cascades responsible for regulating CCL21-driven adhesion and migration through transmembrane-binding partners. Both of the previously described interaction partners of Sema7A, plexin C1 and β₁ integrins, are expressed on mature DCs and, thus, can potentially bind Sema7A in cis (3, 36). Interestingly, DC migration to the LN upon FITC painting was reduced in plexin C1–knockout mice to the same extent as in Sema7A-knockout mice, suggesting that these two molecules could act in concert (36). Soluble recombinant Sema7A-Fc was unable to bind to Sema7A-KD DCs (data not shown), indicating that an interaction partner for soluble Sema7A-Fc is not readily available, irrespective of the presence of endogenous Sema7A that may occupy potential binding sites. Concordantly, Sema7A-Fc could not rescue the defective migration of KD DCs. Thus, it is possible that, on DCs, Sema7A binds its interaction partner only in its GPI-anchored form in cis but not in a soluble form. Such in cis interactions with binding partners were reported for other semaphorin family members (37).

We observed that DCs lacking Sema7A are more adhesive, lack cytoskeletal support of adhesion receptors, form fewer actin-rich protrusions, and are impaired in CCL21-induced migration. Thus, Sema7A on DCs acts by controlling the balance between adhesion/deadhesion, similar to Sema7A function on tumor cells and melanocytes (15, 17, 38). A prominent role for the actin-binding protein cofilin was shown in these studies. Cofilin regulates cell migration downstream of plexin C1, integrins, and CCR7 (15, 39), making it a possible candidate for controlling Sema7A-dependent DC migration. Experiments to unravel a role for cofilin in our studies were hampered by the low number of DCs available for biochemical experiments, as well as by the unconventional mechanisms regulating cofilin activity (40).

CCL21 was shown to facilitate adhesive DC migration by acting on adhesion- and actin-driven migration speed (6, 41, 42). We now demonstrate that Sema7A facilitates CCL21-dependent migration in a three-dimensional environment. Sema7A-KD DCs proved to be defective in actin protrusion formation and displayed an increased adhesion strength, which, especially in the absence of cytoskeletal support of adhesion receptors, slows migration as a result of a lack of force coupling. Importantly, irrespective of ligand engagement of adhesion receptors, the connection between DC membrane proteins and the cytoskeleton facilitates both confinement and adhesion-driven migration (7, 8). In conclusion, our data demonstrate that Sema7A promotes DC migration by enhancing actin-based protrusion formation. In this study, we identified an important novel receptor in the highly complex process of DC migration to the LNs that governs the initiation of adaptive immune responses.

We thank Geert van den Bogaart for discussion and assistance with the analysis of live microscopy data, Youri Adolfs for help with animal experiments, the Radboud Institute for Molecular Life Sciences Department of Cell Biology for use of live cell imaging facilities, and Alessandra Cambi for critical reading of the manuscript.

The authors have no financial conflicts of interest.