Rho Inhibition Induces Migration of Mesenchymal Stromal Cells


  • Bithiah Grace Jaganathan,

    1. Institute of Transfusion Medicine and Immune Hematology, German Red Cross Blood Donor Service, University of Frankfurt, Frankfurt am Main, Germany
    Search for more papers by this author
  • Brigitte Ruester,

    1. Institute of Transfusion Medicine and Immune Hematology, German Red Cross Blood Donor Service, University of Frankfurt, Frankfurt am Main, Germany
    Search for more papers by this author
  • Lars Dressel,

    1. Institute of Transfusion Medicine and Immune Hematology, German Red Cross Blood Donor Service, University of Frankfurt, Frankfurt am Main, Germany
    Search for more papers by this author
  • Stefan Stein,

    1. Chemotherapeutic Research Institute Georg Speyer Haus, Frankfurt, Germany
    Search for more papers by this author
  • Manuel Grez,

    1. Chemotherapeutic Research Institute Georg Speyer Haus, Frankfurt, Germany
    Search for more papers by this author
  • Erhard Seifried,

    1. Institute of Transfusion Medicine and Immune Hematology, German Red Cross Blood Donor Service, University of Frankfurt, Frankfurt am Main, Germany
    Search for more papers by this author
  • Reinhard Henschler M.D.

    Corresponding author
    1. Institute of Transfusion Medicine and Immune Hematology, German Red Cross Blood Donor Service, University of Frankfurt, Frankfurt am Main, Germany
    • Institute of Transfusion Medicine and Immune Hematology, University Hospital Frankfurt, Sandhofstrasse 1, D 60528 Frankfurt, Germany. Telephone: +49 69 6782 191; Fax: +49 69 6782 258
    Search for more papers by this author


Although mesenchymal stromal cells (MSCs) are being increasingly used as cell therapeutics in clinical trials, the mechanisms that regulate their chemotactic migration behavior are incompletely understood. We aimed to better define the ability of the GTPase regulator of cytoskeletal activation, Rho, to modulate migration induction in MSCs in a transwell chemotaxis assay. We found that culture-expanded MSCs migrate poorly toward exogenous phospholipids lysophosphatidic acid (LPA) and sphingosine-1-phosphate (S1P) in transwell assays. Moreover, plasma-induced chemotactic migration of MSCs was even inhibited after pretreatment with LPA. LPA treatment activated intracellular Rho and increased actin stress fibers in resident MSCs. Very similar cytoskeletal changes were observed after microinjection of a cDNA encoding constitutively active RhoA (RhoAV14) in MSCs. In contrast, microinjection of cDNA encoding Rho inhibitor C3 transferase led to resolution of actin stress fibers, appearance of a looser actin meshwork, and increased numbers of cytoplasmic extensions in the MSCs. Surprisingly, in LPA-pretreated MSCs migrating toward plasma, simultaneous addition of Rho inhibitor C2I-C3 reversed LPA-induced migration suppression and led to improved migration. Moreover, addition of Rho inhibitor C2I-C3 resulted in an approximately 3- to 10-fold enhancement of chemotactic migration toward LPA, S1P, as well as platelet-derived growth factor or hepatocyte growth factor. Thus, inhibition of Rho induces rearrangement of actin cytoskeleton in MSCs and renders them susceptible to induction of migration by physiological stimuli.

Disclosure of potential conflicts of interest is found at the end of this article.


Increasing evidence suggests that mesenchymal stromal cells (MSCs) are a promising cell source for tissue engineering, tissue regeneration, and gene therapy applications. MSCs isolated from human bone marrow can differentiate into adipocytes, chondrocytes, and osteocytes [1] but can also acquire markers of other cell types such as myocytes, endothelial cells, cardiomyocytes, and neuron-like cells [2, 3]. These cells could be cultured in an undifferentiated state up to 20–40 population doublings [2]. Apart from bone marrow, MSCs have been isolated from adipose tissue, peripheral blood, cord blood [4, [5]–6], umbilical cord, muscle, and pancreas [7, 8]. In addition, MSCs have been identified in the hematopoietic microenvironment and have been shown to mediate accelerated hematopoietic recovery after hematopoietic stem cell transplantation [9] or to confer immunosuppressive properties [10].

The defective migration potential of MSCs has been discussed to hamper their effective use for gene therapy or tissue regeneration. For example, after i.v. injection, it has been difficult to trace the transplanted MSCs in different organs [11]. Data from other groups as well as our own show that i.v. injected MSCs may be physically trapped in the lungs [12, 13]. However, preclinical models of MSC transplantation have also shown engraftment into various tissues [11, 14]. It has been postulated that for egress from the bloodstream, i.v. injected MSCs are capable of performing at least principally some of the steps that have been recognized to regulate the extravasation of leukocytes as defined by Springer et al. [15, 16]. These imply, in addition to selectin-dependent interactions, also integrin-mediated binding and its enhancement through chemokines [13, 17].

Rho family GTPases have been described as important signaling molecules to the cytoskeleton, regulating the coordinated assembly and activation of actin with actin-binding proteins such as paxillin and α-actinin [18, 19]. Gu et al. have implicated a crucial role of the activation status of Rac GTPases in the proliferation and migration of hematopoietic stem and progenitor cells [20]. We have recently demonstrated that treatment of hematopoietic progenitor cells with stromal cell-derived factor (SDF)-1α results in an intracellular signal that requires intact Rho GTPase signaling [21, 22] and that Rho itself is a regulator of migration responses in hematopoietic progenitor cells [23]. We therefore hypothesized that signaling through Rho family GTPases would influence the migration response in MSCs.

We show here that modulation of Rho GTPase can increase the responsiveness of human MSCs to physiological stimuli such as stimulation by lysophospholipids. This involves a change in their actin cytoskeletal activation status and can turn a signal that negatively regulates MSC chemotaxis into a signal that effectively induces chemotactic migration.

Materials and Methods

Chemicals and Reagents

Isobutylmethylxanthine, β-glycerophosphate, dexamethasone, ascorbic acid, indomethacin, insulin, sphingosine-1-phosphate, lysophosphatidic acid, paraformaldehyde, human laminin, and cell culture tested bovine serum albumin (BSA) were purchased from (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com). Rho kinase inhibitor Y27632 was from (Calbiochem, San Diego, http://www.emdbiosciences.com). Transforming growth factor (TGF)-β1, basic fibroblast growth factor (bFGF), platelet-derived growth factor (PDGF)-bb, and hepatocyte growth factor (HGF), as well as fluorescence-conjugated monoclonal antibodies for CD45 (phycoerythrin-Cy5), CD73 (phycoerythrin), CD90 (fluorescein isothiocyanate), and CD105 (phycoerythrin), were from R&D Systems Inc. (Minneapolis, http://www.rndsystems.com). Trypsin was from Invitrogen (Carlsbad, CA, http://www.invitrogen.com). Phalloidin-tetramethylrhodamine B isothiocyanate (TRITC) and anti-pan-cadherin antibody (CH-19) were from Sigma. Anti-Rho antibody was from Pierce (Rockford, IL, http://www.piercenet.com) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was from Abcam (Cambridge, U.K., http://www.abcam.com).

Isolation of MSCs from Bone Marrow

MSCs were isolated from bone marrow samples of patients undergoing hip surgery after informed consent by a procedure approved by the local ethics committee. The bone marrow cells were suspended in Iscove's modified Dulbecco's medium supplemented with 500 U/ml heparin. Mononuclear cells were separated by density gradient centrifugation at 300g for 20 minutes. The cells were grown in Dulbecco's modified Eagle's medium (DMEM) low glucose with 20% (vol/vol) fetal bovine serum (FBS) supplemented with bFGF (25 ng/ml) at a density of 5 × 106 cells per milliliter. The medium was changed every 2–3 days. After 2–3 weeks, a layer of spindle-shaped cells had formed (MSC). The cells were passaged 1:3 at 80% confluence.

Differentiation of MSCs

To assess their differentiation potential, freshly trypsinized MSCs were seeded into differentiation medium. Briefly, MSCs at passages 3, 6, and 9 were seeded at 1 × 104 per cm2 on tissue culture plastic in the presence of 10 mM β-glycerophosphate, 0.1 μM dexamethasone, and 60 μM ascorbic acid-2-phosphate in DMEM/10% FBS for induction of osteogenic differentiation; 1 μM dexamethasone, 0.2 mM indomethacine, 0.5 mM isobutylmethylxanthine, and 10 μg/ml insulin in DMEM/10% FBS for adipocytic differentiation; or at 2 × 105 cells per milliliter in a 15-ml polypropylene tube (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com) in 0.1 μM dexamethasone, 50 μg/ml ascorbic acid, 6.25 ng/ml selenic acid, 6.25 μg/ml linoleic acid, 6.25 μg/ml insulin, and 10 ng/ml TGF-β in DMEM without FBS in a micromass culture for chondrogenic differentiation. After 3 weeks, differentiation of the cells was assessed after fixation with methanol and subsequent staining with oil red O, alkaline phosphatase, or addition of safranin O for chondrogenic cells. All three, adipogenic, osteogenic, and chondrogenic differentiation, yielded more than 50% of differentiated cells at the different tested passages.

Chemotaxis Assay

The transwell migration assay for human MSCs has been described previously [24]. Briefly, MSCs between passages 3 and 10 were trypsinized, washed once by centrifugation in DMEM/10% FBS (Gibco, Grand Island, NY, http://www.invitrogen.com), and kept in suspension in DMEM with 0.1% BSA at 37°C at a density of 1–5 × 105 cells per milliliter in 15-ml polypropylene tubes in a cell culture incubator for up to 1 hour. To the lower wells of 48-well chambers (Neuro Probe, Gaithersburg, MD, http://www.neuroprobe.com), migration medium (DMEM/1% BSA) and migration-inducing substances were added. Upper wells were filled with migration medium with cells (approximately 32 μl, 20,000 cells per well), in some cases also containing test substances. Filter pore size was 8 μm. The mounting steps were performed within 15 minutes, and the chamber was placed into a humidified incubator. After 6 hours, assays were stopped by removal of the medium from the upper wells and careful removal of the filters. Filters were fixed with methanol by brief submersion and were subsequently wiped on the upper side using the Neuro Probe wiper. Filters were stained with May-Grünwald-Giemsa (Merck & Co., Whitehouse Station, NY, http://www.merck.com) solution for 5 (May-Grünwald) and 15 (Giemsa) minutes each. Evaluation of completed transmigration was performed under a light microscope, and random fields were scanned (2–4 per filter) for the presence of cells at the lower membrane side only. Absolute numbers of migrated cells were calculated by counting migrated cells in parts of the entire filter using scaled grids. Results of the migration experiments were expressed as percentage of input MSCs, calculated by dividing the absolute numbers of migrated cells by input number of MSCs added.


For analysis of cytoskeleton-associated proteins, trypsinized MSCs were seeded on plastic slides precoated with fibronectin (10 μg/ml) (Becton Dickinson), allowed to adhere overnight, rinsed carefully two times with phosphate-buffered saline (PBS), and fixed for 10 minutes in PBS containing 4% paraformaldehyde. Subsequently, cells were permeabilized by incubation with 0.5% Triton X-100 in PBS for 10 minutes. For immunolabeling, the cells were incubated with PBS, 3% BSA, and 0.5% Triton X-100 containing phalloidin-TRITC for 2 hours at room temperature and analyzed using a fluorescence microscope (Carl Zeiss, Jena, Germany, http://www.zeiss.com). Images were taken on a PC using a 1.3 megapixel CCD camera and Axion Software (Zeiss).

Microinjection of cDNA

cDNAs encoding RhoA V14 (kindly provided by Dr. M. Ruthardt, University of Frankfurt, Germany) or C3 transferase from Clostridium limosum (kindly provided by Dr. H. Barth, University of Freiburg, Germany) [23, 25] were subcloned into the PINCO gamma retroviral vector, which is derived from Moloney murine leukemia virus [26]. Plasmids, including an empty PINCO control vector, were dissolved in sterile water at 0.2–1 μg/ml and were microinjected into MSCs that were seeded on sterile glass cover slides precoated with human laminin (100 ng/ml) using an Eppendorf InjectMan NI2 fitted with a microinjection needle and FemtoJet mounted to a Zeiss microscope Axiovert 135. Cells were incubated at 37°C, 5% CO2 for another 6–12 hours before fluorescent microscopic analysis.

Real-Time Polymerase Chain Reaction

Total cellular RNA was isolated from MSCs in TRIzol RNA extraction buffer according to manufacturer's instructions (Invitrogen). Quantification of RNA was performed by measuring the absorbance at 260 nm (NanoDrop, Wilmington, DE, http://www.nanodrop.com). cDNA was produced from 1 μg of RNA by reverse transcription in a 50-μl reaction using SuperScript II (Invitrogen) and oligo(dT) primers at 55°C for 1 hour. The conditions for subsequent real-time polymerase chain reaction (PCR) for Edg receptors were 35 cycles at a denaturation temperature of 94°C for 15 seconds, annealing at 60°C for 45 seconds, and extension at 72°C for 30 seconds using a SYBR Green PCR mix. Primers for Edg-2, -4, -5, and -7 were designed using Primer3 Software (http://www.fokker.wi.mit.edu/primers). The primers used for these were: Edg-2, sense 5′-TGTCTCGGCATAGTTCTGGA-3′ antisense 5′-TTCTTTGTCGCGGTAGGAGT-3′; Edg-4, sense 5′-AGGCTGTGAGTCCTGCAATGT-3′ antisense 5′-TCTCAGCATCTCGGCAAGAGT-3′; Edg-5, sense 5′-GGCCTAGCCAGTTCTGAAA-3′ antisense 5′-GCAATGAGCACCAGAAGGTT-3′; Edg-7, sense 5′-GCATACAAGTGGGTCCATCA-3′ antisense 5′-TCACGACGGAGTTGAGCA-3′; and GAPDH, sense 5′-GGGAAGGTGAAGGTCGGAGT-3′ antisense 5′-GGGTCATTGATGGCAACAATA-3′. PCR fragments were analyzed on a 2% agarose gel stained with ethidium bromide.

Subcellular Fractionation and Detection of Membrane-Bound Rho

MSCs were pretreated with lysophosphatidic acid (LPA) (25 μM) or C2I-C3 (1 μg/ml) for 2 hours. The cells were briefly washed by centrifugation with ice-cold PBS twice. The membrane and cytosolic proteins were separated using Qproteome Cell Compartment Kit (Qiagen, Hilden, Germany, http://www1.qiagen.com) according to manufacturer's instructions. Protein content was quantified using a protein detection kit (Bio-Rad, Hercules, CA, http://www.bio-rad.com). The membrane (30 μg/lane) and cytosolic fractions (40 μg/lane) were run on a 12% polyacrylamide gel. The gels were blotted onto a nitrocellulose membrane and detected with either anti-Rho antibody or anti-pan-cadherin and GAPDH followed by horseradish peroxidase-conjugated anti-mouse IgG antibody (Sigma). The immunoblotted proteins were visualized with the enhanced chemiluminescent reagents (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com).

Lentiviral Production, Titration, and Infection of Human MSCs

The packaging plasmid pCMVΔR8.91 encodes the human immunodeficiency virus-1 regulatory proteins tat and rev as well as Δplasmid pCMV gag and pol precursors [27]. Plasmid pMD.G expresses vesicular stomatitis virus glycoprotein G. Pseudotyped lentiviruses were produced by transient calcium-phosphate transfection of 293T cells with pCMVΔR8.91, pMD.G, and the lentiviral transfer vectors (SIEW) [28] into which cDNAs encoding RhoV14 or C3 transferase (kindly provided by Dr. Holger Barth, University of Freiburg, Germany) were subcloned. Viral supernatants were collected 48–72 hours after transfection. Viral titers were determined on 293T cells as described previously and amounted to 0.1–1 × 108 titer units/ml [29]. Lentiviral transduction of human MSCs was performed by seeding the cells in a 6-well plate (1 × 105 cells per well) in DMEM/20% FBS and bFGF (5 ng/ml). The cells were allowed to attach for 24 hours. Viral supernatants of different vectors containing multiplicity of infection 1–5 were added to the cells in the presence of polybrene (4 μg/ml). After 24 hours, the cells were washed and incubated with fresh medium. The cells were split 72 hours later and expanded. Subsequently, transduced cells were identified by the presence of green fluorescent protein reporter gene by flow cytometry.

Flow Cytometry

For flow cytometric analysis, cells were harvested by trypsinization, washed once with PBS, and resuspended in PBS containing 2% FBS. The cells were incubated with the conjugated antibodies for 30 minutes on ice. The cells were washed by centrifugation and analyzed in a flow cytometer (Becton, Dickinson).

Statistical Evaluation

Analysis was performed through Microsoft (Redmond, WA, http://www.microsoft.com) Excel statistics modules.


To characterize the regulation of chemotactic migration in MSCs by the GTPase Rho in conjunction with several physiological stimuli, culture-expanded MSCs were established from the bone marrow of 10 different donors by adherence selection. As shown in Figure 1A and 1B, MSCs could be expanded up to six log scales and were found to be negative for the hematopoietic marker CD45 and positive for the mesenchymal markers CD73, CD90, and CD105 by flow cytometric analysis. The isolated MSCs could be differentiated into adipocytes, osteocytes, or chondrocytes using specific induction medium as shown by the positive staining using oil red O for adipocytes, which stain the lipid globules, alkaline phosphatase, for osteoblastic cells or safranin O, indicating induction of chondrocytic differentiation (Fig. 1C). We next investigated the chemotaxis of MSCs toward various chemoattractants in the transwell assay. In a positive control, human plasma added to the lower wells resulted in a bell-shaped dose-response curve with mean 4.5% of total MSCs transmigrated, as observed previously (Fig. 1D) [24]. PDGF and bFGF have previously been reported to induce chemotactic migration in MSCs [24, 30, 31]. Similar to our previous study, where PDGF induced chemotactic migration of MSCs with variable efficiency and with on average lower efficiency than plasma [24], we observed a comparatively weaker transmigration of the tested MSCs as measured with plasma (Fig. 1D). SDF-1α, which induced chemotaxis in hematopoietic stem cells, failed to cause detectable chemotactic migration of MSCs (data not shown). Also, the lysophospholipids LPA or sphingosine-1-phosphate (S1P) did not induce MSC migration (Fig. 1D).

Figure Figure 1..

Characterization of mesenchymal stromal cells (MSCs) used. (A): Growth curve of MSCs isolated from human bone marrow of six of the donors used in this study. Values represent means ± SD of three determinations of cell concentrations at each time point. Cumulative cell numbers were calculated. (B): Flow cytometric analysis of cell surface markers on MSCs using antibodies against CD45, CD73, CD90, and CD105. (C): Differentiation of MSCs into adipocytes, osteocytes, and chondrocytes. MSCs were treated with specific induction medium and cultured for 14 days. The cultures were stained with oil red O staining for adipocytes and alkaline phosphatase for osteocytes. MSCs were grown in a micromass culture for chondrogenic differentiation, and differentiation into chondroblasts was revealed by safranin O staining; magnification ×100. (D): Migration of MSCs induced by various stimuli present in the lower wells of transwell chambers. Twenty thousand MSCs were seeded into the upper wells of chemotaxis chambers and allowed to migrate for 6 hours toward various indicated factors at the indicated range of concentrations, and absolute numbers of transmigrated MSCs were determined. The responses varied in percentage among different donors, and representative results are shown. Values are means of four replicate determinations. MSCs were from passages 4 and 6 (B), passage 6 (C), or passage 4 (D). Abbreviations: LPA, lysophosphatidic acid; No., number; PDGF, platelet-derived growth factor; S1P, sphingosine-1-phosphate.

MSCs Express LPA Receptors, but LPA Inhibits Their Plasma-Induced Chemotaxis and Stimulates Actin Stress Fiber Formation

Stimulation with phospholipid mediators LPA and S1P has been reported to regulate migration and adhesion events in several cell types [32, [33], [34], [35], [36]–37]. We therefore investigated whether MSCs would express receptors for these substances. Figure 2 shows the results of semiquantitative real-time PCR reactions from MSC preparations from three different donors. We found three of the LPA receptors (Edg-2, -4, -7) and one S1P receptor (Edg-5) expressed. The expression of these receptors was consistently found in different donors, comparable with human umbilical vein endothelial cells that were included as control. We next pretreated MSCs with LPA and tested them in plasma-induced transwell migration. This resulted in clear inhibition of MSC transmigration (Fig. 2B). Similar data were also obtained with S1P (not shown). These findings led us to investigate whether LPA would also induce alterations in the actin cytoskeleton. Staining of actin with fluorescence-tagged phalloidin revealed the presence of actin stress fibers in untreated control MSCs, which became more prominent after exposure of MSC to LPA (Fig. 3A, 3B). Moreover, at lower magnification, a more intense actin staining was observed after LPA treatment compared with controls. The intensity of the overall actin staining signals between the cells remained relatively homogeneous under each of the conditions (Fig. 3C, 3D).

Figure Figure 2..

MSCs express Edg receptors, and the Edg-2, -4, and -7 cognate ligand LPA modulates chemotaxis and the activation status of Rho in MSCs. (A): Real-time polymerase chain reaction analysis of Edg receptors for LPA (Edg-2, Edg-4, Edg-7) and S1P (Edg-5) in MSCs with HUVECs as positive control. (B): Transwell migration of MSCs against plasma (0.5%) prestimulated with LPA. MSCs were pretreated or not with LPA (25 μM) and allowed to migrate for 6 hours in a transwell chamber. Values are means ± SD (n = 4) of total migrated cells. (C): Analysis of Rho activation status in MSCs. MSCs were treated with LPA (25 μM) or C2I-C3 (1 μg/ml) for 2 hours. Membrane and cytoplasmic fractions were isolated, and equal amounts of protein were analyzed using a Western blot for all three treatments in each cell compartment type with Rho-specific antibody. Staining with anti-pan-cadherin antibody was included to control the membrane fraction, and anti-GAPDH antibody served as loading control; *** p ≤ .0001. MSCs were from passages 4 or 5. Abbreviations: GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HUVEC, human umbilical vein endothelial cell; LPA, lysophosphatidic acid; MSC, mesenchymal stromal cell; NC, negative control (H2O).

Figure Figure 3..

Modification of the actin cytoskeleton by Rho GTPase. Fluorescent microscopic analysis of mesenchymal stromal cells (MSCs) immobilized on fibronectin and 2 hours after treatment with phosphate-buffered saline (A, C) and lysophosphatidic acid (25 μM [B, D]) or 6 hours after microinjection of immobilized MSCs with mock vector (E), RhoAV14 (F), or C3 transferase (G) cDNAs. Actin filaments were stained with phalloidin-tetramethylrhodamine B isothiocyanate and visualized through an inverted microscope. Magnification: (A, B), ×100; (C, D), ×20; (E–G), ×40. MSCs were from passage 5 or 6.

Activation of Rho Regulates Actin Cytoskeleton in MSCs

Since Rho has been shown to control cytoskeletal activation in other adherent cell types such as endothelial cells, we investigated whether LPA would induce Rho activation also in MSCs. The Rho activation assay shown in Figure 2C revealed induction of membrane-bound Rho, corresponding to Rho activation, after exposure of MSCs to LPA. Involvement of Rho in induction of actin stress fiber formation in MSCs was also observed after microinjection of cDNA encoding the dominant active isoform of Rho, RhoAV14. Actin stress fibers became more prominent compared with vector transfected control cells (Fig. 3E, 3F).

The data prompted us to examine the possibility that inactivation of Rho, on the opposite, might suppress stress fiber formation in MSCs. We therefore microinjected cDNA encoding C3 transferase, a known specific inhibitor of Rho, into MSCs. Stress fibers in MSCs almost disappeared, and actin staining revealed much thinner, more flexible-appearing fiber structures (Fig. 3G). These appeared as shorter in length and were included in relatively smaller meshwork-like structures that extended toward the cytoplasmic membrane in more filamentous extensions (Fig. 3G). To investigate whether Rho could be inactivated by the cell-permeable soluble Rho inhibitor C2I-C3, we assessed the degree of intracellular Rho activation by determination of cytoplasmic and membrane-bound Rho. A decrease in membrane-bound Rho was seen after treatment of MSCs with C2I-C3 (Fig. 2C). Taken together, modulation of Rho activity is associated with changes in cytoskeletal activation as revealed by the presence of actin stress fibers in MSCs.

Inhibition of Rho Influences Transwell Migration of MSCs

From the previous findings, we hypothesized that modification of the actin cytoskeleton might alter the migration behavior of MSCs. We therefore investigated the transwell migration of MSCs after Rho inhibition. In a first set of experiments, we analyzed the transwell migration induced by human plasma. Inhibition of Rho with C2I-C3 transferase or Y27632 resulted in significantly increased migration toward plasma (Fig. 4A, 4B). Both treatment of MSCs during the migration period as well as pretreatment and subsequent washout of C2I-C3 before the assay were equally effective in inducing chemotactic migration (data not shown). A different behavior was observed in MSCs under serum-deprived conditions after treatment with LPA; although compared with untreated MSCs, pretreatment with LPA inhibited migration toward plasma (Fig. 4C), migration of LPA pretreated MSCs was enhanced when Rho was inhibited (Fig. 4C). Therefore, the inactivation of Rho can reverse the migratory response of MSCs to the same signal.

Figure Figure 4..

Migration of mesenchymal stromal cells (MSCs) induced by 0.3% plasma present in the lower well of transwell chambers. (A–C): MSCs were pretreated with C2I-C3 (1 μg/ml unless indicated otherwise), Y27632 (10 μM), and/or LPA (25 μM) for 2 hours and seeded at 20,000 into upper wells of transwell chambers. Lower wells contained 0.3% human plasma. Cells were allowed to migrate for 6 hours, and absolute cell numbers were determined. (A): Dose-response curve for C2I-C3 pretreatment; (B): comparison of Rho inhibitor C2I-C3 and Rho kinase inhibitor Y27632 pretreatment; and (C): effect of the simultaneous presence of Rho inhibitor C2I-C3 on migration of LPA pretreated MSCs. Values are means ± SD; n = 4; * p ≤ .01, ** p ≤ .001, *** p ≤ .0001. MSCs were from passage 4 (A), passage 5 (B), or passage 6 (C). Abbreviation: LPA, lysophosphatidic acid.

Since the effects of treatment with soluble inhibitors are expected to be transient in the cells, we performed lentiviral transduction of MSCs with cDNA encoding dominant active Rho (RhoV14) or C3 transferase. Due to the relatively high serum requirements of the transduced MSCs, the transwell migration assays had to be performed in the presence of 10% plasma, yielding a relatively high spontaneous migration compared with the previous serum-deprived conditions (Fig. 5). Consistent with the previous results after short-term modulation of Rho, C3 transferase transduced MSCs, however, showed increased migration capacity compared with the mock transduced controls, whereas MSCs transduced with constitutively active RhoA migrated less efficiently under these conditions (Fig. 5), confirming the data obtained with soluble Rho inhibitor.

Figure Figure 5..

Transwell migration of lentivirally transduced mesenchymal stromal cells (MSCs). MSCs were transduced using cDNAs encoding vector only, RhoV14, or C3 transferase. Thirty-six hours after transduction, 20,000 MSCs were seeded into upper wells of transwell chambers containing Dulbecco's modified Eagle's medium/10% plasma and allowed to migrate for 6 hours, and absolute numbers of migrated cells were determined. Values are means ± SD; n = 4; * p ≤ .01, ** p ≤ .001. MSCs were from passage 5.

In addition to plasma-induced MSC migration, we also tested the chemotactic response of MSCs to phospholipids or growth factors in the absence and presence of Rho inhibition but in the complete absence of serum. Whereas MSCs showed to be nonresponsive or only weakly responsive to PDGF, HGF, LPA, and S1P present in the lower wells in the absence of serum, C2I-C3 treated MSCs showed 3- to 10-fold increased chemotaxis toward PDGF, HGF, LPA, and S1P (Fig. 6).

Figure Figure 6..

Transmigration of mesenchymal stromal cells (MSCs) in serum-free conditions induced by different stimuli present in the lower well. MSCs were pretreated without or with C2I-C3 (1 μg/ml), washed by centrifugation, seeded into chemotaxis chambers that contained the indicated substances in the lower wells, and allowed to migrate for 6 hours. Absolute numbers of transmigrated cells were determined; values are means ± SD; n = 4. MSCs were from passages 4–6. Abbreviations: HGF, hepatocyte growth factor; LPA, lysophosphatidic acid; PDGF, platelet-derived growth factor; S1P, sphingosine-1-phosphate.

In conclusion, our results demonstrate that the inactivation of Rho GTPase by exogenous C2I-C3 or transfection with C3 transferase cDNA alters the actin activation status in MSCs, reducing stress fiber formation and appearance of a more filamentous actin fiber type. In parallel, induction of MSC migration toward plasma is enhanced, and MSCs can now be induced to migrate to a substantial degree toward single physiological stimuli such as LPA, S1P, PDGF, or HGF.


Migration of MSCs to sites of tissue injury is a necessary feature for tissue reconstitution, a process which was found to be guided by specific soluble factors such as growth factors and/or chemokines. Our results show a pivotal role for Rho activation in actin cytoskeletal modification and in the induction of a migratory response in MSCs by soluble factors.

Role of Rho in Chemotactic Migration of Hematopoietic and Mesenchymal Progenitor Cells

Rho GTPases have been shown to be critical regulators of the migration of several cell types, including hematopoietic progenitor cells (HPCs). Absence of the Rac1 and -2 GTPases results in severe migration defects in hematopoietic progenitor cells and impaired engraftment after transplantation [20]. In contrast, inhibition of Rho using C2I-C3 or transfection with C3 transferase cDNA led to increased in vitro migration of HPCs [23]. Ghiaur et al. [38] have furthermore demonstrated that overexpression of the dominant negative RhoA N19 mutant enhances hematopoietic engraftment in a murine bone marrow transplantation model. The results of this study demonstrate that, in MSCs, a cell type which shows much stronger activation of F-actin than HPCs, inhibition of Rho will not only result in a quantitative difference of migration but instead reverse induction of adhesion into induction of migration.

In HPCs, modification of SDF-1α and its receptor CXCR4 has been postulated to be of importance also for Rac GTPase mediated migration induction [20]. Although we and others have demonstrated the presence of the SDF-1α receptor CXCR4 also on human MSCs [17, 24, 39], SDF-1α did not induce chemotactic migration of MSCs in our hands even after pretreatment with Rho inhibitor C2I-C3. Therefore, the SDF-1α signaling pathway seems not to be significantly involved in migration responses of MSCs. Yet, our in vitro data suggest that inhibition of Rho in MSCs and their subsequently increased migratory function might lead to improved in vivo functions of MSCs (e.g., to a more efficient passage through lung capillaries and to increased responsiveness to the physiological signals that induce their extravasation and deposition in target tissues).

Alterations in Actin Cytoskeleton Morphology Induced After Inhibition of Rho

Previous studies have indicated the involvement of Rho GTPases in the cytoskeletal activation in fibroblasts, with overexpression of Rho resulting in formation of stress fibers and of focal adhesion complexes [40, 41]. The cytoskeletal changes brought about by Rho GTPases have been shown to correlate with changes in cell morphology and with altered cell migration [19]. We have shown here the functionality of the cell-permeable C3 transferase C2I-C3 to inhibit Rho and to induce similar changes in actin cytoskeleton in MSCs as seen in fibroblasts after transfection with cDNAs encoding Rho isoforms. Clearly, finer actin structures within the MSCs were associated with an induced chemotactic migration response. We also showed that an opposite effect, that is, induction of a nonmigratory phenotype, was induced by overexpression of an activating mutant of RhoA in MSCs. MSCs are therefore susceptible to modulation of their migratory response through modifications of their Rho GTPase activation.

Activation of Actin Cytoskeleton and Migration in MSCs by Lysophospholipids

S1P and LPA have been described previously as inducers of Rho GTPases that modulate the migration response of various cell types [32, 36, 37, 42]. Meriane et al. have recently implicated RhoA and Rho kinase in the induction of migration in MSCs [43]. In a collagen gel migration assay implying activation of matrix metalloproteinases, the authors demonstrate that S1P leads to an increase in serum-induced MSC migration, which is decreased after treatment with Rho kinase inhibitor. Interestingly, in the migration model of Meriane et al., matrix metalloproteinase-mediated migration stimulated by S1P through gelatin-precoated filters was associated with an increase in actin stress fibers in MSCs, whereas in our transwell model, induction of chemotactic migration was found to be enhanced when Rho activity is downregulated and when stress fibers are decreased. This may be explained by the fact that, during chemotaxis, presence of S1P acting as a chemoattractant in the lower well might not necessarily increase stress fiber formation in the migrating MSCs. In contrast, actin stress fiber formation is required in the migration situation described by Meriane et al., which involves matrix degradation. Therefore, a different activation status of the actin cytoskeleton may be optimal at different stages of migration activation in MSCs, that is, gradient sensing, polarization, and orientation versus subsequent migration through more solid matrix, as reflected by the observations in the two studies.

Our results demonstrate that MSCs express receptors for both S1P and LPA, although the results do not necessarily imply that these receptors are functional. We demonstrated that pretreatment with LPA can induce divergent migration responses depending on the activation status of Rho. Decreased transwell migration responses were observed when RhoA activation was induced in the presence of LPA, but this was reversed to an induction of migration by LPA when C2I-C3 is present. Thus, MSCs can respond bidirectionally to LPA, dependent on whether RhoA can be activated or not.

Taken together, our results suggest that the regulation of chemotactic migration in MSCs is dependent on the activation status of Rho. Substances such as phospholipids LPA and S1P can thus either act to induce the formation of a more flexible actin cytoskeleton or to increase stress fiber formation, dependent on the activation status of Rho in MSCs. These results will be of importance to further decipher the steps that lead to the directional migration activation of mesenchymal progenitor cells, and they may allow us to better influence or engineer MSCs as improved cellular therapeutics for tissue regeneration.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


We thank Heike Nürnberger and Dr. Martin Ruthardt for providing us with the PINCO vectors encoding C3 transferase and RhoA V14, Sabrina Boehme for technical support with migration experiments, and Kristine Eschedor for secretarial assistance. This work was financially supported by Grants 0312625 and 05GN0525 from the German Ministry of Health and Research (BMBF). B.G.J. is currently affiliated with the Hematopoietic Stem Cell Lab, Cancer Research UK, London, U.K.