JAM-A interacts in cis with α3β1 integrin through CD151 to regulate collective cell migration of polarized epithelial cells


 Junctional adhesion molecule (JAM)-A is a cell adhesion receptor localized at epithelial cellcell contacts with enrichment at the tight junctions. Its role during cell-cell contact formation and epithelial barrier formation has intensively been studied. In contrast, its role during collective cell migration is largely unexplored. Here we show that JAM-A regulates collective cell migration of polarized epithelial cells. Depletion of JAM-A in MDCK cells enhances the motility of singly migrating cells but reduces cell motility of cells embedded in a collective by impairing the dynamics of cryptic lamellipodia formation. This activity of JAM-A is observed in cells grown on laminin and collagen I but not on fibronectin or vitronectin. Accordingly, we find that JAM-A exists in a complex with the laminin- and collagen I-binding α3β1 integrin. We also find that JAM-A interacts with CD151, a tetraspanin that forms a stoichiometric complex with α3β1 integrin and that regulates α3β1 integrin activity in different contexts. Mapping experiments indicate that JAM-A associates with both α3β1 integrin and CD151 through its extracellular domain. Similar to depletion of JAM-A, depletion of either α3β1 integrin or CD151 in MDCK cells slows down collective cell migration. Our findings suggest that JAM-A, α3β1 integrin and CD151 exist as a functional complex to regulate collective cell migration of epithelial cells.


Introduction
Collective cell migration is a process in which groups of cells migrate in a coordinated manner.
Movements of cell collectives occur during development, repair and regeneration processes but also during pathophysiological processes like cancer invasion [1,2]. In contrast to collectively migrating cancer cells, which adopt mesenchymal characteristics and which form loose and rather transient cell-cell junctions, collectively migrating epithelial cells maintain their apico-basal polarity and remain tightly connected through stable intercellular junctions [3].
The coordination of directed migration is enabled by the formation of polarized protrusions beneath the cell migrating in front of the cell, so-called cryptic lamellipodia [4,5], while at the same time maintaining regular cell-cell junctions at the apical cell-cell junctions.
The formation of oriented cryptic lamellipodia requires intact adherens junctions which act as scaffolds for the Arp2/3 complex and its regulator Wave [6]. A structure similar to cryptic lamellipodia has been observed in collectively migrating endothelial cells [7][8][9]. Similar to cryptic lamellipodia of migrating epithelial cells, their formation requires an intact cadherin/catenin complex indicating an active role of intercellular junctions in the formation of polarized protrusions during collective cell migration.
Recent evidence strongly suggests a role for intercellular junctions in mechanosensing and transduction. Leader cells at the forefront of a cellular collective respond to chemical cues by front-to-rear polarization, i.e. increased protrusive activity at their front and increased actomyosin-driven contractility at their rear end [2]. The forces generated by migrating leader cells are transmitted to the cells behind the front through intercellular junctions-mediated forcesensing and force-transducing mechanisms [10][11][12][13][14][15]. Molecular components localized at all structures at cell-cell junctions including tight junctions (TJs), adherens junctions (AJs), desmosomes and gap junctions contribute to the behaviour of collectives of migrating cells [16]. Interestingly, physical properties associated with forces such as strain energy or monolayer tension and those associated with kinematic behaviour such as monolayer velocity are differently influenced by adhesion molecules and their associated cytoplasmic proteins [16], indicating that signalling inputs derived from multiple cell-cell adhesion receptor systems are integrated into coordinated cell behaviour during collective cell migration.

Junctional adhesion molecule A (JAM-A) is a member of the immunoglobulin superfamily
(IgSF) with a variety of functions in different cells types [17]. In polarized epithelial cells, JAM-A is localized at intercellular junctions and is enriched at TJs where it is phosphorylated by the Par3 -aPKC -Par6 polarity protein complex [18]. Depletion of JAM-A in SK-CO15 epithelial cells results in reduced cell migration in scratch-wounding assays [19]. In different tumor cell lines, the levels of JAM-A were found to correlate both positively and negatively with cell migration [20][21][22][23] suggesting a context-dependent function of JAM-A during migration. In endothelial cells, JAM-A regulates cell migration in response to bFGF through its interaction with tetraspanin CD9 and αvβ3 integrin [24][25][26].
In this study, we addressed the role of JAM-A during collective cell migration of polarized epithelial cells. Using MDCKII cells in a monolayer expansion model [16], we find that JAM-A is required for efficient migration of MDCKII cell collectives. Depletion of JAM-A increases the migration velocity of singly migrating MDCKII cells, but slows down the migration velocity of MDCKII cell collectives. We find that JAM-A interacts with α3β1 integrin as well as with tetraspanin (Tspan) CD151, a major α3β1 integrin-binding partner and regulator of α3β1 integrin functions. Our findings suggest a tetraspanin-enriched microdomain (TEM) in which CD151 links JAM-A to α3β1 integrin and identify JAM-A as a regulator of collective cell migration in polarized epithelial cells.
Transient transfections of siRNAs were performed using Lipofectamine RNAiMAX (Thermo Fisher Scientific). Transfections of expression vectors were performed using Lipofectamine 2000 (Thermo Fisher Scientific) or Xfect transfection reagent (TaKaRa Bio Europe SAS, Saint-Germain-en-Laye, France) according to the manufacturer's instructions. Lentiviral particles for the generation of stably transfected cell lines expressing either shRNAs or cDNAs were generated by cotransfection of the lentiviral vector and the packaging vectors psPAX2 and pMD2.G (kindly provided by Dr. Didier Trono, Addgene plasmids 12260 and 12259) in a ratio of 3:2:1 into HEK293T cells. Lentiviral transduction of cells was performed as described [29].

Quantitative Polymerase Chain Reaction (qPCR)
To analyze the mRNA level of CD9 and CD151, the total RNA was extracted from the cell lysate using the RNeasy Mini Kit following the instructions from Qiagen (Hilden, Germany) and

8
GFP fusion proteins were precipitated by using the GFP Selector (NanoTag Biotechnologies, Göttingen, Germany). After cell lysis, the postnuclear supernatant was added to 20 µl of GFP Selector slurry. After incubation for 1 h at 4°C, the GFP Selector beads were washed five times with lysis buffer. Bound proteins were eluted by boiling in SDS sample buffer analyzed by Western blotting as described above.

Immunofluorescence microscopy
For immunofluorescence microscopy, cells were grown on collagen-coated glass slides for Plan-Neofluar × 20/0.5 objective for 10 h with images taken every 10 min. The velocity of single cells was determined by tracking the cells using the TrackMate Plugin from ImageJ, which tracks the cell from frame to frame and thereby determines its average velocity.

Collective cell migration
For the analysis of collective cell migration epithelial cell monolayers, a monolayer expansion assay was used in which collective cell migration is triggered by a free surface [13,31]. ImageJ. Briefly, the cell-free area measured at the end of the observation period (t1 = 8h) was subtracted from the cell-free area at the beginning (t0) resulting in the total area covered by migrated cells. This area was divided by the height of the gap resulting in the total distance that the cells had migrated. To take into account that the cells close the gap from both sides the total distance was divided by two resulting in the distance migrated by a single sheet. The migration speed of the cellular collective was calculated by dividing the distance of a single sheet by the observation time and is given in µm/min.

Analysis of single cell motility within cell collectives
Cell motility of single cells embedded in migrating cell collectives was analyzed by seeding cocultures of LA-mCherry-and LA-EGFP-transfected MDCK cells (mixed at ratios of 1:5) on CNIcoated microscope slides used for collective cell migration (see above). After cells had reached full confluency (72 h after seeding) stamps were removed to induce collective cell migration.
6 h after removal of the stamps, cell behaviour was recorded for 10 h with images taken every 3 min using the confocal LSM780 microscope (Carl Zeiss, Jena, Germany) with a Plan-Apochromat x 63/1.4 oil objective. For analysis, the LA-mCherry and LA-EGFP channels, a median filter was applied and cells from one channel were segmented as binary (black and white) images (BW) by manual thresholding.
For single cells movies, cell segemantation was performed acc. to [32]. Briefly, the center of the segmented cell was determined for each frame to calculate the average velocity and directionality using the TrackMate Plugin for ImageJ. To access the dynamic changes of the cell outline, the segmentations of a cell at two different time points was compared using the Jaccard Index J according to the following formula: Where A and B are the segmented cells at time point ( − ∆ ) and ( + ∆ ) respectively: The Jaccard Index is equal to 1 if the cell outline is not changed between the two time points and tends to zero for higher differences, indicating higher protrusion/retraction activity at the cell border. Note, that this index can change, even if the center of the segmented cell does not move at all.
For movies with one or more cells, the protrusion and retraction activity along the segmented region was quantified. To estimate the protrusion activity, the area that was exclusively occupied by the segmented cells at the time point ( + ∆ ) but not at time point ( − ∆ ) was determined and normalized to the perimeter of the segmented area at time point t: In analogy, the retraction activity was calculated: For these calculations ∆ was set to 3 min.
All parameters are represented as mean values over the entire duration of the experiment. The scripts to calculate the Area Increase, the Area decrease and the Jaccard indices are freely available by the authors upon request. Jaccard indices we

Statistics
Results are expressed as arithmetic means ± SD as indicated. To test the normality of data, D'Agostino-Pearson normality test was used. Data with normal distributions were statistically compared by using unpaired, two-tailed student's t-test, whereas Mann-Whitney U test was applied for data without normal distribution. Statistical analyses were performed using GraphPad Prism version 6 (GraphPad Software, San Diego, CA). P-values are indicated as follows: *P<0.05, **P<0.01, ***P<0.001 and ****P<0.0001.

JAM-A limits single cell migration in a cell-autonomous manner in polarized epithelial cells
We first analyzed single cell migration of MDCKII cells after depletion of JAM-A using a doxycycline (dox)-regulatable JAM-A shRNA expression system [27] (Suppl. Fig. 1

JAM-A positively regulates collective cell migration in polarized epithelial cells
We next analyzed the role of JAM-A in collectively migrating cells in a monolayer expansion model [13,31]. To this, cells were grown to confluency on slides in which individual chambers are separated by removable silicon stamps. After reaching confluency, cell sheet migration was induced by removing the stamps. Migration of the monolayer was monitored over a period

JAM-A interacts with α3β1 integrin in MDCKII cells
The lack of efficient locomotion of cells during collective cell migration after depletion of JAM-A together with our observations that JAM-A-regulated single cell and collective cell migration depend on CNI (Fig. 1A, Fig. 2A) suggested that JAM-A might cooperate with a CNI-binding integrin in MDCKII cells. We focussed on α3 integrin, which forms a CNIand LN-binding heterodimer with β1 integrin [35,36]. In co-immunoprecipitation (CoIP) experiments we found that JAM-A and α3β1 integrin exist in a complex in MDCKII cells (Fig. 4A). This interaction was detectable in Brij98-containing lysis buffer but not in NP40-containing lysis buffer. Since Brij98 maintains interactions mediated by tetraspanins (Tspans) [37], a family of integral membrane proteins with four membrane spanning domains [37][38][39][40], these findings suggested that the interaction between JAM-A and α3β1 integrin is indirect and mediated by a member of the Tspan superfamily. To test if the interaction depends on integrin activation by ligand binding, as described for the interaction between JAM-A and αvβ3 integrin in endothelial cells [26,41], To address the question of a functional interaction between JAM-A and α3β1 integrin we analyzed collective cell migration of MDCKII cells after depletion of the α3 subunit by RNAi.
Similar to the depletion of JAM-A, depletion of the α3 integrin chain resulted in a reduced migration velocity of cell collectives grown on CNI (Fig. 4D). Reduced migration velocity was also observed when cells were grown on LN (Fig. 4D), the second major ligand of α3β1 integrin [36]. These observations suggested that JAM-A and α3β1 integrin exist in a complex to cooperate in the regulation of collective cell migration in polarized epithelial cells.

JAM-A interacts with α3β1 integrin-interacting tetraspanins CD151 and CD9
At least 12 different integrin heterodimers have been found in associations with Tspans [40,43]. Tspans interact laterally with several binding partners, and these interactions are frequently sensitive to Triton-X100 or NP40-based lysis conditions [37,39]. Our observation that the interaction of JAM-A with α3β1 integrin is sensitive to NP40-based lysis but detectable in Brij98-based lysis buffers suggested that the JAM-A -α3β1 integrin interaction is indirect and mediated by a member of the tetraspanin family. We focussed on CD151/Tspan24, which directly interacts with α3β1 integrin through an association that involves the extracellular domains of the two proteins [44][45][46], and which regulates α3β1 function in many cell types [47][48][49][50][51][52][53]. Due to the lack of antibodies that recognize canine CD151 in Western blot analyses, we ectopically expressed a EGFP-CD151 fusion protein in MDCKII cells and analyzed its interaction with JAM-A by CoIP. JAM-A readily co-immunoprecipitated with CD151 (Fig. 5A). CD9/Tspan29 has been described as a second α3β1 integrin-interacting Tspan [40,54].

Similar to the interaction of JAM-A with α3β1 integrin, the interaction with CD151 was retained
Previous findings showed that JAM-A interacts with CD9 in platelets and endothelial cells [26,55]. We therefore analyzed the association of JAM-A with CD9 in MDCKII cells. We found that CD9 coimmunoprecipitates with JAM-A (Fig. 5D). This interaction was retained in the absence of the cytoplasmic domain of JAM-A (Fig. 5E) suggesting that in contrast to endothelial cells, in which the interaction with CD9 is mediated through its PDZ domain binding motif [26], JAM-A interacts with CD9 in MDCK epithelial cells through its extracellular domain. Overall, these findings suggested that JAM-A exists with α3β1 integrin in TEM and that the physical interaction of JAM-A with α3β1 integrin can be mediated by Tspan proteins CD151 and CD9 ( Fig. 5F).

Depletion of CD151 attenuates collective cell migration
To address a possible role of the interaction of JAM-A with α3β1 integrin through Tspan CD151 we depleted CD151 in MDCKII cells by RNAi and analyzed the migration velocity during collective cell migration. Knockdown of CD151 significantly attenuated migration velocities of collectively migrating cells (Fig. 6). Similar to what was observed after depletion of α3β1, the depletion of both CD151 and JAM-A reduced the migratory speed of collectively migrating cells on both major α3β1 integrin ligands, i.e. CNI and LN (Fig. 6). Overall, these findings suggested that JAM-A, CD151 and α3β1 integrin co-exist in a TEM to form a functional unit which regulates collective cell migration in polarized epithelial cells.

JAM-A is a cell-cell adhesion receptor that is localized at cell-cell junctions of polarized
epithelial cells with enrichment at the TJs [18,56,57]. Its role in the regulation of the epithelial barrier function, which involves phosphorylation at two sites in the cytoplasmic tail, has been intensively studied [18,27,33,57]. By addressing JAM-A's role in cell migration of polarized epithelial cells, we find that JAM-A limits cell motility when cells migrate as individuals but promotes cell migration when cells migrate as a cellular sheet. The regulation of collective cell migration is most likely mediated through its interaction with α3β1 integrin, which is mediated by Tspan CD151 and perhaps CD9. We propose the existence of a TEM in which Tspan CD151 connects JAM-A to α3β1 integrin to assemble a signalling complex that allows a functional crosstalk of JAM-A and α3β1 integrin during collective cell migration. the follower cells in their center over a long range [11-13, 15, 64], reviewed in [65,66]. Tension

on JAM-A applied through trans-homophilic JAM-A interaction activates RhoA in a Ser285
phosphorylation-dependent manner in CHO cells [67]. Importantly, in MDCK cells JAM-A regulates tensile stress on the TJ scaffolding protein ZO-1 through its interaction with PDZ domain 3 [62,68]. These observation thus suggest the possibility that JAM-A localized at TJs contributes to the regulation of collective cell migration by transmitting tensile forces at the TJs to its directly associated scaffold protein ZO-1 which changes conformation in response to tensile forces [62,69]. This possibility is supported by our observation that Ser285 phosphorylation of JAM-A, which is observed exclusively at the TJs [18], is required for the regulation of collective cell migration of JAM-A (Fig. 2). JAM-A might thus contribute to the regulation of collective cell migration by different mechanisms, one involving the transmission of mechanical strain through TJs and concurrent regulation of actomyosin contractility, traction forces and focal adhesion formation via RhoA [62,69,70], and one involving the regulation of integrin-dependent functions through its association with α3β1 integrin (see below).

The JAM-A -CD151 -α3β1 TEM
As the second major observation of our study, we find that JAM-A interacts with α3β1 integrin i.e. Brij98 [37,40]. We identified Tspans CD151 and CD9 as possible mediaters of this interaction. Both tetraspanins are binding partners of α3β1 integrin [40]. We consider CD151 as more likely to link JAM-A and α3β1 integrin during collective cell migration since CD151 has been found to regulate a number of α3β1 integrin-dependent functions including cell-matrix adhesion, cell motility and migration on LN [46,50,53,71], intercellular junction formation [48,52,72] or cell survival [73]. Our observations that depleting either α3β1 integrin or CD151 results in a similar retardation of collective cell migration in MDCK cells like depleting JAM-A supports the view that JAM-A regulates collective cell migration of MDCK cells through its physical association with α3β1 integrin mediated by CD151 and possibly CD9.
How JAM-A regulates collective cell migration through these associations remains unclear. Both α3β1 integrin and CD151 have been identified at cell-cell junctions of various cell types including tumor cells [52,72], keratinocytes [73,74], and kidney epithelial cells derived from collecting ducts [48]. Therefore, it is likely that the CD151 (and possibly CD9)based TEM containing JAM-A and α3β1 integrin is localized at cell-cell junctions. We speculate that JAM-A regulates α3β1 integrin-associated functions. Studies in platelets have shown that Tyr-phosphorylated JAM-A recruits the Src kinase Csk to inhibit αIIbβ3-associated Src during outside-in signalling [75]. We therefore envisage a scenario for MDCK cells in which the lateral association of JAM-A with α3β1 integrin through a tetraspanin as linker protein might serve to generate spatial proximity between JAM-A and possibly a JAM-A-associated regulatory protein and an α3β1 integrin-associated signalling molecule. Interestingly, we found that a JAM-A construct with a mutation in the known Tyr phosphorylation site (JAM-A/Y281F) [33] fails to restore single cell and collective cell migration in JAM-A null MDCK cells to levels observed after expression of JAM-A/WT (Fig.1, Fig. 2), strongly suggesting that Tyr281 phosphorylation is required to mediate JAM-A's function during single and collective cell migration. This observation is intriguing since phosphorylation/dephosphorylation allows for dynamic regulation of signalling events as they occur during cell migration. Recent observations in integrin-based cell-matrix adhesion complexes indicate that signal propagation within these complexes does not involve gross changes in the composition of the adhesion complex, but is regulated by the relay of phosphotyrosine-dependent signals [76]. The incorporation of JAM-A and α3β1 integrin in a single protein cluster by the activity of CD151 and possibly CD9 could thus serve to rapidly transmit phospho-JAM-A-dependent regulatory events on α3β1 integrindependent signalling pathways.