Regulation of Dendritic Cell Migration by CD74, the MHC Class II-Associated Invariant Chain

See allHide authors and affiliations

Science  12 Dec 2008:
Vol. 322, Issue 5908, pp. 1705-1710
DOI: 10.1126/science.1159894


Dendritic cells (DCs) sample peripheral tissues of the body in search of antigens to present to T cells. This requires two processes, antigen processing and cell motility, originally thought to occur independently. We found that the major histocompatibility complex II–associated invariant chain (Ii or CD74), a known regulator of antigen processing, negatively regulates DC motility in vivo. By using microfabricated channels to mimic the confined environment of peripheral tissues, we found that wild-type DCs alternate between high and low motility, whereas Ii-deficient cells moved in a faster and more uniform manner. The regulation of cell motility by Ii depended on the actin-based motor protein myosin II. Coupling antigen processing and cell motility may enable DCs to more efficiently detect and process antigens within a defined space.

Dendritic cells (DCs) patrol the body periphery by efficiently moving across tissues while retaining a high capacity for antigen (Ag) internalization. Ags are internalized into endosomes, degraded into peptides that are loaded onto major histocompatibility complex (MHC) molecules, and subsequently exposed at the cell surface of DCs (1). If DCs detect an Ag-associated activating signal, they enter into a maturation program that comprises an increase followed by a complete arrest of macropinocytosis and phagocytosis, maximizing uptake and processing of the activating Ag (24), as well as the modification of their migratory properties, enabling them to home to secondary lymphoid organs and present the resulting peptide-MHC complexes to T cells (5). Ag uptake and processing and cell migration must therefore be coordinated for DCs to efficiently activate T lymphocytes, suggesting that common regulatory mechanisms are involved. Although little is known about DC locomotion, it was recently shown that DCs move in the interstitial spaces of tissues by using amoeboid-like migration, which requires actomyosin contractility but not integrin-mediated adhesion (6).

Several lines of evidence point to the MHCII-associated protein invariant chain (Ii or CD74) as a good candidate to regulate DC migration. Indeed, beyond its key role in MHCII trafficking and peptide loading, Ii was shown to associate with two actin-interacting proteins known to regulate cell locomotion: the motor protein myosin II, which controls actomyosin contractility, and the adhesion molecule CD44 (79). To investigate whether Ii regulates DC migration, we compared the in vivo migratory properties of wild-type (WT) and Ii–/– DCs (fig. S1A). Bone marrow–derived DCs (BM-DCs) were loaded with fluorescent dyes [in the presence of lipopolysaccharide (LPS) to induce CCR7 expression] and transferred into the footpad of C57/BL6 mice, and their arrival at the draining lymph node (DLN) was monitored 48 hours after transfer. Increased numbers of Ii-deficient DCs compared with WT cells were found in the DLN of recipient mice (Fig. 1A and fig. S2A). Fluorescein isothiocyanate (FITC)–painting experiments of the ear also showed that enhanced numbers of endogenous DCs had reached the DLN in Ii–/– mice after 24 hours compared with the numbers in WT mice (Fig. 1B). This difference did not result from passive diffusion of FITC because Ii+/+ and Ii–/– mice displayed similar numbers of DCs in LNs at the steady state (fig. S2B). Thus, the absence of Ii facilitates the accumulation in LNs of activated DCs that migrate from the periphery.

Fig. 1.

Ii negatively regulates DC migration in vivo and in collagen-coated transwells. (A) Fluorescently labeled WT and Ii–/– DCs were coinjected into the footpad of C57BL/6 recipient mice. Draining and nondraining (ND) LNs were analyzed by flow cytometry after 48 hours. The presence of migrating DCs is displayed as a percentage of total LN cells (each dot represents one mouse, four independent experiments, ***P < 0.0001). (B) Ears from WT or Ii–/– were painted with a FITC-containing irritating solution. Twenty-four hours later, draining LNs were analyzed by flow cytometry (each dot represents one mouse, three independent experiments **P = 0.004). (C) Migration experiment of transferred WT and CatS–/– BM-DCs performed as described in (A). (D) The numbers of fluorescent DCs remaining in the footpad of recipient mice. For (C) and (D), results are from three independent experiments; **P = 0.0034; *P = 0.0445. (E and F) Migration of WT, Ii–/–, and CatS–/– BM-DCs in uncoated (E) and collagen-coated (F) transwells toward CCL19 and CCL21 (NC, no chemokines). Results are expressed in percentage and normalized to migration of WT DCs (mean ± SD, three independent experiments, **P = 0.001, *P = 0.01). (G) Inhibition of DC migration by the CatS inhibitor LHVS, performed as in (F). The results are represented as the percentage of LHVS-treated DCs that migrated relative to untreated cells (mean ± SD, three independent experiments, ***P = 0.0005). (H) Ii depletion upon infection of day 4 BM-DCs with a lentivirus encoding an Ii shRNA. Infected BM-DCs were collected at day 10, lysed, and analyzed by immunoblotting for Ii and Syk expression. (I) Migration experiment of Ii-depleted BM-DCs performed as described in (F). The results are expressed in percentage and normalized to migration of control WT DCs (mean ± SD, two independent experiments, **P = 0.005, *P = 0.05).

Cathepsin S cleaves the cytosolic tail of Ii when MHCII-Ii complexes have reached endolysosomes. This is required for peptide loading onto MHCII and transport of MHCII-peptide complexes to the cell surface (10, 11). In the absence of CatS, MHCII-Ii complexes accumulate in endolysosomes and, to a minor extent, at the plasma membrane (12). We analyzed the migratory capacity of CatS–/– DCs, which display enhanced Ii protein levels (fig. S1, A to C) (1012). CatS–/– DCs transferred into the footpad of C57/BL6 mice migrated less efficiently from the periphery to LNs than did WT DCs (Fig. 1C). Increased numbers of transferred CatS–/– DCs compared with WT cells were found in the footpad, suggesting that CatS–/– DCs remained at the injection site rather than homed to another tissue (Fig. 1D). Thus, enhanced levels of Ii in CatS–/– DCs correlate with reduced in vivo migration.

We next compared the ability of WT, Ii–/–, and CatS–/– DCs to migrate toward a CCL19/CCL21 gradient in transwells. Ii-deficient cells migrated more efficiently than WT DCs, and CatS–/– cells showed impaired migration in collagen-coated but not in uncoated transwells (Fig. 1, E and F). Transfer of supernatants between WT, Ii–/– and CatS–/– DCs had no effect on cell migration, and Ii levels did not affect the capacity of DCs to degrade gelatin-FITC (fig. S3, A and B). Thus, down-regulation of DC migration by Ii results neither from altered response to CCL21 and CCL19 chemokines nor from differential secretion of soluble factors.

The specific CatS inhibitor LHVS (N-morpholinurea-homophenylalanyl-leucyl-vinylsulfonemethyl) significantly reduced the migration of WT DCs (Fig. 1G) (13). However, incubation of Ii-deficient DCs with LHVS did not compromise their migration (Fig. 1G). This result indicates that decreased migration of CatS–/– DCs resulted from Ii accumulation and not from an Ii-independent action of the protease. This observation was further strengthened by showing that short hairpin-mediated RNA interference (shRNA) depletion of Ii in CatS–/– DCs restored their ability to migrate in collagen-coated transwells (Fig. 1, H and I). Thus, Ii negatively regulates the capacity of DCs to move across a protein matrix in vitro and in vivo.

To unravel the mechanisms by which Ii regulates DC migration, we analyzed DC motility by time-lapse imaging with microfabricated 4-μm-diameter channels (1416) (fig. S4). Nonactivated WT BM-DCs displayed an average median instantaneous velocity of 6.5 μm min–1 ± 3.4 (median ± SD) (Fig. 2A and movie S1). These values are compatible with the ones recently reported for BM-DC migrating in three-dimensional collagen matrices and in the mouse skin (6), validating microchannels as an adequate tool to study DC motility. Strikingly, 71% of Ii–/– DCs migrated at velocities above 6.5 μm min–1, some of them reaching speeds two- to threefold higher (Fig. 2, A to C, and movie S2) (P = 1.2 × 10–6). On the contrary, the velocities of 76% of CatS–/– DCs fell below 6.5 μm min–1 (Fig. 2, A to C, and movie S3) (P = 1.8 × 10–3). Hence, Ii levels regulate the speed of DCs that migrate in a confined microenvironment, with Ii-deficiency favoring fast DC migration.

Fig. 2.

Ii decreases the velocity of DCs migrating in confining microchannels. (A and B) Distribution of median instantaneous speeds for Ii–/–, CatS–/–, and WT DCs treated with 20 μM blebbistatin. In (B), speed distribution is represented in a cumulative graph to allow statistical analysis. (C) Velocity distributions are significantly different from each other with a 95% confidence interval. Velocity distributions of CatS–/– and blebbistatin-treated DCs are not significantly different.

While imaging WT DCs migrating along microchannels, we noticed that most of the cells underwent important speed fluctuations during motion (Fig. 3A and movie S1). Two phases were observed: a motile phase during which pseudopods were extended at the cell front, leading to the formation of big vesicles (most likely macropinosomes), followed by a rather static phase during which these vesicles were transported backward (Fig. 3 and movie S4). Ii–/– DCs showed a more regular, continuous movement, and slow motility phases were rarely observed (Fig. 3A and movie S2). Even when comparing WT and Ii–/– DCs that displayed a similar median velocity, the speed fluctuations undergone by cells during migration were reduced in the absence of Ii (Fig. 3C). These results suggested that Ii regulates DC migration by inducing slowing-down phases during motion. Accordingly, when measuring the local speed variation of individual migrating DCs, we obtained values significantly lower for Ii-deficient cells compared with those of their WT counterpart, verifying that Ii–/– DCs undergo less velocity variation during motion (Fig. 3D) (P = 7.6 × 10–6). Furthermore, 50% of Ii–/– DCs spent most of their time (>80%) migrating faster than 6.5 μm min–1 (the WT median speed), whereas only 24% of WT DCs did so (fig. S6). These results indicate that Ii deficiency favors sustained DC migration at high speed. Thus, we conclude that Ii reduces DC migration by transiently slowing down DC velocity.

Fig. 3.

Ii imposes a discontinuous migration mode to DCs. (A) DCs migrating along a 4-μm fibronectin-coated microchannel, covering a distance of 400 μm. Sequential images (10× objective phase) of temporal sequences were placed one below the other (vertical axis). The interval between two images is 2 min. Scale bar indicates 50 μm. (B) Sequential high resolution (100×) phase contrast images of a WT cell (movie S1) migrating along a microchannel. The channel part displayed is 88 μm long (vertical axis) with images taken over 14 min (40 s between each image, horizontal axis). Scale bar, 20 μm. (C) Representative examples of velocities displayed by individual WT and Ii–/– DCs as a function of time. The data were extracted from movies S1 and S2. WT cell number 1 and Ii–/– cell number 2 correspond to the DCs shown in (A). (D) Cumulative distribution of the relative speed variations of WT and Ii–/– DCs, calculated as the average of the local deviation from the local mean (two points before and two points after each time point) divided by the global median velocity.

We next addressed the molecular mechanisms underlying the regulation of DC migration by Ii. Myosin II is required for DCs to migrate in collagen matrices and in vivo (6). Accordingly, we found that treatment of WT and Ii-deficient DCs with low concentrations of the myosin II inhibitor blebbistatin impaired their migration in collagen-coated transwells (fig. S7, A and B, and movie S5). In microchannels, the speed of WT DCs significantly decreased when inhibiting myosin II activity, with 76% of the cells migrating at velocities below 6.5 μm min–1 (WT median velocity, Fig. 2, A to C) (P = 2.6 × 10–7). Blebbistatin-treated DCs showed a velocity distribution similar to that of CatS–/– cells (Fig. 2, A to C) (P = 8 × 10–2), suggesting that myosin II activity might be compromised in CatS–/– DCs. In view of the ability of myosin II to interact with MHCII-Ii complexes in DCs (7), we hypothesized that Ii retention in CatS–/– DCs alters myosin II localization and activity.

To address this hypothesis, we analyzed myosin II and Ii distribution in CatS–/– DCs plated on rectangular fibronectin-coated micropatterns to obtain a spatial intracellular organization close to the one of elongated DCs migrating along microchannels (fig. S8A). Ii-containing structures are known to accumulate in the perinuclear region of CatS–/– DCs (12). We observed an enrichment of myosin II heavy chain (MyoIIHC) in these structures (Fig. 4A) compared with a weaker colocalization with the more heterogeneous distribution of Ii in WT DCs, (Fig. 4A and fig. S8B). Equivalent results were obtained in WT cells treated with the CatS inhibitor LHVS (fig. S8B). Cryoimmunoelectron microscopy experiments confirmed the presence of myosin II light chain (MLC) in these enlarged endolysosomal compartments of CatS–/– DCs (Fig. 4B). Because these peculiar compartments form as a result of Ii retention in CatS–/– cells (17) and therefore have no structural equivalent to be compared with in WT DCs, it was not possible to provide a rigorous quantification for this qualitative difference. However, we rarely found MLC and Ii colocalizing in endolysosomes of WT DCs (fig. S8C). This result was further verified quantitatively by using biochemistry: The amounts of MyoIIHC and MLC (20 kD) contained in semi-purified endosomal fractions were increased in CatS–/– as compared with WT DCs (fig. S9, A and B). Hence, retention of Ii in endolysosomes of CatS-deficient DCs promotes the association of myosin II to these compartments.

Fig. 4.

The levels of Ii modify the localization and phosphorylation of myosin II as well as DC motility. (A) Immunofluorescence and confocal microscopy analysis of MyoIIHC and Ii subcellular localization in WT and CatS–/– BM-DCs plated on fibronectin-coated micropatterns. Scale bar, 2 μm. (B) Cryoimmunoelectron microscopy of CatS–/– BM-DCs. The colocalization of Ii and MLC was analyzed with secondary Abs coupled to protein A–gold (PAG) of 15-nm and 10-nm diameters, respectively. Scale bar, 200 nm (left) and 500 nm (right). (C) WT and CatS–/– DCs were lysed and analyzed by SDS–polyacrylamide gel electrophoresis and immunoblotting using antiphospho-MLC and antitotal MLC Abs sequentially. (D) MLC phosphorylation immunoblots were quantified by using Image J software; results were normalized to actin or tubulin expression and expressed as percentage of MLC phosphorylation in WT DCs (mean ± SD, three independent experiments, *P = 0.01). (E and F) The motility of LPS-pretreated cells along microchannels was analyzed by time-lapse imaging. Data are displayed as (E) the frequency of direction changes per minute and (F) the mean velocities, calculated by averaging instantaneous cell velocity values while taking into account cell directionality. Values are normalized to the ones of untreated cells from the same type (mean ± SD, 100 to 200 cells per condition from two independent experiments, **P < 0.01, ***P < 0.0001).

When phosphorylated on Thr18 and/or Ser19, MLC binds to and triggers the adenosine triphosphatase (ATPase) activity of the motor, initiating actomyosin contraction (18). To assess whether myosin II retention in endolysosomes from CatS–/– DCs decreases its activity, we analyzed the phosphorylation of MLC. Phosphorylation of the full-length 20-kD MLC was considerably decreased in CatS–/– cells compared with WT (Fig. 4, C and D). We consistently observed a 12-kD fragment of MLC that was verified by mass spectrometry in extracts from CatS–/– DCs (Fig. 4C and fig. S9, B and C). However, this MLC fragment did not react with the antiphosphorylated MLC antibody (Ab) (Fig. 4C), suggesting that it is inactive. Inhibition of CatS activity with LHVS triggered the accumulation of this 12-kD MLC fragment in WT but not in Ii–/– DCs (fig. S9D). Hence, retention of Ii in CatS-deficient DCs induces the specific accumulation of a truncated form of MLC. Thus, accumulation of Ii in DCs lacking CatS affects the subcellular localization and the activity of myosin II, providing a possible mechanism for regulation of DC motility by Ii.

We next investigated whether regulation of Ii expression upon DC activation leads to modification of their migratory capacity. Indeed, DCs undergo a transient increase in Ii and MHCII biosynthesis in the first hour after LPS treatment (4, 19). Accordingly, although Ii-myosin II association was observed in nonactivated DCs, it was further increased after 1 hour of LPS treatment (fig. S10, A and B). We found that WT DCs activated with LPS for 2 hours exhibited strong alterations in their ability to migrate along microchannels (movie S6). Early activated WT DCs displayed significantly reduced instantaneous velocities (fig. S10D) and more fluctuant velocities and changed direction more frequently than untreated DCs (Fig. 4, E and F, and fig. S10C). The effect of LPS on motility was less marked in DCs treated with LPS for 12 hours, which moved slightly slower but with a similar persistence than nonactivated DCs (Fig. 4, E and F). LPS had no significant effect on the velocity and directionality of Ii-deficient DCs (Fig. 4, E and F, fig. S10D, and movie S7). Thus, the motility of DCs is regulated upon activation, and such regulation relies at least in part on Ii.

We show that Ii acts as a brake during DC motion by imposing a discontinuous locomotion mode that alternates between high and low motility phases. Down-modulation of cell migration by Ii is likely to result from its direct or indirect association with myosin II, which is required for polarized transport of MHCII-Ii complexes to endosomes in B cells (7). Ii–myosin II interaction would facilitate the convergence between MHCII and the endocytosed Ag but would temporarily affect the ability of myosin II to promote cell migration, thus leading to a transient decrease in DC velocity. Once in endosomes, Ii–myosin II complexes would dissociate upon Ii cleavage by CatS, restoring high DC motility.

The use of common regulators for Ag processing and cell motility provides a way for DCs to coordinate these two functions in time and space. In immature DCs that patrol peripheral tissues, the periodic low motility phases induced by Ii may enable DCs to efficiently couple Ag uptake and processing to cell migration, facilitating the sampling of the microenvironment. In early-activated DCs, the transient increase in the synthesis and association of Ii with myosin II would reduce the speed and confine the trajectories of the cells by promoting directional changes, enabling DCs to maximize Ag uptake (2). The formation of endocytic vesicles at the front of migrating DCs occurred within defined motility phases, suggesting a possible coupling between macropinocytosis and cell locomotion.

The physiological relevance of myosin II–dependent migration in DCs and other leukocytes was recently demonstrated (6). The molecular mechanism used by DCs to regulate their migratory capacity via myosin II is likely also used by B lymphocytes, in which myosin II and Ii associate upon B cell receptor engagement (7). Indeed, similarly to DCs, activated B cells from Ii–/– mice display enhanced motility ex vivo (fig. S11, A and B). Because myosin II was also shown to respond to Ag receptors in T and natural killer cells (2022), we foresee that modulation of myosin II activity might be a general mechanism used by leukocytes to regulate their specific immune function in time and space.

Supporting Online Material

Materials and Methods

Figs. S1 to S11


Movies S1 to S7

References and Notes

Stay Connected to Science

Navigate This Article