MATERIALS AND METHODS
Human tissue sampling
After informed consent, we obtained liposuction aspirates using an IRB-approved protocol from healthy female donors aged 21 to 59 years who underwent liposuction of the abdomen or thighs. Liposuction aspirates were divided into two portions: a floating adipose portion (also called lipoaspirate) and a denser fluid portion (Fig. 1a). Both portions were used as sources for PLA and LAF cells. For harvesting human vascular endothelial cells (HUVEC) and dermal fibroblasts, umbilical cords and skin were obtained from separate donors under informed consent. Human mesenchymal stem cells derived from bone marrow (frozen at passage 2) were purchased from Cambrex Bio Science Walkersville, Inc. (NJ) and cultured with the same medium used for adipose-derived cells.

Cell isolation
All chemicals were purchased from Wako Pure Chemicals (Osaka, Japan), unless otherwise stated.
PLA cells were separated from the fatty portions of liposuction aspirates using a procedure modified from Zuk et al. (2001). Briefly, the suctioned fat was digested with 0.075% collagenase in PBS for 30 min on a shaker at 37oC. Mature adipocytes and connective tissues were separated from pellets by centrifugation (800 x g, 10 min). Pellets were resuspended in erythrocyte lysis buffer (155mM NH4Cl, 10mM KHCO3, 0.1mM EDTA) and incubated for 5 min at room temperature. The pellets were resuspended and passed through a 100-μm mesh filter (Millipore, MA, USA).
LAF cells were harvested from the fluid portions of liposuction aspirates. The suctioned fluid was centrifuged (400 x g, 10 min), and the pellets were resuspended in erythrocyte lysis buffer. After 5 min at room temperature, lysates were passed through a 100-μm mesh filter. The pellets were then processed for density gradient centrifugation with Ficoll (Amersham Biosciences, NJ, USA). After centrifugation (800 x g, 20 min), cells at the gradient interface were collected, washed with PBS, and passed through a 100-μm mesh filter. For flow cytometry of freshly isolated LAF cells, density gradient centrifugation was not conducted. Nucleated cell counts were performed using a NucleoCounter (Chemometec, Denmark).

Cell culture of adherent cells from adipose tissue
Freshly isolated PLA or LAF cells were plated in medium at a density of 5 x 106 nucleated cells/100-mm gelatin-coated dish. Cells were cultured at 37oC, 5% CO2, in humid air. The culture medium was M-199 containing 10% FBS, 100 IU penicillin, 100 mg/ml streptomycin, 5 ng/ml heparin, and 2 ng/ml acidic FGF. For measurement of doubling time, the same medium was used with a serum concentration of 10% or 15%. Primary cells were cultured for 7 days and were defined as “Passage 0.” The medium was replaced every 3 days, and cells were passaged every week. After primary culture for 7 days, attached cells were passaged by trypsinization and plated in the same medium at a density of 2,000 cells/cm2.

Induced differentiation of cultured PLA and LAF cells
Capacities to differentiate along adipogenic, chondrogenic, and osteogenic lineages were examined. Seven days after seeding PLA or LAF cells at passage 3-5, cell differentiation was initiated by replacing the M-199 culture medium. Cells cultured in control medium (DMEM plus 100 IU penicillin and 100 mg/ml streptomycin) containing 10% FBS were used as negative controls. For adipogenic differentiation, confluent cultures were incubated for 4 weeks in the control medium containing 10% FBS supplemented with 0.5 mM isobutyl-methylxanthine (Sigma, MO), 1 μM dexamethasone, 10 μM insulin (Sigma), and 200 μM indomethacin. Fixed cells (4% paraformaldehyde for 10 min) were washed with 60% isopropanol and incubated for 15 min with Oil-Red O to visualize lipid droplets. Cells were then washed with isopropanol and counterstained with hematoxylin.
For chondrogenic differentiation, two types of evaluations were performed. First, confluent cultures were incubated for 4 weeks with the control medium containing 1% FBS supplemented with 6.25 μg/ml insulin, 10 ng/ml TGFβ1, and 50 nM ascorbate-2-phosphate. Fixed cells (4% paraformaldehyde for 10 min) were washed with 3% acetic acid and incubated for 30 min with 1% Alcian Blue 8 GX (Sigma), 3% acetic acid to visualize the extracellular matrix. Cells were then washed with 3% acetic acid and counterstained with 0.1% Nuclear Fast Red, 5 % Al2(SO4)3 solution. Second, a micromass culture system was used as previously reported (Johnstone et al., 1998). Cells were pelleted in a 15-ml tube and cultured with the chondrogenic medium for 3 weeks.
For osteogenic differentiation, cells were incubated for 4 weeks in the control medium containing 10% FBS supplemented with 0.1 μM dexamethasone, 50 μM ascorbate-2-phosphate, and 10 mM β-glycerophosphate (Nacalai Tesque, Kyoto, Japan). Fixed cells (4% paraformaldehyde for 10 min) were washed and incubated with 2.5% silver nitrate for 20 min in the dark. The cells were then washed, placed in the light for 15 min, and incubated for 2 min with 0.5% hydroquinone. Cells were incubated for 2 min with 5% sodium thiosulphate to visualize calcified deposits. The cells were then washed and counterstained with 0.1% Nuclear Fast Red, 5% Al2(SO4)3 solution.
For adipogenic differentiation, colony-forming unit analysis was also performed to quantify colonies with intracellular lipids. One hundred fifty cells were plated on a 100-mm culture disk and cultured for 10-14 days, followed by incubation for 2 weeks with adipogenic medium and staining with Oil-Red O as described above. The numbers of colonies positive or negative for staining were counted under a microscope.

Flow cytometry and sorting
Freshly isolated PLA and LAF cells were examined for surface and intracellular molecule expression using flow cytometry. In addition, adherent PLA and LAF cells were examined at weeks 1, 2, 4, 6, 8, 10, and 20 of cell culture. The following monoclonal antibodies (MAbs) conjugated to fluorochromes were used: anti-CD4-FITC, CD10-PE, CD13-PE, CD16-PE, CD29-PE, CD31-PE, CD34-PE, CD34-FITC, CD34-PE Cy7, CD36-PE, CD44-PE, CD45-PE, CD45-FITC, CD49d-PE, CD49e-PE, CD54-PE, CD56-PE, CD57-FITC, CD62E-PE, CD62P-PE, CD69-FITC, CD73-PE, CD90-PE, CD106-FITC, CD117-PE, CD135-PE, CD146-PE, CD-151-PE, HLA-A,B,C-PE, Tie-2-PE (BD Biosciences, San Diego, CA, USA), CD31-APC (eBioscience, CA, USA), CD144-PE (Beckman Coulter, CA, USA), CD59-PE　(Ancell, Bayport, MN, USA), CD71-PE, CD105-PE (Serotec, Oxford, UK), CD133-PE, and Flk-1-PE (Techne, NJ, USA). Irrelevant control MAbs were included for all fluorochromes. Cells were incubated with directly conjugated MAbs for 30 min, then washed and fixed in 1% paraformaldehyde. Cells were analyzed using a LSR II (Becton Dickinson, San Jose, CA, USA) or FACS Vantage SE (Becton Dickinson) flow cytometry system. Data acquisition and analysis were then performed (Cell Quest software, Becton Dickinson). Gates were set based on staining with combinations of relevant and irrelevant MAbs so that no more than 0.1% of cells were positive using irrelevant antibodies. Cell sorting and subsequent analyses were performed using a FACSAria cell sorter (Becton Dickinson).

Statistical Analyses
Results were expressed as mean ± SEM. Welch's t-test was used to compare each parameter.

RESULTS
Isolation and expansion of stromal cells from fatty and fluid portions of liposuction aspirates
Compared to the fatty portion of liposuction aspirates, the fluid portion contains more peripheral blood discharged from the suctioned site during liposuction. The volume of peripheral blood varied among patients. The adipose portion was subjected to collagenase digestion followed by filtration for exclusion of extracellular matrix (ECM) fragments and debris, while the fluid portion was centrifuged and processed for lysis of contaminating erythrocytes. After plating on culture dishes, non-adherent cells were discarded by changing the culture medium. Both adherent PLA and LAF cells had fibroblast-like morphologies and proliferated with similar doubling times, although a smaller number of adherent cells were harvested from LAF cells than from PLA cells (Fig. 1b).
Cell yields were normalized by dividing the isolated cell number by the volume (in liters) of the fatty portion of the liposuction aspirate. Normalized numbers of nucleate cells in SVF from the adipose (fresh PLA cells; n=28) and fluid (fresh LAF cells; n=28) portions were 1.31 ± 0.50 x109 and 1.55 ± 0.79 x109 per 1 L of adipose portion (P=0.401), respectively (Fig.1c). However, the cell number varied considerably among patients. Erythrocyte contamination was seen in both fresh PLA and LAF cells, although LAF cells apparently contained a much greater number of cells derived from peripheral blood. After 1 week of cell culture, normalized numbers of adherent PLA (n=28) and LAF (n=28) cells were 9.7 ± 1.7 x107 and 3.0 ± 0.6 x107 (P<0.001), respectively (Fig.1d). Thus, there was no significant difference in cell yield between freshly isolated PLA and LAF cells, but there was a difference between adherent PLA and LAF cells cultured for 1 week.
PLA and LAF cells were cultured in medium with 10% (n=14) or 15% FBS (n=15), and doubling times were measured using cells at passage 0. Doubling times of PLA and LAF cells were 28.5 ± 1.7 h and 31.0 ± 2.6 h, respectively, when cultured with 15% FBS, and 43.3 ± 2.8 h and 40.2 ± 3.0 h, respectively, when cultured with 10% FBS. A statistically significant difference in doubling time was observed between the two serum concentrations in both PLA (P<0.001) and LAF (P<0.05) cells, but not between PLA and LAF cells at either serum concentration (Fig.1e).

In vitro differentiation of PLA and LAF cells
To compare the multipotency of PLA and LAF cells, cell differentiation was induced using cells at passage 3-5 by culturing the cells for 4 weeks with adipogenic, chondrogenic, or osteogenic medium. The results showed that both cell populations have similar capacities to differentiate along the adipogenic (Fig. 2a), chondrogenic (Fig. 2b), and osteogenic lineages (Fig. 2c). In addition, similar cartilage formation was observed by the micromass system (Fig. 2d). Colony-forming unit analysis showed that the percentage of cells staining positive for Oil-Red O was 29.0 ± 7.6% for PLA cells and 24.1 ± 4.3% for LAF cells (P=0.12) (Fig. 2d).

Flow cytometric analysis of PLA and LAF cells
Flow cytometric analysis revealed that freshly isolated LAF cells differ significantly in cell surface marker expression from freshly isolated PLA cells. Compared to fresh LAF cells, fresh PLA cells contained higher percentages of cells positive for CD29 (β1-integrin), CD34, and CD90 (Thy-1) expression and a decreased percentage of CD45+ cells of hematopoietic origin. Conversely, in fresh LAF cells, there were higher percentages of CD31 (PECAM-1)+ cells and CD45+ cells, suggesting that fresh LAF cells contained a larger number of blood-derived cells than fresh PLA cells. In the FACS plot of forward and side scatter characteristics (FSC and SSC) of fresh LAF cells, there were three CD45+ cell clusters corresponding to granulocytes, monocytes/macrophages, and lymphocytes (Figs. 3a, S1). CD34+ cells and CD45- cells were located in the cluster corresponding to monocytes/macrophages (Figs. 3a, S1). Double color analysis of fresh LAF cells for CD34 and CD45 indicated that most of the CD34+ cells were CD45-, suggesting that most of the CD34+ cells were not derived from peripheral blood but from adipose tissue (Fig. 3b). In addition, CD34+CD45- cells, but not CD34+CD45+ cells, proliferated on a culture dish, suggesting that the freshly isolated LAF cells contained adherent stromal cells derived from adipose tissue (Fig. 3b). CD34+ cells (P2) were sorted and cultured using the protocol described above, but after 2 weeks in culture approximately half of the cells were negative for CD34 (Fig. 3b), suggesting that some CD34+ cells lost CD34 expression with increased culture time. The CD34- cells did not reacquire CD34 expression by further culturing. The doubling time of CD34- cells was significantly shorter than that of CD34+ cells (Fig. 3c).
Subsequently, we performed multicolor FACS assays to investigate the SVF (fresh PLA and LAF cells) in more detail, especially with regard to cell composition. Even after processing with hypotonic erythrocyte lysis buffer, the SVF contained a large number of erythrocytes, so erythrocytes and platelets were excluded from analysis by gating them out by cell size. Consequently, only nucleated cells were analyzed (Figs. 4 and 5). The results are summarized in Table 1. We classified freshly isolated cells derived from liposuction aspirates into 11 cell populations [4 blood-derived (CD45+) and 7 adipose-derived (CD45-)] as follows. There were three major (> 1%) and one minor (< 1%) cell populations derived from blood with regard to expression patterns of CD31, CD34, CD45, CD105 (Endoglin), CD14, and CD15 (Figs. 5a, S2): 1) CD31lowCD34-CD45+CD105lowCD14-CD15+ cells (corresponding to granulocytes), 2) CD31lowCD34-CD45+CD105lowCD14+CD15- cells (corresponding to monocytes/macrophages), 3) CD31-CD34-CD45+CD105- cells (corresponding to lymphocytes), and 4) CD31-CD34+CD45+CD105- cells (corresponding to hematopoietic stem cells). Cell compositions varied among samples, with the percentages of the first three populations dependent on the composition of the peripheral blood of each sample. Furthermore, the total percentage of blood-derived populations in SVF depends on the intraoperative hemorrhage volume of each sample.
There were seven adipose-derived populations with regard to expression patterns of CD31, CD34, CD45, CD90, CD105, and CD146: 1) CD31-CD34+CD45-CD90+CD105-CD146- cells (corresponding to ASCs), 2) CD31+CD34+CD45-CD90+CD105low CD146+ cells, 3) CD31-CD34+CD45-CD90+CD105-CD146+ cells, 4) CD31-CD34-CD45-CD90+CD105-CD146+ cells, 5) CD31-CD34+CD45-CD90+CD105low cells, 6) CD31lowCD34+CD45-CD90+CD105lowCD146+ cells,
and 7) CD31-CD34-CD45-CD90+CD105-CD146- cells. The first population corresponded to ASCs and comprised 70-90% of the total adipose-derived (CD45-) cells. Most ASCs were CD29+, CD117 (c-kit)-, CD133/AC133-, CD144 (VE-cadherin)-, and Flk-1(VEGFR-2)- (Fig. 4a). The second population was positive for both CD31 and CD146 (identified as P1 in Fig. 5a; also identified in Figs. 4b and 5b), suggesting that it is composed of vascular endothelial cells or endothelial progenitor cells derived from adipose. The third (CD34+) and fourth (CD34-) populations were CD31-CD146+ (Figs. 4a and 4b), suggesting that they may correspond to pericyte progenitors (CD34+) and pericytes (CD34-), respectively. The fifth population was very small and seemed to correspond to CD105-positive ASCs (P1 in Fig. 5b). The sixth population was also small, with moderate expression of CD31 (P2 in Fig. 5a). Finally, the seventh population was positive only for CD90, suggesting that it may be fibroblasts or ASCs that lost CD34 expression (P3 in Fig. 5a).
Additionally, a multi-color FACS assay was performed for cultured PLA cells (at 5 and 10 days; Fig. 6a, 6b) and cultured BM-MSCs (passage 3; Fig. 6c) for comparison. Most PLA cells cultured for 5 days were CD31-CD34+CD45-CD90+CD105+CD146-, but small percentages of other cell populations such as CD31+CD34+CD45-CD90+CD105+CD146+ cells (endothelial cells or endothelial progenitors), CD31-CD34-CD45-CD90+CD105+CD146+ cells (possibly pericyte progenitors), and CD31-CD34-CD45-CD90+CD105+CD146+ cells (pericytes) were also observed. Unlike freshly isolated ASCs, the major ASC population expressed CD105. Compared to PLA cells cultured for 5 days, the 10-day cultures showed decreased percentages of CD31+ endothelial cells and CD34+ cells and increased percentages of CD31-CD34+CD146+ cells and CD31-CD34-CD146+ cells. Cultured BM-MSCs showed similar surface marker expression patterns. Most cultured BM-MSCs were identified as CD31-CD34-CD45-CD90+CD105+ cells with variable expression of CD146. BM-MSCs contained only a very small percentage (< 1%) of CD34+ cells, which was the only difference from ASCs clearly detected in this assay.
Representative single-color FACS data for freshly isolated and cultured (at 2 weeks) PLA and LAF cells are shown in Fig. 7, and sequential changes in cell surface marker expression are shown in Fig. 8. After plating, the surface marker expression profiles for both PLA and LAF cells changed markedly, and adherent cells of the two cell populations showed quite similar expression profiles. The percentage of CD34+ cells increased in PLA cells, which uniformly expressed mesenchymal markers such as CD13, CD29, CD44, CD73, and CD90. In PLA cells cultured for more than 1 week, CD10, CD49e, CD59, and CD151 were also uniformly expressed. One-week culture of freshly isolated PLA or LAF cells resulted in a dramatic enrichment in CD105 from 1.2 ± 0.6 to 64.1 ± 9.7 (P<0.001) or from 1.4 ± 0.7 to 74.6 ± 7.6 (P<0.001), respectively. No statistically significant difference in CD105 expression was observed between adherent PLA and LAF cells. After 1 week in culture, CD45, Flk-1, Tie-2, CD31, CD117, and CD133/AC133 expression had decreased significantly in PLA and LAF cells. Both cell populations were negative for CD4, CD45, CD62E (E selectin), CD62P (P selectin), CD69, CD135, and CD144 at 1 week and were negative for CD16, CD31, CD57, CD106 (VCAM-1), CD133, Flk-1, and Tie-2 after culture for more than 2 weeks.
CD34 expression decreased with increased culture time, but 10-20% of cells maintained CD34 expression up to 20 and 10 weeks’ culture in PLA and LAF cells, respectively. In addition, consistent expression of the mesenchymal markers (CD13, CD29, CD44, CD73, and CD90) as well as other markers (CD 49d, CD59, CD105, and CD151) was observed in adherent PLA and LAF cells up to 20 and 10 weeks, respectively. After the initial 2 weeks, no remarkable changes in surface marker expression were seen between adherent PLA and LAF cells throughout the culture periods. Taken together, these data suggest that adherent PLA and LAF cells can be expanded using our culture method without losing stem cell-associated surface markers. Our studies further revealed that after the initial 1-2 weeks in culture, adherent PLA and LAF cells have quite similar surface marker expression profiles throughout the culture periods, suggesting that both cell populations can be considered ASCs.

DISCUSSION
Adipose tissue is comprised predominantly of mature adipocytes, connective tissue, ASCs, blood-derived cells, vascular cells such as endothelial (progenitor) cells, smooth muscle cells, and pericytes. Adipocytes represent roughly two-thirds of the total cell number and more than 90% of the tissue volume (van Harmelen et al., 2005). The ratio of adipocytes to ASCs is constant in humans, independent of Body Mass Index (BMI) and age (van Harmelen et al, 2003). In the present study, a correlation between ASC cell yield and age or BMI was not detected (data not shown). ASC cell yield varies among patients and is affected by many factors including donor site and storage duration. It also depends strongly on the isolation method, e.g., duration of collagenase digestion (Bakker et al, 2004; Aust et al., 2004; von Heimburg et al., 2004).
Although only the adipose portion of liposuction aspirates has been used as a source of ASCs, we also isolated cells from the fluid portion of liposuction aspirates and found that a comparable amount of adherent stromal cells can be harvested. By flow cytometric analysis, fresh PLA and LAF cells showed distinct surface marker profiles. Higher percentages of CD31+ and CD45+ cells were observed in fresh LAF cells than in PLA cells, suggesting that fresh LAF cells contain a larger amount of blood-derived cells. However, adherent LAF cells cultured for one week showed surface marker profiles quite similar to cultured PLA cells, suggesting that the fluid portion of liposuction aspirates also contains ASCs. In addition, our results showed that most of the CD34+ cells in freshly isolated LAF cells were derived from adipose tissue (CD45-), and these CD34+CD45- cells expanded in culture dishes, suggesting that they correspond to the same ASC population derived from PLA cells. Why a significant amount of ASCs are isolated from the fluid portion of liposuction aspirates has yet to be determined. Nor is the location of ASCs in adipose tissue clearly understood. Some ASCs are thought to be located in the adipose connective tissue and others between adipocytes or around micro- or macro-vasculature. ASCs located between adipocytes might be released into the fluid by mechanical injury during liposuction procedures, and other ASCs might be released by endogenous proteases during surgery or subsequent storage periods. The extent of mechanical injury may vary among patients because it can be affected by the size of the suction canulla, vacuum pressure strength during liposuction, suction procedure (manual or powered), and other factors.
One of the reasons why adipose tissue is thought to be a promising source of stem cells is that a large volume of adipose tissue can be harvested with minimal morbidity. Thus, ASCs can be used clinically after minimal manipulation (without cell culture). Therefore, examination and characterization of freshly isolated PLA and LAF cells have significant clinical implications. Some recent studies examined freshly isolated SVF from human adipose tissue using magnetic cell sorting (Boquest et al., 2005; Sengenes et al, 2005), and the investigators partly characterized some cell populations in the SVF. CD31-CD34+CD45-CD105+ cells were proposed as typical proliferating ASCs (Boquest et al., 2005). Other studies showed that CD34+CD31- cells differentiated into endothelial cells and contributed to neovascularization (Planat-Bernard et al., 2004b; Miranville et al, 2004). For analysis of SVF, we used multicolor (4 or 5 colors) flow cytometry assays and showed that there are several major and minor cell populations derived from blood and adipose. The typical surface marker expression pattern of fresh ASCs was CD31-CD34+CD45-CD90+CD105-CD146-, which was distinct from that of cultured ASCs in CD105 expression only. This profile of fresh ASCs differs from a previous study (Boquest et al., 2005), possibly because the magnetic cell sorting procedure used in that study might have affected CD105 expression. Our results also showed that most fresh ASCs are CD29+, CD90+, CD117-, CD133-, CD144-, and Flk-1-. It should be noted that there are significant numbers of CD146+ and CD146- cells in both CD34+ and CD34- cell populations in SVF. Given that CD31 is an endothelial cell marker and that CD146 is a known marker of endothelial cells and pericytes, CD31-CD34-CD146+ cells in SVF are likely pericytes, and CD31-CD34+CD146+ cells may function as pericyte progenitors. Nearly all adipose-derived CD31+ cells in SVF were found to be CD34+, and this population was previously reported as endothelial cells (Sengenes et al., 2005). Our results showed that this population was CD90+CD105lowCD146+. However, a major proportion of cultured HUVECs (passage 2) were CD34- and CD90- (Fig. S3b), suggesting that fresh but not cultured endothelial cells express CD34 and CD90. Indeed, freshly isolated HUVEC cells were CD34+ with variable expression of CD90 (Fig. S3a). CD31-CD34-CD45-CD90+CD105-CD146- cells in SVF may be fibroblasts, as suggested by comparison with cultured dermal fibroblasts (Fig. S3c).
PLA and LAF cells showed similar surface marker profiles after being cultured for one week. Notable changes between fresh and cultured states include decreased expression of CD31 and CD45 and increased expression of CD29 and CD105. These changes suggest that cells other than ASCs, such as vascular endothelial cells and blood-derived cells, are selectively excluded during culturing on plastic plates. Even when cultured in endothelial cell growth medium, ASCs quickly outgrow CD31+ endothelial cells (Hutley et al, 2001). As noted above, a high percentage of ASCs began to express CD105 after plating. Our results showed that the expression of most surface markers is sustained throughout the culture period (up to 20 weeks for PLA and 10 weeks for LAF cells), and that significant changes seen after the initial 2 weeks are limited to a gradual increase in CD49d and a decrease in CD71.
Because CD34 is one of the most well established stem cell markers, CD34 expression may indicate ASC clinical usefulness. CD105, known as a mesenchymal stem cell-associated marker, was highly expressed in cultured ASCs and may reflect the capacity of ASCs to differentiate into lineages of mesenchymal origin such as adipose, cartilage, and bone. Flk-1, known to be expressed in hemangioblasts, was recently shown to be expressed in ASCs under certain culture conditions (Martinez-Estrada et al, 2005; Cao et al., 2005). These factors suggest that ASCs may have clinical potential for cell-based therapies. A number of studies characterizing human ASCs have reported that ASCs (freshly isolated or cultured for less than 2 weeks) are CD34+ (Gronthos et al., 2001; Planat-Bernard et al., 2004b; Rehman et al., 2004; Miranville et al., 2004; Boquest et al., 2005). However, CD34 expression was not found in human ASCs cultured for more than 2 weeks with conventional culture methods (Gronthos et al., 2001; Zuk et al., 2002; De Ugarte et al., 2003; Rehman et al., 2004; Lee et al., 2004; Katz et al., 2005; Boquest et al., 2005). In contrast, our results showed that CD34+ ASCs were present at 10-20% of the cell population even after 10-20 weeks of culture (Fig. 7). CD34+ cells were sorted and cultured, but about half of the cells became CD34- after 2 weeks in culture (Fig. 3b), suggesting that cultured ASCs may exist in a variety of stages ranging from CD34+ undifferentiated cells to CD34- partially differentiated cells. The result that CD34+ cells proliferated more quickly than CD34- cells may explain the sustained percentage of CD34+ cells in cultured PLA or LAF cells. Thus, using our culture protocol it is possible to expand CD34+ ASCs taken from liposuction aspirates up to 104-107 times after 4 weeks’ culture. Together with the fact that Flk-1+ ASCs can be obtained in high percentages using another culture protocol (Martinez-Estrada et al, 2005; Cao et al., 2005), ASCs may dramatically change their surface marker profiles depending on the culture media and methods.
It was reported that human ASCs differ from human BM-MSCs in expression of CD49d (expressed only in ASCs) and CD106 (expressed only in BM-MSCs) (Zuk et al., 2002; De Ugarte et al., 2003). In our study using multicolor assays with limited surface markers, the only difference between cultured ASCs and cultured BM-MSCs was CD34 expression (positive in ASCs and negative in BM-MSCs). Human dermal fibroblasts have a surface marker expression profile similar to ASCs but lack expression of CD34 and CD105 (Fig. S3c). Because BM-MSCs (Zuk et al., 2002; De Ugarte et al., 2003) and other stromal progenitors cultured for long periods do not express CD34, ASCs may constitute a unique mesenchymal cell population in view of their CD34 and CD146 expression. This characteristic may contribute to potentialities of ASCs other than mesenchymal progenitors, such as endothelial (Planat-Benard et al., 2004b; Rehman et al., 2004; Miranville et al., 2004) or pericyte progenitors.
In summary, adherent LAF cells have quite similar characteristics with respect to growth kinetics, morphology, surface marker profiles, and capacity for differentiation to adherent PLA cells. A significant amount of ASCs can be isolated from the fluid portion of liposuction aspirates, although in smaller amounts than from the fatty portion. In addition, we found that SVF are composed of heterogeneous cell populations including blood-derived cells, ASCs, endothelial (progenitor) cells, pericytes (and progenitors), and other unknown progenitors. A major population of ASCs in SVF was identified as CD31-CD34+CD45-CD90+CD105-CD146- cells but began to express CD105 after plating. Adipose-derived CD34+ ASCs can be expanded for at least 20 weeks using our culture method. These results suggest that liposuction-derived human ASCs may have significant clinical utility for cell-based therapies.


Figure Legends


Fig. 1. Isolation of adipose-derived stromal cells (ASCs) from the fatty and fluid portions of liposuction aspirates. (a) The two portions of liposuction aspirates in a bottle. The fatty portion floats above the denser fluid portion. Cells isolated from the fatty portion were named PLA cells; those isolated from the liquid portion were termed LAF cells. (b) Primary cultures of freshly isolated PLA and LAF cells. Scale bar = 100 μm. (c) Cell yields of freshly isolated PLA and LAF cells. No statistically significant difference in yield was detected. (d) Adherent cell yields of cultured PLA and LAF cells at 1 week. A significantly higher yield of adherent PLA cells was detected. ?? * P<0.05. (e) Doubling times of adherent PLA and LAF cells. A statistically significant difference was seen between the two serum concentrations (10% and 15%) in both PLA and LAF cells, but not between PLA and LAF cells at either serum concentration. ? * P<0.05


Fig. 2 Cell differentiation analysis of adherent PLA and LAF cells. Representative results are shown in (a)-(d). PLA and LAF cells at passage 3-5 were similarly induced to differentiate into adipogenic, chondrogenic, and osteogenic lineages. (a) Cultures under adipogenic conditions. Adipogenic differentiation was induced, and lipid droplets were visualized with Oil-Red O staining at 4 weeks. (b) Cultures under chondrogenic conditions. Chondrogenic differentiation was induced and visualized with Alcian blue staining at 4 weeks. (c) Cultures under osteogenic conditions. Osteogenic differentiation was induced and visualized with von Kossa staining at 4 weeks. Scale bar = 100 μm. (d) Micromass culture of PLA and LAF cells. Cartilage formation was similarly observed using the micromass system. (e) Colony-forming unit analysis under adipogenic conditions. Between adherent PLA and LAF cells, the difference in percentages of colonies that differentiated into an adipogenic lineage was not statistically significant.


Fig. 3. CD34 expression in freshly isolated and adherent LAF cells.
a) Flow cytometric analysis of freshly isolated LAF cells for cell size, granularity, and expression of CD34 or CD45. In the FACS plot of SSC (granularity) vs. FSC (cell size) for freshly isolated LAF cells, CD34+ cells (left) and CD45- cells (right) are shown in green. Both were located in the same area as monocytes/macrophages, while CD45+ cells (middle) were located in the areas typical for blood cells such as granulocytes, monocytes/macrophages, and lymphocytes (see Fig. S1). b) Double color flow cytometric cell sorting of fresh LAF cells by CD34 and CD45 expression, and the fates of sorted cells after plating. Left: Most CD34+ cells from fresh LAF cells were CD45-. Middle: The CD34+CD45- (P1) and CD34+CD45+ (P2) cell populations were sorted and plated; only CD34+CD45- cells grew on the dish. Scale bar = 100 μm. Right: After 2 weeks’ culture, approximately half of the CD34+CD45- cells had lost CD34 expression. c) Comparison of doubling time between CD34+ and CD34- adherent LAF cells. The doubling time of CD34+ adherent LAF cells was significantly shorter than that of CD34- cells. * P<0.05


Fig. 4. Multicolor FACS analysis of SVF (1). (a) SVF from the adipose portion of liposuction aspirates was analyzed for CD34, CD45, and one of the following markers: CD29, CD90, CD105, CD117, CD133, CD144, CD146, or Flk-1. CD34 and CD45 expression is shown in the graph at left, with blood-derived CD45+ cells in red and adipose-derived CD45- cells in blue. Only adipose-derived CD45- cells (blue dots) were plotted in each of the 8 graphs at right. Most adipose-derived CD34+ cells were CD29+, CD90+, CD105-, CD117-, CD133-, CD144-, and Flk-1-. It should be noted that there are two populations of adipose-derived CD34+ cells with regard to CD146 expression. (b) SVF from the adipose portion was analyzed for CD31, CD34, CD45, and one of the following markers: CD90, CD105, or CD146. Only adipose-derived CD45- cells were plotted. There are two major and one minor populations of CD34+ cells. One major population (endothelial cells or endothelial progenitors) was CD31+CD34+CD90+CD105lowCD146+ and comprised 2-8% of CD34+ cells. The largest population (ASCs) was CD31-CD34+CD90+CD105-CD146-. The minor population was CD31-CD34+CD90+CD105-CD146+. There were two small populations of CD34- cells; one was CD31-CD34-CD90+CD105-CD146+, and the other was CD31-CD34-CD90+CD105-CD146-.


Fig. 5. Multicolor flow cytometric analysis of SVF (2). Freshly isolated SVF from the adipose portion was analyzed with multicolor flow cytometry for CD31, CD34, CD45, and CD105.
(a) CD31 and CD45 expression is shown in the graph at left, with six major populations (P1-P6) identified. Each population was individually plotted for CD34 and CD105 expression in the six graphs at right. P1 was composed of CD31+CD34+CD45-CD105low cells, which are fresh endothelial cells or their progenitors. P2 was an unknown minor population (CD31lowCD34-CD45-CD105low cells; possibly CD31low endothelial cells). P3 contained ASCs, which are CD31-CD34+CD45-CD105-. P4 consisted predominantly of granulocytes and monocytes/macrophages, with a small number of lymphocytes. P5 and P6 consisted of lymphocytes (Fig. S2).
(b) CD45 and CD105 expression is shown in the graph at left, and two adipose-derived (CD45-) populations (P1: CD34+, P2: CD34-) are identified. Each population was individually plotted for CD31 and CD34 expression in the graphs at right. P1 was comprised of CD31+CD34+CD45-CD105low cells (fresh endothelial cells or progenitors) and CD31-CD34+CD45-CD105low cells. P2 was composed of other adipose-derived cells including ASCs.


Fig. 6. Multicolor FACS analysis of cultured PLA cells and cultured BM-MSCs. PLA cells cultured for 5 days (a) and 10 days (b) and BM-MSC (passage 3) (c) were analyzed for CD31, CD34, CD45, and one of following markers: CD90, CD105, or CD146.
(a), (b) In the graphs at far left, all cells were plotted, while only CD45- cells (blue dots) are shown in the six graphs at right. There are several differences in surface marker expressions between freshly isolated and cultured PLA cells; Almost all ASCs started expressing CD105 after plating, and some ASCs lost CD34 expression with increased culture time. CD31+ endothelial cells decreased in percentage with culture time, while CD31-CD34-CD146+ cells (possibly pericyte progenitors) and CD31-CD34-CD146+ cells (pericytes) increased. CD31+CD90+ endothelial cells observed in fresh SVF lost CD90 expression with increased culture time, as seen in HUVEC (Figs. S3a and S3b).
(c) Nearly all BM-MSCs were CD34-. Red and blue dots are CD34+ and CD34- cells, respectively. Cultured BM-MSCs and ASCs showed similar expression of CD31, CD45, CD90, CD105, and CD146, with the only detectable difference being in CD34 expression.


Fig. 7. Comparison of flow cytometric data for freshly isolated and 2-week-cultured PLA and LAF cells. Freshly isolated PLA and LAF cells displayed distinct cell surface marker profiles, while PLA and LAF cells cultured for 2 weeks showed quite similar expression profiles. Adherent, but not freshly isolated, PLA and LAF cells expressed CD105. Results are representative of samples from five patients.


Fig. 8. Sequential changes in representative cell surface marker expression in fresh and cultured PLA and LAF cells. Statistically significant differences in expression of some cell surface markers (data not shown for CD16 and CD49e) were observed between freshly isolated PLA and LAF cells. These differences were not apparent after 1-2 weeks in culture. Data are shown as mean + (LAF) or ? (PLA) standard error.

Fig. S1. FSC and SSC analysis of peripheral blood.
FACS analysis of peripheral blood by FSC (cell size) and SSC (granularity). Blue dots are cells positive for each CD antigen: CD4 (helper T lymphocytes), CD8 (cytotoxic T lymphocytes), CD11b (both granulocytes and monocytes/macrophages), CD14 (monocytes/macrophages), CD15 (granulocytes), and CD19 (B lymphocytes). Granulocytes, monocytes/macrophages, and lymphocytes are located predominantly in the upper, middle, and lower populations, respectively.


Fig. S2. Multicolor FACS analysis of peripheral blood.
Blue dots are cells positive for CD4 (helper T lymphocytes), CD8 (cytotoxic T lymphocytes), CD14 (monocytes/macrophages), and CD15 (granulocytes). All blood-derived cells were CD45+. Granulocytes and monocytes/macrophages were CD31+, while lymphocytes were CD31-.


Fig. S3. Multicolor FACS analysis of freshly isolated HUVEC, cultured HUVEC, and cultured dermal fibroblasts. Freshly isolated HUVEC (a), cultured HUVEC (EGM-2; passage 2) (b), and cultured dermal fibroblasts (DMEM+10%FBS, passage 4) (c) were analyzed for CD31, CD34, CD45, and one of following markers: CD90, CD105, or CD146.
(a) All freshly isolated cells from human umbilical cord vein were plotted in the graph at far left, showing contamination by blood-derived CD45+ cells (red dots). Only adipose-derived CD45- cells (blue dots) were shown in the six graphs at right. Most freshly isolated HUVEC were CD31+CD34+CD45-CD105+CD146+, while CD90 expression varied.
(b) Most cultured HUVEC (passage 2) were CD31+CD34-CD45-CD90-CD105+CD146+. Small percentages of CD31+CD34+CD45-CD90-CD105+CD146+ and CD31+CD34-CD45-CD90+CD105+CD146+ cells were also detected. Red and blue dots are CD34+ and CD34- cells, respectively.
(c) Most of the cultured dermal fibroblasts were CD31-CD34-CD45-CD90+CD105-CD146-, and a small percentage was CD31-CD34+CD45-CD90+CD105-CD146-. Red and blue dots are CD34+ and CD34- cells, respectively, gated in the graph at far left.

Table 1. Summary of cell composition in SVF derived from liposuction aspirates. Results of multi-color FACS assays are summarized. We classified freshly isolated cells from liposuction aspirates into 11 populations according to surface marker expression profiles; 4 blood-derived (CD45+; above) and 7 adipose-derived (CD45-; below) cell populations. Cell composition percentages varied among samples, so a range of values is presented. HSC: hematopoietic stem cell, ASC: adipose-derived stromal cell, fEC: fresh endothelial cell, EPC: endothelial progenitor cell, PC: pericytes, PPC: pericyte progenitor cell.

References
Ashjian PH, Elbarbary AS, Edmonds B, DeUgarte D, Zhu M, Zuk PA, Lorenz HP, Benhaim P, Hedrick MH. 2003. In vitro differentiation of human processed lipoaspirate cells into early neural progenitors. Plast Reconstr Surg 111:1922-1931.
Aust L, Devlin B, Foster SJ, Halvorsen YD, Hicok K, du Laney T, Sen A, Willingmyre GD, Gimble JM. 2004. Yield of human adipose-derived adult stem cells from liposuction aspirates. Cytotherapy 6:7-14.
Awad HA, Halvorsen YD, Gimble JM, Guilak F. 2003. Effects of transforming growth factor beta1 and dexamethasone on the growth and chondrogenic differentiation of adipose-derived stromal cells. Tissue Eng 9:1301-1312.
Bakker AH, Van Dielen FM, Greve JW, Adam JA, Buurman WA. 2004. Preadipocyte number in omental and subcutaneous adipose tissue of obese individuals. Obes Res 12:488-498.
Boquest AC, Shahdadfar A, Fronsdal K, Sigurjonsson O, Tunheim SH, Collas P, Brinchmann JE. 2005. Isolation and transcription profiling of purified uncultured human stromal stem cells: alteration of gene expression following in vitro cell culture. Mol Biol Cell 16: 1131-1141.
Cao Y, Sun Z, Liao L, Meng Y, Han Q, Zhao RC. 2005. Human adipose tissue-derived stem cells differentiate into endothelial cells in vitro and improve postnatal neovascularization in vivo. Biochem Biophys Res Commun. 332:370-9.
Cowan CM, Shi YY, Aalami OO, Chou YF, Mari C, Thomas R, Quarto N, Contag CH, Wu B, Longaker MT. 2004. Adipose-derived adult stromal cells heal critical-size mouse calvarial defects. Nat Biotechnol 22:560-567.
De Ugarte DA, Alfonso Z, Zuk PA, Elbarbary A, Zhu M, Ashjian P, Benhaim P, Hedrick MH, Fraser JK. 2003. Differential expression of stem cell mobilization-associated molecules on multi-lineage cells from adipose tissue and bone marrow. Immunology Letters 89:267-270.
Dragoo JL, Samimi B, Zhu M, Hame SL, Thomas BJ, Lieberman JR, Hedrick MH, Benhaim P. 2003. Tissue-engineered cartilage and bone using stem cells from human infrapatellar fat pads. J Bone Joint Surg Br 85:740-747.
Erickson GR, Gimble JM, Franklin DM, Rice HE, Awad H, Guilak F. 2002. Chondrogenic potential of adipose tissue-derived stromal cells in vitro and in vivo. Biochem Biophys Res Commun 290:763-769.
Garcia-Olmo D, Garcia-Arranz M, Herreros D, Pascual I, Peiro C, Rodriguez-Montes JA. 2005. A phase I clinical trial of the treatment of Crohn's fistula by adipose mesenchymal stem cell transplantation. Dis Colon Rectum 48:1416-1423.
Gronthos S, Franklin DM, Leddy HA, Robey PG, Storms RW, Gimble JM. 2001. Surface protein characterization of human adipose tissue-derived stromal cells. J Cell Physiol 189:54-63.
Halvorsen YC, Wilkison WO, Gimble JM. 2004. Adipose-derived stromal cells--their utility and potential in bone formation. Int J Obes Relat Metab Disord 24 Suppl 4:S41-44.
Hicok KC, Du Laney TV, Zhou YS, Halvorsen YD, Hitt DC, Cooper LF, Gimble JM. 2004. Human adipose-derived adult stem cells produce osteoid in vivo. Tissue Eng 10:371-380.
Huang JI, Zuk PA, Jones NF, Zhu M, Lorenz HP, Hedrick MH, Benhaim P. 2004. Chondrogenic potential of multipotential cells from human adipose tissue. Plast Reconstr Surg 113:585-594.
Hutley LJ, Herington AC, Shurety W, Cheung C, Vesey DA, Cameron DP, Prins JB. 2001. Human adipose tissue endothelial cells promote preadipocyte proliferation. Am J Physiol Endocrinol Metab 281:E1037-1044.
Johnstone B, Hering TM, Caplan AI, Goldberg VM, Yoo JU. 1998. In vitro chondrogenesis of bone marrow-derived mesenchymal progenitor cells. Exp Cell Res 238:265?272
Kang SK, Lee DH, Bae YC, Kim HK, Baik SY, Jung JS. 2003. Improvement of neurological deficits by intracerebral transplantation of human adipose tissue-derived stromal cells after cerebral ischemia in rats. Exp Neurol 183:355-366.
Katz AJ, Tholpady A, Tholpady SS, Shang H, Ogle RC. 2005. Cell surface and transcriptional characterization of human adipose-derived adherent stromal (hADAS) cells. Stem Cells 23:412-423.
Lee RH, Kim B, Choi I, Kim H, Choi HS, Suh K, Bae YC, Jung JS. 2004. Characterization and expression analysis of mesenchymal stem cells from human bone marrow and adipose tissue. Cell Physiol Biochem 14:311-324.
Lendeckel S, Jodicke A, Christophis P, Heidinger K, Wolff J, Fraser JK, Hedrick MH, Berthold L, Howaldt HP. 2004. Autologous stem cells (adipose) and fibrin glue used to treat widespread traumatic calvarial defects: case report.　J Craniomaxillofac Surg 32:370-373.
Martinez-Estrada OM, Munoz-Santos Y, Julve J, Reina M, Vilaro S. 2005. Human adipose tissue as a source of Flk-1+ cells: new method of differentiation and expansion. Cardiovasc Res 65:328-333.
Miranville A, Heeschen C, Sengenes C, Curat CA, Busse R, Bouloumie A. 2004. Improvement of postnatal neovascularization by human adipose tissue-derived stem cells. Circulation 110:349-355.
Mizuno H, Zuk PA, Zhu M, Lorenz HP, Benhaim P, Hedrick MH. 2002. Myogenic differentiation by human processed lipoaspirate cells. Plast Reconstr Surg 109:199-209.
Planat-Benard V, Menard C, Andre M, Puceat M, Perez A, Garcia-Verdugo JM, Penicaud L, Casteilla L. 2004a. Spontaneous cardiomyocyte differentiation from adipose tissue stroma cells. Circ Res 94:223-229.
Planat-Benard V, Silvestre JS, Cousin B, Andre M, Nibbelink M, Tamarat R, Clergue M, Manneville C, Saillan-Barreau C, Duriez M, Tedgui A, Levy B, Penicaud L, Casteilla L. 2004b. Plasticity of human adipose lineage cells toward endothelial cells: physiological and therapeutic perspectives. Circulation 109:656-663.
Peterson B, Zhang J, Iglesias R, Kabo M, Hedrick M, Benhaim P, Lieberman JR. 2005. Healing of critically sized femoral defects, using genetically modified mesenchymal stem cells from human adipose tissue. Tissue Eng 11:120-129.
Rehman J, Traktuev D, Li J, Merfeld-Clauss S, Temm-Grove CJ, Bovenkerk JE, Pell CL, Johnstone BH, Considine RV, March KL. 2004. Secretion of angiogenic and antiapoptotic factors by human adipose stromal cells. Circulation 109:1292-1298.
Rodriguez AM, Pisani D, Dechesne CA, Turc-Carel C, Kurzenne JY, Wdziekonski B, Villageois A, Bagnis C, Breittmayer JP, Groux H, Ailhaud G, Dani C. 2005. Transplantation of a multipotent cell population from human adipose tissue induces dystrophin expression in the immunocompetent mdx mouse. J Exp Med 201:1397-1405.
Safford KM, Hicok KC, Safford SD, Halvorsen YD, Wilkison WO, Gimble JM, Rice HE. 2002. Neurogenic differentiation of murine and human adipose-derived stromal cells. Biochem Biophys Res Commun 294:371-379.
Sengenes C, Lolmede K, Zakaroff-Girard A, Busse R, Bouloumie A. 2005. Preadipocytes in the human subcutaneous adipose tissue display distinct features from the adult mesenchymal and hematopoietic stem cells. J Cell Physiol 205:114-122.
Seo MJ, Suh SY, Bae YC, Jung JS. 2005. Differentiation of human adipose stromal cells into hepatic lineage in vitro and in vivo. Biochem Biophys Res Commun 328:258-264.
Strem BM, Zhu M, Alfonso Z, Daniels EJ, Schreiber R, Begyui R, Maclellan WR, Hedrick MH, Fraser JK. 2005. Expression of cardiomyocytic markers on adipose tissue-derived cells in a murine model of acute myocardial injury. Cytotherapy 7:282-291.
van Harmelen V, Skurk T, Hauner H. 2005. Primary culture and differentiation of human adipocyte precursor cells. Methods Mol Med 107:125-135.
van Harmelen V, Skurk T, Rohrig K, Lee YM, Halbleib M, Aprath-Husmann I, Hauner H. 2003. Effect of BMI and age on adipose tissue cellularity and differentiation capacity in women. Int J Obes Relat Metab Disord 27:889-895.
Van RL, Bayliss CE, Roncari DA. 1976. Cytological and enzymological characterization of adult human adipocyte precursors in culture. J Clin Invest 58:699-704.
von Heimburg D, Hemmrich K, Haydarlioglu S, Staiger H, Pallua N. 2004. Comparison of viable cell yield from excised versus aspirated adipose tissue. Cells Tissues Organs 178:87-92.
Zuk PA, Zhu M, Ashjian P, De Ugarte DA, Huang JI, Mizuno H, Alfonso ZC, Fraser JK, Benhaim P, Hedrick MH. 2002. Human adipose tissue is a source of multipotent stem cells. Mol Biol Cell 13:4279-4295.
Zuk PA, Zhu M, Mizuno H, Huang J, Futrell JW, Katz AJ, Benhaim P, Lorenz HP, Hedrick MH. 2001. Multilineage cells from human adipose tissue: implications for cell-based therapies. Tissue Eng 7:211-228.