|Home | About | Journals | Submit | Contact Us | Français|
We previously demonstrated that human pericytes, which encircle capillaries and microvessels, give rise in culture to genuine mesenchymal stem cells (MSCs). This raised the question as to whether all MSC are derived from pericytes. Pericytes and other cells defined on differential expression of CD34, CD31, and CD146 were sorted from the stromal vascular fraction of human white adipose tissue. Besides pericytes, CD34+ CD31- CD146- CD45- cells, which reside in the outmost layer of blood vessels, the tunica adventitia, natively expressed MSC markers and gave rise in culture to clonogenic multipotent progenitors identical to standard bone marrow-derived MSC. Despite common MSC features and developmental properties, adventitial cells and pericytes retain distinct phenotypes and genotypes through culture. However, in the presence of growth factors involved in vascular remodeling, adventitial cells acquire a pericytes-like phenotype. In conclusion, we demonstrate the co-existence of 2 separate perivascular MSC progenitors: pericytes in capillaries and microvessels and adventitial cells around larger vessels.
The in vitro generated mesenchymal stem cell (MSC) experimentally differentiates into mesodermal lineage cells and can modulate some immune responses [1,2]. Initially extracted from the bone marrow, MSC can be derived from most if not all organs, which may reflect the existence of a systemic reservoir of multipotent progenitor cells [3–5]. However, the in vivo counterpart of the artificial MSC is still not fully characterized [6–8]. An affiliation between MSC and vascular cells has been suggested by the isolation of MSC from artery or vein walls and, less directly, by the correlation between MSC progenitor frequency and vessel density in equine adipose tissue . In the bone marrow, CD146+ reticular cells lining the endothelium in sinusoid walls can self-renew, differentiate into bone, and recapitulate the hematopoietic microenvironment in vivo . Similarities between MSC and perivascular cells have been also described in the dental pulp, endometrium, and several other tissues [11–14]. We recently demonstrated a perivascular origin of human MSC, thus showing that pericytes purified from multiple organs natively display phenotypic and developmental features of MSC . Moreover, cultured pericytes resemble MSC by morphology and growth properties, retain the expression of MSC surface markers, and can differentiate into bone, cartilage, fat, and muscle cells, similar to MSC . Although MSC can clearly derive from cultured pericytes, there is no argument to exclude that other cell types, including other vascular cells, are also at the origin of MSC . In the current study, we investigated whether other cells share with pericytes the ability to originate MSC. To address this question, we purified different subsets of cells distinct from pericytes from the stromal vascular fraction of human white adipose tissue (hWAT) and tested their ability to yield MSC in culture. We here report the identification of a novel perivascular MSC progenitor, typified as CD34+CD31-CD146-CD45- and located in the tunica adventitia of arteries and veins in multiple human tissues, hence distinct from pericytes. Although the presence of multipotent progenitors in the tunica adventitia has been suggested by several studies [17–29], ignorance of their antigenic phenotype has precluded the isolation and characterization of these progenitors from human organs.
Human adipose tissue surgical specimens (n=18) and lipoaspirates (n=15) were obtained from female patients undergoing abdominoplasty and liposuction, respectively. Specimens were obtained as anonymous and unidentified samples; no IRB approval was, therefore, required. Human fetal tissues were obtained after spontaneous, voluntary, or therapeutic pregnancy interruptions performed at Magee-Womens Hospital (University of Pittsburgh), in compliance with Institutional Review Board protocol number 0506176.
The stromal vascular fraction of hWAT was analyzed by flow cytometry, and different populations of cells were sorted. Surgical hWAT was minced, whereas lipoaspirate was diluted with an equal volume of PBS before digestion with Dulbecco's modified Eagle's medium (DMEM) containing 3.5% bovine serum albumin (Sigma) and collagenase II (1mg/mL; Sigma) for 70min under agitation at 37°C. Adipocytes were separated by centrifugation and removed. Pellets were resuspended in erythrocyte lysis buffer (155mM NH4Cl, 10mM KHCO3, and 0.1mM EDTA) and incubated for 10min at room temperature. After centrifugation, pellets were resuspended in PBS and passed through a 70-μm cell filter (BD Biosciences). Fetal tissues were processed as previously described . Cells were incubated with uncoupled anti-CD31 (DAKO; 1:50), then with biotinylated goat anti-mouse Ig (DAKO; 1:500) followed by incubation with streptavidin-PE-Texas Red (BD Biosciences; 1:500). Cells were eventually incubated with a mixture of the following directly conjugated antibodies: anti-CD34-PE (DAKO; 1:100), anti-CD45-APC (Santa Cruz Biotechnologies; 1:100), and anti-CD146-FITC (AbD Serotec; 1:100). All incubations were performed at 4°C for 15min in the dark. Before sorting, DAPI (Invitrogen; 1:1000) was added for dead cell exclusion, the solution was then passed through a 70-μm cell filter and then run on a FACSAria cell sorter (BD Biosciences). Sorted cells were plated at a density of twenty thousand cells per cm2 in 0.2% gelatin-coated plates and cultured at 37°C in EGM2 (Lonza) for the first 2 weeks. Next, cells were cultured in DMEM high glucose, 20% fetal calf serum (FCS; Invitrogen), and 1% penicillin/streptomycin (BD Biosciences).
Total RNAs were extracted from a minimum of 2×104 freshly sorted or cultured cells by using Absolutely RNA nanoprep kit (Stratagene), and cDNAs were synthesized with SuperScript™ II reverse transcriptase (Invitrogen) according to manufacturer's instructions. PCR was performed with Taq polymerase (Invitrogen) per manufacturer's instructions for 35 cycles at 58°C annealing temperature, and PCR products were electrophoresed on 1% agarose gels.
Cultured cells were labeled with the following commercial antibodies: anti-CD34-PE (BD Biosciences; 1:100), anti-CD44-PerCP-Cy5.5 (eBioscience; 1:100), anti-CD73-PE (BD Biosciences; 1:50), anti-CD90-APC (BD Biosciences; 1:50), anti-CD105-PE (Invitrogen; 1:50), and anti-CD146-PE (BD Biosciences; 1:100). Cells were then washed twice, and at least 50,000 events were acquired on an FACSCanto II cytometer (BD Biosciences). Each individual adventitial cell culture (n=8) was analyzed at different passages (p3 to p13).
hWAT samples were impregnated in gelatin/sucrose, frozen in isopentane (Sigma) cooled in liquid nitrogen and embedded in tissue freezing medium (Triangle Biomedical Sciences). Five- to 9-μm sections were cut on a cryostat (Microm) and fixed for 5min with 50% acetone (VWR International) and 50% methanol (Fischer Scientific). Sections were then dried at room temperature and washed thrice for 5min in PBS. Nonspecific binding sites were blocked with 5% goat serum (Invitrogen) in PBS for 1h at room temperature. Sections were incubated with uncoupled primary antibodies overnight at 4°C, or 2h at room temperature in the case of directly coupled antibodies. After rinsing, sections were incubated for 1h at room temperature with a biotinylated secondary antibody, then with fluorochrome-coupled streptavidin, both diluted in 5% goat serum in PBS. The same protocol was used for the staining of cultured cells at different passages. The following uncoupled anti-human primary antibodies were used: anti-CD146 (BD Biosciences; 1:100), anti-CD31 (DAKO; 1:100), anti-CD34 (BD Biosciences; 1:50), anti-CD44, anti-CD73, anti-CD105 (all from Invitrogen; 1:20), anti-CD90 (BD Biosciences; 1:20), anti-NG2 (1:100; BD Biosciences), and anti-PDGFR-β (1:50; Cell Signaling) and anti-vimentin (1:100; Sigma). Coupled antibodies used included the following: anti-α-SMA-FITC (Sigma; 1:100), anti-CD34-FITC (Miltenyi; 1:20), and anti-vWF-FITC (US Biological; 1:100). Human specific anti-dystrophin (Novocastra; 1:20) was used to detect dystrophin-positive myofibers in SCID/mdx mouse muscle sections. Secondary goat anti-mouse antibodies were biotinylated (DAKO; 1:500) or coupled to Alexa 488 (Molecular Probes; 1:500). Streptavidin-Cy3 (Sigma; 1:500) was used after the incubation with biotinylated secondary antibodies. Nuclei were stained with DAPI (4’, 6-diamino-2-phenylindole dihydrochloride, Molecular Probes; 1:2000) for 5min at room temperature. An isotype-matched negative control was performed with each immunostaining. Slides were mounted in glycerol-PBS (1:1; Sigma) and observed and photographed on an epifluorescence microscope (Nikon Eclipse TE 2000-U).
Tissue sections were examined by confocal microscopy after multi-color immunofluorescence staining and DAPI nuclear staining. Human WAT frozen sections were stained with anti-CD34, -CD146, -α-SMA, or -vWF antibodies according to the protocol just described. Images were collected on an Olympus Fluoview 1000 confocal microscope with 40×, 60×, and 100× oil immersion objectives. Excitation wavelengths used for detection included 405, 488, 543, and 633nm for DAPI, FITC, Cy3, and Cy5, respectively. Sequential detection was set to eliminate bleed-through artifacts between fluorophores.
Sorted cells cultured in DMEM+20% FCS for 12 weeks were seeded into 6-well plates at a density of 2.0×103 cells/cm2. Cells were grown for 120h, and the population doubling time was calculated by using the formula: time/no. of doublings, where time=120, and no. of doublings=log2 (N final/N initial). The assay was performed in biological and technical triplicates.
For adipogenic differentiation, adventitial cells at 70% confluence were cultured in DMEM, 10% FCS, 1μM dexamethasone, 0.5μM isobutylmethylxanthine, 60μM indomethacine, and 170μM insulin (all from Sigma). After 14 days, cells were fixed in 2% PFA at room temperature, washed in 60% isopropanol, and incubated with Oil red O for 10min at room temperature for detection of lipid accumulation. Cells at 70% confluence not cultivated in adipogenic differentiation medium were fixed and stained as described for use as a negative control.
For chondrogenesis, high-density pellets were prepared by spinning down 3×105 cultured cells in 15mL conical tubes and grown in serum-free DMEM containing an insulin-transferrin-selenious acid mix (BD Biosciences), 50μg/mL L-ascorbic acid 2-phosphate (Wako), 100μg/mL sodium pyruvate, 40μg/mL L-proline (both from Invitrogen), 0.1μM dexamethasone (Sigma), and 10ng/mL transforming growth factor β1 (TGF-β1) (Peprotech). Pellets cultured in the absence of TGF-β1 were considered as untreated and used as negative controls. After 21 days, pellets were fixed in 10% formalin, dehydrated using a graded series of ethanol washes, and embedded in paraffin. Five micrometers thick sections were rehydrated and stained with Alcian blue and nuclear fast red for the detection of sulfated glycosaminoglycans and nuclei, respectively.
For osteogenic differentiation, cells at 70% confluence were cultivated in osteogenic medium, consisting of DMEM, 10% FCS, 0.1μM dexamethasone, 50μg/mL L-ascorbic acid, and 10mM β-glycerophosphate (all from Sigma). After 21 days, cells were fixed in 4% formaldehyde for 2min and incubated for 10min with alizarin red, pH 4.2 for the detection of calcium deposits. All the cell cultures (n=8) were assayed for their developmental potential in technical triplicate.
Clones were obtained by limiting dilution after plating cultured adventitial cells at a density of 1 cell/well in a 96 multi-well plate. Clones were grown from 2 different specimens at passages 5 and 10. A total of 5 plates (480-wells) were used for each cloning. Clones that reached confluence were expanded, and 5 individual clones obtained from each hWAT specimen were randomly selected and assayed for their phenotype and in vitro developmental potential as just described.
Five thousand cultured adventitial cells were plated in 48-well plates in DMEM high glucose, 20% FCS, and 1% penicillin-streptomycin. When at 70% confluence, cells were cultured in DMEM high glucose, 0.5% FCS, 0.1μM angiopoietin II (ANGPT2), and 1% penicillin-streptomycin for 72h. Cells were then washed with PBS, fixed with cold methanol-acetone (1:1), and stained with anti-α-SMA, anti-CD146, anti-PDGFR-β, and anti-NG2 as described earlier. The assay was performed at different passages (p5 to p10) in biological duplicate and technical triplicate.
Results are expressed as mean±SEM. Unpaired t-test was used to compare the means between 2 groups, and a P value<0.05 was considered significant.
We recently described primary MSCs within human organs as CD146+CD34-CD45-CD56- pericytes. Here, we set up to verify whether the whole potential to give rise to MSC in culture is confined within pericytes. To this end, we investigated the potential of a different sets of cells, distinct from pericytes and isolated from the stromal vascular fraction of hWAT, a well-documented source of adult multipotent cells, to give rise to MSC in culture. After exclusion of hematopoietic (CD45+) and dead (DAPI+) cells, 3 distinct populations could be detected by flow cytometry based on expression of CD34 and CD146. We isolated pericytes as CD146+CD34- cells (Fig. 1a, gray box), as well as a mixture of other cells differentially expressing CD34 and CD146 (Fig. 1a, black box). Outgrowth of MSCs was observed from both cultured populations. As shown in Fig. 1b, we then separated the 2 nonpericyte subpopulations defined as CD34+CD146- and CD34+CD146+. Cultivated CD34+CD146+ cells never gave rise to MSC, whereas MSC-like cells arose from CD34+CD146- cells in vitro. The endothelium-specific antigen CD31 was detected in only CD34+CD146+ cells; thus, no endothelial cells are present within CD34+CD146- progenitors of MSC (Fig. 1c). Long-term cultures of CD34+CD31-CD146- cells, representing 9.8±1.7% of the total stromal vascular fraction, were successfully established as MSC from all specimens processed for this purpose (n=12). To exclude the possibility of a contamination by pericytes, the purity of CD34+CD31-CD146- cells was verified immediately after the sort by RT-PCR analysis, and no CD146 expression was ever detected (Fig. 1d). Further, sorted cells expressed CD34 but not CD31 or CD144 (VE-cadherin), thus confirming that the purified CD34+CD31-CD146- cells are distinct from pericytes and endothelial cells.
We next investigated whether this novel population of MSC progenitors is hWAT specific or rather present in multiple organs. Using the sorting strategy just described for hWAT, we detected the same population of CD34+CD146- cells also in fetal muscle, lung, and bone marrow (Fig. 2a–c). MSC-like cells were obtained in cultures regardless of the tissue of origin, thus demonstrating that CD34+CD146- cells are MSC progenitors distinct from pericytes distributed in different tissues.
MSC are culture-established clonogenic progenitors characterized by a distinct surface phenotype and the ability to differentiate into mesodermal cell lineages. To investigate whether CD34+CD146- cells meet the criteria for being MSC ancestors, we cloned by limiting dilution CD34+CD146- cells obtained from 2 different hWAT samples. Clones developed with an efficiency of 15.6%±1.5%, with differences in proliferation potentials, which suggested that CD34+CD146- cells are heterogeneous. Only 2.7%±1.7% of the clones formed could be serially passaged and further characterized (Fig. 3a). All clones displayed the MSC hallmark markers CD44, CD73, CD90, and CD105 at every passage tested, up to passage 10 (Fig. 3b). All clones were also able to differentiate in vitro into adipocytes, osteocytes, and chondrocytes, as shown in Fig. 3c–e. No significant differences in phenotype and developmental potential were observed at different passages in polyclonal cultures (p3-p10) or among different clones. We, therefore, established that CD34+CD146- cells are genuine progenitors of MSC. The same phenotype and developmental potential was observed in adventitial cells isolated from fetal bone marrow and lung (data not shown).
Immunohistochemistry was performed on sections of WAT, adult and fetal muscle, adult pancreas, and fetal lung to uncover the natural localization of CD34+CD146- cells. We previously confirmed that, in capillaries (diameter: 8–10μm), CD34 is exclusively expressed by endothelial cells . In the current study, the only cells expressing CD34, but not CD31 and CD146, consistent with the FACS plot shown in Fig. 1c (left panel), were located around blood vessels larger than capillaries, spanning arterioles, venules, arteries, and veins, with a diameter greater than 50μm. Together with CD34, we used markers to detect endothelial cells (CD31) and pericytes/smooth muscle cells (CD146 and α-SMA). Double staining confirmed the co-expression of CD34 and CD31 by all endothelial cells. Conversely, cells located in the outmost layer of vessel walls express CD34 but not CD31 (Fig. 4a–c). Double stainings of CD34 and CD146 or α-SMA were then performed to localize nonendothelial CD34+ cells. As shown in Fig. 4d, e, α-SMA+CD146+smooth muscle cells are surrounded by CD34+ cells. No expression of CD34 by pericytes/smooth muscle cells was ever observed, thus confirming our previous observations . By triple staining and confocal microscopy analysis, we have typified the cells that constitute the 3 layers of small to large vessel walls as CD34+CD146-α-SMA- adventitial cells, CD146+α-SMA+CD34- pericytes/smooth muscle cells, and CD34+CD146+/-α-SMA- endothelial cells (Fig. 4f). In conclusion, the anatomic candidate marked as CD34+CD31-CD146- is a vascular adventitial cell.
Flow cytometry analysis on freshly dissociated adipose tissue revealed that stromal vascular fraction derived adventitial cells natively and homogeneously express the typical MSC markers CD44, CD73, CD105, and CD90 (Fig. 5a). Immunohistochemical staining of serial sections from the same adipose tissue confirmed that, besides microvascular pericytes, the only other cells coexpressing CD44, CD90, CD73, and CD105 are, indeed, distributed in the tunica adventitia of larger blood vessels, surrounding α-SMA+ smooth muscle cells (Fig. 5b–e). Altogether, these results indicate that adventitial cells are native progenitors of MSC, anatomically distinct from pericytes and peripheral to blood vessels larger than arterioles and venules.
The observation that adventitial cells and pericytes are anatomically and phenotypically different, yet all give rise to bona fide MSC in culture, prompted us to closely and comparatively follow both cells along extended in vitro expansion. Despite similar morphologies, calculation of population doubling times (PDT) after 12 weeks revealed that adventitial cells proliferate significantly faster than pericytes (Fig. 6a). As of surface phenotype, adventitial cells isolated from 8 distinct hWAT samples expressed no CD146 at any passage tested (up to p13), whereas all cultured pericytes did (Fig. 6b), as already reported . No expression of CD34 was detected in either population by flow cytometry (data not shown). As previously demonstrated , human pericytes express PDGFR-β, CD146, α-SMA, and NG2 (Fig. 6d–g), none of which was ever detected in adventitial cells cultured in regular conditions (Fig. 6i–l), in which the 2 cell types only shared the expression of vimentin (Fig. 6h, m). However, treatment with 0.1μM ANGPT2 for 72h induced expression of all the aforementioned markers in cultured adventitial cells, thus suggesting differentiation into pericyte-like cells (Fig. 5n–q).
Historically, MSCs were isolated on their capacity to selectively proliferate in cultures of total bone marrow and to differentiate in mesodermal cell lineages. Some authors have also described MSC differentiation into nonmesodermal cell lineages, although this is still a matter of debate [30,31]. Remarkably though, MSC did not benefit the theoretical and technologic progress that permitted the prospective identification of hematopoietic and other stem cells. The instrumentalization of MSC for tissue-engineering/cell therapy eclipsed latent questions regarding the identity, origin, and native distribution of these adult stem cells. Only recently were similarities noticed to exist between MSC and perivascular cells [10–14]. Indeed, vascular pericytes purified to homogeneity from multiple organs express MSC markers, can differentiate into mesoderm lineage cells, and give rise in culture to genuine MSC . Are all MSC derived from pericytes, which are exclusively associated with capillaries and microvessels ? The current work addressed this issue along a substractive approach, by investigating the ability of FACS selected nonpericyte cells, within the stromal vascular fraction of hWAT, to give rise to MSC in culture. Our general conclusion is that CD34+CD31-CD146-CD45- cells residing in the outer layer (tunica adventitia) of arteries and veins yield in culture multipotent progenitors functionally and antigenically similar to MSC. Coincidentally, the tunica adventitia is believed to play a role in vascular remodeling and be involved in the development of atherosclerosis [17–20]. Adventitial cells indeed proliferate, differentiate into myofibroblasts, and migrate into the inner layer of blood vessels in response to injury or stress [21–25]. CD34+CD31- cells located in the “vasculogenic zone” of the thoracic artery, between the tunica media and tunica adventitia, can also give rise to endothelium; hence, they could represent a resident pool of endothelial progenitor cells for postnatal vasculogenesis . Multipotent progenitors can originate from larger vessels not surrounded by typical pericytes, such as the human pulmonary artery, fetal aorta wall, and vena saphena [27–29], thus further supporting our present conclusions.
The presence in hWAT of CD34+ MSC progenitors has been reported, although the native identity of these cells has been controversial so far. Traktuev et al. described in adipose tissue CD34+CD31- perivascular cells that, once enriched by plastic adherence and briefly cultured, expressed markers of pericytes (PDGFR-β), MSC (CD90), and smooth muscle cells (α-SMA) . In 2 other studies, CD34+CD31- cells were clearly identified in human adipose tissue as adventitial cells distinct from α-SMA+ smooth muscle cells and PDGFR-β+ or CD146+ pericytes, all of which are invariably CD34 negative [33,34]. Adventitial cells and pericytes from fat, therefore, include 2 anatomically and phenotypically distinct perivascular primary MSC, an observation we have extended to human fetal lung, fetal and adult pancreas, and adult muscle. Importantly, no potential to generate MSC was observed, in any tissue analyzed, outside these 2 perivascular cell compartments. Adventitial fibroblasts can differentiate into α-SMA+myofibroblasts under hypoxia, or when exposed to growth factors involved in vascular remodeling and angiogenesis, TGF-β or angiotensin II [35–37]. Consistently, adventitia-derived MSC treated with angiotensin II upregulate the expression of α-SMA, which typifies fibroblasts, pericytes, smooth muscle cells, and myofibroblasts. We also show for the first time that adventitial cells cultured in the presence of angiotensin II upregulates other pericyte markers: CD146, PDGFR-β, and NG2. The same outcome was observed when adventitial cells were treated with angiopoietin-2, a growth factor produced at the site of vascular remodeling that disrupts interactions between smooth muscle cells and endothelial cells, thus leading to angiogenesis in the presence of VEGF . Altogether, our results indicate the ability of adventitial cells to differentiate into pericytes-like cells under appropriate signaling, thus supporting the previously hypothesized existence of CD34+CD31- pericyte progenitors [39,40]. Differentiation into myofibroblasts suggests that adventitial cells are precursors of smooth muscle cells in homeostasis and disease, in agreement with the emerging theory of vascular remodeling “from the outside-in,” as the adventitia being the sensor of vascular injury or other environmental stress . Given the observed capacity of adventitial cells to differentiate into MSC and pericyte-like cells, here we propose a hypothesis: the centrifugal migration of multipotent adventitial progenitors to surrounding regenerating tissues, in particular at the developing blood vessels.
In final conclusion, we have demonstrated that human blood vessels, depending on their size, are associated with 2 distinct peripheral MSC progenitors, namely pericytes encircling capillaries and microvessels , and adventitial cells surrounding larger arteries and veins. This hypothesis of a dual origin for MSC is also supported by the presence of both CD146+ cells (40%–60%) and CD146- cells within MSC conventionally derived in culture . We propose that, in hWAT and presumably all organs, vascular progenitors of MSC are hierarchically organized with adventitial cells being precursors of pericytes, based on their higher proliferation rate and potential to differentiate into pericytes on proper stimulation.
Mirko Corselli acknowledges the support of the California Institute for Regenerative Medicine Training Grant (TG2-01169) for making this work possible. Our thanks are due to Alison Logar, Stem Cell Research Center, Children's Hospital of Pittsburgh of UPMC, for expert assistance with flow cytometry. The authors also thank Simon Watkins, University of Pittsburgh School of Medicine, for providing access to confocal microscopy, and Dr. David Stoker, Marina Outpatient Surgery Center, Marina del Rey, for the procurement of lipoaspirate samples.
This work was supported by the NIH, Pittsburgh Foundation, University of Pittsburgh and University of California at Los Angeles.
No competing financial interests exist.