SEARCH

SEARCH BY CITATION

Keywords:

  • Human mesenchymal stem cells;
  • Migration;
  • Basic fibroblast growth factor;
  • Growth factors

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Little is known about the migration of mesenchymal stem cells (MSCs). Some therapeutic approaches had demonstrated that MSCs were able to regenerate injured tissues when applied from different sites of application. This implies that MSCs are not only able to migrate but also that the direction of migration is controlled. Factors that are involved in the control of the migration of MSCs are widely unknown. The migratory ability of isolated MSCs was tested in different conditions. The migratory capability was examined using Boyden chamber assay in the presence or absence of basic fibroblast growth factor (bFGF), erythropoietin, interleukin-6, stromal cell-derived factor-β, and vascular endothelial growth factor. bFGF in particular was able to increase the migratory activity of MSCs through activation of the Akt/protein kinase B (PKB) pathway. The results were supported by analyzing the orientation of the cytoskeleton. In the presence of a bFGF gradient, the actin filaments developed a parallelized pattern that was strongly related to the gradient. Surprisingly, the influence of bFGF was not only an attraction but also routing of MSCs. The bFGF gradient experiment showed that low concentrations of bFGF lead to an attraction of the cells, whereas higher concentrations resulted in repulsion. This ambivalent effect of bFGF provides the possibility to a purposeful routing of MSCs.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Mesenchymal stem cells (MSCs) seem to be a useful tool for cellular therapy in cases of injured tissues. Different types of tissue engineering have been examined in the last years using MSCs. These cells showed the ability to differentiate into various kinds of cells [1, 2] and rebuild, for example, bone tissue [3, 4], cartilage tissue [5], or vessels [6]. The construction of vessels, or neovascularization, is of prime importance in a wide range of diseases. One focus of interest is the cure of cardiac infarct. To repair the infarcted regions, different kinds of applications were tested. The stem cells were administered i.v., intracoronary, transendocardial, or transepicardial [7, [8], [9], [10], [11], [12], [13], [14], [15], [16], [17]18] routes. In all cases, the MSCs had to cover a distance to reach the target area. Therefore, the common hypothesis is that MSCs possess a migratory activity. Unfortunately, little is known about the migration of MSCs. Ji et al. demonstrated that the migration of MSCs can be mediated by chemokines such as SDF-1 [19]; vascular endothelial growth factor (VEGF), as well, can mobilize progenitor cells in vivo [20, 21]. This kind of angiogenesis, which occurs through the mobilization of MSCs initiated by chemokines, is essential in therapeutic approaches. Hughes et al. tested whether a treatment with basic fibroblast growth factor (bFGF) or VEGF can increase the myocardial blood flow by neovascularization [22]. They examined whether a therapeutic angiogenesis can be induced in chronically ischemic porcine myocardium using bFGF or VEGF. The myocardial blood flow was found to be significantly increased only by the intramyocardial injection, whereas an i.v. injection did not show any effect [22]. Injection of bFGF or VEGF into the injured area resulted in diffusion of the factors from the point of injection and development of a gradient. This observation suggests that both factors can be chemotactically induced to initiate migration of cells with the capability for local neovascularization in tissue with increased factor concentration. Given the fact that MSCs have the capability for tissue replacement, it would be interesting to see whether these factors can induce a directed migration of the same.

It is known that directed migration depends on a cellular polarization. An early event during cellular polarization following the stimulation of rounded cells by chemo-attractant ligands is a change in filamentous F-actin distribution from azimuthal symmetry around the cell rim to concentrate in a particular region. Additional molecular rearrangements can cause cellular spatial asymmetries involved in migration, such as forward redistribution of chemosensory signaling receptors [23]. To determine whether a cell movement is based on cytoskeleton rearrangements, the cellular distribution of actin and vimentin was analyzed.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Cell Culture and Manipulation

MSCs of human origin were obtained from bone marrow aspirates, as well as from bone marrow from femur and hip head. Material from 10 patients was used. At the time of sampling, patients were between 49 and 84 years, with a mean age of 66.2 years. Before preparing Ficoll-Paque PLUS density gradient centrifugation (Amersham Biosciences, Uppsala, Sweden, http://www.amersham.com), the bone marrow was filtered (70 μm mesh). First change of medium (α-minimal essential medium, 20% [v/v] fetal calf serum [FCS], 200 μM l-glutamine, 100 U/ml penicillin, 100 U/ml streptomycin) was done 2 days after culturing (95% humidity, 5% CO2). Cells were used until passage 3. For every passage or experiment, the cells were plated at 2,000 cells per cm2. The medium was changed twice a week. The study was approved by the local ethics committee and conforms to the Declaration of Helsinki.

MSC Culture Quality Control

The quality of cultured MSCs was controlled by microscopic assessment of the morphology, flow cytometry, CFU-F assay, and differentiation assays.

For flow cytometry analysis, cells were stained with the given volumes of CD106-fluorescein isothiocyanate (FITC), CD105-PE (Ancell, Bayport, MN, http://www.ancell.com/), CD45-ECD, CD14-PC5, and CD34-PC5 (Beckman Coulter, Krefeld, Germany, http://www.beckmancoulter.com) for 20 minutes at room temperature, washed twice, and acquired by means of a Beckman Coulter Epics XL with Expo 32 software (Beckman Coulter). Analyses were performed as previously described [24].

To define the proliferation potential of cultured MSCs, the colony forming unit-fibroblast (CFU-F) assay was performed as described [25].

The differentiation potential of the mesenchymal stem cells were controlled by culturing the cells under conditions that were favorable for adipogenic or osteogenic differentiation as described before [1]. Adipose differentiation was analyzed using Red Oil Staining and osteogenic differentiation by von Kossa staining.

Migration/Boyden Chamber

To analyze migration of MSCs, a modified Boyden chamber with 24-well HTS Fluoro Blok insert system containing 8-μm pores (Falcon; Becton, Dickinson and Company, Heidelberg, Germany, http://www.bd.com) was used. Each insert used 104 cells. After incubation for 8 hours, the cells were fixed using 4% (w/v) paraformaldehyde. Finally, the membrane was transferred onto a coverslide, and cells were stained using mounting medium containing 4,6-diamidino-2-phenylindole (Vectashield; Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). The number of migrated cells was counted.

The additional factors were used in the following concentrations: 500 ng/ml stromal cell-derived factor (SDF)1-β (Upstate, Hamburg, Germany), 10 ng/ml IL-6 (Boehringer Mannheim, Mannheim, Germany, http://www.boehringer.com), 0.5 U/ml erythropoietin (Roche Diagnostics, Mannheim, Germany, http://www.roche-applied-science.com), 20 ng/ml VEGF (PAN Biotech, Nürnberg, Germany), 1 ng/ml up to 400 ng/ml bFGF (Cell Concepts, Umkirch, Germany), 30 μM LY294002 (Calbiochem, Schwallbach Germany, http://www.emdbiosciences.com).

Wound and Healing Assay

For wounding, MSCs were plated at 2,000 cells per cm2 on glass coverslips until they reached a confluence of ∼80%. Afterwards, a pipette tip was used to prepare a scratch across the layer of MSCs as described previously by Faber-Elman et al. [26].

Chemotaxis

To prove the chemotactic behavior of MSCs, a concentration gradient with bFGF was generated using methyl cellulose discs (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com). The discs were imbued with bFGF and dried completely. For controls, the discs were imbued with phosphate-buffered saline (PBS). Cells were plated in a line at the middle of a 10-cm Petri dish. After 6 hours, the Petri dish was washed to remove nonadherent cells. After replacing the medium, the imbued discs were carefully placed in a defined distance between disc and cells. The cells were incubated for 5 days without convulsing the dishes. Because of the consecutive release of bFGF, a concentration gradient was generated (Fig. 1A). After 5 days, cells were fixed with 4% (w/v) paraformaldehyde. All experiments were repeated four times.

thumbnail image

Figure Figure 1.. Cartoon diagram of the bFGF gradient experiment. (A): Mc was imbued with bFGF and dried afterwards. MSCs were placed in a given area with a defined distance to the nitrocellulose. By adding medium, the bFGF diffused in a gradient in the dish. The MSCs migrated in the presence of bFGF (mM), whereas in the absence of bFGF, the MSCs stayed in the placed area. The concentration lines were ascertained using trypan blue-imbued nitrocellulose. (B, C): Comparison of the sideward migration of the MSCs with and without bFGF with cell-to-methyl cellulose distances of 13 mm (B) and 19 mm (C). Abbreviations: bFGF, basic fibroblast growth factor; C, concentration lines; M, mesenchymal stem cells; Mc, methyl cellulose.

Download figure to PowerPoint

Cytoskeleton Staining

For detection of actin filaments, phalloidin-FITC (P5282; Sigma-Aldrich) was used. Vimentin was proven using a Cy3-conjugated mouse anti-human vimentin antibody (C9080; Sigma-Aldrich). Both methods were performed according to the manufacturer's guidelines.

Protein Detection

Akt was detected by using rabbit anti-Akt (immunohistochemistry [IH] 1:500, Western blot [WB] 1:1,000; Upstate), rabbit anti-phosphoAkt (Thr308) (IH 1:500, WB 1:100; Upstate). Actin was detected by using mouse anti-actin (WB 1:1,500; Chemicon, Temecula, CA, http://www.chemicon.com).

For densitometry analysis, the gray scale values were measured using the computer-based system ImageJ Version 1.33 from the NIH [27]. For each test, 25 cells were surveyed.

Western Blot Analysis

Proteins were separated by electrophoresis in a continuous 4% stacking gel and a sodium dodecyl sulfate (SDS)-polyacrylamide 12% separating gel under constant current (70 mA [60 minutes] and 120 mA [180–240 minutes]) and transferred onto a polyvinylidene difluoride membrane (Roche). Detection of proteins was performed as described before [28].

Statistical Analysis

All data are presented as mean ± SD. Data analysis were performed using analysis of variance with Bonferroni post hoc test and/or Student's t test for unpaired data. Data were considered significant at p < .05.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

MSC Mobility

The migration property was tested using Boyden chambers with 8-μm pores. After an incubation time of 8 hours, a total number of 119 ± 11 cells migrated through the membrane without the addition of a stimulus. The highest increase of migrating cells was observed using bFGF. In that case, 268 ± 45 cells migrated through the membrane, which corresponds to a significant 2.3-fold increase compared with the control (p = .018). All other growth factors tested also induced an increase in migration. The increase for erythropoietin (EPO) (1.3-fold, p = .33) was not significant, whereas the increases for IL-6 (1.3-fold, p = .02), SDF1-β (1.7-fold, p = .001), and VEGF (1.8-fold, p = .002) were significant (Fig. 2A).

thumbnail image

Figure Figure 2.. For testing the migratory activity, a modified Boyden chamber assay was performed. After 8 hours, the absolute number of cells on the bottom site of the migration filter was counted. (A): Number of migrated cells in the presence of 20% FCS. (B): Number of migrated cells in the presence of 2% FCS. (C, D): Migratory activity was analyzed in dependence on bFGF concentration (indicated in ng/ml). The assay was performed with and without heparin and with 20% FCS (C) or 2% FCS (D), respectively. *, significant changes. Abbreviations: bFGF, basic fibroblast growth factor; EPO, erythropoietin; FCS, fetal calf serum; SDF, stromal cell-derived factor; VEGF, vascular endothelial growth factor.

Download figure to PowerPoint

These observations depend on the medium concentration of FCS. By decreasing the FCS concentration from 20% to 2%, the absolute number of migrating cells decreased to 49 ± 3 (0.4-fold, p = .001). The mobilizing capacity of all factors except bFGF was directly affected. The factors IL-6 (1.2-fold, p = .04), SDF1-β (1.5-fold, p = .002), and VEGF (1.4-fold, p = .006) still showed the ability to induce a significant increase of migrated cells, but the intensity of mobilization was decreased. In contrast, the number of migrated cells using bFGF was nearly unchanged. Compared with the control, a threefold increase using bFGF was observed (p < .001) (Fig. 2B).

bFGF and Heparin

The glycosaminoglycans heparin and heparan sulfate contain similar structural units in varying proportions, providing considerable biological functions [29]. Through these interactions, heparan sulfate proteoglycans and also heparin participate in many events during cell adhesion, migration, proliferation, and differentiation [30]. The effect of bFGF is directly influenced by heparin [31]. Therefore, we repeated the migration assays in the presence and absence of heparin and with 2% and 20% FCS.

Using 20% FCS, the migratory activity increased in relation to the bFGF concentration. The migration increased steadily and reached a maximum at a concentration of 50 ng/ml. In the range from 50 ng/ml to 200 ng/ml, the migratory activity reached a plateau. In the presence of heparin, this plateau was significantly decreased by 46% (p = .001) (Fig. 2C).

In the presence of 2% FCS, an increase of mobilization by 5 ng/ml bFGF was observed. Further increase of bFGF did not enhance migration; rather, it resulted in a reduced activity. In contrast to the tests performed with 20% FCS, the addition of heparin was without any impact (peak increase: 4-fold with heparin vs. 4.3-fold without heparin; p = .57) (Fig. 2D).

bFGF and Akt

To elucidate whether bFGF activates MSCs, the cells were cultured with and without bFGF (20 ng/ml) for 4 and 48 hours. MSCs were stained for Akt and pAkt and analyzed by densitometry afterwards. After 4 hours of incubation with bFGF, the concentration of Akt increased by 78% (gray-scale value control, 5.5 ± 1.3; bFGF, 9.9 ± 2.9; p = .001) and remained significantly increased by 61% after 48 hours (control, 8.0 ± 1.4; bFGF, 12.9 ± 2.5; p < .001). The amount of phosphorylated Akt was also increased significantly. After 4 hours, the concentration of phosphorylated bFGF was increased by 77% (control, 25.8 ± 6.4; bFGF, 45.9 ± 8.2; p < .001) and 55% after 48 hours (control, 17.7 ± 2.7; bFGF, 27.4 ± 5.6; p < .001). These data were validated by Western blot analysis (Fig. 3).

thumbnail image

Figure Figure 3.. Western blot analysis of human mesenchymal stem cells. Cells were incubated with 20 ng/ml bFGF. After 4 hours of incubation, the cells showed an increased concentration of Akt and pAkt compared with control. Abbreviations: bFGF, basic fibroblast growth factor; ctr, control; pAkt, phosphor-Akt.

Download figure to PowerPoint

To check whether the activation of Akt is essential for the migratory activity of MSCs, we used the phosphatidylinositol 3-kinase (PI3K) inhibitor LY294002, which blocks the activation of Akt. In the presence of the inhibitor LY294002, the migratory activity dropped down to 53% compared with control (p < .001). The decreased mobility was also not affected by adding bFGF (p = .12). The relative number of migrated cells was 63% compared with control.

Wound and Healing

The wound and healing assay was designed as a method of simulating the ability of cells to reconstruct a tissue [26]. A uniform monolayer of MSCs was scratched with a pipette tip. Five hours after scratching, the cells resettled 26% of the wound area in the absence of growth factors, whereas in the presence of 20 ng/ml bFGF, 48% of the area was resettled after 5 hours, because of significant migratory activity (p < .001) (Fig. 4). Furthermore, we monitored the migratory activity by time lapse video. The migrating cells moved independently into the pruned area. The mobility was shown to be not due to pushing of proliferating cells in the unwounded area (supplemental online video).

thumbnail image

Figure Figure 4.. Wound and healing assay with a confluent monolayer of mesenchymal stem cells. (A): After scratching, the surrounding cells migrated into the scratched area (time after scratching is indicated). (B): The time-dependent percentage of resettled area was calculated under control conditions and in the presence of bFGF. *, significant changes. Abbreviation: bFGF, basic fibroblast growth factor.

Download figure to PowerPoint

Gradient bFGF

The data suggest that the migratory activity depends directly on the bFGF concentration. To answer the question of whether the MSCs are migrating toward a preferred bFGF concentration, we performed a two-dimensional migration assay. In a Petri dish, the MSCs were settled into a defined area at the center. To induce a constant bFGF gradient, a dried methyl cellulose disc enriched with bFGF was placed at a defined distance, with a respective PBS control. After 5 days, the cells were fixed, and migration activity was analyzed. The borderline of the migrated cells was measured. The middle of the methyl cellulose disc was used as the origin of migration (Fig. 1A).

Two different distances between the methyl cellulose and the cell frontier were used (13 and 19 mm). In the presence of a bFGF gradient, the prevailing cells migrated sideward along the concentration lines. In the 13-mm distance, there was no migration, and in the distance of 19 mm, there was a slight migration directed to the bFGF center. In both approaches, the cells migrated up to 10 mm to the right angle. The maximum sideward migratory activity was observed at a distance of around 18 mm. This maximum migratory activity was not influenced by the start position of the MSCs but rather by the absolute distance to the methyl cellulose. Under control conditions using PBS-imbued methyl cellulose, the MSCs did not leave the plated area (Fig. 1B, 1C).

Cell Alignment

Migration requires a cell alignment that is dependent on cytoskeletal organization [23]. We therefore analyzed the organization of the cytoskeleton using phalloidin staining and immunohistochemical detection of vimentin. In the control experiments, a large number of cells did not show a common orientation of the actin filament. This phenotype was defined as “round” cells. The filaments were organized in independent directions and were not aligned. In the presence of a bFGF gradient, most of the actin filaments were parallel-oriented. Cells phenotype with a parallel orientation of the actin filaments was defined as “aligned” cells (Figs. 5, 6).

thumbnail image

Figure Figure 5.. Fluorescence microscopy of phalloidin staining (A–D) and immunocytochemical detection of vimentin (E–H). (A, C): Representative nonaligned actin structure under control conditions with detail picture (C). (B, D): Aligned actin structure in the presence of bFGF with detail picture (D). Shown is the distribution of vimentin under control conditions (E, G) and with bFGF (F, H). Abbreviation: bFGF, basic fibroblast growth factor.

Download figure to PowerPoint

thumbnail image

Figure Figure 6.. Fluorescence microscopy of phalloidin and anti-vimentin double staining. Exemplary cells under control conditions (A–D) and in the presence of a bFGF gradient (E–H). The cells were stained with phalloidin-fluorescein isothiocyanate (A, C, E, G) and anti-vimentin-Cy3 (B, D, F, H). Abbreviation: bFGF, basic fibroblast growth factor.

Download figure to PowerPoint

In the absence of bFGF, only 59% of the cells showed an aligned phenotype. In the presence of a bFGF gradient, 96% of the cells exhibited an aligned phenotype (p < .001) (Fig. 7A). We also analyzed the dependence of polarization of the actin filaments on the center of the bFGF gradient. The cell alignment was divided into four groups. Cells that were mostly directed to the center were defined as cells with an angle of 0°. Cells with a right-angled actin orientation to the center were defined as cells with an angle of 90°. In addition, we defined two groups with 30° and 60° angles, respectively. In the control experiment, the aligned cells were oriented randomly. All orientations were distributed equally. In the presence of a bFGF gradient, most of the cells showed an actin filament orientation directed to the center of the bFGF gradient (0°) (Fig. 7B).

thumbnail image

Figure Figure 7.. The phalloidin-stained cells were scaled in round or aligned. (A): The percentage was calculated for control conditions and in the presence of bFGF. (B): The aligned cells were scaled in dependence on the angle to the gradient center. Cells were scaled as oriented directly to the gradient center (0°), 30°, or 60° turned to the center or polarized in a right angle (90°). Abbreviation: bFGF, basic fibroblast growth factor.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

MSCs showed a directed migratory ability. Different cytokines were used to induce a migratory activity. We analyzed the effect of EPO, IL-6, SDF1-β, bFGF, and VEGF. All tested factors showed the ability to increase the migratory activity of MSCs.

The highest effect was observed in the presence of bFGF. By using a modified Boyden chamber assay, the mobility of MSCs was increased at maximum by 2.3-fold compared with control (p = .018). This effect was neutralized in the presence of the PI3K inhibitor LY294002. LY294002 blocks the activity of PI3K and results in inhibition of Akt activity [32]. It is known that bFGF-induced phosphorylation of Akt depends on PI3K [33]. These data indicate the importance of the PI3K/Akt pathway for bFGF induced migration. In addition, the results were supported by the observation that the amount of phospho-Akt increased significantly in the presence of bFGF (p < .001). We therefore conclude that bFGF-induced migratory activity is controlled by the conventional fibroblast growth factor (FGF) receptor pathways.

It is known that activation of FGF receptors is quite sensitive to heparan sulfate and also to heparin. Heparan sulfate proteoglycans act as low-affinity receptors and, together with the fibroblast growth factor receptors (FGFRs), are a requisite part of a dual receptor system that, along with a cell surface tyrosine kinase receptor, transduces ligand activation along appropriate cytosolic and nuclear pathways [34]. Heparin has been demonstrated to be crucial in the interaction of bFGF and FGF receptors [35]. It has been demonstrated that bFGF binding to FGFRs can be inhibited and facilitated by heparin in a concentration-dependent manner such that higher concentrations inhibited binding whereas low concentrations facilitated the binding [36, 37]. Therefore, we performed migration assays with high-dose (20% FCS) and low-dose (2% FCS) serum and also with and without heparin. In the presence and absence of heparin, the cells showed comparable bFGF dependent behavior. In low-dose FCS, heparin seemed to have no impact on the migratory activity of MSCs. Surprisingly in high-dose FCS, the influence of heparin was obvious. The maximum migratory efficiency was decreased by nearly 50% in high-dose FCS in the presence of heparin. These data indicate that an additional factor in the serum modulates the interplay among bFGF, heparin, and FGF receptors.

bFGF is a member of a family of growth factors with a complex network of receptors. Seven variants of the four forms of the FGF receptors (FGFRs) are known [38]. The predominant signaling cascade(s) activated by a particular FGF depend on the cell type and its receptor expression, as well as its immediate environment and stage of development [39]. This complexity accounts for the diverse outcomes of FGF signaling on cell division and migration, affecting developmental processes, angiogenesis, wound healing, and tumorgenesis [36, 40, 41]. bFGFs showed the ability to modulate the mobility of MSCs in a dose-dependent manner. When the concentration of bFGFs was increased, the mobility increased up to maximum. In the presence of a low FCS concentration, the mobility did not reach a plateau; instead, when the bFGF concentration was further increased, the mobility decreased nearly down to the initial mobility. These data suggest that the impact of bFGFs on the mobility of MSCs is mediated by two different pathways, probably mediated by at least two different receptors.

Landgren et al. identified a domain at C-terminal on the FGFR-1 that is essential for chemotaxis mediated by the FGFR-1 [42]. The role of this domain for migration and the according pathway are still unknown. Due to the fact that this domain is conserved in all four FGF receptors, it might be possible that the receptors have different impacts on the cell migration. We propose that at least two receptors are able to control the bFGF-mediated chemotaxis in a two-dimensional manner. Therefore bFGF is able not only to attract but also to control the positioning of MSCs. The gradient experiment demonstrated that an increasing concentration of bFGF attracted the MSCs, whereas higher concentrations of bFGF resulted in a repulsion of the cells. Most of the cells migrated along a continuous concentration level of bFGF.

In cases of hypoxemic stress (e.g., due to cardiac infarction, tissue injury, or insufficient vascularization), fibroblasts are secreting bFGF [43, 44]. We have demonstrated that bFGF is able to attract MSCs. It is known that bFGF is able to induce proliferation and differentiation in various kinds of cells [45, [46], [47]48]. It is possible that bFGF, secreted by cells of injured or insufficient tissue, attracts the MSCs, which migrate and infiltrate into this tissue. The effect of the bFGF secreted might be to attract the MSCs into the injured area and initiate regeneration.

We therefore postulate that neovascularization and also wound healing are closely associated with the interaction between bFGF and the activation of MSCs. This assumption is based on the clinical observation that therapeutic angiogenesis is successfully practicable by using bFGF in, for example, coronary artery diseases [49, [50], [51]52].

Disclosures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

The authors indicate no potential conflicts of interest.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

We thank G. Rajesh Kumar for conscientious reading of the manuscript. This work was supported by a grant from the Novartis Foundation to W.B.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References