Expression of integrins to control migration direction of electrotaxis

Abstract

Proper control of cell migration is critically important in many biologic processes, such as wound healing, immune surveillance, and development. Much progress has been made in the initiation of cell migration; however, little is known about termination and sometimes directional reversal. During active cell migration, as in wound healing, development, and immune surveillance, the integrin expression profile undergoes drastic changes. Here, we uncovered the extensive regulatory and even opposing roles of integrins in directional cell migration in electric fields (EFs), a potentially important endogenous guidance mechanism. We established cell lines that stably express specific integrins and determined their responses to applied EFs with a high throughput screen. Expression of specific integrins drove cells to migrate to the cathode or to the anode or to lose migration direction. Cells expressing αMβ2, β1, α2, αIIbβ3, and α5 migrated to the cathode, whereas cells expressing β3, α6, and α9 migrated to the anode. Cells expressing α4, αV, and α6β4 lost directional electrotaxis. Manipulation of α9 molecules, one of the molecular directional switches, suggested that the intracellular domain is critical for the directional reversal. These data revealed an unreported role for integrins in controlling stop, go, and reversal activity of directional migration of mammalian cells in EFs, which might ensure that cells reach their final destination with well-controlled speed and direction.—Zhu, K., Takada, Y., Nakajima, K., Sun, Y., Jiang, J., Zhang, Y., Zeng, Q., Takada, Y., Zhao, M. Expression of integrins to control migration direction of electrotaxis.

Cell migration direction in our bodies is tightly controlled. For example, immune cells are attracted to infected sites; then some cells reverse migration direction and move back to lymph nodes (1). To achieve proper function in wound healing and regeneration, cells must have proper control to initiate, speed up, slow down, stop, and sometimes reverse migration. The mechanisms determining when to initiate, which direction to move, and when to stop are critical for immune response, wound healing, and cancer metastasis. Loss of this control underlies important disease conditions, such as nonhealing wounds or cancer metastasis. However, the molecular mechanisms are not fully understood, and they are mainly worked out in more linear signaling pathways (2–4). Similar to differentiation of stem cells (5), a molecular network dictates the behavior landscape to bias cell migration directions.

Integrins are one of the most important groups of molecules that could effectively regulate cell migration. They are expressed as transmembrane heterodimers that consist of individual α and β chains, of which at least 24 αβ combinations are known so far in vertebrates. Most integrins are widely distributed, whereas others are limited to certain cell types or tissues (e.g., αIIbβ3 is restricted to platelets, α6β4 to keratinocytes, αEβ7 and α4β7 to T cells, and α4β1 and the β2 subfamily to leukocytes) (6). The expression of integrins undergoes a tightly controlled yet distinct pattern during wound healing (7, 8). For example, α9β1 (9) and αvβ6 (10) are up-regulated upon corneal injury, whereas α3β1 expression is decreased in diabetic wounds (11). Cells in immune response, cancer metastasis, and cancer development also show some drastic patterns of integrin expression (12, 13). As a cluster of specialized transmembrane proteins, integrins have long been proposed to be among the most important modulators in cell migration because they mediate signal transduction bidirectionally through the membrane and play an essential role in the regulation of dynamic interaction between extracellular microenvironment and migration-related molecules. Integrins can be grouped into subfamilies by their evolutionary and functional relationships and because they perform different mechanical or signaling tasks and regulatory roles in cell migration (14, 15). Knockout of individual integrins leads to a wide range of phenotypes, which reflect the unique and nonredundant roles of various integrins (6).

Small electric fields (EFs) are detected in development, wound healing, and cancer metastasis, in which cell migration fluctuates significantly (16–20), and these EFs could induce cell directional migration as well as an increase in migration speed. Remarkably, different types of cells migrate in opposite directions when exposed to applied EFs (18, 21–24), and even the same cell can switch migration directionality depending on EF strength (25). What determines the directional choice remains elusive.

Here, we established a collection of cell lines expressing specific integrins. Using a large-scale testing system and direct current EF as a directional cue, we systematically quantified the roles of individual integrins in regulating the motility and directionality of cell migration in an EF. We confirmed the results with pharmacological inhibition and demonstrated that protein kinase B (Akt)–green fluorescent protein (GFP) polarization was disrupted by expressing specific integrins that reverse migration directionality. By expressing chimeric integrins, we revealed key domains that underlie the directional reversal.

MATERIALS AND METHODS

Materials

DMEM, fetal bovine serum (FBS), nonessential amino acid solution (100×), penicillin-streptomycin, and G418 were purchased from Thermo Fisher Scientific (Waltham, MA, USA). X-tremeGene high-performance DNA transfection reagent was purchased from Roche (Basel, Switzerland). pcDNA3-AKT-PH-GFP (plasmid 18836) (26), chimeric integrin alpha9/alpha4 pcDNA1 neo (plasmid 13587), and alpha9/alpha5 pcDNA1 neo (plasmid 13588) were purchased from Addgene (Watertown, MA, USA). A disintegrin and metalloproteinase 3 (ADAM-3) was synthesized as a glutathione-S-transferase (GST) fusion protein as previously described (27). Anti-α9 antibody Y9A2 was purchased from MilliporeSigma (Burlington, MA, USA). Sylgard 184 silicone elastomer [polydimethylsiloxane (PDMS)] was purchased from Dow DuPont (Midland, MI, USA). Steinberg’s solution was prepared using a recipe previously published (28). Agar was purchased from MilliporeSigma. Silver wires with 99.999% purity were purchased from Advent Research Materials (Oxford, United Kingdom).

Establishment of Chinese hamster ovary cell lines expressing specific human integrins

Chinese hamster ovary (CHO) K1 cells [American Type Culture Collection (ATCC), Manassas, VA, USA] were used as parental cells in this study and cultured in DMEM supplemented with 10% FBS, nonessential amino acid solution, and penicillin-streptomycin at 37°C with air containing 5% CO₂. CHO cells have low integrin background expression. Based on our knowledge, they express hamster integrins α5 (29), αV, β1, and very little β3, β4, α4, α6, or α9. To test the roles of individual integrins in regulating the motility and directionality of cell migration in an EF, we developed CHO cells that express specific human integrins. The cDNA constructs were transfected into CHO cells together with a neomycine-resistant gene. Cell lines expressing human α2 (α2-CHO), human α3 (α3-CHO), human α4 (α4-CHO), human α5 (α5-CHO), human α6 (α6-CHO), human α9 (α9-CHO), human αV (αV-CHO), human β1 (β1-CHO), human β3 (β3-CHO), human αLβ2 (αLβ2-CHO), human αMβ2 (αMβ2-CHO), human α4β7 (α4β7-CHO), human α6β4 (α6β4-CHO), and human αIIbβ3 (αIIbβ3-CHO) have been previously described (30–37). Site-directed mutagenesis for human α9 mutants was carried out using the unique site elimination method (38). The presence of mutations was verified by DNA sequencing. Transfected cells were maintained in DMEM supplemented with 10% fetal calf serum at 37°C in 5% CO₂ for 2 d. Then the cells were transferred to the same medium containing 500–1000 µg/ml G418. After selection with G418 for 10–14 d, cells stably expressing human integrins were cloned by sorting to obtain high expressers. Flow cytometry was performed as previously described (39).

Electrotactic chamber design and fabrication

The multichannel electrotactic chamber is composed of 2 PDMS layers. The bottom layer consists of 15 parallel channels, each 1 cm long and 1 mm wide, for EF-stimulated cell migration, and the top frame contains 2 reservoirs. The 2-layer PDMS structure can be fabricated and assembled by employing standard laser micromachining and oxygen plasma-assisted bonding (40). In brief, PDMS prepolymer at a mixing ratio of 10:1 was thoroughly mixed and degassed in a desiccator for 20 min. The bottom layer was fabricated in a 400 µm–thick PDMS slab, and the top layer was fabricated in a 1.5 mm–thick PDMS slab. After thorough cleaning of the PDMS pieces in an ethanol ultrasonic bath, 2 layers were aligned and permanently bonded together following an oxygen plasma treatment at 90 W for 20 s. The chamber was thoroughly cleaned and affixed to a Petri dish prior to the electrotaxis experiment. We also used the traditional electrotactic chamber, which has been previously described (28), for further experiments.

Electrotaxis assay and time-lapse imaging

Cells expressing specific human integrins were loaded into the electrotactic channels singly with a final density of ∼150 cells per square millimeter and cultured in the chamber overnight at 37°C with air containing 5% CO₂ to allow sufficient attachment. Current was applied to the chamber through agar-salt bridges connecting with silver-silver chloride electrodes in Steinberg’s solution as previously described (28). CO₂-independent medium (Thermo Fisher Scientific) with 10% FBS (∼5 ml) was added into reservoirs to ensure good salt bridge contact and to support cell viability during EF stimulation. A field strength of 300 mV/mm was used unless otherwise noted. EF strengths were monitored at the beginning of the experiment and every 30 min afterward to ensure consistent EF application. Cell migration was observed with an Observer Z1 inverted microscope (Carl Zeiss, Oberkochen, Germany) with the MetaMorph NX program (Molecular Devices, Sunnyvale, CA, USA). The microscope system was able to record serial time-lapse images of multiple locations on the multichannel electrotaxis chamber simultaneously. So, all cell lines in the multichannel chamber could be monitored and tested in 1 single experiment. A regular ×10 phase contrast objective lens was used for microscopy. Images were taken at 5-min intervals.

Quantification of cell migration

For each cell line, over 80 cells were analyzed. The position of a cell was defined by its centroids. Cells that divided, moved in and out of the field, or merged with other cells during the experiment were excluded from analysis. Cell migration was analyzed to determine directedness (cos θ) and track speed by using ImageJ software (National Institutes of Health, Bethesda, MD, USA; http://rsbweb.nih.gov/ij/) with MTrackJ and Chemotaxis tool plugins as previously described (18, 41). Briefly, the directedness was used as an indicator of directionality of cell migration, which is defined as cosine of the angle between the EF vector and a straight line connecting the start and end positions of a cell. A cell migrating directly toward the cathode would have a directedness of 1; a cell migrating directly to the anode would have a directedness of −1. For a group of cells, a mean directedness close to 0 represents random migration. Moreover, cell migration rate was quantified as the track speed, which was presented as the accumulated migration distance per hour.

Immunostaining and fluorescence imaging

Cells were rinsed with chilled PBS immediately after EF stimulation and fixed with 4% paraformaldehyde for 20 min on ice. For immunofluorescence labeling, cells were permeabilized in PBS with 0.1% Triton X-100 for 15 min. Nonspecific binding sites were blocked for 1 h with 10% donkey serum at room temperature. Cells were then incubated overnight at 4°C with the primary antibody. Y9A2 from MilliporeSigma was used to label integrin α9. After incubation with goat anti-mouse secondary antibody–Alexa Fluor 594 for 1 h, samples were stained by phalloidin-FITC (MilliporeSigma) for 30 min to visualize the actin. Each step was followed by 3 5-min washes in PBS. Cell nuclei were labeled with DAPI. Images were acquired with epifluorescence microscopy (Carl Zeiss Observer Z1) and analyzed using MetaMorph NX and ImageJ.

Imaging of AKT-PH-GFP

Cells were seeded at a density of 2.0 × 10⁵ cells per well in a 6-well plate 1 d before transfection. The cells were transfected with 1000 ng per well of pcDNA3-AKT-PH-GFP plasmid using X-tremeGene high-performance DNA transfection reagent according to the manufacturer’s protocol. Twenty-four hours after transfection, cells were trypsinized and reseeded onto the electrotaxis chamber. Serial time-lapse enhanced GFP fluorescence and phase contrast images were captured with or without EF stimulation.

Adhesion assay

Adhesion assays were performed as previously described (42). Briefly, 96-well Immulon microtiter plates (Thermo Fisher Scientific) were coated with 100 µl of 0.1 M NaHCO₃ containing ADAM-3 or GST and were incubated at 37°C for 2 h. Remaining protein-binding sites were blocked by incubating with PBS and 0.1% bovine serum albumin for 30 min at room temperature. After washing with PBS, CHO, α9-CHO, and mutants in 100 µl of DMEM and 0.1% bovine serum albumin were added to the wells and incubated at 37°C for 1 h. After unbound cells were removed by rinsing the wells with DMEM, bound cells were quantified by measuring endogenous phosphatase activity. Wild-type GST was used as a negative control.

Statistical analysis

All data are represented as means ± sem. Statistical analyses were performed using GraphPad Prism v.7.0 (GraphPad Software, La Jolla, CA, USA) with 1-way ANOVA followed by Dunnett’s test or unpaired 2-tailed Student’s t test, and P value was set at 0.05 for rejecting null hypotheses.

RESULTS

Integrin expression influences the basal motility

To investigate the effects of integrin expression on cell migration, we first transfected specific human integrins in plasmid vector individually with a neomycin-resistant gene into CHO cells, which have very low background integrin expression (mainly hamster α5β1 and αvβ1) (29), and we used G418 to select stably expressing clones. Flow cytometry and cell sorting were used to confirm expression of specific integrins and to ensure optimal expression level of specific human integrins (Fig. 1A). We selected 14 lines expressing the following integrins for migration analysis because they are implicated in cell migration, development, and tumor progression (43–48). The α2-CHO, α3-CHO, α4-CHO, α5-CHO, α6-CHO, α9-CHO, and αV-CHO cells homogeneously expressed human α2–, α3–, α4–, α5–, α6–, α9–, and αV–hamster β1 hybrids, respectively. The β1-CHO cells expressed human β1–hamster α5 hybrid. The β3-CHO cells expressed human β3–hamster αV hybrids. The αLβ2-CHO, αMβ2-CHO, αIIbβ3-CHO, α6β4-CHO, and α4β7-CHO cells expressed human αLβ2, αMβ2, αIIbβ3, α6β4, and α4β7, respectively. Cell spreading is dependent upon effective integrin-cytoskeletal interactions. Different integrin cells developed divergent morphology (Fig. 1B). Expression of integrins α5, αV, αMβ2, and β3 led to a flattened, polygonal morphology, whereas α4-, α6-, and β1-expressing cells formed a microspike morphology.

Figure 1.

Integrin expression controls cell basal motility. A) CHO cells transfected with designed human integrin cDNA in plasmid vector with a neomycin resistance gene. After selection by G418 resistance, cells expressing human integrins were sorted by fluorescence-activated cell sorting with specific antibodies. B) Representative images of the pCHO and CHO cells expressing specific human integrins. Scale bar, 100 μm. C) Migration trajectories of pCHO and CHO cells expressing specific integrins. Scale bar, 50 μm. D, E) Migration speed (D) and persistence (E) of pCHO cells and CHO cells expressing specific human integrins. Eighty to 100 cells were analyzed for each cell line. *P < 0.05, **P < 0.01, 1-way ANOVA followed by Dunnett’s test, compared with pCHO.

We then tested the basal motility of these integrin cell lines by measuring migration speed and persistence. Migration speed is the total length traveled (trajectory length) by the cells divided by time, and persistence is the ratio of displacement distance to trajectory length traveled by a cell, which indicates whether a cell migrated directly or had a more wandering pathway. Basal migration speed of parental CHO cells (pCHO) is 30.2 ± 2.7 µm/h. Expression of α2 significantly increased the migration speed to 42.1 ± 4.5 µm/h (P < 0.01), whereas α3 (15.9 ± 1.2 µm/h, P < 0.01), α5 (15.2 ± 1.5 µm/h, P < 0.01), α6 (16.8 ± 1.4 µm/h, P < 0.01), αMβ2 (10.7 ± 1.3 µm/h, P < 0.01), and αV (19.5 ± 2.0 µm/h, P < 0.05) significantly decreased the speed (Fig. 1C, D). In addition, expression of human integrins α4 and α6β4 decreased cell migration persistence (Fig. 1E) when compared with that of the pCHO cells.

Integrin expression profiles dictate electrotaxis

To determine the role of integrins in electrotaxis, we used a multichannel electrotaxis chamber to expose up to 25 strains of cells simultaneously to a 300-mV/mm EF (Fig. 2A, B). Expression of different integrins had significant effects on both the migration speed and directionality in the EF. pCHO cells migrated directionally to the cathode at the speed of 51.4 ± 2.8 µm/h (Fig. 2C, D and Supplemental Table S1). Cells expressing human α2, α4, α6, α9, β1, and α4β7 had significantly increased migration speed (Fig. 2D). Remarkably, expression of integrin α2 increased the migration speed by almost 2-fold (97.5 ± 2.8 µm/h) when compared with the parental cells, which is consistent with the basal motility increase (Fig. 2C, D and Supplemental Fig. S1). Expression of α5, αv, and αIIbβ3 significantly decreased migration speed, whereas expression of α3, αLβ2, αMβ2, β3, and α6β4 did not affect the migration speed in an EF of 300 mV/mm (Fig. 2D). Moreover, EF stimulation increased cell migration speed regardless of integrins (Supplemental Fig. S1). The migration persistence of α2-CHO, α9-CHO, β1-CHO, and α6β4-CHO cells were significantly increased in an EF, whereas that of αLβ2-CHO was decreased.

Figure 2.

Integrin expression determines electrotactic response. A) Multichannel electrotactic chamber design. Simulation confirmed the desired EF geometry. B) Schematic drawing of the EF application. C) Migration trajectories of parental cells and CHO cells express specific human integrins in an EF. Black and red lines indicate trajectories of cells migrating toward cathode and anode, respectively. Scale bars, 100 μm. D, E) Migration speed (D) and directedness (E) of pCHO cells and CHO cells express specific human integrins. Data are shown as means ± sem from 100–130 cells; results were consistent in 3 independent experiments. *P < 0.05, **P < 0.01, 1-way ANOVA followed by Dunnett’s test, compared with pCHO.

Strikingly, expression of integrins β3, α6, and α9 completely changed the migration direction to the anode (Fig. 2C, E and Supplemental Table S1). Expression of α3, αLβ2, and α4β7 decreased, and α4, αV, and α6β4 completely abolished the electrotaxis, whereas expression of αMβ2, α2, α5, β1, and αIIbβ3 did not have significant effects on the directedness of cell migration (Fig. 2C, E). We next confirmed the distinct roles of integrins α2, α4, and α9 in the directional determination of electrotaxis by using a traditional electrotaxis chamber. α2-CHO cells migrated directionally to the cathode, similarly to the parental cells but in a more robust way (Fig. 3A, B and Supplemental Movie S1). Expression of α2 significantly increased both accumulated and Euclidean distance when compared with that of the pCHO cells (Fig. 3C). α4-CHO cells did not show significant electrotactic responses in 3 h. By contrast, α9-CHO cells migrated to the anode, which is opposite to pCHO cells (Fig. 3A, B and Supplemental Movie S1).

Figure 3.

Contrasting directional determination of cathodal and anodal integrins. A) Representative time-lapse images show movement of pCHO, α2-CHO, α4-CHO, and α9-CHO cells in response to indicated EF (see Supplemental Movie S1). The track lines indicate migration paths. B, C) Directedness (B) and migration distance (C). Data are shown as means ± sem from 80–100 cells. **P < 0.01, compared with accumulated distance of pCHO; ^##P < 0.01, compared with Euclidean distance of pCHO.

In addition, we also found that integrin-mediated anodal migration is serum independent (Supplemental Fig. S2). pCHO and α2-CHO cells did not exhibit significant electrotactic responses in serum-free medium. However, α9-CHO and β3-CHO cells showed robust directional migration to the anode.

Integrin α9 specifically mediated the anodal electrotaxis

To confirm that integrins induced reversal of migration direction, we selected cells expressing α9, which showed the strongest directional reversal from the parental line (Fig. 3D). The expression of human integrin α9 was further verified using immunostaining (Fig. 4). We then used an α9-specific antibody, Y9A2, to inhibit α9 function. As expected, cells with blocked α9 resumed cathodal migration like that of the parental cells, albeit with decreased migration speed (Fig. 5 and Supplemental Movie S2).

Figure 4.

Integrin α9 distribution in CHO cells. CHO and α9-CHO cells were fixed and stained with Y9A2 for α9 integrin (red), phalloidin for F-actin (green), and DAPI for nuclei (blue). For α9-CHO in EF, cells were exposed to an EF of 300 mV/mm for 3 h and then fixed and stained immediately. Polarity of the EF was as shown. Yellow arrows indicate the integrin footprints of migration traces.

Figure 5.

Switching migration direction during electrotaxis by inhibition of integrin α9. A) Migration trajectories of α9-CHO cells and Y9A2-treated α9-CHO cells in an EF. Polarity and the strength of the EF are as shown. Red and black lines indicate trajectories of cells migrating toward anode and cathode, respectively. Scale bar, 100 μm. B, C) Directedness (B) and migration speed (C) of α9-CHO cells in the presence and absence of Y9A2. Data are shown as means ± sem from 50 to 80 cells; results were consistent in 3 independent experiments. *P < 0.05, **P < 0.01, Student’s t test.

Next, we used a cell polarity marker, phosphatidylinositol (3,4,5)-trisphosphate (PIP3), to determine the effect of expression of α9 on EF-induced Akt-GFP localization. The location of Akt-GFP in a cell indicates the activation of PIP3. pCHO cells undergoing cathodal migration recruited PIP3 to the leading edge on the cathode-facing side (Fig. 6), which is consistent with previous publications (18, 49). When α9 was expressed in the cells, this cathodal polarization of PIP3 was disrupted (Fig. 6), which suggested that other signaling pathways may take the place of PIP3 to promote anodal migration. We also noticed that α9 integrin can separate from the cell body and remain associated with the substratum as integrin footprints, although it did not show obvious polarization during directional migration (Fig. 4).

Figure 6.

Expression of integrin α9 disrupted cathodal distribution of Akt-GFP, a reporter for PIP3 localization. CHO and α9-CHO cells were transfected with pcDNA3-AKT-PH-GFP plasmid DNA and incubated for 24 h after transfection. Then cells were exposed to an EF of 300 mV/mm, and fluorescence of Akt-GFP was recorded by fluorescence microscope. Right and left white arrows show migration direction of CHO and α9-CHO cells, respectively. Yellow arrows indicate polarized Akt-GFP signals on the cathode-facing side. Polarity of the EF was as shown.

Cytoplasmic domain of integrin α9 determines the directional switch

To elucidate the molecular mechanisms of α9 in the reversal of migration direction, we first identified the putative ligand-binding sites of integrin α9. Our previous study determined that residues 181–190 in the third amino-terminal repeats of α4 and α5 are critical for ligand binding to α4β1 and α5β1, respectively, and suggested that the corresponding region may be ubiquitously involved in ligand binding of non–I domain integrins (50). Thus, we here made 2 α9-mutant cell lines in which Tyr187 was replaced by Ala (Y187A) and Gly190 was replaced by Ala (G190A) (Fig. 7A). These cell lines showed significantly lower adhesion to ADAM-3 (Fig. 7B, C), a disintegrin domain that has been characterized as ligands for integrin α9 (27). These ligand-binding–defective mutant lines, however, maintained the anodal migration to the same extent as the wild-type α9 (Fig. 7D). The migration speed was not affected (Fig. 7E). The binding of α9 appeared not to be involved in the directional reversal.

Figure 7.

Ligand-binding defection does not affect integrin α9–directed anodal migration. A) Flow cytometry quantifications showing transfected CHO cells expressing integrin α9 and its ligand-binding–defective mutants. B, C) Binding assay showing that integrin α9 ligand-binding–defective mutants α9 Y187A and G190A could not bind to ADAM-3 (B), a specific ligand that can be recognized by integrin α9. Wild-type GST protein (C) was used as a negative control. D, E) Directedness (D) and migration speed (E) of CHO, α9-CHO, and α9 mutants in an EF of 300 mV/mm. Data are shown as means ± sem from 50 to 80 cells; results were consistent in 3 independent experiments. pcDNA, plasmid complementary DNA; PE-A, phycoerythrin-conjugated Antibody; WT, wild type.

To further dissect out which domain of α9 is responsible for the reversal, we built 2 chimeric integrin cell lines in which the extracellular and transmembrane domain of the α9 subunit fused to the cytoplasmic domains of α4 or α5 (51). The α9-α4 chimera did not reverse the migration direction, whereas the α9-α5 chimera restored the cathodal directional migration (Fig. 8), which suggested that the cytoplasmic domain of integrin α9 determines the directional switch.

Figure 8.

Requirement of the cytoplasmic domain of α9 is required for the directional reversal. Directedness (A) and migration speed (B) of CHO, wild-type (wt), and chimeric α9-CHO cells in an EF of 300 mV/mm. Data are shown as means ± sem from 50 to 80 cells; the results were consistent in 3 independent experiments.

DISCUSSION

In this study, we established a group of cell lines expressing specific integrins and screened their responses in an applied EF using a multichannel system. Our study reveals the distinct roles of integrins in modulating migration speed, persistence, and direction. We identified 3 categories of electrotaxis phenotypes: cathodal migration, random migration, and anodal migration. Focusing on the most striking directional switch—integrin α9, whose expression makes the cells move to the opposite direction of the parental cells—our results suggest that the cytoplasmic domain is critical for the directional switch.

Integrins can regulate keratinocyte functions during wound healing, including cell migration, survival, and proliferation. Some integrins display persistent or enhanced expression during wound healing (e.g., α2β1, α3β1, α6β4, α9β1, and αvβ5) (52, 53), whereas others are expressed de novo (e.g., α5β1 and αvβ6) (54). Changing the integrin expression profile could also affect cell migration during cancer metastasis; for example, expression of integrins α2β1 (55) and αvβ3 (56) on human cells could enhance metastasis and invasiveness, respectively. α6β4 also involves in epithelial tumor progression in response to specific matrix environments (57) (e.g., laminin) (58), and the ability of α6β4 to cooperate with specific growth factor receptors may be a mechanism for the robust migration of tumor cells (59). In the present study, we expressed human integrins in CHO cells, which is a powerful and frequently used model to study integrin binding and signaling because of the low integrin expression background. Our data showed that expression of α2 in CHO cells significantly increased the basal migration speed of single cells, whereas α3 and α5 decreased migration speed (Fig. 2B). However, expression of α9β1, α6β4, or β3 did not affect the basal motility of CHO cells. In addition, we tested the effects of integrin expression on collective cell migration by a scratch wound assay. The results showed that integrin α2 significantly accelerated the wound closure, whereas α3 slightly decreased the healing rate (Supplemental Fig. S3). Though similar, the results from the scratch assay and single cell random migration do not exactly reflect one another. Specifically, expression of integrins α5, αV, and αMβ2 decreased the random migration speed; however, these cells showed normal or even higher recovery rate in the monolayer scratch assay. These results suggested that integrins may have different roles in random and directional cell migration and that different expression profiles of integrins might contribute to temporal differences of cell motility and migration direction in vivo (7–10).

To precisely control wound healing, angiogenesis, immune response, and development, cells must become motile and also migrate in the correct direction. Endogenous EFs are an effective directional cue for cell migration, which has been demonstrated by showing migratory preference toward the cathode or anode in many cell types, including corneal epithelial cells (18), keratinocytes (60), endothelial cells (23), lymphocytes (21), stem cells (22, 24), and cancer cells (61). However, how a cell perceives external EFs and determines migration direction remains largely unknown. Membrane proteins, such as integrins, have been found to be one of the most important modulators in electrotaxis (49, 62). EFs could redistribute integrin α5β1 toward the trailing edge in cathodal-migrating fibroblasts (63). It is also reported that integrin α4β1 is required for anodal electrotaxis and localizes to the front edge of cardiac progenitor cells (64). Integrin β1, one of the key subunits of integrin receptors, has been found to be involved in the electrotaxis by polarization and activation of several kinds of cells, including fibroblasts (65), epithelial cells (66), keratinocytes (67), and fish keratocytes (68). Knockout of integrin β4 abolished the electrotactic response of keratinocytes, which can be recovered by re-expression of β4 (60). In addition, epidermal growth factor stimulation also recovers the electrotactic response of β4 null cells, which suggests that there are multiple signaling mechanisms underlying the EFs that induce electrotactic cell migration. In our study, we demonstrated that expression profiles of integrins could dictate the cell electrotactic response. Using the directedness value, we grouped these integrin cells into 3 subpopulations: cathodal-migrating, anodal-migrating, and random-migrating cells. Cells expressing α2, α5, β1, αMβ2, and αIIbβ3 migrated to the cathode, whereas cells expressing β3, α6, and α9 reversed the migration direction to the anode. Cells expressing α3, αLβ2, α4β7, αV, and α6β4 migrated at random directions in an EF (Fig. 2C, E). We also tested different clones to confirm the decisive roles of integrins in directional migration. As shown in Supplemental Fig. S4, 2 α2-overexpressed clones migrate consistently to the cathode, whereas overexpression of α6 reversed migration direction in varying degrees. Altogether, this study demonstrated that different members of the integrin family are 1) motility enhancers, whose expression significantly increased the motility; 2) motility suppressors, whose expression inhibited migration; 3) directional switches, whose expression completely reversed cell migration direction from parental cells; and 4) directional restrainers, whose expression restrained cells from directional migration.

Each integrin has a large extracellular domain and a short cytoplasmic tail (14). The extracellular domains can bind a variety of ligands, whereas the intracellular cytoplasmic domains anchor cytoskeletal proteins. This linkage between the cell exterior and interior allows for bidirectional signaling across the plasma membrane. The contribution of the extracellular domain of integrins to cell migration is most likely due to an effect on the avidity of ligand binding. Here, our results showed that ligand-binding defection does not affect integrin α9–mediated anodal migration. Previous reports have demonstrated that the cytoplasmic domain is crucial for α9-mediated enhancement of cell migration (51). We tested the role of α9 cytoplasmic domain in determining migration direction and found that the α9-α5 chimera restored the cathodal directional migration (Fig. 8). The α4 cytoplasmic domain did not affect the α9-mediated anodal migration, which may be due to the sequence similarity.

However, it is still not clear how signals transduced to the cytoskeleton subsequently promote directional migration of the cells. It is reported that an α9 integrin downstream inwardly rectifying K⁺ (Kir) channel, Kir4.2, is involved in α9-enhanced cell migration (69). Kir4.2 colocalizes with α9β1 integrin at the leading edge in glioma and CHO cells (70). It is known to be presented in a wide variety of tissues, and we recently demonstrated that Kir4.2 couples with polyamines to sense extracellular EF for both cathodal- and anodal-migrating cells (71). The commonly proposed mechanism for directional migration of a cell in response to extracellular stimuli is to establish cell polarity (72). It is accomplished through the formation of the front protrusion and rear retraction, which are controlled by distinct cell-signaling events. It is reported that binding of paxillin to the α4 integrin complex can inhibit stable lamellipodium formation by blocking Rac activation to the cell anterior (73). In our present study, we showed that overexpression of the human α4 subunit significantly decreased migration persistence, which is consistent with the previous report. Without the stable lamellipodium formation, α4-CHO cells showed random migration in an EF at a similar or higher speed (Figs. 2 and 3). However, paxillin is not required for α9-dependent enhancement of cell migration (51), which suggests that it not likely to be involved in α9-mediated migration reversal.

In summary, our results reveal a previously unknown mechanism for dictating cell migration direction in an EF by differential expression of integrins. Expression of β3, α6, or α9 resulted in the most striking directional reversal of the cell migration. The intracellular domain, not the extracellular domain, of the molecular directional switch α9 is critical for the directional reversal. The powerful role of the specific integrin subunit in determining directionality might have significant implications in cell migration directional reversal in vivo.

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.

Glossary

ADAM-3

A disintegrin and metalloproteinase 3

Akt

protein kinase B

CHO

Chinese hamster ovary

EF

electric field

FBS

fetal bovine serum

GFP

green fluorescent protein

GST

glutathione-S-transferase

Kir

inwardly rectifying K⁺

pCHO

parental CHO

PDMS

polydimethylsiloxane

PIP3

phosphatidylinositol (3,4,5)-trisphosphate

AUTHOR CONTRIBUTIONS

K. Zhu, Y. Takada, and M. Zhao designed the experiments; K. Zhu, Y. Takada, K. Nakajima, Y. Sun, J. Jiang, Y. Zhang, and Q. Zeng performed the experiments and analyzed the results; and K. Zhu and M. Zhao wrote the manuscript.