Author contributions: C.H.K., G.S., A.J.M., and A.A.-L.: collection and/or assembly of data and final approval of manuscript; K.M., J.R., and M.K.: collection and/or assembly of data; M.S. and S.B.: collection of data; M.Z.R.: conception and design, financial support, and manuscript writing.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS November 29, 2012.
Ceramide-1-phosphate (C1P) is a bioactive lipid that, in contrast to ceramide, is an antiapoptotic molecule released from cells that are damaged and “leaky.” As reported recently, C1P promotes migration of hematopoietic cells. In this article, we tested the hypothesis that C1P released upon tissue damage may play an underappreciated role in chemoattraction of various types of stem cells and endothelial cells involved in tissue/organ regeneration. We show for the first time that C1P is upregulated in damaged tissues and chemoattracts bone marrow (BM)-derived multipotent stromal cells, endothelial progenitor cells, and very small embryonic-like stem cells. Furthermore, compared to other bioactive lipids, C1P more potently chemoattracted human umbilical vein endothelial cells and stimulated tube formation by these cells. C1P also promoted in vivo vascularization of Matrigel implants and stimulated secretion of stromal cell-derived factor-1 from BM-derived fibroblasts. Thus, our data demonstrate, for the first time, that C1P is a potent bioactive lipid released from damaged cells that potentially plays an important and novel role in recruitment of stem/progenitor cells to damaged organs and may promote their vascularization. STEM CELLS2013;31:500–510
Evidence has accumulated that nonhematopoietic stem cells, including, multipotent stromal cells (MSCs), endothelial progenitor cells (EPCs), or very small embryonic-like stem cells (VSELs), are mobilized into peripheral blood (PB) in response to organ injuries (e.g., heart infarct or stroke) and can be subsequently recruited to the damaged tissues [1–14]. These circulating stem cells could potentially contribute to tissue repair directly by supporting formation of new blood vessels (e.g., EPCs) [8–10] or indirectly as a source of several growth factors and membrane-derived microvesicles (e.g., MSCs) that provide trophic signals that inhibit cell apoptosis and stimulate vascularization of damaged tissues [15–18].
It is well established that stem cells circulating in PB are recruited to damaged organs in response to chemotactic factors, including chemokines (e.g., stromal cell-derived factor-1, SDF-1) and some growth factors (e.g., vascular endothelial growth factor, VEGF; basic fibroblast growth factor; or hepatocyte growth factor, HGF) . Evidence has accumulated that, in addition to chemotactic peptides, bioactive lipids also play an important role in the process of recruitment of circulating stem cells [20–23]. The best known of these is sphingosine-1-phosphate (S1P), which chemoattracts various types of stem cells, including MSCs and EPCs [24–27].
Although other mechanisms for ceramide-1-phosphate (C1P) synthesis must exist, C1P is a product of the phosphorylation of ceramide by ceramide kinase (CerK) and, like S1P, belongs to a family of bioactive sphingolipids . Both S1P and C1P are important intracellular signaling molecules. However, in contrast to S1P, C1P is not secreted by intact cells, but released from “leaky” damaged cells . It is also known that C1P, in contrast to its parent molecule ceramide, shows antiapoptotic effects in several types of cells, including macrophages  and keratinocytes , and also plays a role in chemoattracting macrophages  and stimulating proliferation of myoblasts . Recently, we have demonstrated that C1P is a potent chemoattractant for normal hematopoietic stem/progenitor cells (HSPCs) . In particular, we demonstrated that C1P is elevated in bone marrow (BM) tissue damaged by myeloablative treatment and may stimulate homing of transplanted HSPCs .
In contrast to S1P, which binds to five different G protein-coupled receptors , the C1P receptor(s) have not yet been identified. However, based on its signaling sensitivity to pertussis toxin (PTx), the C1P receptor is most likely similar to the Gαi protein-coupled seven-transmembrane spanning S1P receptors . Interestingly, it has been reported that dermal microendothelial cells from mice lacking CerK show profound defects in in vitro angiogenesis assays , which suggests that C1P could also be endowed with proangiopoietic properties.
In this work, we became interested in whether C1P, like S1P [24–27], plays a role in trafficking of MSCs and EPCs. We observed that C1P (a) becomes upregulated in damaged organs, (b) regulates migration and adhesion of stromal (MSCs) and endothelial cells (human umbilical vein endothelial cells [HUVECs]), and (c) directly affects in vitro angiogenesis in the HUVEC model of tube formation and more important it promoted in vivo vascularization of Matrigel implants. Building on these earlier observations, we here demonstrate for the first time that C1P, like S1P , becomes overexpressed at sites of tissue/organ injury and also provides chemotactic homing signals for nonhematopoietic stem cells. Because of this finding, we postulate that modulation of C1P expression and signaling may play an important future role in regenerative medicine. Thus, more work is needed to identify C1P receptor/s and develop compounds that will modify in vivo biological availability of C1P.
MATERIALS AND METHODS
Isolation of BM Cells
Murine BM mononuclear cells (BMMNCs) were isolated by flushing the femurs and tibias of pathogen-free C57BL/6. Whole BM cells were separated by Ficoll gradient, washed, and resuspended in RPMI medium (Thermo Fisher Scientific, South Logan, UT, www.thermofisher.com) containing 0.5% bovine serum albumine (BSA) (Sigma-Aldrich, St Louis, MO, www.sigmaaldrich.com). Human BM samples were obtained from donors after informed consent, and all studies with human cells were approved by U.K. and UofL IRBs.
MSC and HUVEC Culture
BM-derived stromal cells (MSCs) were isolated from whole murine or human BM cells. Whole BM cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 20% fetal bovine serum (FBS) and 50 U/ml penicillin/streptomycin for more than 7 days at 37°C in a 5% CO2 incubator. Nonadherent hematopoietic cells were discarded and adherent cells were maintained. Human MSCs were expanded from BM samples isolated from donors after informed consent. Human umbilical vein endothelial cells (HUVECs) were isolated from umbilical vein using collagenase I (Sigma-Aldrich) digestion and cultured in endothelial basal medium (EBM) (Lonza, Walkersville, MD, www.lonza.com) supplemented with 2% FBS, bovine brain extract, recombinant human epidermal growth factor, hydrocortisone, and GA-1000 (all supplements to EBM reagents from Lonza) as described .
Cell Migration Assay
The 8-μm polycarbonate membranes were coated with 50 μl of 0.5% gelatin for 1 hour. MSCs and HUVECs were detached with Trypsin-EDTA, washed in DMEM (or EBM), resuspended in DMEM (or EBM) with 0.5% BSA, and seeded at a density of 3 × 104 cells per well into the upper chambers of Transwell inserts (Costar Transwell; Corning Costar NY, www.corning.com). The lower chambers were filled with SDF-1 (5, 50, or 300 ng/ml, R&D Systems Inc., Minneapolis, MN, www.rndsystems.com), S1P (0.1 μM, Cayman Chemical, Ann Arbor, MI, www.caymanchem.com), C16-ceramide-1-phosphate (1 μM, Avanti Polar Lipids, Alabaster, AL, avantilipids.com), C18-ceramide-1-phosphate (0.1–10 μM, from bovine brain, Sigma-Aldrich, sonicated in distilled water), lysophosphatidic acid (LPA, 1 μM, Avanti Polar Lipids), or lysophosphatidylcholine (LPC, 1 μM, Avanti Polar Lipids) in 0.5% BSA DMEM or EBM (control). After 24 hours for HUVECs or 48 hours for MSCs, the inserts were removed from the Transwell plates. Cells remaining in the upper chambers were scraped off with cotton wool, and cells that had transmigrated were stained with HEMA 3 according to the manufacturer's instructions (Fisher Scientific, Pittsburgh, PA, www.fishersci.com) and counted either on the lower side of the membranes or on the bottom of the lower Transwell chambers.
In chemotaxis assays performed on sorted by fluorescence-activated cell sorting (FACS) VSELs, EPCs, and MSCs, we used Costar Transwell 24-well plates with 5-μm filters (Costar Corning). Cells after sorting were resuspended in RPMI with 0.5% BSA and loaded to upper chamber, whereas the lower chamber contained medium alone (RPMI + 0.5% BSA) or medium supplemented with C1P (50 μM). After 3 hours incubation in 37°C, 95% humidity, and 5% CO2, cells that migrated to lower chamber were counted using inverted microscope.
Cells migrating in response to C18-C1P were stained in medium containing 2% FBS. All monoclonal antibodies (mAbs) were added at saturating concentrations and the cells incubated for 30 minutes on ice, washed twice, resuspended in staining buffer at a concentration of 5 × 106 cells per milliliter, and analyzed using an LSR II (BD Biosciences, Mountain View, CA, www.bdbiosciences.com). The following anti-mouse mAbs were purchased from BD Pharmingen (San Diego, CA, www.bdbiosciences.com): Ly-GA/E (Sca-1, FITC, clone D7), lineage marker (CD45R [also known as B220], PE, clone RA3-6B2), Ly-6G/Ly-6C (PE, clone RB6-8C5), T-cell receptor β (PE, clone H57-597), T-cell receptor-γδ (PE, clone GL3), CD11b (PE, clone M1/70), Ter119 (PE, clone TER-119), CD51 (biotin, clone RMV-7, with streptavidin-conjugated to PE-Cy5 as secondary Ab), CD31 (APC, clone 390), and CD45 (APC-Cy7, clone 30-F11). The Sca-1+Lin−CD45− cells (VSELs), Lin−CD45+CD31+ (EPCs), and Lin−CD45−CD31−CD51+Sca-1+ (MSCs) were isolated as described .
BM-derived MSCs and HUVECs were starved overnight in DMEM or EBM containing 0.5% BSA at 37°C and stimulated with S1P (0.1 μM), C16-C1P (10 μM), C18-C1P (10, 20, or 50 μM), LPA (1 μM), or LPC (1 μM) for 5 minutes at 37°C and then lysed for 30 minutes on ice in M-PER lysing buffer (Pierce, Rockford, IL, www.piercenet.com) containing protease and phosphatase inhibitors (Sigma-Aldrich). The extracted proteins (30 μg) were separated and analyzed for phosphorylation of MAPKp44/42 and AKT (Ser473). Phospho-specific antibodies were purchased from Cell Signaling Technology Inc. (Danvers, MA, www.cellsignal.com). Equal loading in the lanes was evaluated by stripping the blots and reprobing with the monoclonal or polyclonal antibodies against total MAPKp44/42 and total Akt. For blocking signaling, cells were treated with PTx (0.05 μg/ml, Sigma-Aldrich) for 1 hour. The membranes were developed with an enhanced chemiluminescent reagent (Amersham Life Sciences, Arlington Heights, IL, www.gelifesciences.com) and exposed to film (HyperFilm, Amersham). SDF-1 and COX-2 were analyzed in cell extract prepared from murine stroma after overnight starvation followed 24 stimulation with C18-C1P.
SDF-1 and COX-2 were analyzed in cell extract prepared from murine stroma after overnight starvation followed by 36 hours stimulation with C18-C1P. Extracts were prepared as described above and blots were developed with anti-SDF-1 (R&D Systems Inc.) and anti-COX-2 (Abcam, Cambridge, MA, www.abcam.com). Equal loading in the lanes was evaluated with anti-β-actin antibody (Sigma-Aldrich).
Measurement of C1P in Irradiated Tissues and Myocardium After Experimental Ex Vivo Heart Ischemia
Pathogen-free C57BL/6 mice were used for measurements of C1P in irradiated organs and in the myocardial hypoxia experiments. In irradiation experiments, mice were irradiated by 500 cGy from γ-irradiator source and 24 hours later major organs were extracted.
In ex vivo heart ischemia experiments, whole hearts were excised and cannulated for the Langendorf perfusion technique using perfusion buffer containing 4.7 mM K+ and 1.8 mM Ca2+. The control hearts were allowed to beat for 50 minutes in warmed and oxygenated perfusion buffer, then removed and flash frozen in liquid nitrogen. The ischemic hearts were hung and allowed to beat for 5 or 6 minutes with flow in the warmed and oxygenated perfusion buffer, then the flow was stopped for 30 minutes and the hearts bathed in the warmed buffer. Following ischemia, the flow was restarted, and the hearts started beating in approximately 0.5–1-minute after onset of flow and continued beating for 20 minutes to simulate reperfusion injury. Ischemic hearts were then flash frozen as detailed above. Frozen myocardial samples were used for total C1P measurements using samples of equal weight. Following homogenization, the lipids were extracted as detailed above. In parallel, pathogen-free C57BL/6 mice were irradiated with 500 cGy (γ-irradiation), the organs removed, and, following homogenization, lipid extraction and mass spectrometry measurements performed using liquid chromatography electrospray ionization tandem mass spectrometry (HPLC ESI MS/MS), as described .
To evaluate secretion of matrix metalloproteinase-9 (MMP9) by stromal cells, conditioned media harvested from these cells were analyzed by zymography, as described previously by us .
Messenger RNA Expression Quantification
Total RNA was isolated using the RNeasy Mini Kit (Qiagen Inc., Valencia, CA, www.qiagen.com) from MSCs after stimulation with C18-C1P for 3 hours to evaluate SDF-1 and Cox-2 expression levels. Messenger RNA was reverse transcribed with 500 U of Moloney murine leukemia virus reverse transcriptase primed with oligo-dT. The resulting cDNA fragments were amplified using the SYBR green system (Applied Biosystems, Carlsbad, CA, www.appliedbiosystems.com). The primer sequences for mouse β-2 microglobulin (β2m) were 5′-CAT ACG CCT GCA GAG TTA AGC-3′ (forward) and 5′-GAT CAC ATG TCT CGA TCC CAG TAG-3′ (reverse). For SDF-1, the primer sequences were 5′-CGT GAG GCC AGG GAA GAG T-3′ (forward) and 5′-TGA TGA GCA TGG TGG GTT GA-3′ (reverse). For Cox-2, 5′-TGA GCA ACT ATT CCA AAC CAG C-3′ (forward) and 5′-GCA CGT AGT CTT CGA TCA CTA TC-3′ (reverse). For mouse MMP2, 5′-GGA CCC CGG TTT CCC TAA G-3′ (forward) and 5′-GTC CAC GAC GGC ATC CA-3′ (reverse). For mouse MMP9, 5′-CCT GGA ACT CAC ACG ACA TCT TC-3′ (forward) and 5′-TGG AAA CTC ACA CGC CAG AA-3′ (reverse). For human β-actin, 5′-GGC CCA CAG AAT GCA GAA GGA-3′ (forward), 5′-CGA TCC ACA CGG AGT ACT TG-3′ (reverse). For human MMP2, 5′-TCA AGG GCA TTC AGG AGC TC-3′ (forward), 5′-TGC CAA GGT CAA TGT CAG GAG-3′ (reverse). For human MMP9, 5′-TGG TTA CAC TTCG GGT GGC A-3′ (forward), 5′-GGC CCC AGA GAT TTC GAC TC-3′ (reverse). The relative quantity of the target, normalized to an endogenous control gene (β2m or β-actin) and relative to a calibrator, was expressed as 2−ddCt (fold difference), in which dCt equals the Ct of the target genes minus the Ct of the endogenous control gene, and ddCt equals the dCt of the samples for the target gene minus the dCt of the calibrator for the target gene.
Human Endothelial Tube Formation
HUVECs were assessed on a synthetic basement membrane according to the manufacturer's protocol (Matrigel, BD Biosciences). Briefly, the matrix was thawed at 4°C and polymerized at 37°C for 30 minutes before use. HUVECs (less than passage 4) were resuspended in reduced-serum medium with 0.5% BSA (negative control), FGF-2 (50 ng/ml, Pepro Tech, RockyHill, NJ, www.peprotech.com), SDF-1 (50 ng/ml), S1P (0.1 μM), C16-C1P (1 μμ), C18-C1P (1 μμ), LPA (1 μM), or LPC (1 μμ). The cells were seeded (2 × 104 cells per well, in each of 24 wells) on the Matrigel and placed in a 37°C, 5% CO2 incubator. The tubes in each well were counted and phototraphed. All conditions were tested in duplicate wells in three separate experiments.
Enzyme-Linked Immunosorbent Assay for SDF-1
Level of SDF-1 in cell extracts as well as in conditioned medium harvested from MSCs after 36 hours stimulation with C1P (1 μM) was analyzed by sensitive ELISA assay as described previously by us .
Colony-Forming Unit of Fibroblast Assay
Whole BM cells (1 × 106 cells per well, in each of six wells) were stimulated with C1P (1 or 10 μM). On the next day, nonadherent cells were removed, and each culture was incubated in dulbecco's modified eagle medium (DMEM) plus 20% fetal bovine serum (FBS) for 14 days as described . The adherent cells were washed twice with PBS, fixed, and stained with the HEMA 3 staining kit.
Cell Proliferation Assay
Cells were seeded in culture plates at an initial density of 3 × 104/well in the presence or absence of C18-C1P (0.1 or 1 μM). The cell number was determined at 1, 2, 3, and 4 days after culture initiation. At the indicated time points, cells were harvested from the wells by trypsinization and the number of cells was determined using a Bürker hemocytometer or LSR II flow cytometer (BD Biosciences).
In Vivo Angiogenesis Assay
Severe combined immuno deficiency (SCID) mice were injected subcutaneously in the abdominal region with 0.4 ml Matrigel/mouse (BD Bioscience) containing C1P (1 μM or 10 μM), FGF-2 (50 ng/ml, positive control), or Matrigel without any additives (negative control). After 2 weeks, implants were isolated, homogenized at 4°C, centrifuged, and hemoglobin concentration in supernatant was determined using Drabkin method according to manufacture protocol (Sigma-Aldrich).
Arithmetic means and SDs were calculated using Instat 1.14 software (Graphpad, San Diego, CA, www.graphpad.com). Statistical significance was defined as p < .05. Data were analyzed using Student's t test for unpaired samples and the t test for paired samples.
C1P Chemoattracts BM-Residing MSCs, EPCs, and VSELs
C1P has been demonstrated to be a chemoattractant for macrophages  and BM-residing HSPCs . Thus, we became interested in whether C1P chemoattracts other types of nonhematopoietic stem cells residing in BM, including MSCs, EPCs, and VSELs. Using the Transwell migration assay combined with FACS analysis of cells collected from the lower chambers of Transwell plates after migration, we observed that C1P (chemoattracts BM-residing VSELs (Sca-1+Lin−CD45−, Fig. 1A), MSCs (Sca-1+CD51+CD31−Lin−CD45−, Fig. 1B middle panel), and EPCs (CD31+Lin−CD45−, Fig. 1B lower panel), which suggests that C1P may promote egress of several types of stem cells into PB. Overall, we noticed that the number of murine BM-derived VSELs, EPCs, and MSCs in lower chambers of Transwell system containing C1P was higher × 3.1% ± 0.3%, 1.4% ± 0.1%, and 1.3% ± 0.1%, respectively, as compared to control chambers with no C1P.
Since chemotaxis of rare population of cells could be somehow affected in Transwell assay by admixture of high number of other nucleated cells present in suspension of BM cells, we purified by FACS population of BM-derived VSELs, EPCs, and MSCs and performed chemotaxis of these cells to C1P gradient (Fig. 1C). We noticed that all these cell populations were chemoattracted by C1P.
C1P Is Upregulated in Damaged Tissues
We next became interested in whether C1P, which promotes egress of BM-derived stem cells into PB, could also be potentially involved in chemoattracting circulating stem cells to damaged tissues. Since several clinical trials using stem cells have been performed in patients after acute myocardial infarction [39, 40], we used a murine model of acute heart hypoxia to test this hypothesis. As shown in Figure 2A, using HPLC ESI MS/MS analysis, we observed a significant increase in C1P level in murine hypoxic hearts. Next, based on our observation that C1P and S1P are upregulated in irradiated murine BM , we next analyzed C1P levels in other murine organs by HPLC ESI MS/MS and observed that sublethal irradiation by 500 cGy upregulates the level of this bioactive lipid in irradiated BM and lungs (Fig. 2B, 2C). This data suggest that C1P may play a novel (and until now, underappreciated) role in chemoattracting stem cells circulating in PB to damaged tissues.
C1P Chemoattracts Ex Vivo-Expanded MSCs and HUVECs
Since MSCs and other types of stem cells are used in regenerative medicine [1–14], we performed dose-response chemotactic studies on ex vivo-expanded MSCs and purified HUVECs with C1P and observed that it strongly chemoattracts these cells (Fig. 2C, 2D). As shown in Figure 2C, while the optimal chemotactic dose for MSCs was ∼10 μM, the responsiveness of HUVECs was highest at a 100-fold lower concentration of this bioactive lipid (0.1 μM, Fig. 2D, 2E). Since the receptor(s) for C1P have not yet been identified [29, 32], we could not determine whether these differences were related to differences in receptor density on the target cells. Furthermore, using varying concentrations of C1P in the upper and lower Transwell chambers, we excluded the possibility that the C1P effect is not mediated by directional gradient migration, but is instead due to chemokinesis  of MSCs and HUVECs (data not shown).
In parallel studies, we next compared the chemotactic activity of C1P to that of other bioactive lipids (S1P; lysophosphatidylic acid, LPA; and lysophosphatidylcholine, LPC) against murine MSCs (Fig. 3A), HUVECs (Fig. 3B) as well as human MSCs (Fig. 3C). All these bioactive lipids were used at doses corresponding to their PB plasma levels . In parallel, we also evaluated the responsiveness of these cells to both low (10 ng/ml) and high doses of the α-chemokine SDF-1 (300 ng/ml), which has been demonstrated to be a potent chemoattractant for both MSCs and HUVECs [41, 42].
Figure 3 shows that if used at physiological doses, C1P (including both C16 and C18 analogs) is a very potent chemoattractant for murine MSCs, HUVECs, and human MSCs. Of note, the chemotactic activity of a low-dose of SDF-1 (10 ng/ml) was not effective in our assays to chemoattract these cells. This is an important observation, because the SDF-1 physiological concentration in PB plasma has been reported to be <3 ng/ml . Therefore, our data indicate that C1P, at the physiological level found in PB , is a more potent chemoattractant for these cells.
C1P Increases Adhesion of MSCs
We next became interested in whether C1P increases adhesion of MSCs. We used the colony-forming unit of fibroblasts (CFU-F) assay to evaluate the effect of C1P on adhesion of MSCs present among the populations of murine BMMNCs. Accordingly, BMMNCs were stimulated or not (control) by C1P and plated at a low concentration in Petri dishes. The number of CFU-F colonies established was evaluated after 14 days as described . Figure 3D shows that C1P increases adhesion of clonogenic CFU-F, which are true clonogenic progenitor cells for MSCs, in a dose-dependent manner. This effect was more pronounced for CFU-F with high proliferation potential (Fig. 3E). At the same time, we observed that stimulation of MSCs and HUVECs by C1P does not affect expression of adhesion molecules, including VLA4 or VCAM-1 (data not shown).
C1P Induces Signaling in Ex Vivo-Expanded MSCs and HUVECs
While the receptor(s) for C1P have been not identified, we evaluated whether this bioactive lipid activates signaling in MSCs (Fig. 4A, 4B) and HUVECs (Fig. 4C). We observed that C1P strongly activates phosphorylation of MAPKp42/44 in MSCs (Fig. 5A) and that this signaling is PTx-sensitive (Fig. 4B), which suggests that it is mediated by receptor/s coupled to a Gαi signaling protein. At the same time, we observed stronger phosphorylation of MAPKp42/44 and AKT stimulated by C1P in HUVECs (Fig. 4C) than in MSCs.
C1P Does Not Affect Proliferation of MSCs, But Increases SDF-1, Cox-2, and MMPs Expression
We next became interested in the biological effects of C1P that could be related to organ tissue repair. Our direct proliferation studies, however, revealed that C1P does not significantly affect MSC proliferation (Fig. 5A). At the same time, however, we observed that stimulation of MSCs by C1P increases expression of SDF-1 in these cells at mRNA and protein level (Fig. 5B). Furthermore, increase in level of SDF1 in cell extracts as well as in conditioned media from MSCs stimulated by C1P was subsequently confirmed by ELISA assay (Supporting Information Fig. S1A, S1B). Similarly, C1P also enhanced in MSCs expression of cyclooxygenase-2 (Cox-2) both at mRNA and protein level (Fig. 5C). Of note, both SDF-1 and the Cox-2 product prostaglandin E2 (PGE2)  play an important role in chemoattracting circulating stem cells to the damaged organs, as previously reported [45–48].
We also tested whether C1P enhances expression of MMP-2 and MMP-9, two proteolytic enzymes that are involved in cell trafficking . As shown in Figure 5D, 5E, C1P (the C18 analog) stimulates within 3 hours expression of mRNA for MMP2 and MMP9 in MSCs that were starved before stimulation for 24 hours. Furthermore, we used zymography assay to demonstrate increase in MMP-9 protein in cell supernatants stimulated by C1P (Fig. 5E). Expression of MMP-2 (Fig. 5D) and MMP-9 (Fig. 5E) was upregulated after stimulation by LPA and LPC, but, to our surprise, neither by S1P nor by SDF-1. Both these factors required longer stimulation time (6 hours) to upregulate MMPs in starved MSCs (data not shown).
C1P Does Not Affect Proliferation of HUVECs, But Triggers Capillary-Like Structure Formation in Three-Dimensional-Matrigel Assay and Induces Angiogenesis In Vivo in Matrigel Implants
In parallel experiments, we observed that, as in MSCs (Fig. 5A), C1P does not affect proliferation of HUVECs (Fig. 6A). However, Figure 6B, 6C demonstrate that C1P, in our hands, is a potent inducer of tube formation by HUVECs in vitro in Matrigel assay. In particular, we observed that the C18 isoform, known to be released by cells in response to tissue damage and stress , is the most potent inducer of tube formation. Of note, this tube-formation effect was even stronger than an optimal dose of potent proangiopoietic factor that is FGF-2.
To support this better, we performed in vivo Matrigel implants vascularization assay and noticed again a potent vascularization effect of C1P that was comparable to FGF-2 (Fig. 6D). Finally, we also observed that C1P increases similarly as for MSCs adhesion of HUVECs to plastic dishes (Fig. 6E). Thus, C1P seems to be an underappreciated regulator of angiogenesis.
Mounting evidence indicates that the bioactive sphingolipids, S1P and C1P, play an important role in several biological processes, including cell trafficking [27, 29]. As reported, S1P acts as an intracellular second messenger molecule and is secreted from cells as a potent antiapoptotic factor that stimulates angiogenesis and chemoattracts several types of cells [49, 50]. Murine embryos that lack sphingosine kinase-1 and −2 and thus lack S1P die at early stages of embryonic development [51–53].
Unlike S1P, the biological effects of C1P are still largely unstudied. C1P is a product of the phosphorylation of ceramide, which is a known mediator that promotes cell cycle arrest  and induces apoptosis . By contrast, C1P can act as an intracellular second messenger to promote cell survival [30, 56], or, if released by the cells, can act as an extracellular chemoattractant for monocytes  and hematopoietic stem progenitor cells (HSPCs), as recently demonstrated . Dermal microendothelial cells isolated from ceramide kinase-deficient mice showed profound in vitro defects in angiogenesis , which suggested that C1P is endowed with proangiopoietic properties.
In this report, we demonstrate for the first time that C1P is a strong chemoattractant for several types of BM-residing nonhematopoietic stem cells, including MSCs, EPCs, and VSELs. It may thus play a role in mobilization/egress of these cells from BM into PB. Furthermore, since C1P, as we show here, is released from damaged leaky cells (e.g., heart tissue during acute hypoxia), it may create a chemotactic gradient in damaged organs for stem cells circulating in PB, and such cells could be involved in tissue/organ repair. The same mechanism, however, may also be involved in an unwanted recruitment of circulating EPCs and MSCs to expanding tumor tissue .
Furthermore, we have also shown that C1P upregulates expression of SDF-1 and Cox-2 in stromal cells. Thus, the chemotactic effect of C1P against circulating stem cells may be additionally potentiated by SDF-1, a known chemoattractant for stem cells  and PGE2 (a Cox2 product), which enhances responsiveness of stem cells to an SDF-1 gradient . Thus, we present for the first time evidence that C1P, like other chemoattractants released in damaged tissue, such as SDF-1, VEGF, HGF, or S1P, could be directly or indirectly (by upregulating SDF-1 and Cox-2) involved in recruitment of BM-derived stem cells circulating in PB.
It has been demonstrated in several animal models [58–61] as well as in clinical settings [62–66] that stem cells become mobilized into PB in response to tissue/organ injury. However, the potential involvement of these circulating stem cells in organ tissue regeneration is still not fully understood . While circulating EPCs could supply new vessels in damaged tissues, and VSELs could contribute to replacement of damaged cells, other stem cell types, such as MSCs or HSPCs, could play an important role as a source of several soluble paracrine factors (e.g., growth factors, cytokines, or chemokines) that may play a role in inhibiting apoptosis of damaged cells [68, 69]. Conversely, evidence has accumulated that an important role in stem cell therapies for damaged organs may be played by membrane-derived microvesicles or exosomes, which promote survival of damaged cells and stimulate angiogenesis by delivering to the damaged tissues several proteins, mRNAs, and miRNAs that stimulate angiogenesis [69, 70] and inhibit apoptosis [71, 72]. Microvesicles are highly enriched for S1P , and further studies are needed to measure the content of C1P in these bioactive, spherical membrane fragments.
Our experiments performed in serum-free media indicate that C1P, in contrast to macrophages  and myoblastic cells , does not affect proliferation of MSCs and HUVECs. Thus, the responsiveness of cells to C1P seems to be cell specific. In further support of this observation, neither human nor murine HSPCs respond to C1P by increasing proliferation, as we have already reported . Therefore, in our adhesion assay experiments, the observed increase in the appearance of large CFU-F colonies after 14 days in in vitro culture (Fig. 3E) is best explained by preferential adhesion of CFU-F endowed with high proliferative potential. We also observed that murine and human MSCs and HUVECs, like monocytes , myoblasts , and HSPCs, respond robustly to C1P in chemotactic assays, which supports an important role for C1P in cell trafficking.
In this report, we also demonstrate that C1P enhances adhesiveness of MSCs and HUVECs. Thus, in a similar way as reported for HSPCs , C1P stimulates adhesion of stromal cells. Moreover, like HSPCs, the enhanced adhesion of MSCs and HUVECs did not depend on an increase in expression of VLA4 or VCAM-1 on MSCs and HUVECs .
As mentioned above, receptors for C1P have not yet been identified [29, 32]. However, our signaling studies revealed that stimulation of MSCs and HUVECs by C1P induces phosphorylation of MAPKp42/44 and AKT. The fact that this signaling is PTx-sensitive indicates involvement of Gαi protein-coupled receptors in this phenomenon. The sensitivity of C1P to PTx signaling has been previously demonstrated in monocytes  and in murine and human HSPCs .
An important step in regeneration of damaged organ tissues is promotion of neovascularization or angiogenesis [75, 76]. In this article, for the first time we demonstrate that C1P is a very strong proangiopoietic factor by being a chemoattractant for endothelial cells. Furthermore, in in vitro functional assays, we confirmed that C1P directly promotes tube formation and our data in vivo in murine model demonstrate that C1P promotes also vascularization of Matrigel implants. This observation supports a previous report where it was shown that skin microendothelial cells isolated from CerK-deficient mice show defects in angiogenesis in vitro . Thus, C1P, like S1P , plays an important novel role as a proangiopoietic factor.
Our data demonstrate for the first time that C1P is a potent bioactive lipid released from damaged cells and could play an important and novel role in regulating repair of other damaged organs and tissues. Thus, it may be involved both in recruitment of circulating stem cells to regenerating tissues as well as in unwanted recruitment of EPCs and MSCs to an expanding tumor. This possibility, however, requires further study.
In conclusion, as demonstrated in Figure 7, cells in damaged organs release C1P, which together with other factors not shown in this scheme (e.g., SDF-1, S1P, VEGF, and HGF) creates a chemotactic gradient for MSCs, EPCs, and VSELs circulating in PB. Based on this mechanism, we propose that modification of C1P signaling in damaged tissues will help in developing more optimal therapeutic strategies in regenerative medicine. Conversely, inhibition of a C1P gradient may prevent unwanted recruitment of stem cells circulating in PB into expanding tumor tissue. Therefore, it is important to clone the C1P receptor(s) and then to develop small-molecule C1P receptor agonists and antagonists that could be used in the clinical setting.
This work was supported by NIH Grant 2R01 DK074720, the Stella and Henry Endowment, the European Union structural funds (Innovative Economy Operational Program POIG.01.01.02-00-109/09-00), and Grant Maestro 2011/02/A/NZ4/00035.