Author contributions: M G-O., L.G., and C.S.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; C.B., M.M., L.B., G.T., and C.F.G-P.: collection and/or assembly of data and final approval of manuscript; A.B.: financial support, data analysis and interpretation, and final approval of manuscript; L.C.: conception and design, financial support, data analysis and interpretation, manuscript writing, and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
first published online in STEM CELLSEXPRESS March 26, 2013.
Adipose tissue (AT) has become accepted as a source of multipotent progenitor cells, the adipose stromal cells (ASCs). In this regard, considerable work has been performed to harvest and characterize this cell population as well as to investigate the mechanisms by which transplanted ASCs mediate tissue regeneration. In contrast the endogenous release of native ASCs by AT has been poorly investigated. In this work, we show that native ASCs egress from murine AT. Indeed, we demonstrated that the release of native ASCs from AT can be evidenced both using an ex vivo perfusion model that we set up and in vivo. Such a mobilization process is controlled by CXCR4 chemokine receptor. In addition, once mobilized from AT, circulating ASCs were found to navigate through lymph fluid and to home into lymph nodes (LN). Therefore, we demonstrated that, during the LN activation, the fat depot encapsulating the activated LN releases native ASCs, which in turn invade the activated LN. Moreover, the ASCs invading the LN were visualized in close physical interaction with podoplanin and ER-TR7 positive structures corresponding to the stromal network composing the LN. This dynamic was impaired with CXCR4 neutralizing antibody. Taken together, these data provide robust evidences that native ASCs can traffic in vivo and that AT might provide stromal cells to activated LNs. STEM Cells2013;31:1309–1320
Adipose tissue (AT) regulates and coordinates energy homeostasis and performs many key endocrine functions [1–3]. It exhibits this fascinating ability to expand or regress in response to alterations in energy balance, which is rather unique among organs in adults. AT is composed of adipocytes and various cell types among which are a population of multipotent progenitors, termed adipose stromal cells (ASCs) . ASCs display similar but not identical features with their bone marrow (BM) counterparts, mesenchymal stromal cells (MSCs) [5–9]. Like MSCs, ASCs are endowed with multilineage mesodermal differentiation potentials and exhibit extensive paracrine and immune-modulatory properties [10–15]. Most current research is directed at investigating their putative therapeutic properties after their transplantation in damaged and/or degenerating tissues . While some reports described their homing into injured tissues after their injection into the circulation, the mobilization of the endogenous cells from AT is poorly documented, although some in vitro reports suggest that AT may release cells [17, 18]. Previously, we have reported that human native ASCs (i.e., freshly harvested) [11, 13] express functional chemokine CXCR4 receptor, the activation of which by CXCL12 (or stromal derived factor or SDF-1) induces their migration in vitro . In BM, the disruption of the CXCL12/CXCR4 axis induces the rapid mobilization of hematopoietic stem/progenitor cells (HSPCs) from the BM to the circulation . Therefore, we postulated that the CXCL12/CXCR4 axis might regulate the migration and/or retention of ASCs in AT.
In this study, we aimed to examine the ability of AT to release ASCs in vivo. We show for the first time that native ASCs egress from AT both ex vivo and in vivo, and that the CXCL12/CXCR4 axis might regulate such an event. Moreover, we identified that circulating native ASCs might navigate through lymphatics and invade lymph nodes (LN) and might be recruited by LN stroma during the establishment of an immune response.
MATERIALS AND METHODS
Animals and Treatment Procedures
Ten-week old male C57BL/6 mice (Janvier, Le Genest-Saint-Isle, France, http://www.janvier-europe.com) were kept under controlled light (12-hour light/dark cycles) and temperature (22°C–24°C) conditions. All studies were carried out under the INSERM Animal Care Facility guidelines and local ethical approval from Toulouse Rangueil Hospital. A group of mice was treated with a CXCR4 antagonist AMD3100 (5 mg/kg, i.p., Cayman Chemical, Montigny-le-Bretonneux, France, https://www.caymanchem.com) or saline solution (NaCl 0.9%). After 90 or 180 minutes, animals were sacrificed and blood, BM, spleen, inguinal and mesenteric LNs (ILN and MLN), and subcutaneous (Sc), mesenteric (Mes), and perigonadic (PG) ATs were dissected.
Isolation of the Stroma Vascular Fraction from AT
Sc, Mes, and PG ATs were minced and digested with collagenase (type II, Sigma-Aldrich, Lyon, France, http://www.sigmaaldrich. com; 250 U/mL in phosphate-buffered saline [PBS], 2% bovine serum albumin [BSA], pH = 7.4) for 30–45 minutes at 37°C, under constant agitation. After centrifugation (300g, 10 minutes, room temperature [RT]), the floating mature adipocytes were removed and the pellet containing the stroma vascular fraction (SVF) was suspended in erythrocyte lysis buffer (ELB; 155 mmol/L NH4Cl; 5.7 mmol/L K2HPO4; 0.1 mmol/L EDTA, pH = 7.3) for 10 minutes at RT. After filtration through a 37-μm sieve, cells were resuspended in Flow Cytometry (FACS) buffer (PBS; 2 mmol/L EDTA; 0.5% BSA) for further flow cytometry analyses or kept at −20°C for further RNA isolation.
Isolation of BM, Blood, Spleen, and LN Mononuclear Cells
BM cells were flushed from femurs with RPMI-1640 (Invitrogen, Cergy Pontoise, France, http://www.invitrogen.com). After centrifugation (300g, 10 minutes, 4°C), red blood cell lysis was performed with ELB (10 minutes, RT), and BM-mononuclear cells (MNCs) were filtered through 37 μm sieves. BM-MNCs were resuspended in FACS buffer or in an appropriate buffer for further RNA isolation (buffer RLT from Qiagen, Courtaboeuf, France, http://www1.qiagen.com).
Blood was collected into ELB (10 minutes, RT). Samples were then centrifuged (300g, 10 minutes, 4°C), and MNCs were resuspended in FACS buffer.
Spleen, ILN, popliteal (PLN), and MLN were dissected and digested with liberase (0.125 mg/mL in RPMI-1640 medium, 0.2 mg/mL DNase I, Roche Diagnostics, Mannheim, Germany, http://www.roche-applied-science.com) for 30 minutes at 37°C under constant agitation. Splenic and LN derived-MNCs were isolated by crushing the spleen or the LN over a 37-μm sieve. After centrifugation (300g, 10 minutes, RT), red blood cells were lysed in ELB for 10 minutes (RT) and resuspended in FACS buffer.
Collection of Lymph in the Thoracic Duct and Isolation of Lymph MNCs
The lymph was collected in the thoracic duct as previously described by Ionac et al. , in animals treated with AMD3100 or saline 90 minutes before surgery. In order to easily identify the lymphatic vessels, 300 μL of milk cream (15% fat) was administered intragastrically 30 minutes prior to lymph collection. Briefly, mice were anesthetized with a mixture of isofluran and oxygen (3–3.5 vol.% for induction and 2–2.5 vol.% for maintenance), 5 U of heparin was injected subcutaneously and body temperature was monitored and maintained between 36.5°C and 37.5°C. Thereafter, the thoracic duct was localized and catheterized by using silicon tubing (0.64 mm outer diameter and 0.3 mm inner diameter) previously flushed with sterile heparinized saline solution. After cannulation, the lymph was collected for 30 minutes (final volume collected of 20–30 μL), resuspended in FACS buffer for flow cytometry analyses.
AT Cell sorting, Adipogenic and Osteogenic Differentiation
Murine crude AT-derived SVF was depleted in CD45+ cells by use of mouse CD45 microbeads and LD columns (MACS Cell Separation, Miltenyi Biotec SAS, Paris, France, http://www.miltenyibiotec.com) according to the manufacturer's instructions (CD45− cell fraction purity was 99.5% ± 0.2%, n = 8). CD45− cell population was centrifuged (300g, 4°C), and Sca-1+ cells were selected using mouse anti-Sca-1 microbeads and LS columns (MACS) according to the manufacturer's instructions (CD45−/Sca-1+ cell population purity was 98.9% ± 0.2%, n = 8).
Crude SVF cells or magnetically sorted CD45+ or CD45−/Sca1+ cells were plated at a density of 80,000 cells per cm2 in ECBM (endothelial basal cell medium) medium supplemented with 10% fetal calf serum (FCS). After 24 hours at 37°C and 5% CO2, cells were induced to undergo adipogenic or osteogenic differentiation. Adipogenic differentiation was induced with ECBM medium supplemented with 2% FCS, 66 nmol/L insulin, 1 nmol/L triiodothyronine, 100 nmol/L cortisol. After 10 days of culture, cells were rinsed with PBS and either fixed with 10% formaldehyde for further lipid accumulation staining or lysed (V/V) with PBS supplemented with 0.2% Tween 20 for further triglyceride content determination. Lipid droplets were stained with BODIPY 493/503 (10 μg/mL, Molecular Probes, Invitrogen, Villebon sur Yvette, France), and nuclei were stained with Hoescht 33342. Fluorescence analysis was performed using a Nikon inverted Eclipse TE300 microscope. Triglyceride content was determined by use of Triglycerides Enzymatique PAP 150 kit (bioMérieux, Marcy l'Etoile, France).
Osteogenic differentiation was induced by culturing cells in α-MEM medium supplemented with 10% FCS, 250 μM ascorbate-2-phosphate,10 mM β-glycero-phosphate, and 2.5 μM all-trans retinoic acid for 21 days . Extracellular matrix mineralization was evaluated after 21 days of culture by using Alizarin Red staining .
Ex Vivo Perfusion of Mouse Mesenteric Adipose Fat Pad
Ten-week-old male C57BL/6 mice were euthanized, bled, and the superior mesenteric artery was cannulated at its junction with the abdominal aorta. The whole mesenteric fat pad was dissected from the intestine, placed in an organ chamber, and perfused using a peristaltic pump (Gilson, Middleton) at a constant flow (0.25–0.3 mL/minute) with oxygenated (95% O2/5% CO2) Krebs–Henseleit (KH) solution (115 mM NaCl, 4.6 mM KCl, 2.5 mM CaCl2, 25 mM NaHCO3, 1.2 mM KH2PO4, 1.2 mM MgSO4, 0.01 mM EDTA, 11.1 mM glucose, 10 mM HEPES, pH 7.4). After an equilibration period of 90 minutes, the mesenteric fat pad was perfused with increasing concentrations of the CXCR4 antagonist,AMD3100 (10−8 M and 10−6 M, Cayman Chemicals), or KH solution alone, as control. Perfusates were collected every 30 minutes and centrifuged (300g, 10 minutes, 4°C). Cells were resuspended in FACS buffer for flow cytometry analysis.
In Vitro AMD3100 Treatment of Adipose-Derived SVF
Freshly harvested SVF cells from Sc, Mes, and PG ATs were incubated in ECBM/0.1% BSA (37°C, 5% CO2) in the presence or absence of AMD3100 (10−6 M) for 180 minutes. After centrifugation (300g, 10 minutes, 4°C), cells were resuspended in FACS buffer for further flow cytometry analysis.
Native ASCs Transplantation
Murine crude AT-derived SVF from green fluorescent protein (GFP)-expressing male mice (10-week-old, C57BL/6 TgN(act-EGFP)OsbC15-001-FJ001, kindly provided by Prof. M. Okabe ) was depleted in CD45 cells by use of mouse CD45 microbeads and LD columns (MACS Cell Separation, Miltenyi Biotec SAS) as previously described. Immediately after isolation, native ASCs were injected in the right hind footpad (300,000 cells per footpad) of C57BL/6 mice. Twelve hours later, mice were sacrificed and PLN were dissected and fixed in 4% paraformaldehyde (PFA) for 24 hours. Thereafter, samples were included in 2% agarose and cut into 300 μm slices for immunohistochemical analyses.
Ten-week-old male C57BL/6 mice (Janvier) were immunized in the right hind footpad with 50 μg of ovalbumin (OVA, Sigma-Aldrich, Saint-Quentin Fallavier, France) in incomplete Freund's adjuvant (IFA). Left hind footpads were used as controls. A group of mice received also two subcutaneous injections (4 and 7 days after immunization) of a total of 50 μg mouse anti human CXCR4 Ab (12G5, R&D Systems, Abigndon, U.K., http://www.rndsystems.com) in the immunized hind limb or matched control IgG Ab in the control hind limb. Nine days after immunization, mice were euthanized, and the ScAT, the BM and the ILN and PLN from both the right and left sides were dissected and managed as previously described for flow cytometry analyses.
Flow Cytometry Analysis
Isolated cells from ATs, blood, LNs, lymph, and spleen or mesenteric perfusates were incubated (25 minutes, 4°C, dark) with Fluorescein isothiocyanate (FITC)-CD105, FITC-CD34, Phycoerythrin (PE)-CD31, PE-CD3, PE-CD140b, PE-CD117, Peridinin chlorphyll protein (PerCP)-CD34, Allophycocyanin (APC)-Cy 7-CD29 (Biolegend, Saint Quentin Yvelines, France), PE-CD90, APC-CXCR4, APC-CD31, CD45-V450, Sca-1-V500 (BD Biosciences, Le Pont de Claix, France, http://www.bdbiosciences.com), live and dead-V450 (Invitrogen) antibodies or the appropriate isotype controls (BD Biosciences). For intracellular CXCR4 staining, previously stained cells were fixed (20 minutes, 4°C, dark) with Cytofix/Cytoperm buffer (BD Biosciences). After centrifugation (400g, 10 minutes, RT), cells were resuspended into perm/wash buffer (10 minutes, 4°C, dark) (BD Biosciences). Thereafter, cells were stained with an APC-CXCR4 antibody or the appropriate isotype control (25 minutes, 4°C, dark) in perm/wash buffer. After washing, the labeled cells were quantified on a CANTO II flow cytometer and analyzed using FACSDiva software (BD Biosciences).
Cell Number Quantification
The number of cells was calculated by using DNA concentrations determined by fluorimetric assays . Cells isolated from the ScAT, MesAT, PGAT, LNs, and spleen were resuspended in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH = 7.5)/0.1% Triton X-100, sonicated. Quant-iT PicoGreen dsDNA reagent (Invitrogen) was added to samples and DNA standard, and fluorescence was detected on a Fluoroskan Ascent FL fluorometer (Labsystems, Cergy Pontoise, France) according to the manufacturer's instructions.
Intact mice AT pieces (2–3 mm3) were fixed in neutral buffered 4% (w/v) PFA (1 hour, RT), blocked in PBS/3% BSA (30 minutes, RT), and incubated in PBS/0.1% BSA/0.2% Triton/0.05% Tween (Immunofluorescence (IF) buffer) with primary antibodies overnight at 4°C (Supporting Information data 3). After washing steps in IF buffer, AT samples were incubated for 1 hour with secondary antibodies or with antibodies directed against lectin, Sca1-, and CD45. Nuclei were stained with Hoescht 33342. Fluorescence analysis was performed using a Zeiss LSM510 NLO confocal microscope.
PLN slices were blocked in PBS/2% horse serum/0.2% Triton (2 hours, RT) and incubated in the same buffer containing the primary antibodies anti-CD3 (rat, 1:50, BD Pharmingen, Le Pont de Claix, France), anti-B220 (rat, 1:50, BD Pharmingen), anti-ER-TR7 (rat, 1:50, Abcam, Paris, France), and anti-podoplanin (hamster, 1:50, Abcam) (overnight, 4°C). After washing in PBS/0.2% Tween, samples were incubated with secondary antibodies (anti-rat or anti hamster [Alexa fluor 546, 1:200, Invitrogen] in PBS/2% horse serum/0.2% Triton (1 hour, RT, light protected). Nuclei were stained with Hoescht 33342 (1:1,000, Invitrogen) Fluorescence analysis was performed using a Zeiss LSM510 NLO confocal microscope.
Total RNA was extracted from murine AT-SVF and BM-MNCs isolated cells using the RNeasy kit (Qiagen). RNA concentrations were determined by fluorimetric assays (Ribogreen, Invitrogen). Reverse transcription was performed over 0.5 μg of RNA by using Superscript II, random hexamers, and dNTPs according to the manufacturer's instructions (Invitrogen). The primer for CXCR4 was provided by Applied Biosystems (Assays on demand: Mm01292123-m1, Courtaboeuf, France, http://www. appliedbiosystems.com). Amplification reaction was carried out with 15 ng complementary DNA samples in 96-well plates (Applied Biosystems) in a GeneAmp 7500 sequence detection system. The polymerase chain reaction (PCR) mixture contained 5 μL of Taqman primers (5× prediluted in water) and 10 μL of 2× Taqman PCR master Mix (Applied Biosystems). All reactions were performed at 50°C (2 minutes), 95 °C (10 minutes), then 40 cycles of 95°C (15 seconds) and 60°C (1 minute). The results were analyzed by use of GeneAmp 7500, and all values were normalized to 18S rRNA levels.
Protein Extraction and Western Blot Analysis
Protein extraction was performed in murine PLNs from mice previously immunized with OVA and sacrificed 3 or 6 days after immunization. Tissues were homogenized in ice-cold buffer containing 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% Nonidet P40, 1% sodium deoxycholate, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM Na3VO4 and supplemented with complete protease inhibitor tablets (Roche Diagnostics, Basel, Switzerland). Tubes containing homogenates were frozen at −80°C and thawed at 37°C three consecutive times then centrifuged (10 minutes, 14,000g, 4°C). Samples were purified and concentrated by using centrifugal filter units of 3 and 100 kDa (Millipore, Tullagreen, Ireland, http://www.millipore.com) to obtain proteins between 3 and 100 kDa. Equivalent amounts of proteins (15 μg) were loaded in Laemmli buffer (60 mM Tris-HCl [pH 6.8], 10% glycerol, 2% SDS, 5% β-mercaptoethanol, 0.01% bromophenol blue) and size-separated in 12% SDS-polyacrylamide gel electrophoresis. Proteins were transferred to PolyVinyl DiFluoride (PVDF) membranes (Amersham, Velizy-Villacoublay, France, http://www.amersham.com) using a transblot apparatus (Bio-Rad, Marnes-la-Coquette, France, http://www.bio-rad. com). For immunoblotting, membranes were blocked with 5% nonfat dried milk in Tween-PBS for 1 hour. A primary antibody against CXCL12 (Cell Signaling Technology, Boston, MA, http://www.cellsignal.com, 1/1,000 final dilution) was applied overnight at 4°C. After washing, an appropriate secondary antibody (anti-rabbit IgG peroxidase conjugated) was applied for 1 hour at a dilution of 1/2,000. Blots were washed, incubated in commercial enhanced chemiluminescence reagents (Bio-Rad) and analyzed with Chemidoc XRS (Bio-Rad). To prove equal loadings of samples, blots were reincubated with β-actin antibody (Sigma-Aldrich). Values for CXCL12 were normalized with β-actin to account for variations in gel loading.
CXCL12 ELISA Assay
ILNs (D9) originating from animals vaccinated or not with OVA-IFA were isolated and suspended in Assay Diluent according to the manufacturer's instructions (CXCL12 ELISA kit (R&D Systems, Lille, France). CXCL12 expression levels were quantified using a CXCL12 ELISA kit according to the manufacturer's guide.
Data are expressed as mean ± SEM of n independent experiments. Statistical analyses were performed with GraphPad Prism Software (San Diego, CA). Comparisons among different groups were analyzed by ANOVA, then post hoc Dunnet multiple comparison tests, Mann–Whitney U test for two groups or Wilcoxon matched paired t test for two groups when paired observations. Differences were considered statistically significant at p < .05.
Native Immunophenotype, In Situ Localization, and Morphology of Murine ASCs
We previously reported that CXCL12 was able to trigger human native ASCs migration in vitro, we hypothesized that like BM, the CXCL12/CXCR4 axis might control in vivo the retention and/or migration of native ASCs in AT. Therefore, we first aimed to characterize native murine ASC immunophenotype. To do so, SVF originating from Sc, Mes, and PG AT were harvested and analyzed by flow cytometry using CD45, Sca-1, and CD31 antibodies. Regardless of the fat depot, SVF was composed of leukocytes (CD45+ cells) and Sca-1+/CD45− cell population (Fig. 1A; Supporting Information Fig. S1A–S1C), the latter included endothelial cells (Sca-1+/CD45−/CD31+ cells) and Sca-1+/CD45−/CD31− cells (Fig. 1A) that represented 48.6% ± 2.8% of the ScAT-SVF and roughly 30% of both the PGAT and the MesAT-SVF (Supporting Information Fig. S1A–S1C). Based on our previous work on human native ASCs [13, 25, 26] and in accordance with Rodeheffer et al. , we identified native murine ASCs as the Sca-1+/CD45−/CD31− cell population. The further immunophenotyping of Sca-1+/CD45−/CD31− cells showed that murine native ASCs were positive for CD29, CD34, CD90 (Fig. 1B) and were negative for CD105 and CD117 (c-kit) marker distinguishing them from HSPCs .
We next investigated their spatial organization in intact pieces of ScAT, PGAT, and MesAT by the use of Sca-1 antibodies and were distinguished from leukocytes and endothelial cells by the use of CD45 and lectin antibodies, respectively. Murine native ASCs presented a branched morphology with several lateral protrusions and were detected both in the stroma and close to vessels whatever the fat depot (Fig. 1C). The adipogenic potential of the SVF subpopulations was tested. Conversely to leukocytes where no lipid formation was observed (Supporting Information Fig. S1E), both lipid staining and triglyceride content after 10 days of adipogenic culture conditions, revealed a higher lipid accumulation in the Sca-1+/CD45− cell fraction as compared to the crude SVF (Fig. 1D, 1E). Additionally, osteogenic potential of the Sca-1+/CD45−/CD31− cells was evaluated and compared to the crude SVF. As for adipogenesis, osteogenesis was found dramatically enriched in the Sca-1+/CD45− cell fraction as compared to the crude SVF (Fig. 1F, 1G). To note, the number of Sca-1+/CD45−/CD31− cells was significantly higher in the ScAT compared to both the PGAT and the MesAT (Supporting Information Fig. S1D).
Finally the expression of CXCR4 receptor on native murine ASCs was studied. RT-qPCR studies revealed that AT-SVF expressed similar levels of CXCR4 mRNAs regardless of the fat depot studied (Fig. 2A), the expression of which was about twice higher in BM-MNCs used as control (Fig. 2A). The further analysis of CXCR4 expression in both intact pieces of AT (immunohistochemistry approach) or in freshly harvested AT-SVF (flow cytometry analysis) demonstrated that as for human native ASCs, CXCR4 was expressed at the cell surface of murine ASCs (Fig. 2B, 2C respectively).
ASCs Are Released After Ex Vivo Perfusion of AT with CXCR4 Antagonist
To investigate whether AT was able to release native ASCs, we set up a model consisting in the ex vivo microperfusion of an isolated intact fat pad with a saline solution (KH buffer) in the presence or absence of a CXCR4 antagonist, that is, AMD3100. MesAT was chosen because it is irrigated both by the blood and lymphatic systems and it is anatomically connected to the abdominal aorta by a single artery (the superior mesenteric artery) allowing the perfusion of the complete fat depot after its catheterization. Therefore, ASC release was assessed by quantifying by flow cytometry the number of Sca-1+/CD45−/CD31− cells released by MesAT in the perfusate (Fig. 3A). The ex vivo perfusion of MesAT with AMD3100 significantly enhanced the number of native ASCs in the perfusate compared to the control (Fig. 3B), demonstrating that native ASCs egress from AT. In contrast, the number of CD45+ cells released from MesAT was not modified by AMD3100 perfusion (Fig. 3C). All together these results indicate that ASCs are released from AT ex vivo.
In Vivo, the Acute Administration of AMD3100 Induces ASC Decrease in ScAT and MesAT
Because we evidenced ex vivo the release of ASCs from AT, we asked whether AT was able to release native ASCs in vivo. To do so, the ASC content of ScAT, MesAT, and PGAT was determined by flow cytometry in mice, 90 minutes and 180 minutes after AMD3100 injection. In accordance with the ex vivo data, the acute administration of AMD3100 induced a time-dependent significant decrease in ASC content in the ScAT (decrease of 32.9% ± 8.1% corresponding to 106 cells per depot after 180 minutes) and the MesAT (decrease of 27.4% ± 8.1% corresponding to 2 × 105 cells per depot after 180 minutes) (Fig. 4A). In contrast, PGAT was not responsive to AMD3100 administration (Fig. 4A). Moreover and regardless of the tissue, AMD3100 treatment did not modify the number of CD45+ cells which confirmed our data obtained in the ex vivo perfusion model (Fig. 4B). Very importantly no fat depot weight changes were observed after AMD3100 administration (Fig. 4C). To rule out that the reduction in ASC number observed in AT after AMD3100 administration was due to a cell death increase, cell viability was evaluated by flow cytometry. After AMD3100 exposure, cell survival was modified neither in vivo (Fig. 4D) nor in vitro (Supporting Information Fig. S2A). To exclude that the decrease in ASC content following AMD3100 treatment was attributable to a modification of cell surface marker expression, rather to cell release, SVF from ScAT, MesAT, and PGAT was isolated from control mice and was exposed in vitro to AMD3100 for 180 minutes. In that context, no ASC immunophenotype modifications were observed after AMD3100 exposure in vitro (Fig. 4E; Supporting Information Fig. S2B). Taken together these data demonstrate that in vivo, AT is able to release native ASCs.
After AMD3100 Treatment, Native ASCs Traffic in the Lymphatic System and Accumulate in LNs
Next, we examined the route by which native ASCs traffic after their release from AT in vivo. Therefore, the presence of circulating Sca-1+/CD45−/CD31− cells was determined in blood by flow cytometry 30, 60, 90, and 180 minutes after the administration of AMD3100. As previously reported, the number of circulating CD45+ cells increased 90 minutes after AMD 3100 administration (threefold increase, from 1.3 ± 0.2 to 3.6 ± 0.3 × 106 cells per mL). Surprisingly no Sca-1+/CD45−/CD31− cells could be detected in the blood (Supporting Information Fig. S3A–S3C). Consequently, we surmised that native ASCs might egress AT via the lymphatic system and, therefore, we collected lymph fluid from murine thoracic duct (TD) after AMD3100 administration. As expected, lymph fluid collected from murine TD exhibited a different profile compared to the blood and contained up to 75% of CD3 lymphocytes (Fig. 5A). As we hypothesized, ASCs were detected in the lymph fluid (Fig. 5B) and their percentage was strongly increased 90 minutes after the AMD3100 administration (Fig. 5B, 4.5-fold increase, n = 3, p < .05). We next surveyed murine secondary lymphoid organs such as ILN and MLN for the presence of ASCs. Ninety minutes after AMD3100 treatment, an increase in the number of Sca-1+/CD45−/CD31− cells within ILN and MLN was observed (Fig. 5C). Because ASCs exhibit colony-forming unit fibroblasts (CFU-Fs) activity, we cultured ILN-derived cells to assay for the frequency of CFU-Fs in the ILN of AMD3100-treated mice. Interestingly, the increase in Sca-1+/CD45−/CD31− cells in ILN after AMD3100 treatment was associated with an increase in the CFU-F activity of ILN-derived cells (from 52 per 1 × 106 ILN cells in control mice to 87 per 1 × 106 ILN cells in AMD3100-treated mice, Fig. 5D).
Transplanted Native ASCs Home to LNs
To visualize and monitor the infiltration of ASCs in LNs, native (i.e., freshly purified) ASCs isolated from ScAT-derived GFP transgenic mice (C57BL/6) were used. Moreover, to get closer to the number of ASCs egressing AT 180 minutes after AMD3100 treatment (about 2 × 105 cells, Fig. 4A), 1.5 × 105 of freshly harvested native ASCs-GFP+ were injected in the right hind footpad of C57BL/6 mice. The presence of ASCs-GFP+ in both right PLN and ILN was investigated and left PLN and ILN were used as control. Only 12 hours after their injection, native ASCs-GFP+ were visualized in LNs (Supporting Information Fig. S4A, video 1). We then investigated whether ASCs home into specific LN microstructures. To do so, we first studied LN stromal microarchitecture, by performing fluorescence immunohistochemistry of whole LNs. The distinct LN compartments were visualized using antibodies against B220, CD3, and ER-TR7 antigens detecting, respectively, B lymphocytes, T lymphocytes, and LN-reticular network (RN) (Supporting Information Fig. S4B–S4D). Whatever the time (12 hours and 36 hours postinjection) ASC-GFP+ cells were found in close interaction to podoplanin+ (Fig. 5E) and ER-TR7+ (Fig. 5F, video 2) structures identifying the fibroblastic reticular cells composing the RN of LNs.
Immunization Triggers the Specific Decrease of ASC Content in AT Together with their Accumulation in LNs
Because circulating native ASCs were able to home into LNs, we hypothesized that LN activation might trigger their release from surrounding AT. For this purpose, C57BL/6 mice were immunized with OVA-IFA in the right hind limb footpad (Fig. 6A). Nine days later, antigen-specific T-cell response was analyzed in the immunized right hind limb draining LNs (ILN and PLN) and was compared to the one of the contralateral left hind footpad that was used as control. The expression of activation markers such as CD44 and CD62L on CD4 and CD8 T cells was evaluated. As expected, frequencies of CD44+effector CD4 and CD8 T cells increased while frequency of CD62L+naïve T cells decreased in immunized ILN and PLN when compared to control nonimmunized LNs (Supporting Information Fig. S5A). In parallel, native ASC contents were assessed in the ScAT, ILN, and PLN from immunized hind limb and were compared to those from control hind limb. The specific T-cell response was associated with a very strong and specific increase in Sca-1+/CD45−/CD31− cell content in both immunized ILN and PLN (30- and 40-fold increase, respectively, Fig. 6B) while, the content in Sca-1+/CD45−/CD31− cells was concomitantly reduced in immunized hind limb-derived ScAT (Fig. 6B). None of these changes were observed in the contralateral control left hind limb. To note, the leukocyte content was not modified regardless of the fat pad studied conversely to immunized ILN and PLN where immunization was associated with a dramatic increase in leukocyte content (Fig. 6C). To determine the relative contribution of BM as a putative source of stromal cells for immunized LNs, the content in Sca-1+/CD45−/CD31− cells was evaluated in the femur-derived BM from vaccinated and control hind limbs by flow cytometry 9 days after immunization procedure. Interestingly and conversely to ScAT, no changes were observed neither in BM total cell number nor in the number of Sca-1+/CD45−/CD31− cells from vaccinated hind limbs compared to control ones (Supporting Information Fig. S5B, S5C).
In parallel, CXCL12 protein expression was investigated in LNs during the immunization procedure with OVA-IFA. While no changes were detected in control LNs, the CXCL12 levels were found to be dramatically increased in the immunized LNs (Fig. 6D, 6E). Finally, to elucidate the role played by the axis CXCL12/CXCR4, mice were immunized with OVA-IFA in right hind footpad and CXCR4 neutralizing antibodies were injected in the right ScAT after the immunization. The left ScAT was injected with isotypic IgG antibodies and was used as control. As expected and consistent with the data obtained with AMD3100, CXCR4 neutralizing antibodies were able to trigger the release of ASCs from ScAT. Indeed, the Sca-1+/CD45−/CD31− cell content was more reduced in the immunized ScAT when ScAT were treated with CXCR4 neutralizing antibodies (Fig. 6F). Interestingly and consistent with our hypothesis, while ASC release by ScAT was reinforced with CXCR4 neutralizing antibodies, their accumulation within the LN 9 days after immunization was also reduced (Fig. 6G), suggesting that CXCR4/CXCL12 axis might account for the tropism and/or retention of ASCs toward activated LNs. Collectively, these results show that activated LNs may attract ASCs via CXCR4 dependent pathway.
Our data show for the first time that native ASCs are mobilized from white AT, ex vivo and in vivo, and that the CXCL12/CXCR4 axis mediates such a process. Moreover, our results suggest that during LN activation following the induction of a specific immune response, ASCs are attracted to and migrate toward the corresponding activated LN, where antigen-specific immune response takes place.
While AT has become accepted as a source of multipotent progenitor cells , their putative release to the circulation has been poorly investigated. Indeed, the release of stem/progenitor cell populations, the so-called mobilization process, has been essentially viewed as originating from BM although some experiments in rodents have demonstrated that a percentage of circulating progenitors originate from extramedullary sites [29, 30]. Cell mobilization is a finely controlled process which involves various partners such as chemoattractants and adhesion molecules, leading to the release of stem/progenitor cells followed by their active migration across the endothelial wall [31, 32]. Among these factors, the chemokine CXCL12 and its receptor CXCR4 play key roles [33–35]. To note, CXCL12 was recently, shown to bind with high affinity to chemokine receptor CXCR7 . However, the CXCR7-CXCL12 complex does not seem to activate the canonical Gi-mediated signaling network . Indeed, CXCR7 exhibits a scavenging activity which might control the availability of CXCL12 .We previously demonstrated that human native ASCs express CXCR4 receptor and that CXCL12 is produced by human AT capillary endothelial cells . In this study, we confirmed that CXCR4 is present at the cell surface of native murine ASCs that we identified as Sca1+/CD29+/CD34+/CD90+/CD31−/CD45− . Therefore, we speculated that, as already stated for HSPCs from BM, mechanisms disrupting the CXCL12/CXCR4 axis may promote the egress of ASCs from AT [39, 40]. Here, we show for the first time that the acute administration of CXCR4 antagonist induces a rapid, specific, and robust reduction of ASC content in ScAT and MesAT. Moreover, we demonstrated that such a reduction was attributable neither to cell death nor to putative effects of AMD3100 on the phenotype of native ASCs but to the specific and time dependent release of ASCs from AT. Moreover, to definitively demonstrate that the decrease in ASC content in vivo was indeed the result of the mobilization of endogenous ASCs from AT, we specifically developed an ex vivo perfusion model of an intact fat pad which ultimately confirmed that AMD3100 elicits the specific release of endogenous ASCs from AT. AT has been proposed as a source of recruitable cells, however, the experiment approaches which were used to test such a hypothesis have been always performed based on a transplantation approach . Our study is, to our knowledge, the first one to provide both in vivo and ex vivo robust evidences showing that native ASCs can naturally egress AT.
The discovery of naturally circulating MSCs has been attempted by several groups. However, these studies show conflicting results which may be attributed to differences in mobilization procedures, MSC immunophenotype (mainly based on passaged cells rather than native cells), and also to the fact that peripheral blood was always chosen as a natural route for circulating MSCs . In this work, we were unable to identify the presence of a significant proportion of blood circulating ASCs after AMD3100 administration. Interestingly, it has been recently reported that HSPCs home to peripheral organs via the blood but leave them predominantly via the draining lymphatics [43, 44]. Therefore, we tested such a hypothesis for ASCs and found that circulating native ASCs were detected in lymph fluid after AMD3100 administration. Moreover, we reported that native ASCs content was diminished in response to AMD3100 in ScAT and MesAT but not in PGAT. Very importantly and consistent with the presence of ASCs in the lymph fluid, PGAT, unlike both ScAT and MesAT, is not connected to lymphatic system . Together with their circulation in the lymph fluid, we also reported that after AMD3100 administration, the LNs embedded in the fat pads specifically accumulate Sca1+/CD31−/CD45− cells, and we further demonstrated that ASCs exhibit LN homing after their injection. The genetic marking and tracking of labeled ASCs in vivo would bring a complementary demonstration but and as already stated, it is technically hardly possible because, like MSCs, no unique and specific marker is expressed by ASCs [46, 47]. Therefore, native ASCs can only be tracked by a combination of cell surface markers. The fact that native ASCs could egress from ScAT and MesAT by trafficking trough the lymphatics reinforces the close relationship, which exists between lymphatic system and AT biology . First, LNs are nearly always found anatomically associated with AT. Second, extensive works performed by Pond and colleagues have revealed that AT surrounding LNs serve as a reservoir of energy to support local immune responses [49–51]. Third, in a mouse model devoid of LNs, the LN-associated fat pads fail to develop . Fourth, a recent work published by Benezech et al.  showed in mouse embryos that a common precursor cell can give rise to both LN intrinsic organizer stromal cells and the adipocytes that reside in the adjacent fat pads. Stromal cells are present at the first phase of LN development and are essential for establishing the formation of LNs during murine embryogenesis . Moreover, upon inflammation or immune response, there is a dramatic increase in the LN stromal compartment, which raises the question about the origin and plasticity of LN resident stromal cells. While local proliferation has been suggested , circulating stromal cell precursors, such as MSCs, have been proposed to contribute to stromal compartment expansion [54, 56, 57]. Our data strongly support that ASCs are recruited during the establishment of an immune response. This is also consistent with the observation that a robust decrease of ASC content in the corresponding fat depot (i.e., ScAT) occurs concomitantly as the immune response takes place, the LN size increases, and the LN gets enriched in ASCs. Interestingly, the cellular content of BM was not modified in the immunized hind limb, since neither total cell number nor Sca1+/CD31−/CD45− cell number were changed conversely to ScAT. Such a result may suggest that conversely to ScAT, BM contributes poorly to the stromal compartment expansion. Furthermore, our findings support the idea that the functional relevance of ASC incorporation in activated LN might be related to the remodeling of the LN stroma. Indeed, the injected ASCs-GFP+ are found in close association with ER-TR7+ and podoplanin+ structures, corresponding to the LN-RN, known to assist the LN during expansion and contraction phases but also to facilitate antigen or antibody transport .
As already very well-demonstrated by the various studies of Pond et al., the products of AT lipolysis may nourish and/or regulate the activated lymphoid cells, therefore, offering local, almost instant provisioning of immune responses. Our study shows that together with providing the fuel to immune cells AT could also provide a local source of stromal progenitors needed to mount a prompt and effective immune response. Although ASCs guidance signals provided by lymph fluid and activated LNs need further investigations, it should be noted that CXCL12 might represent a key element. Indeed, the immune response establishment is accompanied by a time-dependent increase in CXCL12 expression in the activated LN, and the use of CXCR4 neutralizing antibody in ScAT was able to block ASCs recruitment in the activated LN. Hence, together with patterning the immune system the CXCL12/CXCR4 axis might also coordinate interactions between the immune system and AT, as already stated for both the immune and nervous systems . Our study highlighting that recruited ASCs participate to the remodeling of LN-stroma raises, however, several issues. Indeed, since an emerging body of data indicates that MSCs possess immunomodulatory properties, it is tempting to speculate that recruited ASCs control T- and/or B-cell expansion during LN activation. If so, do ASCs reside definitively in the activated LN which in turn could induce AT remodeling or is it a reversible process leading to the return of ASCs into AT?
Whatever the answer to these issues, altogether our findings provide new insights into the role of AT since we propose that besides being a great reservoir of ASCs, AT could release this cell population in vivo according to the needs of the organism and that the axis CXC12/CXCR4 plays a central role in this process. Moreover, since such a mobilization can be achieved pharmacologically and due to the size of the adipose organ this might open a completely unexplored and very promising approach regarding the regenerative potentials of mobilized ASCs.
This work was supported by Inserm, “La ligue nationale contre le cancer”, the “Association Française d'étude et de recherches sur l'obésité” (Afero) and Roche (2007). We thank Christophe Guissard for his invaluable technical assistance. We are grateful to Denis Calise for sharing his surgical expertise for lymph fluid collection; to the “Institut des Technologies Avancées en sciences du Vivant” (ITAV) UMS 3039; and to Dr. J. Rouquette for LN imaging. We also thank Pr. Rémy Burcelin for fruitful conversations. We thank Dr. D. Thompson for his precious and invaluable support.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.