We have shown previously that T1α/podoplanin is required for capillary tube formation by human lung microvascular lymphatic endothelial cells (HMVEC-LLy) and that cells with decreased podoplanin expression fail to properly activate the small GTPase RhoA shortly after the beginning of the lymphangiogenic process. The objective of this study was to determine whether podoplanin regulates HMVEC-LLy migration and whether this regulation is via modulation of small GTPase activation. In analysis of scratch wound assays, we found that small interfering RNA (siRNA) depletion of podoplanin expression in HMVEC-LLy inhibits VEGF-induced microtubule-organizing center (MTOC) and Golgi polarization and causes a dramatic reduction in directional migration compared with control siRNA-transfected cells. In addition, a striking redistribution of cortical actin to fiber networks across the cell body is observed in these cells, and, remarkably, it returns to control levels if the cells are cotransfected with a dominant-negative mutant of Cdc42. Moreover, cotransfection of a dominant-negative construct of Cdc42 into podoplanin knockdown HMVEC-LLy completely abrogated the effect of podoplanin deficiency, rescuing MTOC and Golgi polarization and cell migration to control level. Importantly, expression of constitutively active Cdc42 construct, like podoplanin knockdown, decreased RhoA-GTP level in HMVEC-LLy, demonstrating cross talk between both GTPases. Taken together, the results indicate that polarized migration of lymphatic endothelial cells in response to VEGF is mediated via a pathway of podoplanin regulation of small GTPase activities, in particular Cdc42.
- cell polarization
- microtubule-organizing center
- Rho small GTPases
- vascular endothelial growth factor
- wounding assays
lymphangiogenesis, the process of new lymphatic vascular formation, is used by normal tissues for development and repair of lymphatic vasculature, whereas tumors depend on it for establishment of lymphatic vascular connections so vital for metastasis. Polarized migration of lymphatic endothelial cells is considered crucial to the process, but the molecular regulatory pathway mediating this event in this cell type remains to be fully uncovered.
Migration of endothelial cells is stimulated by growth factors, notable among which is VEGF that also induces their proliferation and organization into capillary tubes. The migration process itself is a complex event involving dynamic reorganization of the actin cytoskeleton assembly and cell-matrix adhesion. It comprises sequential cycles of 1) lamellipodia extension and adherence to the substrata at the leading edge of the cell, 2) forward propulsion of the cell body, and 3) detachment from the substratum and retraction of the tail end (14, 29). The propulsive force for locomotion comes from an intricate, cablelike network of bundled actin and myosin filament associations, F-actin fibers, that contract to exert tension. A particular form of these fiber cables termed stress fiber is manifested when cells are undergoing stress or are stimulated by growth factors, and these terminate at specialized sites on the cell membrane, focal adhesions, where they are also anchored to the substrata via integrins. Modification of the nature and distributive location of actin fibers and their association with membrane-matrix anchors such as integrins and focal adhesion complexes permits cell movement (30).
The establishment/maintenance of cell polarity and actin cytoskeletal dynamics necessary for directional migration are regulated by the Rho family of small GTPases (reviewed in Refs. 12, 22, 33). Activation of Rho signaling increases actomyosin contractility and, as a consequence, the formation of stress fibers and focal adhesions (1, 8). Stimulation with VEGF induces stress fiber and focal adhesion formation in endothelial cells, and RhoA/Rho kinase (ROCK) signaling is involved in VEGF-induced endothelial cell migration and angiogenesis in vitro (25, 39). In lymphatic endothelial cells that were only recently recognized to differ from blood vascular endothelial cells in certain respects, the molecular regulatory pathways for some of these events observed in the presence of VEGF are yet to be defined.
Podoplanin is a type 1 transmembrane protein expressed by a number of cell types but is restricted in the endothelium to lymphatic endothelial cells. The functions of podoplanin are yet to be fully uncovered, but it has been implicated in tumor aggressive behavior by promoting epithelial-to-mesenchymal transition, migration, invasiveness, and metastasis to the pulmonary microvasculature (4, 5, 40, 41). Podoplanin-deficient mice have impaired lung and lymphatic development and die shortly after birth from respiratory failure; blood vascular development is, however, normal (27). Moreover, it is reported that in developing mouse embryo, interaction between podoplanin and platelets from the cardinal vein is required for the differentiation of the lymphatic system from the blood vascular system (36). We have previously reported that in human lung lymphatic microvascular endothelial (HMVEC-LLy) cells, early activation of RhoA in the lymphangiogenic process, which is required for the successful establishment of the capillary network, is dependent on podoplanin expression (18). Moreover, we (17) reported in abstract form that podoplanin is required for a full migratory response to VEGF in these cells and that complete reorientation of the microtubule-organizing center (MTOC) in response to VEGF in scratch wound assays, necessary for directional migration, is achieved only if the cells express podoplanin. In the present study, our main objective was to determine how podoplanin-dependent changes in GTPase activity contribute to polarization and migration in lymphatic endothelial cells.
MATERIALS AND METHODS
Cell culture and source of reagents.
Primary HMVEC-LLy cells were obtained from Lonza Walkersville (Walkersville, MD) and grown in EGM-2MV media, which consists of EBM-2 basal media, 5% FBS, and SingleQuots containing growth factors (VEGF, FGF-B, IGF-I, and EGF). Cells were trypsinized routinely using reagents recommended by Lonza Walkersville when they reached 70–80% confluency and then plated at 1:4 dilution in fresh complete media. Passages 3-7 were used for all experiments. Rho inhibitor CT04 was from Cytoskeleton (Denver, CO). VEGF used for wounding assays was recombinant VEGF165 from PeproTech (Rocky Hill, NJ).
Small interfering RNA and transfection.
Sequences corresponding to the silencing and the control RNA were manufactured by Invitrogen (Carlsbad, CA), and transfection protocol was performed exactly as described in Ref. 18 using Lipofectin.
EGFP, RhoA, and Cdc42 plasmids and transfections.
Enhanced green fluorescent protein (EGFP) plasmid pcDNA3-EGFP (Addgene plasmid 13031) was developed by Dr. Doug Golenbock; pcDNA3-EGFP-RhoA-T19N (Addgene plasmid 12967), pcDNA3-EGFP-RhoA-Q63L (Addgene plasmid 12968), pcDNA3-EGFP-Cdc42-T17N (Addgene plasmid 12976), and pcDNA3-EGFP-Cdc42-Q61L (Addgene plasmid 12986) were developed in Dr. Gary Bokoch's laboratory (32). All plasmids were obtained through the plasmid repository Addgene. For plasmid transfections, cells were plated onto 100-mm diameter dishes and then transfected the following day when confluency was ∼40% using PrimeFect I reagent from Lonza Walkersville according to the manufacturer's instructions, using 120 μl of PrimeFect I reagent and 7.5 μg of DNA per dish (added to 4 ml of complete media on the dish), and allowed to transfect for only 3 h (longer incubation times cause toxicity and cell death). Media were replaced with complete media, and cells were allowed to recover in the same dish for at least 24 h before being plated for wounding assays. For double transfections with small interfering RNA (siRNA) and plasmid, cells were transfected at a higher confluency (60%), and Lipofectin was used (instead of PrimeFect I reagent), with a combination of 100 nM siRNA and 2.5 μg of plasmid per 100-mm dish, following the same protocol described for siRNA transfections in Ref. 18, allowing mixture to transfect the cells for 4 h.
HMVEC-LLy cells or cells that were transfected with control siRNA, podoplanin siRNA, or plasmid DNA were trypsinized 24–36 h after transfection and plated on 12-well plates at a density of 200,000 cells per well in EGM-2MV media and allowed to reach confluency (usually 1–2 days). At that time, cells were starved in EBM-2-0.5% FBS media overnight, and then a vertical line across the middle of the wells (wound) was made using a 200-μl pipette tip (average wound width: 700–900 μm). Media were replaced with starvation media, complete EGM-2MV, or EBM-2 MV-0.5% FBS-10 ng/ml VEGF. Pictures were taken immediately after the wound (at 0 h) and at 24 and 48 h after the wound using an Olympus IX50 microscope and QCapture software controlling a cooled charge-coupled device (CCD) camera (×40 magnification). Three pictures were taken per well, and at least triplicate wells per condition were analyzed using ImageJ free software. Wounded area per field was individually assessed and averaged per well, and percentage of area covered was calculated for each well on an individual basis. To determine scale, a picture was taken of a micrometer, and calibration was performed with the Analysis tool in ImageJ. Experiments were performed at least 3 different times, with triplicate or quadruplicate wells individually analyzed.
Immunofluorescence staining of wounded cells.
Cells were plated onto glass coverslips, allowed to reach confluency, and then starved overnight in EBM-2-0.5% FBS before wounding the monolayers. After the wound, media were replaced with either EBM-2-0.5% FBS containing VEGF or starvation media alone (unless indicated differently in figure legends) and allowed to migrate for 3 h. Coverslips were then rinsed once in PBS, cells were fixed and permeabilized by immersion in absolute methanol at −20°C for 3 min, and after two 5-min washes in PBS and blocking in 3% goat serum for 20 min, staining was performed. For the staining of the MTOC or Golgi, polyclonal antibodies directed against pericentrin or giantin were used at 1:500 and 1:1,000, respectively (antibodies from Covance, Emeryville, CA). Monoclonal anti-α-tubulin DM1A antibody was from Sigma-Aldrich (St. Louis, MO). Secondary antibodies goat anti-rabbit Alexa Fluor 594 and goat anti-mouse Alexa Fluor 488 were from Molecular Probes-Invitrogen and used at 1:1,000 dilution. After staining, samples were mounted on slides using VECTASHIELD mounting media with 4′,6′-diamidino-2-phenylindole (DAPI; Vector Laboratories, Burlingame, CA) and observed under an Olympus BX60 fluorescence microscope.
Staining of actin fibers.
For the staining of actin fibers in cells, control siRNA- or podoplanin siRNA-transfected cells were plated onto glass coverslips, allowed to recover for 2 days, starved for 2 h in EBM-2-0.5% FBS, and then left untreated or treated for 1 h with 50 ng/ml VEGF in starvation media. Cells were then rinsed in PBS, fixed using 4% paraformaldehyde in PBS for 10 min at room temperature, rinsed twice with PBS, permeabilized with 0.1% Triton X-100 in PBS for 5 min, and rinsed twice with PBS. If staining for podoplanin expression was performed, coverslips were first blocked with 3% BSA in PBS and then stained using a rat monoclonal antibody against human podoplanin (AngioBio, Del Mar, CA) at a 1:100 dilution, after which cells were rinsed, incubated with a secondary anti-rat Alexa Fluor 488 (Molecular Probes-Invitrogen), and then stained with rhodamine phalloidin from Cytoskeleton according to the manufacturer's recommendations (3–5 units per coverslip, in PBS containing 0.5% BSA, for 30 min at room temperature). After washing 3 times with PBS, coverslips were mounted onto slides using DAPI-containing VECTASHIELD mounting media (Vector Laboratories) and observed using a fluorescence microscope (Olympus BX60).
RhoA activity assays.
The protocol for pull-down assays was based on our previous method described in Ref. 18, with slight modifications. Basically, cells plated onto p100 dishes were control siRNA- or podoplanin siRNA-transfected and incubated for 48 h. Cells were then starved overnight in EBM-2-0.5% FBS media and left untreated or treated with 50 ng/ml VEGF (or left in complete media without starvation). After washing with ice-cold PBS, pH 7.4, cells were lysed in 0.4 ml of ice-cold lysis buffer [50 mM Tris·HCl, pH 7.5, 10 mM MgCl2, 500 mM NaCl, 1% Triton X-100, 10 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 10 mM β-glycerol phosphate, 1 mM vanadate, and protease inhibitor cocktail without EDTA from Roche Diagnostics (Indianapolis, IN)]. After lysis for 20 min on ice, cell debris was removed by centrifugation at 300 g for 10 min at 4°C, and lysates (100–200 μg protein) were mixed with 15 μl of Rhotekin-RBD protein agarose beads (50 μg of protein; Cytoskeleton) or 20 μl of PAK-GST protein agarose beads (20 μg; Cytoskeleton) and incubated for 1 h at 4°C with rotation. Samples were then centrifuged (5,000 rpm for 1 min at 4°C), washed twice in ice-cold wash buffer (25 mM Tris·HCl, pH 7.5, 30 mM MgCl2, and 40 mM NaCl), resuspended in 30 μl of Laemmli buffer, heated at 100°C for 5 min, separated on 12% polyacrylamide gels, and processed for Western blot after transferring to PVDF membranes. Rabbit monoclonal antibodies against RhoA and Cdc42 were from Cell Signaling Technology (Danvers, MA). Monoclonal anti-α-tubulin antibody (clone B-5-1-2) for normalization was from Sigma-Aldrich. Mouse monoclonal against Rac1 was from Cytoskeleton.
Immunoprecipitation of active Cdc42.
Protocol used was based on the availability of a mouse monoclonal antibody directed against the active form of Cdc42, Cdc42-GTP, commercially available through NewEast Biosciences (Malvern, PA). Control siRNA- or podoplanin siRNA-transfected cells were grown in complete media for 3–4 days. After washing twice with ice-cold PBS, pH 7.4, cells were lysed in 0.4 ml of ice-cold lysis buffer (50 mM Tris·HCl, pH 7.5, 10 mM MgCl2, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 10 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 10 mM β-glycerol phosphate, 1 mM vanadate, and protease inhibitor cocktail without EDTA from Roche Diagnostics). After lysis for 10 min on ice, cell debris was removed by centrifugation at 12,000 g for 10 min at 4°C, 0.35 ml of lysate was diluted to 1 ml with lysis buffer, 1 μl of anti-active Cdc42 monoclonal antibody was added, 20 μl of Protein A/Protein G mix was added, and samples were incubated for 1 h with rotation at 4°C. Beads were pelleted by centrifugation for 1 min at 5,000 g and then washed three times with 0.5 ml of lysis buffer, resuspended in 25 μl of 2× reducing SDS-PAGE sample buffer, heated at 100°C for 5 min, separated on 16% polyacrylamide gels, and processed for Western blot after transferring to PVDF membranes. Rabbit polyclonal antibody against total Cdc42 from NewEast Biosciences was used for the Western blot.
Results are expressed as means ± SD of data obtained. Statistical analysis was performed using Student's t-test for paired comparisons.
Podoplanin knockdown inhibits HMVEC-LLy cell polarization and delays wound closure.
To begin addressing the possibility of a role for podoplanin in lymphatic endothelial cell migration, we performed wounding of confluent starved cell monolayers in which podoplanin expression was decreased by siRNA and followed the recovery of the monolayer with time in EBM-2 media containing 0.5% FBS and 10 ng/ml VEGF. We (18) have previously used this same siRNA to downregulate the expression of podoplanin in these cells with success. As shown in Fig. 1, A and B, control cells completely repaired the endothelial monolayer wound by 48 h, but cells with decreased podoplanin expression did not (wound recovered 40%).
It is known that microtubules orient toward the leading edge of a migratory cell as an indicator of the disposition of the cell to move in that direction, and in that respect scratch wound assays allow the study of cell polarization. We studied cell polarization in HMVEC-LLy cells early on after the wounding by staining for pericentrin, a protein that is present in centrosomes and is an indicator of the position of the MTOC. In these experiments, a cell is considered to be polarized toward the front if the MTOC is located in a region that comprises a 120° sector facing the wound. If cells are randomly oriented and do not show a preference toward a specific direction, about 1:3 (or 33%) of the MTOCs at the front of the wound would be scored as polarized toward the wound. As seen in Fig. 1, C and D, podoplanin siRNA-transfected cells have decreased MTOC polarization in the presence of VEGF, <60 vs. 80% in control siRNA-transfected cells at 3 h after the wounding. As indicated by the white arrows, many of the MTOC in cells that are transfected with podoplanin siRNA are oriented away from the wound and toward the back of the cell.
Cell polarity in the wound scratch model can also be assessed by measuring the movement of the Golgi complex to the side of the cell facing the wound, just as in MTOC polarization. We assessed localization of the Golgi apparatus in migrating cells by staining for giantin, a Golgi resident protein that is commonly used to determine the position of this organelle in a given cell. As shown in Fig. 1, E and F, only 50% of cells transfected with podoplanin siRNA show polarization of their Golgi at 3 h after wounding compared with 90% polarization in control siRNA-transfected cells. Thus depletion of podoplanin impedes polarization of MTOC and Golgi complex in HMVEC-LLy cells. Double staining of podoplanin and pericentrin or Golgi in control or podoplanin siRNA-transfected cells is shown in Supplemental Fig. S1 (available in the data supplement online at the AJP-Lung Cellular and Molecular Physiology web site).
Podoplanin knockdown results in reorganization of HMVEC-LLy actin cytoskeleton assembly with stress fibers formed.
Cell locomotion involves rearrangement of the actin cytoskeleton, and the nature of F-actin assembly in a cell gives some indication of its readiness to move. To examine the effect of lack of podoplanin on F-actin assembly in HMVEC-LLy, with or without VEGF stimulation, we analyzed F-actin stress fibers present in control siRNA- vs. podoplanin siRNA-transfected cells. As shown in Fig. 2A, when cells that express podoplanin are starved, the majority of the polymerized actin is present as cortical actin around the cell body, and depletion of podoplanin by use of siRNA causes a dramatic redistribution and a significant increase in stress fiber formation in these cells (Fig. 2B). Treatment with VEGF is effective at increasing stress fiber formation in control cells and has some effect in podoplanin siRNA-transfected cells. Clearly, lack of podoplanin expression dramatically increases stress fiber formation in HMVEC-LLy cells. Transfection efficiency in these cells is high, as described in our earlier work (18).
RhoA-GTP level is decreased in podoplanin knockdown cells, and precise regulation of the activity is important for polarized migration of HMVEC-LLy.
Rho small GTPases are known to regulate dynamic processes involved in cell migration, and we know from our previous work (18) that lack of podoplanin prevents appropriate activation of RhoA-GTP when lymphatic cells are plated onto Matrigel. Therefore, to determine the contribution of RhoA activity to podoplanin regulation of HMVEC-LLy migration, we first performed pull-down assays for RhoA activity in cultured cells. Analysis of RhoA-GTP levels revealed that siRNA depletion of podoplanin expression in HMVEC-LLy decreases RhoA-GTP activity, both in growing (Fig. 3A) and serum-starved cells, basally or stimulated with VEGF (Fig. 3B). Analysis of Rac1-GTP or Cdc42-GTP activity did not show a significant difference in basal level of these small GTPases after downregulation of podoplanin (Supplemental Fig. S2, A and B). Using a new antibody commercially available and developed against the active form of Cdc42, we were able to show an increase in the amount Cdc42-GTP in cells with reduced podoplanin expression (Supplemental Fig. S2C). Next, we analyzed in wounding assays cells that have been pretreated with 1 μg/ml Rho inhibitor (blocks RhoA but not the related Cdc42 and Rac1 proteins) for 2 h in serum-free media before wounding and allowed to recover from the wound in response to VEGF treatment. As shown in Fig. 3, C and D, control cells (no inhibitor treatment) in the presence of VEGF cover ∼60% of the wounded area by 48 h vs. only 30% for cells treated with the inhibitor. This result is not due to cytotoxicity since cell viability was not affected by the Rho inhibitor treatment [as assessed by WST-1 assays (Roche Diagnostics) performed, data not shown].
To assess whether RhoA activation level influences MTOC and Golgi polarization, cells pretreated with the Rho inhibitor were wounded and stained for MTOC and Golgi positioning after 3 h of treatment with VEGF. The results are shown in Fig. 4, A–D. As clearly seen in the pictures, pretreatment of HMVEC-LLy with Rho inhibitor prevents MTOC polarization (40% in inhibitor-treated vs. 80% in control cells) and Golgi polarization (35 vs. 60%).
Finally, to confirm the data obtained with the Rho inhibitor and further explore the effects of RhoA in lymphatic endothelial cell migration, we used available plasmids containing different versions of EGFP-tagged RhoA in transfection experiments testing the effect of dominant-negative vs. constitutively active constructs of RhoA in HMVEC-LLy. Moreover, we verified that transfection with the RhoA constructs does not result in altered levels of the other GTPases, Rac1 and Cdc42, as in the case of HeLa cells recently described in the literature (Ref. 7; see Supplemental Fig. S3). As shown in Fig. 4, E and F, cells expressing a constitutively active (EGFP-RhoA-Q63L) or dominant-negative form (EGFP-RhoA-T19N) of RhoA have a significantly impaired migration response to VEGF. HMVEC-LLy expressing EGFP-RhoA-Q63L repair only 30% of the monolayer wound, and those expressing EGFP-RhoA-T19N only ∼20% of the wounded area, by 48 h. In contrast, control cells expressing EGFP alone recover >50% of the wounded area by that time. The experiment was repeated three separate times with similar results, and the results shown correspond to one set of samples.
High Cdc42 activity explains the increased level of actin polymerization observed in podoplanin knockdown cells and contributes to the low RhoA-GTP levels and poor migratory phenotype.
It has been shown that RhoA, Rac, and Cdc42 are involved in stress fiber formation in response to growth factors (5, 25). We sought to determine whether the increase in actin polymerization could be due to an increase in active Cdc42. To test this hypothesis, we transfected EGFP or EGFP fused to a constitutively active mutant Cdc42 (EGFP-Cdc42-Q61L) into HMVEC-LLy and analyzed the stress fibers present. As shown in Fig. 5A, top, transfection of the constitutively active mutant of Cdc42 into normal lymphatic endothelial cells causes significant increase in the number of stress fibers present in the cell body compared with EGFP alone. More importantly, as shown in Fig. 5A, bottom, in podoplanin knockdown cells, the dominant-negative mutant of Cdc42 (EGFP-Cdc42-T17N) is able to restore the initially very high level of stress fibers to those resembling controls. For comparison, transfection with constitutively active RhoA (EGFP-RhoA-Q63L) causes a small increase in stress fibers in podoplanin siRNA-transfected cells, whereas the cotransfection with dominant-negative RhoA (EGFP-RhoA-T19N) is not able to exert the dramatic effect seen with the Cdc42 construct (Supplemental Figs. S4 and S5).
In endothelial cells, Cdc42 has been shown to inhibit RhoA activity (13). We were interested in addressing a possible cross talk between Cdc42 and RhoA in lymphatic endothelial cells that could partly explain the nonmigratory phenotype, so we transfected normal cells with EGFP alone or with the constitutively active mutant of Cdc42-EGFP construct (EGFP-Cdc42-Q61L) and determined RhoA-GTP levels using a pull-down assay. As shown in Fig. 5B, increasing active levels of Cdc42 inhibits RhoA activity in these cells and can explain some of the phenotype observed since we have seen that RhoA inhibition causes defective cell polarization and migration. Also, transfection of siRNA podoplanin together with the dominant-negative Cdc42 construct (EGFP-Cdc42-T17N) increases RhoA-GTP levels (Supplemental Fig. S6).
Cdc42 itself could also have effects on migration, so we studied the role of Cdc42 constructs in wound closure. As shown in Fig. 5C, expression of the EGFP-tagged, dominant-negative version of Cdc42 (EGFP-Cdc42-T17N) has no effect at all in the ability of HMVEC-LLy to cover the wound in response to VEGF. Interestingly, when analyzing wound-healing response in HMVEC-LLy cells transfected with the EGFP-tagged, constitutively active Cdc42 mutant (EGFP-Cdc42-Q61L), we could observe a slightly diminished ability of the cells to recover from the wound. As shown in Fig. 5, C and D, cells are able to cover ∼80% of the wounded area in 48 h, when control cells expressing EGFP alone are able to recover 100%.
Expression of a dominant-negative mutant version of Cdc42 in HMVEC-LLy cells with decreased podoplanin expression restores VEGF-induced polarization response rescuing their impaired wound closure ability.
Since our results point to a change in active levels of Cdc42 as responsible for the actin reorganization seen in the cells that lack podoplanin, we therefore questioned the polarization phenotype of cells that lack podoplanin but are expressing a dominant-negative mutant Cdc42. As shown in Fig. 6, A and B, wounding assay results for cell polarization in EGFP-expressing podoplanin knockdown cells were 32% in starvation media, 30% in complete media, and 31% in media with no VEGF. In contrast, results for podoplanin knockdown cells expressing a dominant-negative mutant of Cdc42 (EGFP-Cdc42-T17N) were 37% in starvation media, 65% in complete media, and 30% in media with no VEGF. Thus expression of dominant-negative Cdc42 in podoplanin knockdown cells restores polarized response to VEGF stimulation.
In the experiments shown thus far, we have seen that expression of a dominant-negative construct of Cdc42 is able to restore the actin stress fibers in podoplanin knockdown cells to those of control cells and that VEGF-induced polarization signals are also restored to normal. Next, we assessed the ability of the cells to migrate and repair an endothelial monolayer wound in response to VEGF stimulation. The results are presented in Fig. 6, C and D. As can be seen in the figure, with podoplanin siRNA-transfected cells, only ∼25% of the wounded area is recovered by 48 h. In contrast, HMVEC-LLy cotransfected with podoplanin siRNA and dominant-negative construct of Cdc42 are able to recover as much as 50% of the wounded area. Thus cotransfection of dominant-negative construct of Cdc42 in podoplanin knockdown cells rescues their impaired migratory response to VEGF. Interestingly, levels of RhoA-GTP appear to increase in podoplanin siRNA-transfected cells that express a dominant-negative version of Cdc42 (EGFP-Cdc42-T17N), as shown in Supplemental Fig. S6.
The major finding of this study is that directional migration of HMVEC-LLy cells in response to VEGF stimulation in vitro is dependent on podoplanin regulation of Rho GTPases, in particular Cdc42 activity. Our data indicate that in lymphatic endothelial cells, proper activation of small Rho GTPases and complete reorientation of the MTOC and Golgi in response to VEGF, which are necessary for directional migration in scratch wound assays, are achieved only if the cells express podoplanin.
Although earlier investigators have provided great insight into the biological consequence of lack of podoplanin, the physiological function of podoplanin and its mediator pathways remains largely unknown (27). It has been well-described in the literature that podoplanin expression in tumors is linked to acquisition of an invasive phenotype with increased potential for metastasis. Consequently, a number of investigators using forced expression in tumor-derived or immortalized cell lines, and cell types other than lymphatic endothelial cells have been used to explore a role for podoplanin in cell migration (4, 28, 41). Scholl et al. (28) reported that ectopic expression of PA2.26 in immortalized, nontumorigenic keratinocytes induces an epithelial fibroblastoid morphological conversion with increased plasma membrane extensions concomitantly to a major reorganization of the actin cytoskeleton, redistribution of the actin-binding protein ezrin to cell surface projections, and enhanced motility, findings that suggest involvement of PA2.26 in cell migration. In their in vitro experiment, Wicki et al. (40, 41) assessed migratory properties of MCF 7 breast epithelial carcinoma cell line after enforced expression of podoplanin and reported increased cell migration independent of epithelial-mesenchymal transition (EMT) and with downregulation of the activities of the small Rho GTPases. In contrast, Martín-Villar et al. (15) using Madin-Darby canine kidney (MDCK) type II epithelial cells transfected with human wild-type podoplanin construct reported increased migration in association with EMT and increased RhoA activation. These results are interesting and point to a podoplanin-mediated pathway of cell migration in conditions of abnormal expression but with apparent contradictory findings on direction of change in RhoA activity. Although a direct comparison of these earlier works with the current study may not be appropriate given the differences in model, our observation of decreased RhoA activity consequent to lowering podoplanin expression appears consistent with that of Martín-Villar et al. (15), who found increased RhoA activity following forced expression of podoplanin.
Our findings are unique because using primary HMVEC-LLy cells, known to naturally express podoplanin, we could show in vitro that the physiological response of migration in the presence of VEGF by these cells is crucially dependent on precise regulation of Cdc42 and RhoA activities by podoplanin. This study is the first to demonstrate that podoplanin is crucial for directional migration of primary human lymphatic endothelial cells and, moreover, that this effect is achieved through podoplanin regulation of Cdc42 and RhoA activities.
Directional or polarized migration appears to be a consequence of reorganization and alignment of the actin and microtubular cytoskeleton along a vectorial axis induced by intrinsic or extrinsic cues such as growth factors resulting in orientation of the centrosome or MTOC and Golgi bodies to the front of the cells, between the leading edge and the nucleus, which moves posteriorly (19–21). In a number of cell types, it has been shown that the small GTPases Cdc42 and mDia regulate MTOC localization (20, 21), and cross talk and regulatory feedback exist with RhoA (9, 10, 19, 24). Thus Cdc42 appears to be essential for different aspects of cellular polarization and contributes to maintain proper orientation and polarized morphology of subcellular structures. We found in our experiments that deficient podoplanin expression in HMVEC-LLy cells results in failure of MTOC and Golgi translocation to the front of the cell toward the wound edge and, as a consequence, impairment in cell migration. Moreover, we demonstrated that this lack of polarization in MTOC and Golgi distribution seen in podoplanin knockdown cells can be induced in control cells by treatment with a chemical Rho inhibitor and that, importantly, decreasing Cdc42 activity in podoplanin knockdown cells by use of a dominant-negative construct completely restores polarized migration. Furthermore, we showed using a constitutively active mutant of Cdc42 that Cdc42-GTP drives inhibition of RhoA in HMVEC-LLy. Since podoplanin knockdown cells expressing a dominant-negative mutant of Cdc42 are able to polarize their Golgi properly in response to VEGF and migrate to close the wound to the extent of control cells, it appears that the main reason for the deregulation in migratory ability of HMVEC-LLy lacking podoplanin is increased Cdc42 activity.
In its active form, Cdc42 binds to and activates WASp (which then recruits and activates the Arp2/3 complex) and interacts with PAR-6, IQGAP1, and others (11), all downstream effectors involved in the regulation of actin polymerization. In this report, after staining for actin stress fibers, we show that the establishment of cortical actin in human lymphatic endothelial cells is dependent on podoplanin expression, and using constructs that have a mutant Cdc42 capable of blocking endogenous Cdc42 activity, we show that the stress fibers in podoplanin knockdown cells can be returned to those of control cells. Thus increased Cdc42 activity appears responsible for the generation of a multitude of stress fibers across the cell body in podoplanin knockdown cells. In our experiments, we observed that HMVEC-LLy have high levels of expression of Cdc42 with robust expression also seen even with podoplanin knockdown cells. Interestingly, although we could not detect significant changes in Cdc42 activity following podoplanin depletion with siRNA by pull-down experiments, using a more sensitive, commercially available assay based on immunoprecipitating Cdc42-GTP with a specific mouse monoclonal antibody, we were able to show a significant increase in the level of active Cdc42 in cells with decreased podoplanin expression (Supplemental Fig. S2, B and C); this and the data provided by the use of the dominant-negative construct allow us to conclude that most of the observed effects involve perturbations in this pathway.
Rho GTPases can also contribute to regulate the actin cytoskeleton through the phosphorylation of ezrin/radixin/moesin (ERM) family of actin-binding proteins (4, 35, 37). In endothelial cells, moesin is the most abundant ERM protein (6), although there is no report specific to the nature of ERM proteins in endothelial cells of a lymphatic nature. Phosphorylation of moesin by ROCK results in depolymerization and distribution of actin to cortical locations in the cell membrane, allowing cross-linking between actin, integrins, and focal adhesion complexes at cell attachment sites to the substrata (at membrane protrusions). In a previous study, we reported that phosphorylated ERM proteins fail to distribute to plasma membrane protrusions in podoplanin knockdown cells after adhesion to Matrigel without a significant change in their phosphorylation level. Unfortunately, in our (18) experiments leading to that observation, we did not look at actin distribution in podoplanin knockdown cells. Although it is likely that the early events that take place after initial attachment to Matrigel differ from those that occur in the process of migration and are related to the extensive remodeling taking place to allow capillary tubelike formation by endothelial cells, it will be interesting to determine how ERM proteins are regulated by small GTPases in lymphatic endothelial cells, in particular whether a Cdc42-dependent pathway redistributes ERM to plasma membranes allowing to recover cortical actin and increasing motility.
RhoA has been implicated in formation of the lamellipodia at the leading edge of the cell as well as the formation of stress fibers under the influence of the growth factor VEGF (1, 13, 26), and the other Rho family members, Cdc42 and Rac1, have been implicated in formation of membrane ruffles and events leading to the retraction of the rear or trailing end of cells during migration. A definite cross talk exists between the different Rho GTPases regulating cell behavior, since Cdc42 is known to activate Rac1 and inhibit RhoA simultaneously and directly, and Rac1 has been described to downregulate RhoA, in both fibroblasts and microvascular endothelial cells (13, 26). Moreover, it is known that positive and negative regulation of RhoA is important for cell migration as both activation and inhibition of RhoA activity have been shown to suppress cell migration (2, 3, 19, 23, 34). Of interest in our findings is that HMVEC-LLy expressing constitutively active RhoA, similar to those expressing a dominant-negative RhoA mutant construct, showed a decrease in their ability to migrate and the fact that cells treated with an inhibitor of Rho failed to polarize in the scratch wound assays. This suggests that tight regulation of RhoA activity in these cells is crucial for proper migration.
In summary, this study found that directional migration of HMVEC-LLy cells in response to VEGF stimulation in vitro is dependent on podoplanin regulation of Cdc42 and RhoA GTPase activation. While this paper was in the final stages of preparation, a report was published describing that Csk-associated substrate (Cas)-dependent induction of podoplanin expression in Src-transformed cells is responsible for their increased migration (31). However, no mechanism was provided on how this increased migration is achieved. The data in our report indicate that podoplanin present at the plasma membrane of lymphatic endothelial cells negatively regulates Cdc42. The importance of the findings lies in the fact that directional migration of lymphatic endothelial cells is crucial for lymphangiogenesis, an important process in normal tissue morphogenesis that also contributes to development and propagation of pathologies such as fibrosis and tumor spread to distant sites. Thus a better understanding of the mechanisms involved may offer an opportunity to advantageously manipulate the process for therapeutic purposes.
This study was supported by a Children's Mercy Hospital Physician Scientist Award to I. I. Ekekezie.
No conflicts of interest, financial or otherwise, are declared by the author(s).
We thank Christine Concepción for secretarial assistance in preparation of this paper.
- Copyright © 2011 the American Physiological Society