Research ArticleDENDRITIC CELLS

Lysosome signaling controls the migration of dendritic cells

See allHide authors and affiliations

Science Immunology  27 Oct 2017:
Vol. 2, Issue 16, eaak9573
DOI: 10.1126/sciimmunol.aak9573

Lysosomal calcium powers dendritic cell migration

Activation of dendritic cells (DCs) by stimulus such as bacterial sensing promotes DC maturation. In contrast to random migratory patterns of immature DCs, mature DCs migrate in a continuous and directional manner. Here, Bretou et al. have examined the role of lysosomes in regulating migration of mature DCs. They report that release of lysosomal calcium connects DC activation with directional migration of DCs. Deletion of the lysosomal calcium channel transient receptor potential cation channel, mucolipin subfamily, member 1 (TRPML1) impaired directional DC migration and DC chemotaxis to lymph nodes. The authors propose the release of lysosomal calcium to be a key node that links stimulus sensing with DC migration.

Abstract

Dendritic cells (DCs) patrol their environment by linking antigen acquisition by macropinocytosis to cell locomotion. DC activation upon bacterial sensing inhibits macropinocytosis and increases DC migration, thus promoting the arrival of DCs to lymph nodes for antigen presentation to T cells. The signaling events that trigger such changes are not fully understood. We show that lysosome signaling plays a critical role in this process. Upon bacterial sensing, lysosomal calcium is released by the ionic channel TRPML1 (transient receptor potential cation channel, mucolipin subfamily, member 1), which activates the actin-based motor protein myosin II at the cell rear, promoting fast and directional migration. Lysosomal calcium further induces the activation of the transcription factor EB (TFEB), which translocates to the nucleus to maintain TRPML1 expression. We found that the TRPML1-TFEB axis results from the down-regulation of macropinocytosis after bacterial sensing by DCs. Lysosomal signaling therefore emerges as a hitherto unexpected link between macropinocytosis, actomyosin cytoskeleton organization, and DC migration.

INTRODUCTION

Dendritic cells (DCs) play a key role in the initiation of adaptive immune responses. Immature DCs patrol the tissues and check for the presence of danger-associated antigens by continuously internalizing extracellular material via macropinocytosis (1). The sensing of a danger signal triggers DC maturation, reduces antigen uptake, and up-regulates the surface expression of CCR7. This chemokine receptor recognizes CCL21 and CCL19 gradients and drives DCs to lymphatic vessels (LVs) and lymph nodes (LNs), where the DCs present antigen to (and thus activate) T cells (2, 3).

As DCs mature, their migratory behavior changes (4). Whereas immature DCs display random-like migration that alternates between fast and slow phases, mature DCs migrate in a continuous, directional manner (5). This difference results from changes in the regulation of actin nucleation after microbial sensing. The migration of both immature and mature DCs involves actin nucleation at the cell rear by the formin mDia1 (mammalian diaphanous 1), which couples to myosin II contractility (6 ). However, in immature DCs, a second actin pool is nucleated by the Arp2/3 (actin-related protein 2/3) complex at the cell front for macropinosome formation, which periodically reduces the cell’s migration speed and directionality. Hence, down-regulation of Arp2/3 activity upon DC maturation concomitantly down-regulates macropinocytosis and increases the migration speed (5). However, the nature of the signals that control these changes in actin nucleation upon microbial sensing has yet to be identified.

Antigens taken up by macropinocytosis are internalized in endocytic compartments, which then fuse with lysosomes (7 ). Exposure to the lysosome’s proteolytic environment facilitates antigen processing into peptides for loading onto major histocompatibility complex class II (MHC II) molecules (7, 8). Accordingly, it was recently reported that sensing the bacterial product lipopolysaccharide (LPS) enhances the transcription of lysosomal genes (9). This phenomenon relies on the translocation of transcription factor EB (TFEB) from the cytosol to the nucleus (10). TFEB is a master regulator of lysosomal gene expression, and its nuclear translocation results in the expression of a network of genes involved in lysosome activity, biogenesis, and secretion (11, 12). In the steady state, TFEB is phosphorylated by the mammalian target of rapamycin complex 1 (mTORC1) and therefore remains in the cytosol (13). Translocation of TFEB to the nucleus can be triggered in response to nutrient deprivation, which inhibits mTORC1 activity (13). Nuclear import of TFEB also occurs after calcium efflux from lysosomes through the transient receptor potential cation channel, mucolipin subfamily, member 1 (TRPML1). The latter activates calcineurin, which, in turn, dephosphorylates TFEB (14). Given that the trpml1 promoter is a target of TFEB, it has been suggested that the mutual regulation between TRPML1 and TFEB establishes a positive feedback loop for maintenance of a high level of lysosomal activity (14). Inhibition of macropinocytosis in macrophages compromises the arrival of amino acids to lysosomes and thereby reduces mTORC1 activity (15). However, it is not known whether down-regulation of macropinocytosis also results in TFEB translocation to the nucleus.

Here, we investigated the role of lysosomal signaling in DC migration. We found that activation of the TFEB-TRPML1 axis is (i) sufficient for triggering fast DC migration and (ii) required for DC chemotaxis to the LNs. Lysosomal signaling is activated after microbial stimulation of DCs and the down-regulation of macropinocytosis, revealing a hitherto unexpected role for macropinocytosis in linking nutrient sensing to cell locomotion.

RESULTS

The lysosome-associated TFEB regulates DC migration

In DCs, endolysosomes are the most abundant endocytic compartments and constitute the main site for MHC II antigen processing (16 ). We have previously shown that cell migration is coupled to the activity of the antigen processing machinery (17). Hence, we hypothesized that endolysosomes might act as a signaling platform to control DC migration upon bacterial sensing. To test this hypothesis, we focused on TFEB, the master regulator of lysosome biogenesis and function (9, 11). First, we found that DCs derived from the bone marrow of TFEB–green fluorescent protein (GFP) knock-in mice exhibited elevated levels of nuclear TFEB after treatment with the microbial product LPS (Fig. 1, A and B). This translocation to the nucleus was also accompanied by up-regulation of the TFEB target gene TPP1 (fig. S1A). Nuclear localization of TFEB was also observed after the inhibition of mTORC1 with Torin 2 (Fig. 1C and fig. S1B) (13). This finding indicates that (i) TFEB’s localization in the cytoplasm relies on mTORC1 activity and (ii) LPS sensing promotes the translocation of functional TFEB to the nucleus in DCs, as recently reported (10).

Fig. 1 TFEB translocates to the nucleus and is required for fast DC migration.

(A) Immunofluorescence images of TFEB-GFP knock-in immature DCs (iDCs) and LPS-DCs. Red dotted circles show nuclei (×60 magnification; scale bar, 10 μm; one plane is shown). (B) The normalized TFEB nucleus-to-cytosol ratio from cells shown in (A) (n = 60 cells per condition, pooled from N = 4 independent experiments). (C) The normalized TFEB nucleus-to-cytosol ratio from immature DCs treated with Torin 2 (2.5 nM; n = 78 cells per condition, pooled from N = 2 independent experiments). (D to F) Analysis of BMDC migration in 4 μm–by–5 μm fibronectin-coated microchannels. Cells were imaged between 6 and 16 hours after LPS treatment (100 ng ml−1 for 30 min). (D) Sequential images of wild-type (WT) and TFEB-inducible KO LPS-DCs migrating in microchannels (×10 magnification; one image every 2 min; scale bar, 50 μm). (E) Mean instantaneous velocities of cells shown in (D) (from left to right: n = 57, 68, 77, and 54 cells; representative results from one of two experiments are shown). ns, not significant. (F) Mean instantaneous velocities of BMDCs treated with DMSO or Torin 2 (2.5 nM; n = 200 cells per condition; representative results from one of four experiments are shown). (B and C) For each experiment, data were normalized against the mean TFEB nucleus-to-cytosol ratio observed in immature DCs (A) or DMSO-treated immature DCs (B). Data are quoted as means, and error bars correspond to SEM. (E and F) Data are represented as boxes and whiskers; bars include 90% of the points, the line represents the median, and the box contains 75% of the data. Mann-Whitney test was applied.

We next sought to determine whether TFEB influences DC migration. Using microfabricated channels that mimic the confined geometry of the interstitial space in tissues (17, 18), we found that LPS-treated DCs (referred to hereafter as LPS-DCs) increased their migration speed (Fig. 1, D and E), as previously reported (5). This speed increase was not observed with DCs from a tamoxifen-inducible TFEB knockout (KO) mouse (referred to hereafter as TFEB KO DCs) or with TFEB-silenced DCs (Fig. 1, D and E, and fig. S1, C to E), showing that TFEB is required for the LPS-triggered switch to fast DC migration. The TFEB KO DCs expressed normal surface levels of the CD86 activation marker upon LPS treatment (fig. S1, F and G). LPS-induced lysosome clustering (19) was less pronounced in TFEB KO DCs than in their wild-type mature counterparts (fig. S1, H and I). Treatment of immature DCs with the mTORC1 inhibitor Torin 2 increased their migration speed (up to the values seen in LPS-DCs) but did not change the surface expression of CD86 (Fig. 1F and fig. S1J). We conclude that the induction of fast DC migration upon LPS sensing requires the master regulator of lysosomal gene transcription TFEB.

The TFEB target gene TRPML1 controls DC chemotaxis and in vivo migration

We next sought to determine whether TFEB’s target genes control DC migration. The lysosomal calcium channel TRPML1 was a likely candidate because (i) expression of its gene was up-regulated in LPS-DCs in a TFEB-dependent manner (fig. S2A) (11, 20, 21) and (ii) intracellular calcium levels are known to regulate cell motility (22, 23). The lysosomal localization of TRPML1 in mature DCs was probed by ectopic expression of TRPML1-GFP (24) because cognate antibodies are not available for immunofluorescence analyses (Fig. 2A). The observation of enlarged lysosomes in Trpml1−/− DCs was consistent with reports on other cell types (25) and with the mucolipidosis type IV phenotype in patients carrying mutations in the trpml1 gene (26). Although the lysosomes were tightly clustered at the rear of wild-type DCs, they were more dispersed in Trpml1−/− cells (fig. S2, B and C). To investigate the dynamics of lysosome localization during DC migration, we labeled them with fluorescent wheat germ agglutinin (WGA). This label was found to colocalize with the Lamp1 marker (fig. S2D). We found that the fraction of lysosomes localized at the rear of migrating Trpml1+/+ DCs increased slightly upon maturation (fig. S2E). No such increase was observed in Trpml1−/− cells, suggesting that lysosomal calcium promotes lysosome clustering at the rear of LPS-DCs.

Fig. 2 TRPML1 is required for fast DC migration, chemotaxis, and arrival at the LNs.

(A) Localization of TRPML1-GFP (green), Lamp1 (red), and the nucleus (N) (DAPI; blue) in transfected, migrating LPS-DCs (8 μm–by–5 μm fibronectin-coated microchannels; spinning disk microscope, ×100 magnification; scale bar, 5 μm; one plane is shown). (B) Mean instantaneous velocities of BMDCs migrating in 4 μm–by–5 μm fibronectin-coated microchannels (n = 240 cells per condition; representative results from one of three experiments are shown). (C to F) Chemotaxis of LPS-DCs migrating in collagen gels (representative results from one of three experiments are shown). (C) Directionality of LPS-DCs migrating in the presence or absence of CCL21 (n = 455 and 532 cells for Trpml1+/+ and Trpml1−/− LPS-DCs with CCL21, respectively; n = 533 and 394 cells for Trpml1+/+ and Trpml1−/− LPS-DCs without CCL21, respectively). (D) Trajectories of LPS-DCs undergoing chemotaxis. The starting point of each trajectory was translated to the origin of the plot (n = 50 random trajectories per condition). (E) The mean instantaneous speed measured from the trajectories obtained in (C). (F) The mean square displacement (MSD) of LPS-DCs, measured from the trajectories obtained in (C). (G) In vivo migration of LPS-DC to LNs. DCs were stained with carboxyfluorescein diacetate succinimidyl ester or orange [5-(and-6)-(4-chloromethyl(benzoyl)amino]tetramethylrhodamine and co-injected into the footpad of wild-type recipients; 16 hours after the injection, flow cytometry was used to assess the proportion of DCs that had reached the popliteal LNs. An example of a fluorescence-activated cell sorting (FACS) dot plot is shown. (H) LN homing index for LPS-DCs (n = 6 mice per condition, pooled from two independent experiments). (I and J) Analysis of mDC populations in mouse inguinal LNs at steady state. (I) Frequency of mDCs as a proportion of live cells (n = 3 mice per condition; representative results from one of three experiments are shown). (J) Frequency of mDC2 cells as a proportion of live cells (n = 7 mice in total, from three independent experiments). (B and E) Data are represented as boxes and whiskers; bars include 90% of the points, the line represents the median, and the box contains 75% of the data. (H) Data are quoted as means, and error bars correspond to SEM. (B, H, and I) Mann-Whitney test was applied. (E) Welch two-sample t test was applied (P = 1.358 × 10−7). (J) Paired t test was applied.

As observed in TFEB-deficient DCs (i.e., TFEB KO or TFEB-silenced cells), LPS-treated Trpml1−/− DCs migrated more slowly (Fig. 2B). Furthermore, immature Trpml1−/− DCs also migrated more slowly than wild-type cells (which was not observed in TFEB-deficient DCs). This might be because neither tamoxifen-induced deletion of the tfeb gene nor TFEB silencing completely abrogated expression of this protein (fig. S1, C and D), whereas Trpml1−/− DCs were derived from total KO mice (27). Hence, TRPML1 (whose gene expression is controlled by TFEB) is required for fast DC migration.

We have previously shown that the fast DC migration triggered by microbial stimulation is required for sensing CCL21 gradients and reaching the LNs in vivo (5). Accordingly, Trpml1−/− mature DCs sensed haptotactic CCL21 gradients less efficiently than wild-type cells, although CCR7 surface expression was unaltered (Fig. 2C and fig. S2F). The trajectories exhibited by Trpml1−/− LPS-DCs along CCL21 gradients were less persistent, and their speed was relatively low (Fig. 2, D to F), indicating that TRPML1 is needed for chemotaxis. Accordingly, TRPML1-deficient mature DCs migrated less to the LNs when transferred into the footpad of wild-type mice (Fig. 2, G and H). Moreover, analysis of migratory DCs in inguinal LNs showed that Trpml1−/− mice had fewer cells from the dermal mDC2 subset (MHC IIhighCD11c+CD11b+CD103EpCAM DCs), indicating that the latter rely on TRPML1 to migrate from the periphery to LNs (Fig. 2, I and J, and fig. S2G) (28, 29). Together, these results show that the lysosomal calcium channel TRPML1 is required for fast DC motility, chemotaxis, and in vivo migration.

Calcium release by TRPML1 from lysosomes controls fast DC migration

Next, we used the TRPML1 activator 2-[2-(3,4-dihydro-2,2,4-trimethyl-1(2H)-quinolinyl)-2-oxoethyl]-1H-isoindole-1,3(2H)-dione (MLSA1) to establish whether the role of TRPML1 in DC locomotion depended on its calcium channel activity (30). Treatment of the calcium probe Fluo4-AM–loaded DCs with MLSA1 induced a calcium peak in the cytosol. This peak was lower in Trpml1−/− DCs, showing that MLSA1 is rather specific (Fig. 3, A and B). There were no differences in spontaneous calcium oscillations between wild-type and Trpml1−/− DCs (fig. S3, A to C). Treatment of immature DCs with MLSA1 was sufficient to increase their migration speed to the values reached by LPS-DCs (Fig. 3C and fig. S3D), although this increase was not seen in Trpml1−/− LPS-DCs. Treatment of LPS-DCs with MLSA1 did not enhance their speed, suggesting that TRPML1 is already fully activated in these cells (fig. S3D). Neither trpml1 gene deletion nor MLSA1 treatment modified the surface expression levels of CD86 (fig. S3, E and F). Hence, calcium release from lysosomes by TRPML1 triggers fast DC migration, as does LPS sensing.

Fig. 3 Fast DC migration relies on calcium release by TRPML1.

(A) A typical calcium response by Fluo4-AM–loaded Trpml1+/+ (black line) or Trpml1−/− (gray line) immature DCs. Cells were challenged with MLSA1 (10 μM) and thapsigargin (1 μM). Dotted lines indicate the treatment time. To prevent overlap between the two responses, we added an artificial offset (Trpml1+/+ signal + 100). (B) Quantification of calcium responses (data were pooled from N = 2 independent experiments with n = 29 and 30 cells for Trpml1+/+ and Trpml1−/− immature DCs, respectively). (C and D) Analysis of BMDC migration in 4 μm–by–5 μm fibronectin-coated microchannels. (C) Mean instantaneous velocities of DMSO- or MLSA1 (10 μM)–treated immature DCs (data were pooled from N = 2 independent experiments, with n = 316, 291, 206, and 163 cells from left to right). (D) Mean instantaneous velocities of DMSO- or Torin 2 (2.5 nM)–treated immature DCs (data were pooled from N = 2 independent experiments, with n = 215, 202, 201, and 166 cells from left to right). (E) Immunofluorescence images of TFEB-GFP knock-in control (siCtrl) or TRPML1 knockdown (siTRPML1-A) LPS-DCs (×60 magnification; scale bar, 10 μm; one plane is shown). (F) The normalized TFEB nucleus-to-cytosol ratio from cells shown in (E) (n = 40 cells per condition, pooled from N = 2 independent experiments). (G) Normalized TFEB nucleus-to-cytosol ratio from DMSO- or MLSA1-treated immature DCs (n = 60 cells per condition, pooled from N = 3 independent experiments). (B, F, and G) Data are quoted as means, and error bars correspond to SEM. (B) MLSA1 values were normalized according to the thapsigargin values, which were set to 100% in immature DCs. (F and G) For each experiment, data were normalized with respect to the mean TFEB nucleus-to-cytosol ratio obtained in siCtrl immature DCs (F) or DMSO-treated immature DCs (G). (C and D) Drugs (MLSA1 or Torin 2) were present during the recording. Data are represented as boxes and whiskers; bars include 90% of the points, the line represents the median, and the box contains 75% of the data. Mann-Whitney test was applied.

TFEB and TRPML1 are part of a positive feedback loop, within which TRPML1-mediated calcium release triggers the activation of calcineurin. The latter dephosphorylates TFEB, which results in TFEB nuclear translocation and sustained trpml1 transcription (14). We thus sought to determine whether this type of positive feedback loop operates in DCs. First, we found that the TFEB nuclear translocation induced by Torin 2 did not increase the migration speed of immature Trpml1−/− DCs (Fig. 3D and fig. S3G). Furthermore, TFEB translocation to the nucleus was impaired when TRPML1 was silenced in mature DCs (Fig. 3, E and F, and fig. S3H) but was induced upon TRPML1 activation in immature DCs (Fig. 3G and fig. S3I). These results suggest that the switch from slow to fast migration observed in DCs upon microbial sensing requires both TRPML1-dependent translocation of TFEB to the nucleus and TFEB-dependent TRPML1 transcription. We conclude that the TFEB-TRPML1 axis is activated upon microbial sensing and controls DC migration ex vivo and in vivo.

Lysosomal signaling regulates front-back F-actin distribution in migrating DCs

We next investigated the mechanisms downstream of lysosomal signaling that trigger fast DC migration. We have previously shown that LPS enhances the migration speed of DCs by regulating the nucleation and subcellular distribution of F-actin (5). Immature DCs have a large pool of F-actin at their front; this is nucleated by Arp2/3 and promotes antigen capture by macropinocytosis while reducing cell speed. In contrast, LPS-DCs display a predominant, cortical patch of F-actin at their rear, which is nucleated by the formin mDia1 and is required for cell motility (5). We observed that this pool of F-actin remained close to the lysosomes in migrating LPS-DCs (Fig. 4A and movie S1). Lysosome labeling with WGA did not alter the migration speed or maturation of DCs (figs. S2D and S4, A and B) (6 ).

Fig. 4 TFEB controls actin cytoskeleton reorganization in response to microbial sensing.

(A) LifeAct-GFP immature DCs and LPS-DCs migrating in 8 μm–by–5 μm fibronectin-coated microchannels. The middle (M) and cortical (C) planes were imaged. The inset (rectangle with a dashed boundary) highlights the proximity of the lysosomes (in red, labeled with WGA AF-647) to the F-actin patch observed in LPS-DCs. The red arrowhead indicates an actin cable that originates in the actin patch and extends toward the cell rear (spinning disk microscope, ×100 magnification; scale bars, 5 μm). (B) Sequential images of LifeAct-GFP siCtrl or TFEB-silenced (siTFEB) LPS-DCs (8 μm–by–5 μm fibronectin-coated channels; epifluorescence microscope, ×20 magnification; one image every 2 min; scale bars, 10 μm). (C) Mean LifeAct-GFP distribution in siCtrl or siTFEB DCs (from top to bottom, n = 40, 38, 29, and 35 cells; representative results from one of two experiments are shown). (D) Dynamic analysis of the fraction of time spent by the cells depicted in (C) with LifeAct-GFP at their front. (E) The mean F-actin front-back ratio. (D) Data are quoted as means, and error bars correspond to SEM. (E) Data are represented as boxes and whiskers; bars include 90% of the points, the line represents the median, and the box contains 75% of the data. Mann-Whitney test was applied.

We next looked at whether the TFEB-TRPML1 axis induces fast DC migration by facilitating the formation and/or maintenance of F-actin at the cell rear. TFEB-silenced LPS-DCs had abnormally low levels of F-actin at their rear but higher levels at their front (relative to control cell; Fig. 4, B and C; fig. S4C for map generation; and movie S2). Quantification of migrating cells using LifeAct-GFP density maps (5) confirmed this result on the population level; migrating LPS-DCs spent longer with F-actin at the cell front and had a higher front-back F-actin ratio when TFEB was silenced (Fig. 4, D and E). Furthermore, TFEB knockdown did not modify the distribution of F-actin in immature DCs (Fig. 4, C to E, and fig. S4D). Accordingly, forced translocation of TFEB to the nucleus (using Torin 2) resulted in an accumulation of F-actin at the cell rear (fig. S4E).

LifeAct-GFP–expressing TRPML1-deficient LPS-DCs had much the same phenotype as TFEB knockdown cells: The F-actin structure normally observed at the cell rear was less prominent, and F-actin remained at the cell front for a longer period, similar to immature DCs (Fig. 5, A to C; fig. S5, A to E; and movie S3). Accordingly, the treatment of immature DCs with MLSA1 shortened the period during which F-actin accumulated at the cell front (Fig. 5, D to F); this finding also agrees with the effect of MLSA1 on migration speed. MLSA1 had no effect on the distribution of F-actin in Trpml1−/− DCs (Fig. 5, E and F, and fig. S5F). The absence of TRPML1 had no effect on global lysosome localization (fig. S5G). We conclude that lysosomal signaling through the TRPML1-TFEB axis regulates DC migration through organization of the actin cytoskeleton.

Fig. 5 TRPML1 controls the organization of the DC actin cytoskeleton.

(A) Sequential images of LifeAct-GFP LPS-DCs (8 μm–by–5 μm fibronectin-coated channels; epifluorescence microscope, ×20 magnification; one image every 2 min; scale bars, 10 μm). (B) Mean LifeAct-GFP distribution of DCs (from top to bottom: n = 76, 97, 69, and 71 cells; representative results from one of three experiments are shown). (C) Fraction of time spent by the cells depicted in (B) with LifeAct-GFP at their front. (D) Cortical LifeAct-GFP signal obtained in DMSO- and MLSA1 (10 μM)–treated DCs (8 μm–by–5 μm fibronectin-coated channels; spinning disk microscope, ×100 magnification; scale bar, 5 μm). (E) Mean LifeAct-GFP distribution in MLSA1-treated cells (from top to bottom: n = 41, 42, 45, 63, 78, and 57 cells; representative results from one of two experiments are shown). (F) The fraction of time spent by the cells depicted in (D) with LifeAct-GFP at their front (n = 58 and 60 cells for Trpml1+/+ and Trpml1−/− MLSA1-treated LPS-DCs, respectively). (C and F) Data are quoted as means, and error bars correspond to SEM. Mann-Whitney test was applied.

The TRPML1-TFEB axis controls F-actin dynamics by promoting myosin IIA activity

To link the release of lysosomal calcium to reorganization of the actin cytoskeleton, we decided to focus on the actin-based motor protein myosin II. Calcium is known to regulate myosin II activity by activating myosin light-chain kinase, which, in turn, phosphorylates the light chain of myosin II (23, 31). Myosin IIA (the only isoform of the motor protein expressed in DCs; www.immgen.org) has also been shown to be essential for leukocyte migration in confined environments (6, 23). Time-lapse imaging of myosin IIA–GFP showed that it was enriched at the rear of LPS-DCs, where it was found in the vicinity of lysosomes (Fig. 6A and movie S4). Note that in TRPML1-deficient LPS-DCs, the amount of myosin IIA–GFP associated with lysosomes was abnormally low and the front-back ratio was abnormally high (Fig. 6, A to D; fig. S6, A and B; and movie S4). Hence, altered F-actin reorganization in Trpml1−/− DCs that have sensed LPS is linked to defective myosin IIA distribution.

Fig. 6 Calcium release through TRPML1 controls myosin IIA retrograde flow at the rear of LPS-DCs.

(A) Lysosomes are clustered in the vicinity of the myosin IIA–GFP pool located at the cell rear. The cortical plane of a migrating Trpml1+/+ LPS-DC (scale bar, 5 μm; lysosomes stained with WGA AF-647) is shown. (B) The normalized myosin IIA–GFP signal (myosin IIA–GFP concentration) measured in the region where lysosomes are located (data were pooled from N = 2 independent experiments, with n = 95, 81, 93, and 76 cells from left to right). (C) The mean myosin IIA–GFP distribution of Trpml1+/+ and Trpml1−/− BMDCs (from top to bottom: n = 74, 68, 58, and 44 cells; representative results from one of three experiments are shown). (D) The myosin IIA–GFP front-back ratio measured in the cells depicted in (C). (E) Cortical LifeAct-GFP signal obtained in DMSO- or para-nitroblebbistatin (p-blebb; 50 μM)–treated LPS-DCs (scale bars, 5 μm). (F) Left: Mean LifeAct-GFP distribution in LPS-DCs, DMSO- or para-nitroblebbistatin–treated cells (from top to bottom: n = 51 and 67 cells; representative results from one of three experiments are shown). Right: The fraction of time spent by these cells with LifeAct-GFP at their front. (G) Myosin IIA localizes on actin cables. Left: The overlay shows myosin IIA–GFP (green) and utrophin–red fluorescent protein (RFP; red) in an LPS-DC (scale bar, 5 μm; the cortical plane shown, and the nucleus indicated by white dashed lines). Right: The insets show sequential images of utrophin-RFP and myosin IIA–GFP (light blue rectangle). To highlight the myosin IIA retrograde flow without any contribution of cell velocity, all time points were aligned with respect to the cell rear (retrograde flow is indicated by white arrowheads; scale bars, 1 μm). Images were corrected for background and bleaching, and a median filter (2-pixel radius) was applied. (H) Myosin IIA–GFP at the cell rear undergoes retrograde flow, as indicated by red rectangles. Left: The cortical myosin IIA–GFP signal (C) obtained in a Trpml1+/+ LPS-DC is depicted; all time points were aligned as in (G) (an image was recorded every 400 ms; an image every 8 s is depicted; scale bar, 5 μm.) Right: Kymographs showing myosin IIA–GFP retrograde flow in migrating LPS-DCs. All time points were aligned as in (G). The tangent of the angle between the red dotted line (myosin IIA–GFP) and the y axis (cell rear) enables one to calculate the myosin IIA–GFP retrograde flow velocity (one image per 400 ms). (I) Myosin IIA–GFP retrograde flow in LPS-DCs (data were pooled from N = 2 to 4 independent experiments, with n = 14, 22, 11, and 12 cells from left to right). (J) Myosin IIA–GFP retrograde flow in immature DCs (n = 12 cells per condition, pooled from N = 2 independent experiments). (A to J) Experiments were carried out in 8 μm–by–5 μm fibronectin-coated microchannels; for (A), (E), (G), and (H), images were acquired on a spinning disk microscope (×100 magnification). (B) Data were normalized for each experiment with respect to the median ratio obtained in Trpml1+/+ immature DCs. (B and D) Data are represented as boxes and whiskers; bars include 90% of the points, the line represents the median, and the box contains 75% of the data. (F, I, and J) Data are quoted as means, and error bars correspond to SEM. (I and J) For each experiment, data were normalized against the mean retrograde flow for myosin IIA–GFP in immature DCs. Mann-Whitney test was applied.

We therefore hypothesized that the release of lysosomal calcium by TRPML1 might promote sustained myosin IIA activity at the rear of the DC and thus regulate actin dynamics locally. Accordingly, we found that the predominant F-actin structure localized at the rear of LPS-DCs was barely observed upon myosin II inhibition. Myosin II inhibition was also associated with a low cell speed (Fig. 6, E and F; fig. S6C; and movie S5). Analyses of myosin IIA–GFP dynamics at high spatial and time resolutions showed that the motor was moving toward the cell rear along actin cables (Fig. 6G and movie S6). Kymograph analysis showed that this myosin IIA–GFP retrograde flow was low in Trpml1−/− LPS-DCs (i.e., similar to the levels observed in immature cells; Fig. 6, H and I, and movie S4). Inversely, stimulation of lysosomal calcium release by MLSA1 produced a relative increase in the retrograde flow of myosin IIA–GFP in immature DCs (i.e., giving values seen in LPS-DCs; Fig. 6J). Together, these results suggest that TRPML1-triggered calcium release from lysosomes activates myosin IIA, stabilizes F-actin at the rear of LPS-DCs, and promotes fast DC migration. Our results also highlight the TRPML1-TFEB lysosomal signaling axis as being critical for migration mode switching in DCs upon LPS sensing.

The TRPML1-TFEB axis is activated by down-modulation of macropinocytosis in mature DCs

We next investigated the nature of the signals that activate lysosome signaling upon LPS recognition by DCs. It has recently been shown that in macrophages, macropinocytosis facilitates amino acid supply to lysosomes and leads to local mTORC1 activation (15). These results fit well with the well-known down-regulation of macropinocytosis after the loss of Cdc42 and Arp2/3 activities in mature DCs (5). We therefore hypothesized that activation of the TRPML1-TFEB axis (occurring downstream of mTORC1 inhibition) might result from the down-regulation of macropinocytosis in LPS-DCs.

To test this hypothesis, we analyzed the effects of macropinocytosis inhibitors on the nuclear translocation of TFEB-GFP. Treatment of immature DCs with rottlerin or a Cdc42 inhibitor (ML141) (5) triggered TFEB nuclear localization and increased the speed of immature DCs (like LPS; Fig. 7, A and B, and fig. S7A). Accordingly, these inhibitors promoted myosin IIA–GFP accumulation at the rear of Trpml1+/+ immature DCs but had no influence on Trpml1−/− cells, highlighting the specific role of this calcium channel in the activation of lysosomal signaling downstream of macropinocytosis inhibition (Fig. 7, B to F). Similar results were obtained using the macropinocytosis inhibitor 5-(N-ethyl-N-isopropyl)amiloride (EIPA) (fig. S7, B and C) (6 ), although this compound decreased DC survival. Together, these results suggest that down-regulation of macropinocytosis downstream of LPS sensing activates the TRPML1-TFEB axis, probably as a result of decreased nutrient delivery to lysosomes. Calcium release by TRPML1 then promotes myosin IIA activity and stabilizes F-actin filaments at the rear of mature DCs, increasing their speed and persistence for arrival at the LNs.

Fig. 7 Inhibition of macropinocytosis activates the TFEB-TRPML1 axis.

(A) Normalized TFEB nucleus-to-cytosol ratio from immature DCs treated with rottlerin (3 μM) or ML141 (50 μM) (data were pooled from N = 2 independent experiments, with n = 40, 45, 43, and 45 cells from left to right). (B) Mean instantaneous velocities of immature DCs migrating in 4 μm–by–5 μm fibronectin-coated microchannels, treated with rottlerin (3 μM) or ML141 (50 μM) (n = 109, 136, 140, 79, 77, and 120 cells from left to right). (C to F) Analysis of the localization of myosin IIA–GFP in immature DCs migrating in 8 μm–by–5 μm fibronectin-coated microchannels. (C) Mean myosin IIA–GFP distribution in immature DCs treated with rottlerin (3 μM) (from top to bottom: n = 50, 40, 47, and 50 cells; representative results from one of two experiments are shown). (D) The myosin IIA–GFP front-back ratio measured in the cells depicted in (C). (E) The mean myosin IIA–GFP distribution in immature DCs treated with ML141 (50 μM) (from top to bottom: n = 66, 49, 50, and 50 cells; representative results from one of two experiments are shown). (F) The myosin IIA–GFP front-back ratio measured in the cells depicted in (E). (A to F) Drugs were present during the recording. (A) Data were normalized for each experiment with respect to the mean TFEB ratio obtained in DMSO-treated immature DCs. Data are quoted as means, and error bars correspond to SEM. (B, D, and F) Data are represented as boxes and whiskers; bars include 90% of the points, the line represents the median, and the box contains 75% of the data. Mann-Whitney test was applied.

DISCUSSION

In peripheral tissues, the sensing of microbial components by DCs modifies the cells’ migratory behavior, promotes their trafficking to LNs, and initiates adaptive immune responses. The signaling pathways and molecular mechanisms underlying these changes in DC migration have nonetheless remained elusive. Here, we showed that signaling from lysosomes controls this process via a positive feedback loop mediated by the TFEB-TRPML1 axis. Upon LPS sensing, lysosomal calcium efflux by TRPML1 (i) promotes fast DC migration (by reorganizing the actomyosin cytoskeleton) and (ii) initiates TFEB translocation to the nucleus (which maintains trpml1 gene expression). Accordingly, our results show that TRPML1 is needed for DC chemotaxis to the LNs in vivo. We also found that the mere inhibition of macropinocytosis is sufficient to induce TFEB nuclear translocation and the cytoskeleton rearrangements required for fast DC migration. Macropinocytosis therefore emerges as an unexpected link between nutrient sensing, lysosomal signaling, and DC migration.

Our results showed that the mere activation of TRPML1 stimulates TFEB nuclear translocation and fast DC migration; this finding strongly suggests that the TRPML1 calcium channel is the main molecular player in this process. The expression of fascin [an actin-bundling molecule that regulates the adhesion of DCs (32)] was shown to be up-regulated in microarray analyses of TFEB-overexpressing cells (33). We found that fascin was up-regulated in migrating LPS-DCs but not in MLSA1-treated immature cells (fig. S8, A and B), indicating that calcium release from lysosomes is not sufficient to induce fascin expression. Lysosomal calcium release by TRPML1 induced myosin IIA retrograde flow and F-actin reorganization; however, these findings do not rule out a possible contribution of the endoplasmic reticulum to DC migration. It has been suggested that calcium in the endoplasmic reticulum is the main calcium source for lysosome refilling (34) and thus is needed to maintain the myosin IIA back-front gradient (6, 23). In addition to the role of TRPML1 in calcium dynamics, a lack of this protein is associated with lysosomal storage disorders (26). It was shown that this type of lysosomal defect can compromise macrophage migration by inhibiting endocytic recycling (35). However, we did not find a correlation between lysosome size and the migration speed of Trpml1−/− DCs (fig. S8C). Last, chronic loss of TRPML1 activity correlates with cholesterol accumulation in lysosomal compartments (30, 36), which can affect CCR7 oligomerization and signaling (37) as well as trafficking of skin DCs to LNs (38). However, we found that cholesterol distributions were similar in Trpml1+/+, Trpml1−/−, and MLSA1-treated DCs (fig. S8, D to F), suggesting that altered cholesterol homeostasis is not responsible for the impaired migration of TRPML1-deficient cells.

It remains to be seen how the TFEB-TRPML1 axis controls the migration of DCs upon infection in vivo. Analysis of DC subpopulations showed that TRPML1 deficiency has a specific impact on the number of mDC2 cells in LNs under homeostatic conditions (Fig. 2I). To further define the contribution of lysosome signaling to adaptive immunity, one must determine whether mDC2 cells exhibit elevated levels of TFEB and/or TRPML1 activity and how TFEB and TRPML1 KO mice respond to infections that usually activate these DCs. In summary, we found that activation of the TRPML1-TFEB axis is a critical signal that allows DCs to switch from a tissue-patrolling mode to the fast migration mode required for chemotaxis to LVs and homing to LNs. We further characterized the underlying mechanism by demonstrating that TRPML1 controls myosin IIA activity and maintains F-actin nucleation at the rear of DCs (5). Our data further suggest that (i) the inhibition of macropinocytosis after DC maturation activates lysosomal signaling and (ii) macropinocytosis might link nutrient sensing to cell locomotion.

MATERIALS AND METHODS

Study design

The objective of this study was to investigate the role of lysosomal signaling in the migration of DCs at the immature stage or upon microbial stimulation. We focused on the TFEB-TRPML1 axis and used microchannels as a tool to model the confined environment where DCs naturally evolve in tissues. DC migration, chemotaxis, and cytoskeleton dynamics were evaluated in TFEB- or TRPML1-silenced and KO DCs as well as in DCs treated with macropinocytosis inhibitors.

There was no predefined study component. Experiments were not randomized, and the investigators were not blinded to allocation during experiments and outcome assessment.

Cell culture

Mouse bone marrow–derived DCs (BMDCs) were cultured as previously described (5, 17 ). To obtain mature DCs, we incubated cells (day 10; 2 × 106 cells) for 30 min with LPS (100 ng ml−1; L6511, Sigma) and washed them three times (5). For migration experiments, cells were usually imaged between 6 and 16 hours after LPS treatment.

Preparation of microchannels and velocity measurements

Microchannels were prepared as previously described (17, 39, 40). For velocity measurements (carried out in 4 μm–by–5 μm microchannels), phase contrast images of migrating cells were acquired during 16 hours (frame rate of 2 min) on an epifluorescence Nikon Ti-E video microscope equipped with a cooled charge-coupled device (CCD) camera (HQ2, Photometrics) and a 10× objective. Kymographs of migrating cells were generated and analyzed using a custom program (17 ).

Actin, lysosomes, and myosin IIA density map generation and analysis

BMDCs generated from LifeAct-GFP or myosin IIA–GFP knock-in mice were incubated with WGA (AF-647) to label the lysosomes before any activation (2 × 106 cells were incubated with 0.25 μg ml−1 for 30 min at 37°C, 5% CO2). Cells were subsequently activated or not and loaded in 8 μm–by–5 μm microchannels. Migrating cells were then imaged for 16 hours using an epifluorescence Nikon Ti-E video microscope equipped with a cooled CCD camera (HQ2, Photometrics) and a 20× objective. Images were processed and analyzed using ImageJ software. Density maps were generated via a routine developed in the laboratory, as described previously (see fig. S4C and the Supplementary Materials) (5).

Myosin II retrograde flow

Myosin IIA–GFP–expressing Trpml1+/+ or Trpml1−/− immature and LPS-DCs were loaded in 8 μm–by–5 μm microchannels and were imaged using a spinning disk microscope (Yokagawa CSU-X1 spinning head mounted on a Nikon Eclipse Ti inverted microscope) equipped with a CoolSNAP HQ2 camera (Photometrics) and a 100× 1.4–numerical aperture (NA) oil immersion objective (acquisition of one image per 400 ms). For TRPML1 activation, myosin IIA–GFP immature DCs were loaded in microchannels and treated for 6 hours with 10 μM MLSA1 [dimethyl sulfoxide (DMSO) was used as a control] before imaging. To quantify the myosin IIA–GFP retrograde flow at the cell rear, we generated kymographs using ImageJ software and extracted the velocity of myosin IIA patches toward the cell rear. For each experiment, flow velocities were normalized to the mean retrograde flow of myosin IIA–GFP measured in the corresponding immature DC.

TFEB-GFP ratio

TFEB-GFP knock-in BMDCs, activated or not, were seeded on glass coverslips. For drug experiments, Torin 2 (2.5 nM) or MLSA1 (10 μM) was added, with DMSO being used as a control. Six hours after initial LPS activation or drug treatments, cells were fixed with 4% paraformaldehyde and labeled with anti-GFP antibody (Ab) (secondary Ab coupled to AF-488), anti-Lamp1 (secondary Ab coupled to AF-647), and Alexa Fluor phalloidin (AF-546). Z-stack images (1-μm spacing) were acquired on an inverted spinning disk confocal microscope (Roper/Nikon) using a 60× oil immersion objective (1.4 NA). For each cell, a single plane corresponding to the center of the nuclei was selected. The TFEB-GFP nucleus-to-cytosol ratio was obtained by dividing, after background subtraction, the TFEB-GFP signal detected in the nucleus [segmentation on 4′,6-diamidino-2-phenylindole (DAPI) signal] over the TFEB-GFP signal detected in the cytosol (segmentation on phalloidin signal). Data were normalized for each experiment to the mean TFEB ratio obtained in immature DCs at t = 6 hours.

Migration in collagen gels

Collagen experiments were performed as previously described (see the Supplementary Materials) (5). Cells were imaged (phase contrast, ×10 magnification; frequency of acquisition, one image every 2 min) during 16 hours. Images were processed using Fiji to extract cell tracks. The mean image (mean projection on all time points) of the movie was subtracted from every time point to remove static structures. Resulting movies were processed using the Fiji plugin Filter Mean (radius, 3 pixels), and cells were tracked using custom software (41).

Analysis of migratory DCs

Isolation of DCs

Inguinal LNs were collected from wild-type or Trpml1−/− mice and digested with Liberase TL Research Grade (0.15 mg ml−1; Roche) and deoxyribonuclease (0.15 mg ml−1; Sigma) in a CO2-independent medium (Invitrogen) for 15 min at 37°C. Digestion was stopped by washes in cold phosphate-buffered saline–bovine serum albumin (PBS-BSA) + 0.5% EDTA (2 mM) (Invitrogen). A single-cell suspension was obtained by gentle pipetting and filtering through a 70-μm cell strainer.

Flow cytometry

A multiparameter analysis was performed on the Fortessa (BD Biosciences), and data were analyzed with FlowJo software (Tree Star). Monoclonal Abs specific to mouse CD8α (53-6.7) allophycocyanin (APC; 553035, BD Biosciences) MHC II (I-A/I-E) (M5/114.15.2) Alexa Fluor 700 (56-5321-82, eBioscience), CD103 (2E7) PerCP-eFluor 710 (46-1031-82, eBioscience), CD11b (M1/70) Brilliant Violet 605 (101237, BioLegend), CD11c (N418) PE (phycoerythrin)–Cy7 (cyanine 7) (25-0114-82, eBioscience), and EpCAM (epithelial cell adhesion molecule) (G8.8) APC/Cy7 (118218, BioLegend) were purchased from BD Biosciences or eBioscience. Cells were stained on ice for 30 min in PBS-BSA + 0.5% EDTA (2 mM). Before acquisition, cells were washed three times and resuspended in PBS-BSA + 0.5% EDTA (2 mM) solution with DAPI (1 μg ml−1) to exclude dead cells. The detailed gating strategy is described in fig. S2G. Isotype controls were not performed because staining for CD11c, MHC II, CD103, CD8a, CD11b, and EpCAM by the antibodies listed above were validated in previous publications (42, 43).

Statistics and reproducibility

Experiments shown in the figures correspond to representative experiments (n, number of cells analyzed; N, number of experiments) or a pool of experiments. An internal control was systematically included in migration experiments. All graphs and statistical analysis were performed using GraphPad Prism (version 5 or 7), and whenever possible, exact P values are indicated in the figures. No statistical method was used to predetermine sample size. Comparisons between any two groups were made using the nonparametric Mann-Whitney statistical test or the paired t test (Fig. 2J). In the collagen experiment, where the number of tracks was extremely high (>2000), a Welch two-sample t test was applied. Boxes in box plots extend from the 25th to the 75th percentile, with the median indicated by a line, and the whiskers extend from the 10th to the 90th percentile. Bar graphs show the means, and error bars correspond to SEM. Additional information can be found in Supplementary Materials and Methods.

SUPPLEMENTARY MATERIALS

immunology.sciencemag.org/cgi/content/full/2/16/eaak9573/DC1

Materials and Methods

Fig. S1. TFEB translocates to the nucleus and is required for fast migration of LPS-DCs (related to Fig. 1).

Fig. S2. TRPML1 is required for fast DC migration, chemotaxis, and arrival at LNs (related to Fig. 2).

Fig. S3. Fast DC migration relies on calcium release by TRPML1 (related to Fig. 3).

Fig. S4. TFEB controls actin cytoskeleton reorganization in response to microbial sensing (related to Fig. 4).

Fig. S5. TRPML1 controls the organization of the DC actin cytoskeleton (related to Fig. 5).

Fig. S6. Calcium released through TRPML1 controls myosin IIA retrograde flow at the rear of LPS-DCs (related to Fig. 6).

Fig. S7. Inhibition of macropinocytosis activates the TRPML1-TFEB axis (related to Fig. 7).

Fig. S8. Fascin expression, lysosomal area, and cholesterol homeostasis in migrating DCs.

Table S1. Source data for all graphs with n < 25.

Movie S1. F-actin is in proximity to the endolysosomes at the rear of the LPS-DCs.

Movie S2. F-actin dynamics in control (siCtrl) and TFEB-silenced (siTFEB) LPS-DCs.

Movie S3. F-actin dynamics in Trpml1+/+ and Trpml1−/− LPS-DCs.

Movie S4. Myosin IIA–GFP dynamics in Trpml1+/+ and Trpml1−/− LPS-DCs.

Movie S5. F-actin dynamics in LPS-DCs treated with DMSO or para-nitroblebbistatin.

Movie S6. Myosin IIA moves along actin cables.

References (4452)

REFERENCES AND NOTES

Acknowledgments: We acknowledge the BioImaging Cell and Tissue Imaging (PICT-IBiSA) Core Facility of the Institut Curie (ANR10-INBS-04), the Imaging Core at the Institut Pierre-Gilles de Gennes (ANR-10-EQPX-34), and the Institut Curie animal facility. We thank M. Sixt (for LifeAct-GFP mice), H. Xu (for the TRPML1-GFP construct), and D. Obino and G. Duménil (for critical reading of the manuscript). Funding: M.B.: a fellowship from the Association pour la Recherche contre le Cancer and the Institut Curie; P.V. and M.C.: fellowships from the Fondation pour la Recherche Médicale; L.B.: a PhD fellowship from the French government. This work was supported by grants from the European Research Council (Strapacemi 243103) and the Institut National du Cancer to A.-M.L.-D., the DCBIOL Labex (ANR-10-IDEX-0001-02-PSL and ANR-11-LABX-0043), and the Association Nationale pour la Recherche to P.V. (ANR-16-CE13-0009). P.V. is an INSERM staff researcher. Author contributions: M.B. designed, performed, and analyzed most experiments; prepared manuscript figures; and participated in drafting the manuscript. P.J.S. helped with in vivo and Ca2+ experiments, performed FACS experiments, and analyzed the cholesterol distribution. D.S. and J.H. analyzed DC populations in LNs, and D.S. performed immunofluorescence analyses of fascin. M.M. coded the software routines for image analysis (calcium, actomyosin, and organelle dynamics). D.L. performed immunofluorescence staining analyses. M.C. observed that DCs accelerated upon inhibition of macropinocytosis. O.M. helped with FACS experiments. L.B. analyzed actin dynamics in DCs treated with Torin 2. P.M. developed the software for analyzing and modeling cell trajectories in collagen experiments. C.S. and A.B. generated and provided TFEB-GFP and TFEB-inducible KO precursors. S.M. provided Trpml1−/− mice. M.P. provided support for microfabrication and helped with experiment design. P.V. performed collagen and in vivo experiments. P.V. and A.-M.L.-D. designed the overall research and wrote the manuscript. Competing interests: A.-M.L.-D., P.V., and M.B. hold a patent application (publication no. WO2015118167 A1, Modulation of lysosomal calcium channel Mcoln1 regulates DC migration). The other authors declare that they have no competing interests.
View Abstract

Navigate This Article